Make great documentation goal 1.2, and renumber goals 13 to 16 as 12 to 15 wherever they are cited

.github/workflows/packages.yml

@@ -0,0 +1,162 @@
+name: Packages
+
+on:
+  workflow_call:
+    inputs:
+      tag:
+        type: string
+        required: true
+    secrets:
+      TAP_GITHUB_TOKEN:
+        required: true
+      PACKAGES_GPG_KEY:
+        required: true
+      PACKAGES_RSA_KEY:
+        required: true
+      DOCKERHUB_TOKEN:
+        required: true
+  workflow_dispatch:
+    inputs:
+      tag:
+        description: The released tag to package, such as v0.1.0
+        type: string
+        required: true
+
+permissions:
+  contents: read
+
+concurrency:
+  group: packages
+  cancel-in-progress: false
+
+jobs:
+  packages:
+    name: Binaries, packages and image for ${{ inputs.tag }}
+    runs-on: ubuntu-26.04
+    permissions:
+      contents: write
+      packages: write
+    outputs:
+      latest: ${{ steps.latest.outputs.latest }}
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          ref: ${{ inputs.tag }}
+          path: src
+          fetch-depth: 0
+          persist-credentials: false
+      # Check out the packaging config at the workflow's commit, so a tag cut before the
+      # config existed can still be packaged.
+      - uses: actions/checkout@v7
+        with:
+          path: tooling
+          persist-credentials: false
+      - name: The Go version the gate tests
+        id: go
+        run: sed -n 's/^ARG GO_VERSION=/version=/p' tooling/Dockerfile >> "$GITHUB_OUTPUT"
+      - uses: actions/setup-go@v7
+        with:
+          go-version: ${{ steps.go.outputs.version }}
+          cache: false
+      - name: Channels holding one version
+        id: latest
+        env:
+          GH_TOKEN: ${{ github.token }}
+          TAG: ${{ inputs.tag }}
+        run: |
+          latest="$(gh release view --repo "$GITHUB_REPOSITORY" --json tagName --jq .tagName)"
+          if [ "$latest" = "$TAG" ]; then
+            echo "latest=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "skip=--skip=homebrew,scoop" >> "$GITHUB_OUTPUT"
+            echo "OLDER_TAG=1" >> "$GITHUB_ENV"
+          fi
+      - uses: docker/setup-buildx-action@v4
+      - uses: docker/login-action@v4
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - uses: goreleaser/goreleaser-action@v7
+        with:
+          version: v2.18.3
+          workdir: src
+          args: release --clean --config ../tooling/release-tooling/goreleaser.yaml ${{ steps.latest.outputs.skip }}
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          TAP_GITHUB_TOKEN: ${{ secrets.TAP_GITHUB_TOKEN }}
+      - name: The built binary and the published image print the tag
+        env:
+          TAG: ${{ inputs.tag }}
+        run: |
+          test "$(dist/fejkdata_linux_amd64_v1/fejkdata --version)" = "fejkdata $TAG"
+          test "$(docker run --rm "ghcr.io/larvit/fejkdata:${TAG#v}" --version)" = "fejkdata $TAG"
+
+  dockerhub:
+    name: Docker Hub image for ${{ inputs.tag }}
+    needs: packages
+    runs-on: ubuntu-26.04
+    permissions:
+      packages: read
+    steps:
+      - uses: docker/login-action@v4
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - uses: docker/login-action@v4
+        with:
+          username: ${{ vars.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+      - name: Copy the GHCR image to Docker Hub, and check it prints the tag
+        env:
+          LATEST: ${{ needs.packages.outputs.latest }}
+          TAG: ${{ inputs.tag }}
+        run: |
+          tags=(-t "docker.io/larvit/fejkdata:${TAG#v}")
+          if [ "$LATEST" = true ]; then
+            tags+=(-t docker.io/larvit/fejkdata:latest)
+          fi
+          docker buildx imagetools create "${tags[@]}" "ghcr.io/larvit/fejkdata:${TAG#v}"
+          test "$(docker run --rm "docker.io/larvit/fejkdata:${TAG#v}" --version)" = "fejkdata $TAG"
+
+  repos:
+    name: apt, dnf and apk repositories for ${{ inputs.tag }}
+    needs: packages
+    if: needs.packages.outputs.latest == 'true'
+    runs-on: ubuntu-26.04
+    permissions:
+      contents: read
+      pages: write
+      id-token: write
+    environment:
+      name: github-pages
+      url: ${{ steps.deploy.outputs.page_url }}
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - name: The release's Linux packages
+        env:
+          GH_TOKEN: ${{ github.token }}
+          TAG: ${{ inputs.tag }}
+        run: gh release download "$TAG" --repo "$GITHUB_REPOSITORY" --dir packages --pattern '*.deb' --pattern '*.rpm' --pattern '*.apk'
+      - uses: docker/setup-buildx-action@v4
+      - name: Install fejkdata from each repository, then export them
+        env:
+          PACKAGES_GPG_KEY: ${{ secrets.PACKAGES_GPG_KEY }}
+          PACKAGES_RSA_KEY: ${{ secrets.PACKAGES_RSA_KEY }}
+          TAG: ${{ inputs.tag }}
+        run: |
+          build() {
+            docker buildx build -f release-tooling/package-repos.Dockerfile \
+              --secret id=gpg,env=PACKAGES_GPG_KEY --secret id=rsa,env=PACKAGES_RSA_KEY \
+              --build-context packages=packages --build-arg "EXPECTED_VERSION=$TAG" "$@" .
+          }
+          build --target checks --output type=local,dest=checks
+          build --target site --output type=local,dest=site
+      - uses: actions/upload-pages-artifact@v5
+        with:
+          path: site
+      - id: deploy
+        uses: actions/deploy-pages@v5

.github/workflows/test.yml

@@ -4,6 +4,9 @@ on:
   pull_request:
   push:
     branches: [main]
+  workflow_dispatch:
+  schedule:
+    - cron: '17 5 * * *'
 
 permissions:
   contents: read
@@ -11,17 +14,127 @@ permissions:
 jobs:
   test:
     name: vet + fmt + tests
-    runs-on: ubuntu-24.04
+    if: github.event_name != 'schedule'
+    runs-on: ubuntu-26.04
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
         with:
           persist-credentials: false
+          fetch-depth: 0
+      - name: Relevant changes
+        id: changes
+        run: |
+          if [ "$GITHUB_EVENT_NAME" = "pull_request" ]; then
+            git fetch --no-tags origin "$GITHUB_BASE_REF"
+            base="origin/$GITHUB_BASE_REF"
+          else
+            base='${{ github.event.before }}'
+          fi
+          if git diff --quiet "$base"...HEAD -- '*.go' go.mod go.sum Dockerfile .dockerignore .github data release-tooling testdata vulnerability-filing README.md todo.md; then
+            echo "code=false" >> "$GITHUB_OUTPUT"
+          else
+            echo "code=true" >> "$GITHUB_OUTPUT"
+          fi
+          if git diff --quiet "$base"...HEAD -- '*.go' go.mod go.sum release-tooling Dockerfile .dockerignore LICENSE DATA-LICENSES.md .github/workflows/packages.yml .github/workflows/test.yml; then
+            echo "packaging=false" >> "$GITHUB_OUTPUT"
+          else
+            echo "packaging=true" >> "$GITHUB_OUTPUT"
+          fi
+          echo "base=$base" >> "$GITHUB_OUTPUT"
+      - name: CHANGELOG entry for data changes
+        run: |
+          base='${{ steps.changes.outputs.base }}'
+          if ! git diff --quiet "$base"...HEAD -- data testdata/shipped_shape.txt && git diff --quiet "$base"...HEAD -- CHANGELOG.md; then
+            echo "::error::the shipped data changed without a CHANGELOG.md entry (see the README's Versioning)"
+            exit 1
+          fi
       # `docker build` streams the context to the daemon. A compose bind-mount of
       # `.` mounts an empty host dir instead, because the job is itself a container.
       - name: Latest supported Go
+        if: ${{ steps.changes.outputs.code == 'true' }}
         run: docker build .
       # Runs even when the step above failed, so a red build says whether the
       # failure is version-specific.
       - name: Lowest supported Go
-        if: ${{ !cancelled() }}
-        run: docker build --build-arg GO_VERSION=1.22.12 .
+        if: ${{ !cancelled() && steps.changes.outputs.code == 'true' }}
+        run: docker build --build-arg GO_VERSION=1.22.12 --target portable .
+      - name: Package repositories, signed with throwaway keys
+        if: ${{ !cancelled() && steps.changes.outputs.packaging == 'true' }}
+        run: |
+          docker build -f release-tooling/package-repos.Dockerfile --target throwaway-keys --output "$RUNNER_TEMP/keys" .
+          PACKAGES_GPG_KEY="$(cat "$RUNNER_TEMP/keys/gpg.asc")" PACKAGES_RSA_KEY="$(cat "$RUNNER_TEMP/keys/rsa.pem")" \
+            docker build -f release-tooling/package-repos.Dockerfile --target checks \
+            --secret id=gpg,env=PACKAGES_GPG_KEY --secret id=rsa,env=PACKAGES_RSA_KEY .
+
+  vulnerabilities:
+    name: Known vulnerabilities in this Go version and the required modules
+    runs-on: ubuntu-26.04
+    timeout-minutes: 10
+    permissions:
+      actions: write
+      contents: write
+      pull-requests: write
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - name: Fetch govulncheck's report from vuln.go.dev
+        run: docker build --no-cache-filter vulnerabilities --target vulnerability-report --output "$RUNNER_TEMP/vulnerabilities" .
+      - name: File each one in todo.md
+        id: file
+        run: python3 vulnerability-filing/file_vulnerabilities.py "$RUNNER_TEMP/vulnerabilities/report.json" todo.md
+      - name: Propose the filed items
+        if: failure() && steps.file.outcome == 'failure' && github.ref == 'refs/heads/main'
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          if git diff --quiet todo.md || [ "$(gh pr list --head vulnerabilities --state open --json number --jq length)" != 0 ]; then
+            exit 0
+          fi
+          git config user.name 'github-actions[bot]'
+          git config user.email '41898282+github-actions[bot]@users.noreply.github.com'
+          git switch -C vulnerabilities
+          git commit -qam "File the known vulnerabilities govulncheck reports"
+          git push -qf "https://x-access-token:${GH_TOKEN}@github.com/${GITHUB_REPOSITORY}.git" vulnerabilities
+          gh pr create --head vulnerabilities --title "File the known vulnerabilities govulncheck reports" \
+            --body "govulncheck reports known vulnerabilities in the Go version or a required module, and no release is cut until they are gone. This files them atop the next release in todo.md, scored 0.0 until rank_todo.py scores them; each item names the bump that fixes it."
+          # A pull request the workflow's token opens starts no workflow.
+          gh workflow run test.yml --ref vulnerabilities
+
+  release:
+    name: Release from CHANGELOG.md
+    needs: [test, vulnerabilities]
+    if: github.event_name == 'push'
+    runs-on: ubuntu-26.04
+    permissions:
+      contents: write
+    outputs:
+      tag: ${{ steps.publish.outputs.tag }}
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - id: publish
+        run: python3 release-tooling/publish_release.py
+        env:
+          FORGE_API_URL: ${{ github.api_url }}
+          FORGE_REPOSITORY: ${{ github.repository }}
+          FORGE_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          SHA: ${{ github.sha }}
+
+  packages:
+    needs: release
+    if: needs.release.outputs.tag != ''
+    permissions:
+      contents: write
+      packages: write
+      pages: write
+      id-token: write
+    uses: ./.github/workflows/packages.yml
+    with:
+      tag: ${{ needs.release.outputs.tag }}
+    secrets:
+      TAP_GITHUB_TOKEN: ${{ secrets.TAP_GITHUB_TOKEN }}
+      PACKAGES_GPG_KEY: ${{ secrets.PACKAGES_GPG_KEY }}
+      PACKAGES_RSA_KEY: ${{ secrets.PACKAGES_RSA_KEY }}
+      DOCKERHUB_TOKEN: ${{ secrets.DOCKERHUB_TOKEN }}

.gitignore

@@ -1,4 +1,4 @@
 .claude
 *.out
 __pycache__/
-todo.md
+data-import/cache/

AGENTS.md

@@ -1,9 +1,137 @@
 # Rules
 
-- Go runs only through `docker compose run --rm <test|vet|fmt|build>`; the merge gate is `docker build .`.
-- Tests first, in their own commit; the implementation follows in the next. A re-pin of seeded output is its own commit.
+- Go runs only through `docker compose run --rm <test|vet|fmt|build|cyclo|generate|vulncheck>`; the merge gate is what the README's Development section lists.
+- Tests first, in their own commit; the implementation follows in the next. A re-pin of seeded output or of `testdata/shipped_shape.txt` is its own commit.
+- A change under `data/` or to the shape pin adds its `CHANGELOG.md` entry under `Unreleased` in the same PR, and so does a change to a flag, an exit code, an exported name, a fence, a builtin or the lowest Go; what is major is the README's Versioning table.
 - One-line commit messages: no ticket prefix, no repo name, no authorship trailers.
+- GitHub is where the project lives, so `gh` is the tool: merges are fast-forward
+  only, `gh pr merge --rebase` where the branch is already on top of main.
+- CodeRabbit skips every pull request until the repository has 10 GitHub stars: check
+  its status once when a PR is marked ready, answer a review if one is there, and never
+  wait for one.
+- From 0.2.0, README goal 3 is gated at 7.0 on the `comprehension-panel` skill, and below it
+  nothing else merges — no feature, no category, no data — bar a security fix, a
+  dependency bump, a repair of behaviour the README documents, the infrastructure
+  the round itself runs on, and a chunk the comprehension-floor decision lists as
+  shipped under the floor. `todo.md` carries the round: a PR per item it plans, the nine-seat
+  findings at depth 1 or a restructure the maintainer approved, then the nine-seat run
+  again, until the score passes. At or above 7.0
+  every pull request scores with the four-seat run and answers it in one run. No merge
+  lowers the last score its own panel recorded: the nine-seat round and the four-seat
+  run are two ratchets, never compared with each other. A lower score blocks the merge
+  even where it may be panel noise: fix what the seats name until the score rises. A PR
+  that only records a run's scores and files what its seats name merges even when a
+  score fell. Through 0.1.0, a merge is blocked only by a paired ruling that rules worse.
 - Hard tabs. No comment by default; delete a restatement, a rationale, history, or a file preamble.
-- One spelling per result: reject the other at `New`, and let the error name the spelling to use.
-- A standing choice a reader would relitigate goes under Decisions in the README, not in a comment.
+- Prose naming a source states what that source states: read the register's own field
+  before paraphrasing it, and name the register you read before claiming none
+  publishes a spelling.
 - A README example is a `json` block that loads and renders as a category; `readme_test.go` runs every one.
+- Cyclomatic complexity is gated at 14: split a function that would go over it. A table-shaped dispatch, a switch with one case per kind of its input, may sit at 14 and stays whole.
+
+# Decisions
+
+In [docs/decisions.md](docs/decisions.md):
+
+- GitHub is canonical, and the module path names it
+- The Gitea copy stays, as a pull mirror of GitHub
+- The vocabulary sits below `doc.go`'s package clause, not in the package doc
+- Options and fields share one namespace
+- `{a|b}` stays beside nested choices
+- Flags follow getopt_long
+- A struct tag is a template by its shape
+- The CLI renders exactly the template on its stdin, and adds nothing
+- The CLI reads only the library's public API
+- A template that does not compile is misuse (exit 2), including a reference that resolves to nothing, save a lone reference under the CLI's `--format` until item 129
+- Whitespace around a JSON object, array, string or null is dropped, and a bare number or boolean is a format string
+- `FakeTemplate` and `NewTemplate` both stay
+- The shipped data is embedded, not discovered
+- The shipped data is Go packages a library imports by choice, and the CLI carries every one, from 0.1.0
+- A shipped package's folder spells its namespace, its Go name is that namespace in lower case without separators, and its manifest mounts it with `namespace`
+- A source whose manifest carries an index loads each category on first reach; any other loads in `New`, with what it reads
+- Every `{…}` draws afresh, and only a name keeps a pick
+- Reference sigils follow the filesystem
+- A change to what exists is a major; a minor only adds
+- An error is a contract by what it names, not its bytes
+- A load reports every mistake at once
+- Formats are checked after all data is loaded, so their errors name the category
+- Raising the lowest supported Go is a major
+- The format check runs on the latest Go only
+- The changelog heading is the one spelling of a release; CI cuts the tag
+- A `--data-path` override rebinds every reference to the category it replaces
+- The repeat cap bounds renders, not bytes
+- 64-bit targets only
+- A constant zero divisor is a load error; in a string column a divisor that is not constant prints `Inf`
+- Samples say what they emit, transforms what they do
+- A record is a template seen as columns; a Go struct is the one second schema
+- A struct's records follow Go's field access, and compile on first use
+- A category may reference itself, and a record's fences run at load
+- A record's column set is fixed before the first draw
+- Null is a `null` item, not a rate
+- A typed column holds one value, not composed text
+- A column that only reads one reference or name is the column it reads
+- A typed column's calc is refused unless proven
+- `Column` carries text, not a Go value
+- A unit takes the stem of what it is, and a file the stem of the units it holds
+- One name, one meaning
+- The performance gate asserts allocations, not wall-clock time
+- Rows live in a TSV, the shape in JSON
+- A selector is bracketed, and a dot inside it is literal
+- `parent` names the link column and the table alike
+- After a row, a path names a column or a linked table
+- A selected row is fixed, not drawn
+- `..` steps up to the parent row, and a step down after it draws afresh, or once per name's pick
+- A table selects by one key column, a code, never a free-form name
+- A parent row with no child row is a load error
+- A table is a record of string columns
+- The key map is built at load, the rest on first draw
+- Two categories may name one TSV
+- A cell may read a reference, and each row its own
+- A path is walked once without drawing before it is walked for real
+- A country's postal codes and streets are siblings under its locality
+- A locale package imports its country's geo package, and its `Modules()` returns both
+- The default embed holds every Swedish postort the import can place, with all its streets, and, of the US places that hold most of a postal code's land, every state capital, every place of 100,000 or more and a population-weighted tenth, each with a quarter of its streets, at least ten
+- A locale's `address` binds its country's record once and prints it whole
+- A postort's kommun comes from its name, its tätort or its codes, never from distance
+- A highway designation is not a street, and a US postal code belongs to the place holding most of its land inside places
+- `--list` stays a plain list of paths
+- A layout is always quoted
+- A title is a table under `sex`
+- A table owns the spelling of a selector on it
+- No builtin reads the clock, so a date is bounded by days, never by an age
+- `misc` is what every locale shares
+- A register's canonical spelling loses to the one its domain writes
+- `misc.territory` is the spine, and a `misc` table naming a territory links to it
+- A table whose register publishes no frequency draws evenly
+- `misc.timezone` weighs a zone by the people living in it
+- `misc.car` is one flat table, not a make linked to its models
+- `misc.territory` names its sovereign in a column, and there is no `misc.country` table
+- `misc.loglevel` is a table keyed by POSIX's keyword, carrying the code
+- `misc.tld` keys carry the leading dot, where other tables key on a bare code
+- `misc.tld` is a table of its own, and `misc.territory.tld` stays a column
+- `misc.territory` carries a currency code, it does not link to `misc.currency`
+- An extension may name two media types
+- The Swedish ids draw Skatteverket's test series
+- The US given names come from a mirror of the SSA file
+- `List` advertises direct descents only
+- A path draws through its compiled steps
+- A name lives in the category binding it, or in the repeat or choice item binding it, and is drawn on its first read
+- comprehension floor: every dimension and the overall at 7.0 or above, from 0.2.0
+- The template engine stays the root package until item 107; what reads no engine type, a table's rows included, sits in `internal/`
+- Goal 5.1's `{…}` covers `{{`, `}}` and a lone `}`, so the goal names no escape
+- Goal 7.2 holds while a first template can be written with no escape; a literal brace still takes `{{` or `}}`
+- The locales are Ethnologue 2026's ten most-spoken languages, each in the country with the most speakers of it, the Nordic countries and Germany
+- A country with no open address register covering it whole ships its places and postal codes without streets
+- The apt, dnf and apk repositories are static files on GitHub Pages, holding the latest release only
+- The image is pushed to GHCR and copied to Docker Hub, whose name the README shows
+- A known vulnerability in the Go version or a required module blocks a release, even in code fejkdata never calls, and blocks no merge
+
+# Scoring run
+
+The project values for `~/.claude/skills/comprehension-panel/scoring-run.md`:
+
+- `{language}`: Go
+- `{kind}`: a Go library and CLI that renders fake data from JSON templates
+- `{domain}`: fake test-data generation and template languages
+- `{domain docs}`: the documentation of any fake-data library, such as Faker
+- `{3am question}`: a seeded render of `sv_SE.address` prints the postal code of one postort beside a street of another

CHANGELOG.md

@@ -0,0 +1,106 @@
+# Changelog
+
+Format: [Keep a Changelog](https://keepachangelog.com/en/1.1.0/). A release's
+`Breaking` section comes first and names each rejected spelling with its
+replacement, and each removed path, column or flag.
+
+## [Unreleased]
+
+### Added
+
+- Prebuilt binaries for Linux, macOS and Windows on amd64 and arm64, packages for apt,
+  dnf, apk, Homebrew and Scoop, and the image `larvit/fejkdata` on Docker Hub, also at `ghcr.io/larvit/fejkdata`, so
+  the CLI installs without Go. 0.1.0 ships them too; the README's
+  [Install](https://github.com/larvit/fejkdata#install) lists each command.
+
+## [0.1.0] - 2026-10-13
+
+The first release: a Go library and a CLI that render locale-aware fake data from JSON
+templates, and the data they ship with. The
+[README](https://github.com/larvit/fejkdata/blob/v0.1.0/README.md) documents each part.
+Install the CLI with `go install github.com/larvit/fejkdata/cmd/fejkdata@v0.1.0`, or the
+library with `go get github.com/larvit/fejkdata@v0.1.0`. Before 1.0.0 a minor may break,
+and any release may change what a seed draws, so pin one exact version
+([Versioning](https://github.com/larvit/fejkdata/blob/v0.1.0/README.md#versioning)).
+
+### Added
+
+- The CLI, `fejkdata`, renders exactly the template on its stdin and adds nothing:
+  `echo '{/sv_SE.person} hihi' | fejkdata` prints e.g. `Sara Eriksson hihi`. `--seed`
+  repeats a run's output while the version and data stay the same, and without it each
+  run takes a new seed; `-n` renders the template that many times, and `--list`
+  prints the paths the data offers.
+  `--data-path` layers a directory of your own data over the shipped set, and
+  `--no-shipped-data` loads that directory alone. `--format json|ndjson|csv|sql` writes
+  a record's columns, one record per row. The CLI exits `0` on success, `1` on a runtime
+  error such as data that fails to load, and `2` on misuse.
+- The library, `github.com/larvit/fejkdata`, for Go 1.22 or later on a 64-bit target.
+  `New` loads the modules `WithDataFS` names and the folders `WithDataPath` names, drawing
+  from the seed `WithSeed` gives. Without a seed, `New` fails and names `WithSeed`;
+  without data, it fails and names `WithDataFS`. The same seed repeats the draws
+  while the version and data stay the same. `Fake` renders a path, and `FakeTemplate`
+  and `NewTemplate` render a template. `FakeRecord`, `FakeRecordTemplate` and
+  `NewRecordTemplate` render a record, whose `JSON`, `CSVHeader`, `CSVLine` and
+  `SQLInsert` methods write its columns as the CLI does. `FakeStruct` fills a struct's
+  `fake:"…"` tagged fields. `List` gives the paths `Fake` renders, direct descents only,
+  and `IsTemplate` tells an inline template from a path. Data that fails to load gives
+  an error matching `ErrLoad`, and a record with no fields gives one matching
+  `ErrNoColumns`.
+- Package `github.com/larvit/fejkdata/cli` is the `fejkdata` command:
+  `cli.Main(modules...)` runs it over the modules given, so
+  `cli.Main(append(data.Modules(), mydata)...)` builds a CLI carrying your own data.
+  That CLI names itself by the file it runs from, and its `--version` prints its own
+  module's version before fejkdata's.
+- The template language
+  ([Data format](https://github.com/larvit/fejkdata/blob/v0.1.0/README.md#data-format)).
+  Every value in a data file is a string, a choice, a template, a table, a number or a
+  boolean. A `{…}` token reads a sibling field, a name, or a reference from the data
+  root, the file's folder or the folder above (`{/sv_SE.person.last}`, `{.x}`, `{..x}`),
+  or it calls a builtin. Every `{…}` draws afresh, and a name is the one way to keep a
+  pick: `{/sv_SE.person as p}{p.first} {p.last}` is one person. The builtins draw check
+  digits (`luhn`, `mod11`, `ean`), samples (`digits`, `int`, `float`, `uuid`, `ulid`,
+  `iban`, `date`, `time` and more), a counter (`seq`), arithmetic (`calc`) and
+  transforms (`lowercase`, `uppercase`, `ascii`). `New` refuses data that breaks the
+  grammar, could mean two things or cannot render a valid value.
+- Records ([Records](https://github.com/larvit/fejkdata/blob/v0.1.0/README.md#records)):
+  a template seen as columns. A column is a string unless it declares a `datatype`,
+  `integer`, `number` or `boolean`, and a `null` item draws it as null.
+- Tables ([Table](https://github.com/larvit/fejkdata/blob/v0.1.0/README.md#table)): a
+  category naming a `rows` TSV beside it, with the options `key`, `name`, `weight` and
+  `parent`. A path selects a row by key or name, `misc.territory[SE]`, descends to a
+  linked table, `geo.SE.locality[Lund].street`, and steps up to the parent row,
+  `{/misc.timezone[Europe/Oslo]..territory.name}`.
+- Data sources ([Data](https://github.com/larvit/fejkdata/blob/v0.1.0/README.md#data)):
+  a directory is a namespace, and each JSON file in it a category. A `.fejkdata.json`
+  manifest at a source's root may carry a `namespace`, the dot path the source mounts
+  under, and an `index`, with which each category loads on the first call reaching it.
+- The shipped data, in Go packages that each export `FS` and `Modules()`:
+  `github.com/larvit/fejkdata/data/en_US` (package `enus`), `…/sv_SE` (`svse`),
+  `…/misc`, `…/geo/SE` (`geose`) and `…/geo/US` (`geous`). `data.Modules()`, from
+  `github.com/larvit/fejkdata/data`, lists every one, and the CLI carries them all.
+  [`DATA-LICENSES.md`](https://github.com/larvit/fejkdata/blob/v0.1.0/DATA-LICENSES.md)
+  names each source.
+  - `sv_SE` and `en_US`: an `address`, a `person` and more per locale. `sex`,
+    `first-name` and `last-name` are tables weighted by bearers, from SCB, the SSA and
+    the Census Bureau. A person's first name and sex agree, and so does an `en_US`
+    person's title; `person.prefix` is null where a person has no title.
+    `sv_SE.personnummer` and `sv_SE.samordningsnummer` draw Skatteverket's test series.
+    `en_US.ssn` and `en_US.itin` draw the ranges the SSA and the IRS assign, with the
+    columns `area`, `group` and `serial`, and `en_US.phone` draws an exchange NANPA may
+    assign.
+  - `misc`: IANA's registries of HTTP methods, protocols, ports and root-zone TLDs, the
+    HTTP status codes, ISO 3166-1 territories, ISO 4217 currencies with their CLDR
+    symbols, ISO 639 languages, the syslog severities, mime-db's media types, tzdb's
+    zones weighted by the people living in each, user agents and cars, all as tables;
+    and coordinates, credit card numbers, RFC 3339 instants, emoji, MAC addresses,
+    ObjectIds and UUIDs.
+  - `geo.SE` and `geo.US`: five linked tables per country, `region`, `municipality`,
+    `locality`, `postal-code` and `street`, weighted by population and address counts
+    and built from SCB, GeoNames, Trafikverket NVDB, the US Census Bureau and Wikidata,
+    and an `address` record that draws a street and a postal code inside one
+    locality; `sv_SE.address` and `en_US.address` print it. Every Swedish postort
+    (postal town) the import can place ships with all its streets. A US place ships
+    only if it is incorporated and holds most of some postal code's land. Of those
+    places, every state capital but Honolulu ships, as does every place of 100,000
+    people or more and a tenth of all of them drawn by population. Each US place
+    carries a quarter of its streets, at least ten.

DATA-LICENSES.md

@@ -0,0 +1,31 @@
+# Data licenses
+
+Every shipped dataset, its source, its licence and the attribution it asks for. A
+`data-import/` script rebuilds each sourced table, run as the README's
+[Development](README.md#development) section says; a curated one is hand-written.
+
+| Table | Source | Licence | Attribution | Rebuild |
+|-------|--------|---------|-------------|---------|
+| `geo/SE/locality.tsv`, `postal-code.tsv` | [GeoNames](https://www.geonames.org/) postal codes for SE; populations from SCB tätorter 2023 | [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/); CC0 1.0 | "Postal codes from GeoNames, www.geonames.org" | `data-import/geo-se.py` |
+| `geo/SE/region.tsv`, `municipality.tsv` | [SCB](https://www.scb.se/) län and kommun codes 2026 and population 2024 | [CC0 1.0](https://creativecommons.org/publicdomain/zero/1.0/) | none required | `data-import/geo-se.py` |
+| `geo/SE/street.tsv` | [Trafikverket NVDB](https://www.trafikverket.se/) Gatunamn, through the open API | CC0 1.0 | none required | `data-import/geo-se.py` |
+| `geo/US/region.tsv`, `municipality.tsv`, `locality.tsv` | [Census Bureau](https://www.census.gov/) Gazetteer 2026 and population estimates 2025; which places are state capitals from [Wikidata](https://www.wikidata.org/) | [public domain](https://www.usa.gov/government-works); CC0 1.0 | none required | `data-import/geo-us.py` |
+| `geo/US/postal-code.tsv`, `street.tsv` | Census Bureau ZCTA to place relationships 2020 and TIGER/Line 2025 address ranges and feature names | public domain | none required | `data-import/geo-us.py` |
+| `sv_SE/first-name.tsv`, `last-name.tsv` | [SCB](https://www.scb.se/) names with at least two bearers, 31 December 2022 | CC0 1.0 | "Källa: SCB" | `data-import/names-se.py` |
+| `en_US/first-name.tsv` | [SSA](https://www.ssa.gov/oact/babynames/) baby names, births 1930 to 2020, through [hackerb9/ssa-baby-names](https://github.com/hackerb9/ssa-baby-names) | public domain | none required | `data-import/names-us.py` |
+| `en_US/last-name.tsv` | Census Bureau surnames occurring 100 or more times, 2010 | public domain | none required | `data-import/names-us.py` |
+| `sv_SE/sex.tsv`, `sv_SE/birth-number.tsv`, `sv_SE/title.tsv`, `en_US/sex.tsv`, `en_US/title.tsv`, `misc/car.tsv` | curated; Skatteverket's test birth numbers are facts, and `car.tsv` waits for an international source (goal 13) | — | — | — |
+| `misc/currency.tsv` | [datasets/currency-codes](https://github.com/datasets/currency-codes); symbols from [Unicode CLDR](https://github.com/unicode-org/cldr) `en.xml` and `root.xml` | PDDL 1.0; [Unicode License v3](https://www.unicode.org/license.txt) | CLDR: "Copyright © 1991-2025 Unicode, Inc. Unicode and the Unicode Logo are registered trademarks of Unicode, Inc. in the United States and other countries." | `data-import/currency.py` |
+| `misc/httpmethod.tsv` | [IANA HTTP Method Registry](https://www.iana.org/assignments/http-methods/) | [public domain](https://www.iana.org/help/licensing-terms) | none required | `data-import/httpmethod.py` |
+| `misc/httpstatus.tsv` | [IANA HTTP Status Code Registry](https://www.iana.org/assignments/http-status-codes/) | [public domain](https://www.iana.org/help/licensing-terms) | none required | `data-import/httpstatus.py` |
+| `misc/language.tsv` | [datasets/language-codes](https://github.com/datasets/language-codes), the [Library of Congress](https://www.loc.gov/standards/iso639-2/) ISO 639-2 register | [PDDL 1.0](https://opendatacommons.org/licenses/pddl/1-0/) | none required | `data-import/language.py` |
+| `misc/loglevel.tsv` | [RFC 5424](https://www.rfc-editor.org/rfc/rfc5424.txt) Table 2, the syslog message severities; keywords from the [POSIX](https://pubs.opengroup.org/onlinepubs/9799919799/basedefs/syslog.h.html) `<syslog.h>` severity macros | facts | none required | `data-import/loglevel.py` |
+| `misc/mimetype.tsv` | [mime-db](https://github.com/jshttp/mime-db), the IANA media type registry with filename extensions | [MIT](https://github.com/jshttp/mime-db/blob/master/LICENSE) | "Copyright (c) 2014 Jonathan Ong, Copyright (c) 2015-2022 Douglas Christopher Wilson" | `data-import/mimetype.py` |
+| `misc/port.tsv` | [IANA Service Name and Transport Protocol Port Number Registry](https://www.iana.org/assignments/service-names-port-numbers/) | [public domain](https://www.iana.org/help/licensing-terms) | none required | `data-import/port.py` |
+| `misc/protocol.tsv` | [IANA Protocol Numbers](https://www.iana.org/assignments/protocol-numbers/) | [public domain](https://www.iana.org/help/licensing-terms) | none required | `data-import/protocol.py` |
+| `misc/territory.tsv` | [datasets/country-codes](https://github.com/datasets/country-codes) | [PDDL 1.0](https://opendatacommons.org/licenses/pddl/1-0/) | none required | `data-import/territory.py` |
+| `misc/timezone.tsv` | [IANA tzdb](https://www.iana.org/time-zones) 2026d `zone.tab` and the standard offset of each zone; weights from [GeoNames](https://www.geonames.org/) `cities15000` populations | [public domain](https://data.iana.org/time-zones/tzdb/LICENSE); [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) | "Populations from GeoNames, www.geonames.org" | `data-import/timezone.py` |
+| `misc/tld.tsv` | [IANA Root Zone Database](https://www.iana.org/domains/root/db), checked against the root zone's [TLD list](https://data.iana.org/TLD/tlds-alpha-by-domain.txt) | [public domain](https://www.iana.org/help/licensing-terms) | none required | `data-import/tld.py` |
+| `misc/useragent.tsv` | [top-user-agents](https://github.com/microlinkhq/top-user-agents) desktop and mobile lists | [MIT](https://github.com/microlinkhq/top-user-agents/blob/master/LICENSE.md) | "Copyright © 2020 Kiko Beats" | `data-import/useragent.py` |
+
+Every other category is hand-written under [`data/`](data), MIT like the code.

Dockerfile

@@ -1,18 +1,43 @@
-# The merge gate: `docker build .` fails if vet, formatting or tests fail, and CI
-# runs exactly this. Day-to-day, prefer `docker compose run` (bind-mounts source,
-# no rebuilds). GO_VERSION defaults to the latest stable Go; override it to test
-# the lowest supported version: docker build --build-arg GO_VERSION=1.22.12 .
-ARG GO_VERSION=1.26.4
-FROM golang:${GO_VERSION}
+ARG GO_VERSION=1.27.2
+FROM golang:${GO_VERSION} AS portable
 
 WORKDIR /app
 
-# Module layer cached separately from source.
 COPY go.mod ./
 RUN go mod download
 
 COPY . .
 RUN go vet ./... && \
-	{ unformatted="$(gofmt -l .)"; test -z "$unformatted" || \
-	  { echo "unformatted files:"; echo "$unformatted"; exit 1; }; } && \
+	go run github.com/fzipp/gocyclo/cmd/gocyclo@v0.6.0 -over 14 -ignore _test . && \
 	go test -race ./...
+
+FROM golang:${GO_VERSION} AS vulnerabilities
+WORKDIR /app
+COPY . .
+RUN go run golang.org/x/vuln/cmd/govulncheck@v1.8.0 -scan module -format json > /report.json
+
+FROM scratch AS vulnerability-report
+COPY --from=vulnerabilities /report.json /
+
+FROM python:3.14.8-slim AS release-tooling
+COPY release-tooling/ /release-tooling/
+RUN python3 -m unittest discover -s /release-tooling -p '*_test.py' && touch /tested
+
+FROM python:3.14.8-slim AS vulnerability-filing
+COPY vulnerability-filing/ /vulnerability-filing/
+# The filer's test files an item into the plan itself.
+COPY todo.md /todo.md
+RUN python3 -m unittest discover -s /vulnerability-filing -p '*_test.py' && touch /tested
+
+FROM goreleaser/goreleaser:v2.18.3 AS packaging
+WORKDIR /check
+COPY release-tooling/goreleaser.yaml ./
+RUN git init -q && git remote add origin https://github.com/larvit/fejkdata.git && goreleaser check --config goreleaser.yaml && touch /tested
+
+FROM portable
+# BuildKit skips a stage nothing copies from.
+COPY --from=release-tooling /tested /tmp/release-tooling-tested
+COPY --from=vulnerability-filing /tested /tmp/vulnerability-filing-tested
+COPY --from=packaging /tested /tmp/packaging-tested
+RUN unformatted="$(gofmt -l .)"; test -z "$unformatted" || \
+	{ echo "unformatted files:"; echo "$unformatted"; exit 1; }

LICENSE

@@ -1,6 +1,7 @@
 MIT License
 
 Copyright (c) 2026 Timewave
+Copyright (c) 2026 Larvit
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

arm.go

@@ -0,0 +1,117 @@
+package fejkdata
+
+import (
+	"fmt"
+
+	"github.com/larvit/fejkdata/internal/builtinfunc"
+	"github.com/larvit/fejkdata/internal/grammar"
+	"github.com/larvit/fejkdata/internal/invariant"
+)
+
+// arm is one alternative of a {a|b} token or one operand, split into the head,
+// whose node `template.startOf` finds, and the tail of a dotted path into it.
+type arm struct {
+	spelling string // as written, for messages
+	head     string
+	tail     []string
+	levels   []pickKey // the key of each level the steps pass, from the head they start at to the leaf; a step's at indexes it
+	steps    []pathStep
+	leaves   []node // every node the path may land on, one per variant it passes
+	kind     armKind
+	named    *nameBinding // namedRead: the binding of the name it reads through
+}
+
+// armKind is how expand reads an arm, fixed at compile.
+type armKind uint8
+
+const (
+	freshRead armKind = iota // drawn afresh at every read
+	namedRead                // read through a name, kept in its pick
+)
+
+// splitArm splits one name into head and tail. refs maps a reference to what
+// resolveRefs resolved it to; before that, a reference is whole.
+func splitArm(name string, refs map[string]resolvedRef) arm {
+	if grammar.IsRef(name) {
+		if r, resolved := refs[name]; resolved {
+			return pathArm(name, r.head, r.tail)
+		}
+		return pathArm(name, name, nil)
+	}
+	segs, err := grammar.SplitPath(name)
+	if err != nil || len(segs) == 1 {
+		return pathArm(name, name, nil)
+	}
+	return pathArm(name, segs[0], segs[1:])
+}
+
+func pathArm(name, head string, segs []string) arm {
+	return arm{spelling: name, head: head, tail: segs, levels: levelKeys(append([]string{head}, segs...), 1)}
+}
+
+// key is the key of the leaf a lands on, the one key every way of writing this read shares.
+func (a arm) key() pickKey { return a.levels[len(a.levels)-1] }
+
+func checkSegments(a arm) error {
+	if len(a.tail) == 0 {
+		return nil
+	}
+	return grammar.CheckSegments(append([]string{a.head}, a.tail...))
+}
+
+// op is one compiled unit of a format string: a literal run, a name read,
+// or a builtin already prepared with its args.
+type op struct {
+	grammar.Token
+	arms []arm // grammar.PathRead: the '|' alternatives, split into head and tail once
+	call builtinfunc.Call
+	// operands are the fields the builtin reads, in the order its operands func
+	// fixed; expand reads them before the call. nil for a builtin that reads none.
+	operands []arm
+}
+
+func (t *template) compileArms(names []string, targets map[*nameBinding]nameTarget) ([]arm, error) {
+	if len(names) == 0 {
+		return nil, nil
+	}
+	arms := make([]arm, len(names))
+	for i, name := range names {
+		a, err := t.compileArm(name, targets)
+		if err != nil {
+			return nil, err
+		}
+		arms[i] = a
+	}
+	return arms, nil
+}
+
+// compileArm compiles one read into a path: from the head it names, or, for a read through a
+// name, from what that name binds.
+func (t *template) compileArm(name string, targets map[*nameBinding]nameTarget) (arm, error) {
+	a := splitArm(name, t.refs.byName)
+	if !t.isName(a.head) {
+		start := t.startOf(a.head)
+		if start == nil {
+			panic(invariant.Broken("{%s} reads a head nothing resolved", a.spelling))
+		}
+		w := compilePath(start, a.tail)
+		a.steps, a.leaves = w.steps, w.leaves
+		return a, nil
+	}
+	b := t.nameScope.lookup(a.head)
+	if grammar.HasSelector(a.tail) {
+		if b.bindsField() {
+			return a, fmt.Errorf("a path through name %q may not select a row; bind the row to a name of its own", a.head)
+		}
+		return a, fmt.Errorf("a path through name %q may not select a row; read the row without the name, {%s}, or bind it to a name of its own", a.head, grammar.JoinSegments(append([]string{b.ref}, a.tail...)))
+	}
+	target := targets[b]
+	full := append(target.tail[:len(target.tail):len(target.tail)], a.tail...)
+	if err := provePath(target.start, full, a.head, len(target.tail)); err != nil {
+		return a, err
+	}
+	w := compilePath(target.start, full)
+	a.kind, a.named, a.steps, a.leaves = namedRead, b, w.steps, w.leaves
+	a.levels = levelKeys(full, 0)
+	return a, nil
+}

bench_test.go

@@ -6,17 +6,17 @@ import (
 	"testing"
 )
 
-func benchGenerator(b *testing.B, dir string) *Generator {
+func benchGenerator(b *testing.B, data Option) *Generator {
 	b.Helper()
-	f, err := New(WithoutShippedData(), WithDataPath(dir), WithSeed(1))
+	f, err := New(data, WithSeed(1))
 	if err != nil {
 		b.Fatal(err)
 	}
 	return f
 }
 
-func benchPath(b *testing.B, dir, path string) {
-	f := benchGenerator(b, dir)
+func benchPath(b *testing.B, data Option, path string) {
+	f := benchGenerator(b, data)
 	b.ReportAllocs()
 	b.ResetTimer()
 	for i := 0; i < b.N; i++ {
@@ -26,12 +26,12 @@ func benchPath(b *testing.B, dir, path string) {
 	}
 }
 
-func BenchmarkPerson(b *testing.B)     { benchPath(b, "data/sv_SE", "person") }
-func BenchmarkAddress(b *testing.B)    { benchPath(b, "data/sv_SE", "address") }
-func BenchmarkWord(b *testing.B)       { benchPath(b, "data/sv_SE", "word") }
-func BenchmarkCreditcard(b *testing.B) { benchPath(b, "data/misc", "creditcard") }
-func BenchmarkSSN(b *testing.B)        { benchPath(b, "data/sv_SE", "ssn") }
-func BenchmarkUUIDv7(b *testing.B)     { benchPath(b, "data/misc", "uuid") }
+func BenchmarkPerson(b *testing.B)       { benchPath(b, withShipped(), "sv_SE.person") }
+func BenchmarkAddress(b *testing.B)      { benchPath(b, withShipped(), "sv_SE.address") }
+func BenchmarkWord(b *testing.B)         { benchPath(b, withShipped(), "sv_SE.word") }
+func BenchmarkCreditcard(b *testing.B)   { benchPath(b, withShipped(), "misc.creditcard") }
+func BenchmarkPersonnummer(b *testing.B) { benchPath(b, withShipped(), "sv_SE.personnummer") }
+func BenchmarkUUIDv7(b *testing.B)       { benchPath(b, withShipped(), "misc.uuid") }
 
 func tmpData(b *testing.B, name, body string) string {
 	b.Helper()
@@ -44,52 +44,64 @@ func tmpData(b *testing.B, name, body string) string {
 
 func BenchmarkCalc(b *testing.B) {
 	dir := tmpData(b, "inv", `{"format":"{net} x {qty} = {calc(net * qty, 2)}","net":"19.99","qty":"3"}`)
-	benchPath(b, dir, "inv")
+	benchPath(b, WithDataPath(dir), "inv")
 }
 
 func BenchmarkLongLiteral(b *testing.B) {
 	dir := tmpData(b, "sql", `{"format":"INSERT INTO customers (id, name, city) VALUES (123, '{word}', '{word}');","word":["alpha","beta","gamma","delta"]}`)
-	benchPath(b, dir, "sql")
+	benchPath(b, WithDataPath(dir), "sql")
 }
 
 func BenchmarkRepeat(b *testing.B) {
 	dir := tmpData(b, "many", `{"format":"{word}","repeat":20,"separator":", ","word":["alpha","beta","gamma","delta"]}`)
-	benchPath(b, dir, "many")
+	benchPath(b, WithDataPath(dir), "many")
 }
 
-// BenchmarkBound measures a format that binds a head and reads two paths from it —
-// what a correlated pair costs against BenchmarkUnbound, the same output drawn from
-// two independent fields.
-func BenchmarkBound(b *testing.B) {
-	dir := tmpData(b, "addr", `{"format":"{place.postal-code} {place.locality}","place":[{"format":"{locality}","locality":"Stockholm","postal-code":"1{digits(2)} {digits(2)}"},{"format":"{locality}","locality":"Tranås","postal-code":"573 {digits(2)}"}]}`)
-	benchPath(b, dir, "addr")
+// BenchmarkNamed measures a format that binds a field to a name and reads two paths through
+// it — what a correlated pair costs against BenchmarkUnnamed, the same output drawn from two
+// independent fields.
+func BenchmarkNamed(b *testing.B) {
+	dir := tmpData(b, "addr", `{"format":"{place as p}{p.postal-code} {p.locality}","place":[{"format":"{locality}","locality":"Stockholm","postal-code":"1{digits(2)} {digits(2)}"},{"format":"{locality}","locality":"Tranås","postal-code":"573 {digits(2)}"}]}`)
+	benchPath(b, WithDataPath(dir), "addr")
 }
 
-func BenchmarkUnbound(b *testing.B) {
+func BenchmarkUnnamed(b *testing.B) {
 	dir := tmpData(b, "addr", `{"format":"{postal-code} {locality}","postal-code":["1{digits(2)} {digits(2)}","573 {digits(2)}"],"locality":["Stockholm","Tranås"]}`)
-	benchPath(b, dir, "addr")
+	benchPath(b, WithDataPath(dir), "addr")
 }
 
-// BenchmarkBoundDeep reads two paths through an intermediate level, the shape the
-// depth question is about.
-func BenchmarkBoundDeep(b *testing.B) {
-	dir := tmpData(b, "addr", `{"format":"{p.addr.city} {p.addr.zip}","p":{"format":"{addr}","addr":{"format":"{city}","city":["Stockholm","Tranås"],"zip":"1{digits(2)} {digits(2)}"}}}`)
-	benchPath(b, dir, "addr")
+// BenchmarkNamedDeep reads two paths through a name and an intermediate level.
+func BenchmarkNamedDeep(b *testing.B) {
+	dir := tmpData(b, "addr", `{"format":"{p as q}{q.addr.city} {q.addr.zip}","p":{"format":"{addr}","addr":{"format":"{city}","city":["Stockholm","Tranås"],"zip":"1{digits(2)} {digits(2)}"}}}`)
+	benchPath(b, WithDataPath(dir), "addr")
 }
 
-// BenchmarkBoundWide reads ten paths from one row: the point where the maps an
-// expansion holds outgrow a single bucket.
-func BenchmarkBoundWide(b *testing.B) {
-	dir := tmpData(b, "row", `{"format":"{r.a}{r.b}{r.c}{r.d}{r.e}{r.f}{r.g}{r.h}{r.i}{r.j}","r":[
+// BenchmarkNamedWide reads ten paths through one name: the point where the maps a pick keeps
+// outgrow a single bucket.
+func BenchmarkNamedWide(b *testing.B) {
+	dir := tmpData(b, "row", `{"format":"{r as s}{s.a}{s.b}{s.c}{s.d}{s.e}{s.f}{s.g}{s.h}{s.i}{s.j}","r":[
 		{"format":"x","a":"1","b":"2","c":"3","d":"4","e":"5","f":"6","g":"7","h":"8","i":"9","j":"0"},
 		{"format":"y","a":"A","b":"B","c":"C","d":"D","e":"E","f":"F","g":"G","h":"H","i":"I","j":"J"}]}`)
-	benchPath(b, dir, "row")
+	benchPath(b, WithDataPath(dir), "row")
+}
+
+func BenchmarkFirstFake(b *testing.B) {
+	b.ReportAllocs()
+	for i := 0; i < b.N; i++ {
+		f, err := New(withShipped(), WithSeed(1))
+		if err != nil {
+			b.Fatal(err)
+		}
+		if _, err := f.Fake("sv_SE.address"); err != nil {
+			b.Fatal(err)
+		}
+	}
 }
 
 func BenchmarkNew(b *testing.B) {
 	b.ReportAllocs()
 	for i := 0; i < b.N; i++ {
-		if _, err := New(WithSeed(1)); err != nil {
+		if _, err := New(withShipped(), WithSeed(1)); err != nil {
 			b.Fatal(err)
 		}
 	}

bound_test.go

@@ -1,714 +0,0 @@
-package fejkdata
-
-import (
-	"fmt"
-	"regexp"
-	"strings"
-	"testing"
-)
-
-// This file pins the one-draw rule for dotted sibling tokens. A format that
-// addresses a sibling by path ({place.locality}) draws that sibling once per
-// expansion, so every {place.*} token in the same format reads the same variant.
-// That is what makes a correlated pair — a locality and a postal code that agree
-// — expressible in data, without the row having to be the whole node.
-
-// swedishPlaces is a two-variant sibling whose variants pair a locality with the
-// postal-code prefix that really belongs to it.
-const swedishPlaces = `{"format":"%s","place":[{"format":"{locality}","locality":"Stockholm","postal-code":"1{digits(2)} {digits(2)}"},{"format":"{locality}","locality":"Tranås","postal-code":"573 {digits(2)}"}]}`
-
-// agree reports whether a rendered "postcode locality" pair is a real pairing.
-var agree = regexp.MustCompile(`^(1[0-9]{2} [0-9]{2} Stockholm|573 [0-9]{2} Tranås)$`)
-
-func places(format string) string {
-	return strings.Replace(swedishPlaces, "%s", format, 1)
-}
-
-func TestDottedTokenCorrelatesSiblingFields(t *testing.T) {
-	f := engine(1)
-	seen := map[string]bool{}
-	for i := 0; i < 500; i++ {
-		got := mustRender(t, f, places("{place.postal-code} {place.locality}"))
-		if !agree.MatchString(got) {
-			t.Fatalf("draw %d = %q, want a postcode and locality that pair", i, got)
-		}
-		seen[got[len(got)-4:]] = true
-	}
-	if len(seen) < 2 {
-		t.Fatalf("500 draws produced only %v, want both localities", seen)
-	}
-}
-
-func TestRenderingALevelAndReadingIntoItIsRejected(t *testing.T) {
-	// One draw, one spelling. A token that renders a level and one that reads a
-	// path into it name the same draw two ways, and only the path reads what the
-	// other rendered — so the pair is a load error rather than a shape whose two
-	// halves can disagree.
-	rejected := map[string]string{
-		"bare head beside a path": `{"format":"{p} + {p.first}","p":{"format":"{first}","first":["Anna","Bo"]}}`,
-		"prefix path beside a path": `{"format":"{p.addr}|{p.addr.city}","p":{"format":"{addr}","addr":[` +
-			`{"format":"{city}","city":["Kiruna","Boden"]},{"format":"{city}","city":["Malmö","Lund"]}]}}`,
-		"either order": `{"format":"{p.first} + {p}","p":{"format":"{first}","first":["Anna","Bo"]}}`,
-	}
-	for name, file := range rejected {
-		_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{"cat": file})))
-		if err == nil {
-			t.Errorf("%s: New = nil error, want the overlapping tokens rejected", name)
-			continue
-		}
-		if !strings.Contains(err.Error(), "reads a path into") {
-			t.Errorf("%s: New = %v, want it to name the overlap", name, err)
-		}
-	}
-	// The same path twice is one spelling, so it stays legal.
-	if _, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"{p.first} + {p.first}","p":{"format":"{first}","first":["Anna","Bo"]}}`,
-	}))); err != nil {
-		t.Errorf("New = %v, want one path read twice accepted", err)
-	}
-}
-
-func TestCalcOperandNamingABoundLevelIsRejected(t *testing.T) {
-	// A calc operand renders its field, so naming a bound level in one is the same
-	// overlap as a bare token: {calc(item * 1)} renders what {item.price} reads a
-	// path into, and the two disagree.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"{item.price}|{calc(item * 1)}","item":[` +
-			`{"format":"{price}","price":"10"},{"format":"{price}","price":"20"}]}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), "reads a path into") {
-		t.Fatalf("New = %v, want the calc operand rejected as an overlap", err)
-	}
-	// Arithmetic over fields no path names is untouched.
-	if _, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"{net} x {qty} = {calc(net * qty, 2)}","net":"19.99","qty":"3"}`,
-	}))); err != nil {
-		t.Errorf("New = %v, want plain arithmetic accepted", err)
-	}
-}
-
-func TestReferenceNamingABoundLevelIsRejected(t *testing.T) {
-	// A reference can name a bound level from the data root, which renders it
-	// afresh beside the path that reads its held draw — the same overlap by
-	// another spelling.
-	rejected := map[string]string{
-		"reference names the head": `{"format":"{p.first}|{/cat.p}","p":{"format":"{first}","first":["Anna","Bo"]}}`,
-		"reference names the leaf": `{"format":"{p.addr}|{/cat.p.addr}","p":{"format":"x","addr":["A","B","C","D"]}}`,
-		// The reference need not sit in the format that binds: any field it renders
-		// reaches the level just the same, however deep.
-		"reference from a sibling field": `{"format":"{p.first}|{inner}","p":[` +
-			`{"format":"{first}-{last}","first":"A","last":"1"},{"format":"{first}-{last}","first":"B","last":"2"}],` +
-			`"inner":"{/cat.p}"}`,
-	}
-	for name, file := range rejected {
-		_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{"cat": file})))
-		if err == nil || !strings.Contains(err.Error(), "reads a path into") {
-			t.Errorf("%s: New = %v, want the reference rejected as an overlap", name, err)
-		}
-	}
-	// A reference to anything this format does not bind is untouched.
-	if _, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat":     `{"format":"{p.first} {/surname}","p":{"format":"{first}","first":["Anna","Bo"]}}`,
-		"surname": `["Eriksson","Lindqvist"]`,
-	}))); err != nil {
-		t.Errorf("New = %v, want a reference outside the bound level accepted", err)
-	}
-}
-
-func TestCycleReachedOnlyByAPathTokenIsRejected(t *testing.T) {
-	// A path token renders what it lands on, not the level it started from, so the
-	// cycle walk has to follow it there. A head whose own format names nothing
-	// would otherwise hide the cycle until render, where it is fatal.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"a": `{"format":"{p.x}","p":{"format":"static","x":"{/a}"}}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), "reference cycle") {
-		t.Fatalf("New = %v, want the cycle through {p.x} rejected", err)
-	}
-}
-
-func TestALevelRenderedOnlyByAPathTokenIsHeld(t *testing.T) {
-	// {p.a} renders q, so it is a route to the level {q.x} holds — even though p's
-	// own format names nothing.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"thing": `{"format":"{p.a} {q.x}","p":{"format":"static","a":"{/thing.q}"},` +
-			`"q":{"format":"{x}","x":["1","2"]}}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), "reads a path into") {
-		t.Fatalf("New = %v, want the second route to q rejected", err)
-	}
-}
-
-// TestACalcOperandIsHeldAgainstEveryRoute pins what a {calc()} operand's hold
-// fences. A calc renders its operand whole, so the draw it holds is that one node's
-// value: another route conflicts only by naming that same node, and then the value
-// shown is not the value computed. Two names that merely draw from one source are
-// two draws, as {word} {word} is — the rule everywhere else in the engine.
-func TestACalcOperandIsHeldAgainstEveryRoute(t *testing.T) {
-	rejected := map[string]struct {
-		files map[string]string
-		want  string // the route the error names, spelled as the author wrote it
-	}{
-		"a reference beside the operand": {
-			map[string]string{
-				"cat": `{"format":"{/cat.net} x 2 = {calc(net * 2, 2)}","net":["10.00","20.00"]}`,
-			},
-			`{/cat.net} renders "net"`,
-		},
-		"a reference one level down": {
-			map[string]string{
-				"cat": `{"format":"{calc(net * 2, 2)} {q}","net":["10.00","20.00"],` +
-					`"q":"{/cat.net}"}`,
-			},
-			`{q} renders "net"`,
-		},
-		// A reference ending at the operand binds the choice itself, not a variant,
-		// so wrapping the operand in one changes nothing.
-		"a reference to an operand wrapped in a choice": {
-			map[string]string{
-				"cat": `{"format":"{calc(n * 2, 2)} {q}","n":{"format":"{v}","v":["1","2"]},` +
-					`"q":"{/cat.n}"}`,
-			},
-			`{q} renders "n"`,
-		},
-		// The reaching route can be another operand, which is not a token — so the
-		// error must name it as one rather than invent a {b} the format never wrote.
-		"an operand reaching another operand": {
-			map[string]string{
-				"cat": `{"format":"{calc(a + b, 0)}","a":{"format":"{x}","x":["1","2"]},` +
-					`"b":"{/cat.a}"}`,
-			},
-			`calc operand "b" renders "a"`,
-		},
-		// Reaching *into* the operand is the same shape as naming it: the draw the
-		// calc holds settled that value too. The token spelling {net.v} is already
-		// rejected, so the reference spelling has to be.
-		"a reference into the operand": {
-			map[string]string{
-				"cat": `{"format":"{calc(net * 2, 2)}|{/cat.net.v}",` +
-					`"net":{"format":"{v}","v":["10.00","20.00"]}}`,
-			},
-			`{/cat.net.v} renders "net"`,
-		},
-		"a reference into the operand one level down": {
-			map[string]string{
-				"cat": `{"format":"{calc(net * 2, 2)}|{q}","net":{"format":"{v}","v":["10.00","20.00"]},` +
-					`"q":"{/cat.net.v}"}`,
-			},
-			`{q} renders "net"`,
-		},
-		// Each head is walked against its own seen set. Sharing one across heads
-		// would let the first head's walk mark the only route to the second's draw,
-		// and this violation sits on the later head in sorted order.
-		"a violation on the second of two held heads": {
-			map[string]string{
-				"cat": `{"format":"{calc(a + b, 0)} {w}","a":{"format":"{x}","x":["1","2"]},` +
-					`"b":{"format":"{y}","y":["3","4"]},"w":"{/cat.b}"}`,
-			},
-			`{w} renders "b"`,
-		},
-	}
-	for name, c := range rejected {
-		_, err := New(WithoutShippedData(), WithDataPath(writeData(t, c.files)))
-		if err == nil || !strings.Contains(err.Error(), "a {calc()} also reads") {
-			t.Errorf("%s: New = %v, want the second route to the operand rejected", name, err)
-			continue
-		}
-		if !strings.Contains(err.Error(), c.want) {
-			t.Errorf("%s: New = %v, want it to name %q", name, err, c.want)
-		}
-	}
-
-	accepted := map[string]map[string]string{
-		// The spellings that read the one draw: the operand itself, and a bare token.
-		"only the operand and a bare token": {
-			"cat": `{"format":"{net} x 2 = {calc(net * 2, 2)}","net":["10.00","20.00"]}`,
-		},
-		// A sibling the operand's format never renders is no part of its value. A
-		// path head differs — a path may read into anything the level contains —
-		// which is why that half of the fence covers containment.
-		"a reference to a sibling the operand never renders": {
-			"cat": `{"format":"{calc(net * 2, 2)} {unit}",` +
-				`"net":{"format":"{v}","v":["1","2"],"spare":["kg","lb"]},` +
-				`"unit":"{/cat.net.spare}"}`,
-		},
-		// Two operands drawing from one source are two names, so two draws: each is
-		// held under its own name and shown once, and neither can disagree.
-		"two operands sharing one source": {
-			"die": `["1","2","3","4","5","6"]`,
-			"cat": `{"format":"{d1} + {d2} = {calc(d1 + d2, 0)}","d1":"{/die}",` +
-				`"d2":"{/die}"}`,
-		},
-		// Likewise a reference drawing from what the operand draws from: {/common}
-		// and net are two names, not two spellings of one field.
-		"a reference to what an operand renders through": {
-			"common": `["1","2"]`,
-			"cat":    `{"format":"{calc(net * 2, 2)} {/common}","net":"{/common}"}`,
-		},
-		// A fixed string cannot disagree with itself, so it needs no fence.
-		"a literal operand named twice": {
-			"cat": `{"format":"{calc(n * 2, 0)} {/cat.n}","n":"5"}`,
-		},
-		// One node reached twice while walking the operand: the walk must not
-		// revisit it, and the repeat is not a second route to anything.
-		"an operand that renders one field twice": {
-			"cat": `{"format":"{calc(v * 2, 0)}","v":{"format":"{a}{a}","a":["1","2"]}}`,
-		},
-	}
-	for name, files := range accepted {
-		if _, err := New(WithoutShippedData(), WithDataPath(writeData(t, files))); err != nil {
-			t.Errorf("%s: New = %v, want it accepted", name, err)
-		}
-	}
-}
-
-func TestAPathReachesEveryVariantItMightDraw(t *testing.T) {
-	// A path through a choice may land in any variant, so both walks have to see
-	// all of them. Reaching only the first leaves whatever hides in a later
-	// variant to be found at render — a cycle there is fatal, and a second route
-	// to a held level disagrees silently.
-	rejected := map[string]struct{ file, want string }{
-		"a cycle in a later variant": {
-			`{"format":"{p.x}","p":[{"format":"h","x":"safe"},{"format":"h","x":"{/cat}"}]}`,
-			"reference cycle",
-		},
-		"a second route in a later variant": {
-			`{"format":"{p.x} {q.y}","p":[{"format":"h","x":"safe"},{"format":"h","x":"{/cat.q}"}],` +
-				`"q":{"format":"{y}","y":["1","2"]}}`,
-			"reads a path into",
-		},
-	}
-	for name, c := range rejected {
-		_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{"cat": c.file})))
-		if err == nil || !strings.Contains(err.Error(), c.want) {
-			t.Errorf("%s: New = %v, want it to mention %q", name, err, c.want)
-		}
-	}
-}
-
-func TestALevelAPathNeverRendersIsAccepted(t *testing.T) {
-	// A path token does not expand its head's format, so a reference sitting in
-	// that format is not a second route to anything: it is never rendered by the
-	// path at all. Both orders must load.
-	accepted := map[string]string{
-		"reference in the head's own format": `{"format":"{p.first} {q.a}",` +
-			`"p":{"format":"{first} {/thing.q}","first":["A","B"]},"q":{"format":"{a}","a":["1","2"]}}`,
-		"the mirror shape": `{"format":"{p.first} {q.a}",` +
-			`"p":{"format":"{first}","first":["A","B"]},"q":{"format":"{a} {/thing.p}","a":["1","2"]}}`,
-	}
-	for name, file := range accepted {
-		if _, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{"thing": file}))); err != nil {
-			t.Errorf("%s: New = %v, want it accepted", name, err)
-		}
-	}
-}
-
-func TestADeepDiamondChainLoads(t *testing.T) {
-	// A diamond chain is 2^n routes through n nodes. The search past a bound level
-	// must walk each node once, not once per route, or a deep chain never loads.
-	files := map[string]string{"l0": `"x"`}
-	for i := 1; i <= 30; i++ {
-		files[fmt.Sprintf("l%d", i)] = fmt.Sprintf(
-			`{"format":"{a}{b}","a":"{/l%d}","b":"{/l%d}"}`, i-1, i-1)
-	}
-	files["thing"] = `{"format":"{p.first} {/l30}","p":{"format":"x","first":["A","B"]}}`
-	if _, err := New(WithoutShippedData(), WithDataPath(writeData(t, files))); err != nil {
-		t.Fatalf("New = %v, want a deep diamond chain to load", err)
-	}
-}
-
-func TestASharedNodeIsWalkedOnce(t *testing.T) {
-	// Two fields reaching one node make the render graph a diamond, not a tree.
-	// The search past a bound level must take that in its stride rather than walk
-	// the shared node once per route.
-	f, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"{p.first}|{q}","p":{"format":"{first}","first":["Anna","Bo"]},` +
-			`"q":{"format":"{a}{b}","a":"{/shared}","b":"{/shared}"}}`,
-		"shared": `"x"`,
-	})), WithSeed(1))
-	if err != nil {
-		t.Fatalf("New = %v, want a shared node accepted", err)
-	}
-	if got, err := f.Fake("cat"); err != nil || !strings.HasSuffix(got, "|xx") {
-		t.Fatalf("Fake(cat) = %q, %v, want it to end in xx", got, err)
-	}
-}
-
-func TestTheEarlierReaderIsNamed(t *testing.T) {
-	// A token and a calc operand can name one level. The one the format writes
-	// first is the one reported, so the error points at the same place a reader
-	// looking at the format would start.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"{calc(p * 1)} {p} {p.first}","p":{"format":"{first}","first":["1","2"]}}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), `calc operand "p"`) {
-		t.Fatalf("New = %v, want the calc operand named, being written first", err)
-	}
-}
-
-func TestReferenceToAMatchingStringIsAccepted(t *testing.T) {
-	// A literal renders one fixed string, so no draw of it can disagree with a
-	// held one. Two unrelated literals that merely spell the same text must not
-	// read as the same node — the trap when comparing a value type.
-	if _, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat":   `{"format":"{p.city} {/other.tag}","p":{"format":"{city}","city":"Stockholm"}}`,
-		"other": `{"format":"x","tag":"Stockholm"}`,
-	}))); err != nil {
-		t.Fatalf("New = %v, want a reference to a matching string accepted", err)
-	}
-}
-
-func TestNestedChoiceDrawsOneVariant(t *testing.T) {
-	// A choice item may itself be a choice, so drawing a bound head unwraps until
-	// it reaches a value. Stopping at one level leaves a choice where the walk
-	// expects a template.
-	f := engine(21)
-	seen := map[string]bool{}
-	for i := 0; i < 200; i++ {
-		seen[mustRender(t, f, `{"format":"{p.x}","p":[{"format":"{x}","x":"1"},{"format":"{x}","x":"2"}]}`)] = true
-	}
-	if !seen["1"] || !seen["2"] || len(seen) != 2 {
-		t.Fatalf("200 draws produced %v, want 1 and 2", seen)
-	}
-}
-
-func TestRepeatingLevelIsNamedInTheError(t *testing.T) {
-	// The level carrying the repeat is the one to fix, so the error names it
-	// rather than the head the path started from.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"[{p.a.b}]","p":{"format":"{a}","a":{"format":"{b}","repeat":3,"separator":",","b":"z"}}}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), `"p.a"`) {
-		t.Fatalf("New = %v, want it to name the level p.a that carries the repeat", err)
-	}
-}
-
-func TestPathIntoARepeatingLevelBehindAChoiceIsRejected(t *testing.T) {
-	// A choice of rows is the shape this feature is for, so the repeat rule has to
-	// reach inside one — otherwise the direct spelling is a load error and the same
-	// mistake behind a choice silently drops the repeat.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"[{p.a}]","p":[{"format":"{a}","repeat":3,"separator":",","a":"x"},{"format":"{a}","a":"y"}]}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), "repeat") {
-		t.Fatalf("New = %v, want the repeat behind a choice rejected", err)
-	}
-}
-
-func TestPathIntoARepeatingLevelIsRejected(t *testing.T) {
-	// A path reads one level's draw, so it can never apply that level's repeat.
-	// The engine rejects options that cannot take effect, and this is one.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"[{p.a}]","p":{"format":"{a}","repeat":3,"separator":",","a":"z"}}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), "repeat") {
-		t.Fatalf("New = %v, want a path into a repeating level rejected", err)
-	}
-}
-
-func TestPathIntoAPlainTemplateNamesTheMissingField(t *testing.T) {
-	// A head that is one template, not a choice, reports the missing segment
-	// directly — there are no variants to compare.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"cat": `{"format":"{a.nope}","a":{"format":"x","b":"1"}}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), `no field "nope"`) {
-		t.Fatalf("New = %v, want it to name the missing field", err)
-	}
-}
-
-func TestFieldWithoutAPathTokenStillDrawsEachTime(t *testing.T) {
-	// A format with no dotted token keeps today's behaviour: every {word} is an
-	// independent draw.
-	f := engine(3)
-	varied := false
-	for i := 0; i < 200 && !varied; i++ {
-		got := mustRender(t, f, `{"format":"{w}{w}{w}","w":["a","b","c","d","e"]}`)
-		varied = got[0] != got[1] || got[1] != got[2]
-	}
-	if !varied {
-		t.Fatal("200 draws of {w}{w}{w} never varied, want independent draws")
-	}
-}
-
-func TestTwoBoundHeadsDrawIndependently(t *testing.T) {
-	// Binding is per head, so two correlated sets in one format compose instead
-	// of needing a cross product of variants.
-	f := engine(4)
-	seen := map[string]bool{}
-	for i := 0; i < 400; i++ {
-		seen[mustRender(t, f, `{"format":"{p.v}{q.v}",
-			"p":[{"format":"{v}","v":"A"},{"format":"{v}","v":"B"}],
-			"q":[{"format":"{v}","v":"1"},{"format":"{v}","v":"2"}]}`)] = true
-	}
-	for _, want := range []string{"A1", "A2", "B1", "B2"} {
-		if !seen[want] {
-			t.Fatalf("400 draws produced %v, want every combination including %q", seen, want)
-		}
-	}
-}
-
-func TestRepeatRedrawsTheBinding(t *testing.T) {
-	// The binding lives for one expansion, so each repetition is a fresh draw —
-	// and each repetition is internally consistent.
-	f := engine(5)
-	tmpl := strings.Replace(places("{place.postal-code} {place.locality}"), `"format":"{place.postal-code} {place.locality}"`,
-		`"format":"{place.postal-code} {place.locality}","repeat":6,"separator":"\n"`, 1)
-	varied := false
-	for i := 0; i < 40; i++ {
-		got := mustRender(t, f, tmpl)
-		lines := strings.Split(got, "\n")
-		if len(lines) != 6 {
-			t.Fatalf("repeat 6 produced %d lines", len(lines))
-		}
-		for _, line := range lines {
-			if !agree.MatchString(line) {
-				t.Fatalf("repetition %q does not pair", line)
-			}
-		}
-		if lines[0] != lines[1] || lines[1] != lines[2] {
-			varied = true
-		}
-	}
-	if !varied {
-		t.Fatal("40 renders of repeat 6 never varied within a render, want a redraw per repetition")
-	}
-}
-
-func TestNestedTemplateKeepsItsOwnBinding(t *testing.T) {
-	// A binding belongs to the expansion that made it: an inner template's own
-	// place is not the outer one's.
-	f := engine(6)
-	tmpl := `{"format":"{place.v}{inner}",
-		"place":[{"format":"{v}","v":"A"},{"format":"{v}","v":"B"}],
-		"inner":{"format":"{place.v}","place":[{"format":"{v}","v":"A"},{"format":"{v}","v":"B"}]}}`
-	seen := map[string]bool{}
-	for i := 0; i < 400; i++ {
-		seen[mustRender(t, f, tmpl)] = true
-	}
-	if !seen["AB"] && !seen["BA"] {
-		t.Fatalf("400 draws produced %v, want the inner binding to be independent", seen)
-	}
-}
-
-func TestBoundDrawIsSeedStable(t *testing.T) {
-	tmpl := places("{place.postal-code} {place.locality}")
-	a := mustRender(t, engine(42), tmpl)
-	b := mustRender(t, engine(42), tmpl)
-	if a != b {
-		t.Fatalf("same seed gave %q and %q, want an identical draw", a, b)
-	}
-}
-
-func TestDottedTokenErrors(t *testing.T) {
-	// A path token is validated where every other token is: at New, naming what
-	// is wrong, so a typo can never reach a render.
-	rejected := map[string]struct {
-		file string
-		want string
-	}{
-		"unknown head": {
-			`{"format":"{nope.x}","place":{"format":"{v}","v":"A"}}`,
-			`no field "nope"`,
-		},
-		"unknown tail": {
-			`{"format":"{place.nope}","place":[{"format":"{v}","v":"A"},{"format":"{v}","v":"B"}]}`,
-			`"nope"`,
-		},
-		"tail only some variants carry": {
-			`{"format":"{place.locality}","place":[{"format":"{locality}","locality":"A"},"x"]}`,
-			`locality`,
-		},
-		"path into a literal": {
-			`{"format":"{x.y}","x":"plain"}`,
-			`"y"`,
-		},
-		"dotted field key": {
-			`{"format":"[{a.b}]","a.b":"V"}`,
-			`contains "."`,
-		},
-	}
-	for name, c := range rejected {
-		_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{"cat": c.file})))
-		if err == nil {
-			t.Errorf("%s: New = nil error, want it rejected at load", name)
-			continue
-		}
-		if !strings.Contains(err.Error(), c.want) {
-			t.Errorf("%s: New = %v, want it to mention %q", name, err, c.want)
-		}
-	}
-}
-
-func TestSamePathReadTwiceReadsOneValue(t *testing.T) {
-	// A path is drawn once per expansion, so reading it twice reads one value —
-	// what a name shown in a display form and again in an address needs.
-	f := engine(7)
-	for i := 0; i < 300; i++ {
-		got := mustRender(t, f, `{"format":"{p.first}|{p.first}","p":{"format":"{first}","first":["Anna","Astrid","Elin","Karin"]}}`)
-		parts := strings.Split(got, "|")
-		if parts[0] != parts[1] {
-			t.Fatalf("draw %d = %q, want one value read twice", i, got)
-		}
-	}
-}
-
-func TestDifferentTailsUnderOneHeadShareTheRow(t *testing.T) {
-	// Two tails of one head stay in the same variant, each keeping its own value.
-	f := engine(8)
-	for i := 0; i < 300; i++ {
-		got := mustRender(t, f, `{"format":"{p.a}{p.b}",
-			"p":[{"format":"x","a":"A","b":"1"},{"format":"y","a":"B","b":"2"}]}`)
-		if got != "A1" && got != "B2" {
-			t.Fatalf("draw %d = %q, want a row's own pair", i, got)
-		}
-	}
-}
-
-// deepPlaces nests the correlated facts a level down: the row holds an addr, and
-// the addr holds the city and the zip that belongs to it.
-const deepPlaces = `{"format":"%s","p":{"format":"{addr}","addr":[{"format":"{city}","city":"Stockholm","zip":"11111"},{"format":"{city}","city":"Kiruna","zip":"98100"}]}}`
-
-func TestPathHoldsItsDrawAtEveryLevel(t *testing.T) {
-	// A choice below the head is drawn once too, so two paths sharing a prefix
-	// share every level of it — not just the row they started from.
-	f := engine(11)
-	seen := map[string]bool{}
-	for i := 0; i < 500; i++ {
-		got := mustRender(t, f, strings.Replace(deepPlaces, "%s", "{p.addr.city}/{p.addr.zip}", 1))
-		if got != "Stockholm/11111" && got != "Kiruna/98100" {
-			t.Fatalf("draw %d = %q, want a city and the zip that belongs to it", i, got)
-		}
-		seen[got] = true
-	}
-	if len(seen) < 2 {
-		t.Fatalf("500 draws produced only %v, want both", seen)
-	}
-}
-
-func TestPathHoldsItsDrawThreeLevelsDown(t *testing.T) {
-	// The rule does not run out at two: every level a path passes through is held.
-	f := engine(12)
-	tmpl := `{"format":"{p.geo.town.name}/{p.geo.town.zip} {p.geo.region}","p":{"format":"{geo}","geo":[{"format":"{town}","region":"Norrbotten","town":[{"format":"{name}","name":"Kiruna","zip":"98100"},{"format":"{name}","name":"Luleå","zip":"97200"}]},{"format":"{town}","region":"Skåne","town":[{"format":"{name}","name":"Malmö","zip":"21100"},{"format":"{name}","name":"Lund","zip":"22100"}]}]}}`
-	want := map[string]bool{
-		"Kiruna/98100 Norrbotten": true, "Luleå/97200 Norrbotten": true,
-		"Malmö/21100 Skåne": true, "Lund/22100 Skåne": true,
-	}
-	seen := map[string]bool{}
-	for i := 0; i < 600; i++ {
-		got := mustRender(t, f, tmpl)
-		if !want[got] {
-			t.Fatalf("draw %d = %q, want a town with its own zip and region", i, got)
-		}
-		seen[got] = true
-	}
-	if len(seen) != len(want) {
-		t.Fatalf("600 draws produced %v, want all four towns", seen)
-	}
-}
-
-func TestPathWalksVariantsOfDifferentShape(t *testing.T) {
-	// One variant holds addr as a plain template where another holds a choice.
-	// Both carry the path, so both must walk — the walk reads what it drew, never
-	// assuming a shape.
-	f := engine(13)
-	tmpl := `{"format":"{p.addr.city}/{p.addr.zip}","p":[
-		{"format":"{addr}","addr":{"format":"{city}","city":"Kiruna","zip":"98100"}},
-		{"format":"{addr}","addr":[
-			{"format":"{city}","city":"Malmö","zip":"21100"},
-			{"format":"{city}","city":"Lund","zip":"22100"}]}]}`
-	for i := 0; i < 400; i++ {
-		got := mustRender(t, f, tmpl)
-		if got != "Kiruna/98100" && got != "Malmö/21100" && got != "Lund/22100" {
-			t.Fatalf("draw %d = %q, want a city with its own zip", i, got)
-		}
-	}
-}
-
-func TestDeepPathsUnderOneHeadStayIndependentWhereTheyDiverge(t *testing.T) {
-	// Two paths share only what they actually share: a common prefix is one draw,
-	// and each keeps its own draw past the point they part.
-	f := engine(14)
-	tmpl := `{"format":"{p.a.v}{p.b.v}","p":{"format":"x","a":[{"format":"{v}","v":"1"},{"format":"{v}","v":"2"}],"b":[{"format":"{v}","v":"1"},{"format":"{v}","v":"2"}]}}`
-	seen := map[string]bool{}
-	for i := 0; i < 400; i++ {
-		seen[mustRender(t, f, tmpl)] = true
-	}
-	for _, want := range []string{"11", "12", "21", "22"} {
-		if !seen[want] {
-			t.Fatalf("400 draws produced %v, want every combination including %q", seen, want)
-		}
-	}
-}
-
-func TestEmptyPathSegmentIsRejected(t *testing.T) {
-	// "{a.}", "{.b}" and "{a..b}" are unfinished paths, and the segment naming
-	// nothing is reported as that rather than as a missing field. An empty name is
-	// rejected where it is authored, so no data can make these resolve.
-	rejected := map[string]string{
-		"trailing dot": `{"format":"[{a.}]","a":"x"}`,
-		"leading dot":  `{"format":"[{.b}]","a":{"format":"{b}","b":"V"}}`,
-		"double dot":   `{"format":"[{a..b}]","a":"x"}`,
-	}
-	for name, file := range rejected {
-		_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{"cat": file})))
-		if err == nil {
-			t.Errorf("%s: New = nil error, want the unfinished path rejected", name)
-			continue
-		}
-		if !strings.Contains(err.Error(), "empty segment") {
-			t.Errorf("%s: New = %v, want it to name the empty segment", name, err)
-		}
-	}
-}
-
-func TestCycleThroughAPathTokenIsRejected(t *testing.T) {
-	// A path token is a render edge like any other, so a cycle routed through one
-	// must be caught at New. Reaching render would be fatal: the recursion never
-	// terminates, and a stack overflow cannot be recovered.
-	_, err := New(WithoutShippedData(), WithDataPath(writeData(t, map[string]string{
-		"a": `{"format":"{p.x}","p":{"format":"{x}","x":"{/a}"}}`,
-	})))
-	if err == nil || !strings.Contains(err.Error(), "reference cycle") {
-		t.Fatalf("New = %v, want the cycle through {p.x} rejected", err)
-	}
-}
-
-func TestBoundPathIsReachableByFake(t *testing.T) {
-	// Binding changes how a format reads a sibling, not what List and Fake offer:
-	// the sub-fields stay addressable on their own.
-	dir := writeData(t, map[string]string{"address": places("{place.postal-code} {place.locality}")})
-	f, err := New(WithoutShippedData(), WithDataPath(dir), WithSeed(9))
-	if err != nil {
-		t.Fatalf("New = %v", err)
-	}
-	for _, path := range []string{"address", "address.place", "address.place.locality", "address.place.postal-code"} {
-		if _, err := f.Fake(path); err != nil {
-			t.Errorf("Fake(%q) = %v, want it to render", path, err)
-		}
-	}
-}
-
-func TestRepeatedBareTokenOfAHeldNameIsRejected(t *testing.T) {
-	for src, want := range map[string]string{
-		`{"format":"{w} {w} {uppercase(w)}","w":["a","b"]}`:                                    "write {w} once",
-		`{"format":"{w} {w|x} {uppercase(w)}","w":["a","b"],"x":["c","d"]}`:                    "write {w} once",
-		`{"format":"{n} + {n} = {calc(n * 2)}","n":["1","2"]}`:                                 "write {n} once",
-		`{"format":"{p.a} {q} {q} {lowercase(q)}","p":{"format":"{a}","a":"1"},"q":["A","B"]}`: "write {q} once",
-	} {
-		_, err := compile(parse(t, src))
-		if err == nil || !strings.Contains(err.Error(), want) || !strings.Contains(err.Error(), "holds") {
-			t.Errorf("compile(%s) = %v, want the repeated token rejected naming %s", src, err, want)
-		}
-	}
-	for _, ok := range []string{
-		`{"format":"{w} {w}","w":["a","b"]}`,
-		`{"format":"{w} {uppercase(w)}","w":["a","b"]}`,
-		`{"format":"{net} x {qty} = {calc(net * qty, 2)}","net":["19.99","5.00"],"qty":["3","7"]}`,
-		`{"format":"{uppercase(w)} {uppercase(w)}","w":["a","b"]}`,
-	} {
-		if _, err := compile(parse(t, ok)); err != nil {
-			t.Errorf("compile(%s) = %v, want it accepted", ok, err)
-		}
-	}
-}

builtins.go

@@ -1,416 +0,0 @@
-package fejkdata
-
-import (
-	"encoding/base64"
-	"fmt"
-	"math"
-	"strconv"
-	"strings"
-	"unicode"
-)
-
-// maxLen caps sample output lengths (hex, nanoid, base64) and the renders a repeat
-// multiplies to along any path; maxDecimals caps float/calc decimal places. So a
-// fat-fingered or overflowing argument fails at New instead of trying to allocate
-// gigabytes — or panicking — at render.
-const (
-	maxLen      = 1 << 20 // 1,048,576 chars/bytes
-	maxDecimals = 1024
-)
-
-// builtins is the registry of {name(args)} functions. Derivations read the digits
-// emitted so far in the current expansion (place them after their payload);
-// samples read only the rng. A time-based id (uuid v7, ulid) draws its timestamp
-// from the rng, not the wall clock, so seeded output stays reproducible.
-var builtins = map[string]builtin{
-	"luhn":   {arity: 0, prep: derive(func(e string) string { return string(rune('0' + luhnCheck(e))) })},
-	"mod11":  {arity: 0, prep: derive(mod11Check)},
-	"ean":    {arity: 0, prep: derive(eanCheck)},
-	"uuid":   {arity: 0, prep: sample(uuidV7)},
-	"ulid":   {arity: 0, prep: sample(ulid)},
-	"nanoid": {arity: 1, check: posIntArg, prep: chars(nanoidAlphabet)},
-	"hex":    {arity: 1, check: posIntArg, prep: chars(hexDigits)},
-	"digits": {arity: 1, check: posIntArg, prep: chars("0123456789")},
-	"upper":  {arity: 1, check: posIntArg, prep: chars("ABCDEFGHIJKLMNOPQRSTUVWXYZ")},
-	"lower":  {arity: 1, check: posIntArg, prep: chars("abcdefghijklmnopqrstuvwxyz")},
-	"base64": {arity: 1, check: posIntArg, prep: func(a []string) callFn {
-		n := atoi(a[0])
-		return func(s *session, _ string, _ []string) string {
-			return base64.StdEncoding.EncodeToString(randBytes(s, n))
-		}
-	}},
-	"int": {arity: 2, check: intRangeArgs, prep: func(a []string) callFn {
-		lo, span := atoi(a[0]), atoi(a[1])-atoi(a[0])+1
-		return func(s *session, _ string, _ []string) string { return strconv.Itoa(lo + s.IntN(span)) }
-	}},
-	"float": {arity: 3, check: floatArgs, prep: func(a []string) callFn {
-		lo, hi, dp := atof(a[0]), atof(a[1]), atoi(a[2])
-		return func(s *session, _ string, _ []string) string {
-			return strconv.FormatFloat(lo+s.Float64()*(hi-lo), 'f', dp, 64)
-		}
-	}},
-	"iban": {arity: 1, check: ibanArg, prep: func(a []string) callFn {
-		cc := a[0]
-		return func(s *session, _ string, _ []string) string { return iban(s, cc) }
-	}},
-	"calc":      {arity: -1, check: checkCalc, prep: calcPrep, operands: calcOperands},
-	"lowercase": {arity: 1, check: transformArg, prep: transformPrep(strings.ToLower), operands: transformOperand},
-	"uppercase": {arity: 1, check: transformArg, prep: transformPrep(strings.ToUpper), operands: transformOperand},
-	"ascii":     {arity: 1, check: transformArg, prep: transformPrep(asciiFold), operands: transformOperand},
-	// seq is the one stateful builtin: a per-session counter from 1, advancing on
-	// each call. An optional name selects an independent counter; no name uses the
-	// default one. Deterministic by construction, so seeded output stays stable.
-	"seq": {arity: -1, check: seqArg, prep: func(a []string) callFn {
-		key := ""
-		if len(a) == 1 {
-			key = a[0]
-		}
-		return func(s *session, _ string, _ []string) string {
-			return strconv.FormatUint(s.next(key), 10)
-		}
-	}},
-}
-
-// derive and sample are the two argument-free builtin shapes: a derivation reads
-// the output emitted so far, a sample reads only the rng. chars is the shape of a
-// sample of n characters drawn from an alphabet.
-func derive(f func(emitted string) string) func([]string) callFn {
-	return func([]string) callFn {
-		return func(_ *session, emitted string, _ []string) string { return f(emitted) }
-	}
-}
-
-func sample(f func(rng) string) func([]string) callFn {
-	return func([]string) callFn {
-		return func(s *session, _ string, _ []string) string { return f(s) }
-	}
-}
-
-func chars(alphabet string) func([]string) callFn {
-	return func(a []string) callFn {
-		n := atoi(a[0])
-		return func(s *session, _ string, _ []string) string { return randChars(s, n, alphabet) }
-	}
-}
-
-const hexDigits = "0123456789abcdef"
-
-// transforms are the builtins that rewrite one operand's value; they nest, so
-// {lowercase(ascii(x))} folds then lowers.
-var transforms = map[string]func(string) string{
-	"ascii":     asciiFold,
-	"lowercase": strings.ToLower,
-	"uppercase": strings.ToUpper,
-}
-
-// unwrapTransform peels nested transform calls off an operand arg, returning the
-// field it finally names and the transforms to apply, innermost last.
-func unwrapTransform(arg string) (leaf string, chain []func(string) string, err error) {
-	for {
-		name, args, isCall := funcCall(arg)
-		if !isCall {
-			return arg, chain, nil
-		}
-		fn, isTransform := transforms[name]
-		if !isTransform {
-			return "", nil, fmt.Errorf("%s(%s) is not a transform, so it cannot be an operand", name, strings.Join(args, ","))
-		}
-		if len(args) != 1 {
-			return "", nil, fmt.Errorf("%s takes 1 arg, got %d", name, len(args))
-		}
-		chain = append(chain, fn)
-		arg = args[0]
-	}
-}
-
-func transformArg(fields map[string]node, a []string) error {
-	leaf, _, err := unwrapTransform(a[0])
-	if err != nil {
-		return err
-	}
-	if isRef(leaf) {
-		if leaf == refPrefix {
-			return fmt.Errorf("reference has no path")
-		}
-		return nil
-	}
-	return checkArm(leaf, fields)
-}
-
-func transformOperand(a []string) []string {
-	leaf, _, err := unwrapTransform(a[0])
-	if err != nil {
-		return nil
-	}
-	return []string{leaf}
-}
-
-func transformPrep(outer func(string) string) func([]string) callFn {
-	return func(a []string) callFn {
-		_, chain, err := unwrapTransform(a[0])
-		if err != nil {
-			panic(fmt.Sprintf("fejkdata: transform arg %q reached prep unvalidated: %v", a[0], err))
-		}
-		return func(_ *session, _ string, operands []string) string {
-			v := operands[0]
-			for i := len(chain) - 1; i >= 0; i-- {
-				v = chain[i](v)
-			}
-			return outer(v)
-		}
-	}
-}
-
-// asciiFolds maps the Latin letters with diacritics or ligatures to ASCII.
-var asciiFolds = map[rune]string{
-	'À': "A", 'Á': "A", 'Â': "A", 'Ã': "A", 'Ä': "A", 'Å': "A", 'Æ': "AE", 'Ç': "C",
-	'È': "E", 'É': "E", 'Ê': "E", 'Ë': "E", 'Ì': "I", 'Í': "I", 'Î': "I", 'Ï': "I",
-	'Ð': "D", 'Ñ': "N", 'Ò': "O", 'Ó': "O", 'Ô': "O", 'Õ': "O", 'Ö': "O", 'Ø': "O",
-	'Ù': "U", 'Ú': "U", 'Û': "U", 'Ü': "U", 'Ý': "Y", 'Þ': "Th", 'ß': "ss", 'Œ': "OE",
-	'à': "a", 'á': "a", 'â': "a", 'ã': "a", 'ä': "a", 'å': "a", 'æ': "ae", 'ç': "c",
-	'è': "e", 'é': "e", 'ê': "e", 'ë': "e", 'ì': "i", 'í': "i", 'î': "i", 'ï': "i",
-	'ð': "d", 'ñ': "n", 'ò': "o", 'ó': "o", 'ô': "o", 'õ': "o", 'ö': "o", 'ø': "o",
-	'ù': "u", 'ú': "u", 'û': "u", 'ü': "u", 'ý': "y", 'þ': "th", 'ÿ': "y", 'œ': "oe",
-}
-
-// asciiFold rewrites s to ASCII: folded Latin letters stay, any other non-ASCII
-// rune is dropped.
-func asciiFold(s string) string {
-	var b strings.Builder
-	for _, r := range s {
-		if r < unicode.MaxASCII {
-			b.WriteRune(r)
-		} else {
-			b.WriteString(asciiFolds[r])
-		}
-	}
-	return b.String()
-}
-
-// atoi parses an arg a builtin's check already validated. It panics rather than
-// returning zero, so a check that stops covering its own args is a stack trace and
-// not a silently wrong length, range or decimal count.
-func atoi(s string) int {
-	n, err := strconv.Atoi(s)
-	if err != nil {
-		panic(fmt.Sprintf("fejkdata: builtin arg %q reached prep unvalidated: %v", s, err))
-	}
-	return n
-}
-
-// atof is atoi for a float arg, and reports an unvalidated one the same way.
-func atof(s string) float64 {
-	f, err := strconv.ParseFloat(s, 64)
-	if err != nil {
-		panic(fmt.Sprintf("fejkdata: builtin arg %q reached prep unvalidated: %v", s, err))
-	}
-	return f
-}
-
-func randBytes(r rng, n int) []byte {
-	b := make([]byte, n)
-	for i := range b {
-		b[i] = byte(r.IntN(256))
-	}
-	return b
-}
-
-func randChars(r rng, n int, alphabet string) string {
-	b := make([]byte, n)
-	for i := range b {
-		b[i] = alphabet[r.IntN(len(alphabet))]
-	}
-	return string(b)
-}
-
-func posIntArg(_ map[string]node, a []string) error {
-	n, err := strconv.Atoi(a[0])
-	if err != nil || n < 1 {
-		return fmt.Errorf("count %q must be a positive integer", a[0])
-	}
-	if n > maxLen {
-		return fmt.Errorf("count %d exceeds the maximum %d", n, maxLen)
-	}
-	return nil
-}
-
-func intRangeArgs(_ map[string]node, a []string) error {
-	lo, e1 := strconv.Atoi(a[0])
-	hi, e2 := strconv.Atoi(a[1])
-	if e1 != nil || e2 != nil {
-		return fmt.Errorf("int(min,max) needs integer args, got %q,%q", a[0], a[1])
-	}
-	if lo > hi {
-		return fmt.Errorf("int(min,max): min %d > max %d", lo, hi)
-	}
-	if uint64(hi)-uint64(lo) >= uint64(math.MaxInt64) { // span hi-lo+1 would overflow int -> IntN panic
-		return fmt.Errorf("int(min,max): range %d..%d is too wide", lo, hi)
-	}
-	return nil
-}
-
-func floatArgs(_ map[string]node, a []string) error {
-	lo, e1 := strconv.ParseFloat(a[0], 64)
-	hi, e2 := strconv.ParseFloat(a[1], 64)
-	dp, e3 := strconv.Atoi(a[2])
-	if e1 != nil || e2 != nil || e3 != nil {
-		return fmt.Errorf("float(min,max,dp) needs numeric args, got %q,%q,%q", a[0], a[1], a[2])
-	}
-	if lo > hi {
-		return fmt.Errorf("float(min,max,dp): min %v > max %v", lo, hi)
-	}
-	if math.IsInf(hi-lo, 0) { // an overflowing span would render as "+Inf"
-		return fmt.Errorf("float(min,max,dp): range %v..%v is too wide", lo, hi)
-	}
-	if dp < 0 || dp > maxDecimals {
-		return fmt.Errorf("float(min,max,dp): decimals %d out of range 0..%d", dp, maxDecimals)
-	}
-	return nil
-}
-
-func seqArg(_ map[string]node, a []string) error {
-	if len(a) > 1 {
-		return fmt.Errorf("seq takes at most one name, got %d args", len(a))
-	}
-	if len(a) == 1 && a[0] == "" {
-		return fmt.Errorf("seq name must not be empty")
-	}
-	return nil
-}
-
-// nanoidAlphabet is the 64-char URL-safe set Nano IDs use.
-const nanoidAlphabet = "_-0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"
-
-// uuidV7 builds an RFC 9562 v7 UUID. The 48-bit timestamp field is drawn from
-// the rng (not the clock) to stay reproducible, then the version (7) and variant
-// (10) bits are forced; the rest is random.
-func uuidV7(r rng) string {
-	b := randBytes(r, 16)
-	b[6] = b[6]&0x0f | 0x70
-	b[8] = b[8]&0x3f | 0x80
-	var sb strings.Builder
-	for i, x := range b {
-		if i == 4 || i == 6 || i == 8 || i == 10 {
-			sb.WriteByte('-')
-		}
-		sb.WriteByte(hexDigits[x>>4])
-		sb.WriteByte(hexDigits[x&0x0f])
-	}
-	return sb.String()
-}
-
-// crockford is the ULID/Crockford base32 alphabet (no I, L, O, U).
-const crockford = "0123456789ABCDEFGHJKMNPQRSTVWXYZ"
-
-// ulid builds a 26-char ULID: 128 random bits (timestamp from the rng) encoded
-// big-endian into base32, the 130-bit stream left-padded with two zero bits.
-func ulid(r rng) string {
-	b := randBytes(r, 16)
-	out := make([]byte, 26)
-	for i := range out {
-		v := 0
-		for j := 0; j < 5; j++ { // 5 bits per char; the two leading pad bits are zero
-			real := 5*i + j - 2
-			bit := 0
-			if real >= 0 {
-				bit = int(b[real/8]>>(7-uint(real%8))) & 1
-			}
-			v = v<<1 | bit
-		}
-		out[i] = crockford[v]
-	}
-	return string(out)
-}
-
-// luhnCheck returns the Luhn check digit (0-9) over the digits of s; non-digit
-// runes are skipped. Doubling runs from the rightmost digit, so the result is
-// correct whatever the payload length.
-func luhnCheck(s string) int {
-	sum, double := 0, true
-	for i := len(s) - 1; i >= 0; i-- {
-		c := s[i]
-		if c < '0' || c > '9' {
-			continue
-		}
-		d := int(c - '0')
-		if double {
-			if d *= 2; d > 9 {
-				d -= 9
-			}
-		}
-		double = !double
-		sum += d
-	}
-	return (10 - sum%10) % 10
-}
-
-// mod11Check returns the weighted mod-11 check character over the digits of s
-// (weights 2..7 cycling from the right). A would-be value of 10 emits 'X', as in
-// ISBN-10 / ISO 7064; non-digits are skipped.
-func mod11Check(s string) string {
-	sum, w := 0, 2
-	for i := len(s) - 1; i >= 0; i-- {
-		c := s[i]
-		if c < '0' || c > '9' {
-			continue
-		}
-		sum += int(c-'0') * w
-		if w++; w > 7 {
-			w = 2
-		}
-	}
-	if chk := (11 - sum%11) % 11; chk != 10 {
-		return string(rune('0' + chk))
-	}
-	return "X"
-}
-
-// eanCheck returns the EAN-13 / UPC-A / ISBN-13 / GTIN check digit over the
-// digits of s: weights 3 and 1 alternating from the rightmost digit, mod 10.
-func eanCheck(s string) string {
-	sum, w := 0, 3
-	for i := len(s) - 1; i >= 0; i-- {
-		c := s[i]
-		if c < '0' || c > '9' {
-			continue
-		}
-		sum += int(c-'0') * w
-		w = 4 - w // 3 <-> 1
-	}
-	return string(rune('0' + (10-sum%10)%10))
-}
-
-// ibanLen maps a supported country code to the full IBAN length. The check digits
-// sit between the country code and the BBAN, so — unlike luhn/ean — iban can't be
-// a left-to-right derivation; it generates the whole value instead.
-var ibanLen = map[string]int{"BE": 16, "DE": 22, "DK": 18, "ES": 24, "FI": 18, "NO": 15, "SE": 24}
-
-func ibanArg(_ map[string]node, a []string) error {
-	if _, ok := ibanLen[a[0]]; !ok {
-		return fmt.Errorf("iban(%q): unsupported country code", a[0])
-	}
-	return nil
-}
-
-// iban generates a structurally valid IBAN for cc: a numeric BBAN of the right
-// length, then mod-97 check digits. Real bank/branch structure isn't modelled —
-// the result passes length and checksum validation, which is what fake data needs.
-func iban(r rng, cc string) string {
-	bban := make([]byte, ibanLen[cc]-4)
-	for i := range bban {
-		bban[i] = byte('0' + r.IntN(10))
-	}
-	rem := 0
-	feed := func(d int) { rem = (rem*10 + d) % 97 }
-	for _, c := range bban {
-		feed(int(c - '0'))
-	}
-	for i := 0; i < len(cc); i++ { // letters A-Z -> 10..35, fed as two digits
-		v := int(cc[i]-'A') + 10
-		feed(v / 10)
-		feed(v % 10)
-	}
-	feed(0)
-	feed(0)
-	return fmt.Sprintf("%s%02d%s", cc, 98-rem, bban)
-}

builtins_test.go

@@ -2,9 +2,14 @@ package fejkdata
 
 import (
 	"encoding/base64"
+	"fmt"
 	"regexp"
 	"strconv"
+	"strings"
 	"testing"
+	"time"
+
+	"github.com/larvit/fejkdata/internal/invariant"
 )
 
 func TestBuiltinIDGenerators(t *testing.T) {
@@ -41,6 +46,7 @@ func TestBuiltinSamplesReproducible(t *testing.T) {
 		`"{uuid()}"`, `"{ulid()}"`,
 		`"{nanoid(12)}"`, `"{int(1,1000000)}"`,
 		`"{float(0,1,6)}"`, `"{base64(12)}"`, `"{iban(SE)}"`,
+		`"{date(2000-01-01,2020-12-31,'2006-01-02 15:04:05')}"`, `"{time('15:04:05')}"`,
 	} {
 		if a, b := mustRender(t, engine(7), tmpl), mustRender(t, engine(7), tmpl); a != b {
 			t.Fatalf("%s not reproducible: %q != %q", tmpl, a, b)
@@ -69,6 +75,9 @@ func TestBuiltinFloat(t *testing.T) {
 		if got := mustRender(t, f, `"{float(1,2,3)}"`); !re.MatchString(got) {
 			t.Fatalf("float(1,2,3) = %q, want d.ddd in [1,2]", got)
 		}
+		if got := mustRender(t, f, `"{float(-1,1,0)}"`); got == "-0" {
+			t.Fatalf("float(-1,1,0) = %q, want a zero printed unsigned", got)
+		}
 	}
 }
 
@@ -95,10 +104,10 @@ func TestBuiltinChecksums(t *testing.T) {
 	}
 }
 
-// TestBuiltinSeqPerSession pins seq's contract: a counter from 1, advancing on
+// TestBuiltinSeqPerGenerator pins seq's contract: a counter from 1, advancing on
 // each call, named counters independent, and the whole thing scoped to one
-// session so a fresh Generator restarts at 1.
-func TestBuiltinSeqPerSession(t *testing.T) {
+// draw state so a fresh Generator restarts at 1.
+func TestBuiltinSeqPerGenerator(t *testing.T) {
 	f := engine(1)
 	for i := 1; i <= 5; i++ {
 		if got := mustRender(t, f, `"{seq()}"`); got != strconv.Itoa(i) {
@@ -115,9 +124,9 @@ func TestBuiltinSeqPerSession(t *testing.T) {
 	if got := mustRender(t, engine(2), `{"format":"{seq()}","repeat":3,"separator":","}`); got != "1,2,3" {
 		t.Fatalf("repeat seq = %q, want 1,2,3", got)
 	}
-	// ...and a fresh session restarts from 1.
+	// ...and a fresh generator restarts from 1.
 	if got := mustRender(t, engine(2), `"{seq()}"`); got != "1" {
-		t.Fatalf("new session seq = %q, want 1", got)
+		t.Fatalf("new generator seq = %q, want 1", got)
 	}
 }
 
@@ -136,7 +145,7 @@ func TestBuiltinArgLimits(t *testing.T) {
 		`{"format":"{x}","x":"1","repeat":2000000000}`, // repeat -> multi-GB join
 	}
 	for _, tmpl := range bad {
-		if _, err := compile(parse(t, tmpl)); err == nil {
+		if _, err := resolved(t, tmpl); err == nil {
 			t.Errorf("compile(%s) = nil error, want a limit rejection", tmpl)
 		}
 	}
@@ -181,52 +190,229 @@ func ibanValid(s string) bool {
 	return rem == 1
 }
 
-// TestRegistryShapes pins the builtin contract compileOps relies on: every entry
-// supplies prep, and args parsed at compile only behind a check.
-func TestRegistryShapes(t *testing.T) {
-	for name, b := range builtins {
-		if b.prep == nil {
-			t.Errorf("builtin %q has no prep: compileOps would call a nil func", name)
-		}
-		if b.arity != 0 && b.check == nil {
-			t.Errorf("builtin %q parses args in prep with no check", name)
-		}
-	}
-}
-
-// TestArgGuardsPanic pins the guards that report a builtin arg its check should
-// have rejected. No data reaches them — checkFunc runs a builtin's check before
-// compileOps ever calls prep — so they are exercised directly.
-func TestArgGuardsPanic(t *testing.T) {
-	for name, call := range map[string]func(){
-		"atoi on an unvalidated arg": func() { atoi("nope") },
-		"atof on an unvalidated arg": func() { atof("nope") },
-	} {
-		mustPanic(t, name, call)
-	}
-}
-
-// mustPanic fails unless call panics, which is what separates a reported
-// invariant break from a silently wrong value.
+// mustPanic fails unless call panics with invariant.Broken's phrase, which is what
+// separates a reported invariant break from a silently wrong value.
 func mustPanic(t *testing.T, name string, call func()) {
 	t.Helper()
 	defer func() {
-		if recover() == nil {
-			t.Errorf("%s: no panic, want the invariant reported", name)
+		if r := recover(); r == nil || !strings.HasPrefix(fmt.Sprint(r), invariant.Broken("")) {
+			t.Errorf("%s: recovered %v, want the invariant reported as an internal error", name, r)
 		}
 	}()
 	call()
 }
 
 func TestClassBuiltinArgs(t *testing.T) {
-	for _, bad := range []string{`"{digits(0)}"`, `"{digits(2000000000)}"`, `"{upper(-1)}"`, `"{lower(x)}"`, `"{digits()}"`, `"{upper(1,2)}"`} {
-		if _, err := compile(parse(t, bad)); err == nil {
+	for _, bad := range []string{`"{digits(2000000000)}"`, `"{upper(-1)}"`, `"{lower(x)}"`, `"{digits()}"`, `"{upper(1,2)}"`} {
+		if _, err := resolved(t, bad); err == nil {
 			t.Errorf("compile(%s) = nil error, want the arg rejected", bad)
 		}
 	}
 	for _, ok := range []string{`"{digits(1048576)}"`, `"{upper(1)}"`, `"{lower(26)}"`} {
-		if _, err := compile(parse(t, ok)); err != nil {
+		if _, err := resolved(t, ok); err != nil {
 			t.Errorf("compile(%s) = %v", ok, err)
 		}
 	}
 }
+
+// TestBuiltinDateAndTime pins the two clock-free samples: date draws a second in
+// [from 00:00:00, to 23:59:59], both days reachable, and renders it in the quoted Go
+// layout, commas and English names included; time draws a second within one day.
+func TestBuiltinDateAndTime(t *testing.T) {
+	f := engine(1)
+	seen := map[string]bool{}
+	for i := 0; i < 500; i++ {
+		got := mustRender(t, f, `"{date(1990-01-01,1990-12-31,'2006-01-02')}"`)
+		if d, err := time.Parse("2006-01-02", got); err != nil || d.Year() != 1990 {
+			t.Fatalf("date = %q, want a 1990 calendar date (err %v)", got, err)
+		}
+		seen[got] = true
+	}
+	if len(seen) < 200 {
+		t.Fatalf("date drew %d distinct days of 365 in 500, want a uniform spread", len(seen))
+	}
+	lo, hi := false, false
+	for i := 0; i < 200; i++ {
+		got := mustRender(t, f, `"{date(2020-02-28,2020-02-29,'2006-01-02')}"`)
+		if got != "2020-02-28" && got != "2020-02-29" {
+			t.Fatalf("date(2020-02-28,2020-02-29) = %q, out of range", got)
+		}
+		lo, hi = lo || got == "2020-02-28", hi || got == "2020-02-29"
+	}
+	if !lo || !hi {
+		t.Fatalf("date never hit a bound: lo=%v hi=%v (bounds must be inclusive)", lo, hi)
+	}
+	if got := mustRender(t, f, `"{date(2020-07-04,2020-07-05,'January 2, 2006')}"`); got != "July 4, 2020" && got != "July 5, 2020" {
+		t.Fatalf("date with a comma in its layout = %q", got)
+	}
+	rfc := regexp.MustCompile(`^2021-\d\d-\d\dT\d\d:\d\d:\d\dZ$`)
+	clock := regexp.MustCompile(`^([01]\d|2[0-3]):[0-5]\d$`)
+	ampm := regexp.MustCompile(`^(1[0-2]|[1-9]):[0-5]\d (AM|PM)$`)
+	seconds := map[string]bool{}
+	for i := 0; i < 300; i++ {
+		got := mustRender(t, f, `"{date(2021-01-01,2021-12-31,'2006-01-02T15:04:05Z07:00')}"`)
+		if !rfc.MatchString(got) {
+			t.Fatalf("date in an RFC 3339 layout = %q, want %s", got, rfc)
+		}
+		seconds[got[17:19]] = true
+		if got := mustRender(t, f, `"{time('15:04')}"`); !clock.MatchString(got) {
+			t.Fatalf("time('15:04') = %q, want %s", got, clock)
+		}
+		if got := mustRender(t, f, `"{time('3:04 PM')}"`); !ampm.MatchString(got) {
+			t.Fatalf("time('3:04 PM') = %q, want %s", got, ampm)
+		}
+		got = mustRender(t, f, `"{date(1950-01-01,2000-12-31,'060102')}-238{luhn()}"`)
+		if d := digitsOnly(got); len(d) != 10 || !luhnValid(d) {
+			t.Fatalf("a personnummer over date() = %q, want ten Luhn-valid digits", got)
+		}
+	}
+	if len(seconds) < 30 {
+		t.Fatalf("date drew %d distinct seconds in 300, want the whole day, not midnight", len(seconds))
+	}
+}
+
+// TestBuiltinDateArgs pins the New-time checks: bounds are calendar dates in order, and the
+// layout is quoted.
+func TestBuiltinDateArgs(t *testing.T) {
+	for tmpl, want := range map[string]string{
+		`"{date(1990-13-01,1990-12-31,'2006-01-02')}"`:        "1990-13-01",
+		`"{time('2006-01-02 15:04')}"`:                        "write date(from,to,layout)",
+		`"{date(1990-12-31,1990-01-01,'2006-01-02')}"`:        "is after",
+		`"{date(1990-01-01,1990-12-31,2006-01-02)}"`:          "'2006-01-02'",
+		`"{date(1990-01-01,1990-12-31,'January 2, 2006)}"`:    "'",
+		`"{date(1990-01-01,1990-12-31,\"January 2, 2006\")}"`: "quoted: 'January 2, 2006'",
+		`"{date(1990-01-01,1990-12-31)}"`:                     "3 arguments",
+		`"{date(1990-01-01,1990-12-31,January 2, 2006)}"`:     "'January 2, 2006'",
+		`"{time(3:04 PM, Mon)}"`:                              "'3:04 PM, Mon'",
+		`"{time(15:04)}"`:                                     "'15:04'",
+	} {
+		_, err := resolved(t, tmpl)
+		if err == nil || !strings.Contains(err.Error(), want) {
+			t.Errorf("compile(%s) = %v, want an error mentioning %q", tmpl, err, want)
+		}
+	}
+}
+
+// TestBuiltinLayoutErrorsNameARunnableSpelling pins the two layout errors a user
+// can follow: a double-quoted layout is named single-quoted, without its own
+// quotes carried into the suggestion, whether it split on a comma or not.
+func TestBuiltinLayoutErrorsNameARunnableSpelling(t *testing.T) {
+	_, err := compile(parse(t, `"{date(1990-01-01,1990-12-31,\"2006-01-02\")}"`))
+	if err == nil || !strings.Contains(err.Error(), "write '2006-01-02'") {
+		t.Fatalf("a double-quoted layout = %v, want it named single-quoted", err)
+	}
+	if strings.Contains(err.Error(), `'"`) {
+		t.Errorf("%v names a layout that renders its own quotes", err)
+	}
+	if !strings.Contains(err.Error(), "is double-quoted") {
+		t.Errorf("%v does not say which quotes were wrong", err)
+	}
+	_, err = compile(parse(t, `"{time(15:04)}"`))
+	if err == nil || !strings.Contains(err.Error(), "write '15:04'") {
+		t.Fatalf("a bare layout = %v, want it named quoted", err)
+	}
+	// The comma hint belongs to a layout that split, not to a call given extra args.
+	_, err = compile(parse(t, `"{time(0,12,'15:04')}"`))
+	if err == nil || !strings.Contains(err.Error(), "takes 1 argument, got 3") {
+		t.Fatalf("time with three args = %v, want the count named in the singular", err)
+	}
+	if strings.Contains(err.Error(), "holding a comma") {
+		t.Errorf("%v offers the comma hint though the layout is already quoted", err)
+	}
+}
+
+// TestBuiltinDateSpansOneDay pins that from == to is a day: with a clock layout it
+// draws every second of it, and without one it prints that day.
+func TestBuiltinDateSpansOneDay(t *testing.T) {
+	f := engine(1)
+	seen := map[string]bool{}
+	for i := 0; i < 500; i++ {
+		got := mustRender(t, f, `"{date(2026-01-01,2026-01-01,'2006-01-02 15:04:05')}"`)
+		if !strings.HasPrefix(got, "2026-01-01 ") {
+			t.Fatalf("date over one day = %q, out of range", got)
+		}
+		seen[got] = true
+	}
+	if len(seen) < 400 {
+		t.Fatalf("date over one day drew %d distinct seconds in 500, want the whole day", len(seen))
+	}
+	if got := mustRender(t, f, `"{date(2026-01-01,2026-01-01,'2006-01-02')}"`); got != "2026-01-01" {
+		t.Fatalf("one day in a date-only layout = %q, want that day", got)
+	}
+}
+
+func TestClassBuiltins(t *testing.T) {
+	cases := map[string]*regexp.Regexp{
+		`"{digits(1)}"`:          regexp.MustCompile(`^[0-9]$`),
+		`"{digits(3)}"`:          regexp.MustCompile(`^[0-9]{3}$`),
+		`"{int(1,9)}"`:           regexp.MustCompile(`^[1-9]$`),
+		`"{upper(1)}"`:           regexp.MustCompile(`^[A-Z]$`),
+		`"{lower(1)}"`:           regexp.MustCompile(`^[a-z]$`),
+		`"{upper(2)}{lower(2)}"`: regexp.MustCompile(`^[A-Z]{2}[a-z]{2}$`),
+	}
+	f := engine(7)
+	for tmpl, re := range cases {
+		for i := 0; i < 100; i++ {
+			if got := mustRender(t, f, tmpl); !re.MatchString(got) {
+				t.Fatalf("%s produced %q, want %s", tmpl, got, re)
+			}
+		}
+	}
+}
+
+func TestFunctionTokenLuhn(t *testing.T) {
+	// {luhn()} appends a Luhn check digit over the digits emitted so far in the
+	// current expansion (non-digits skipped but kept). It reads the output
+	// buffer, so a value is never re-rendered. Bodies are escaped to fix input.
+	f := engine(1)
+	cases := map[string]string{
+		`"811218987{luhn()}"`:                      "8112189876",  // personnummer body
+		`"7992739871{luhn()}"`:                     "79927398713", // classic Luhn vector
+		`"811218-987{luhn()}"`:                     "811218-9876", // '-' skipped, kept
+		`{"format":"{n}{luhn()}","n":"811218987"}`: "8112189876",  // over a rendered token
+	}
+	for tmpl, want := range cases {
+		if got := mustRender(t, f, tmpl); got != want {
+			t.Fatalf("%s = %q, want %q", tmpl, got, want)
+		}
+	}
+}
+
+func TestBuiltinCompileErrors(t *testing.T) {
+	for _, bad := range []string{
+		`"{upper(x)}"`,         // count must be an integer
+		`"{int(a,b)}"`,         // non-integer args
+		`"{int(5,1)}"`,         // min > max
+		`"{nanoid(-1)}"`,       // negative count
+		`"{float(1,2,-1)}"`,    // negative decimals
+		`"{float(NaN,NaN,2)}"`, // bounds must be finite
+		`"{float(Inf,Inf,2)}"`, // same-sign infinities
+		`"{float(1,NaN,2)}"`,   // one NaN bound
+		`"{float(-Inf,1,2)}"`,  // one infinite bound
+		`"{iban(US)}"`,         // unsupported country
+		`"{seq(a,b)}"`,         // seq takes at most one name
+	} {
+		if _, err := resolved(t, bad); err == nil {
+			t.Errorf("compile(%s) = nil error, want error", bad)
+		}
+	}
+}
+
+func TestArgErrorsNameTheSpelling(t *testing.T) {
+	if _, err := resolved(t, `"{hex(99999999999999999999)}"`); err == nil || !strings.Contains(err.Error(), "exceeds the maximum") {
+		t.Errorf("compile(hex past the range) = %v, want the maximum named", err)
+	}
+}
+
+func TestASignOrLeadingZeroIsTheInteger(t *testing.T) {
+	for long, short := range map[string]string{
+		`"{float(1,2,02)}"`:                 `"{float(1,2,2)}"`,
+		`{"format":"{calc(a,02)}","a":"1"}`: `{"format":"{calc(a,2)}","a":"1"}`,
+		`"{digits(+5)}"`:                    `"{digits(5)}"`,
+		`"{digits(05)}"`:                    `"{digits(5)}"`,
+		`"{int(007,9)}"`:                    `"{int(7,9)}"`,
+		`"{int(+1,5)}"`:                     `"{int(1,5)}"`,
+	} {
+		sameRenders(t, long, short)
+	}
+}

calc.go

@@ -2,316 +2,149 @@ package fejkdata
 
 import (
 	"fmt"
-	"math"
+	"slices"
 	"strconv"
 	"strings"
-	"unicode"
+
+	"github.com/larvit/fejkdata/internal/builtinfunc"
+	"github.com/larvit/fejkdata/internal/grammar"
+	"github.com/larvit/fejkdata/internal/invariant"
 )
 
-// calcNode is a parsed expression node. It evaluates over the operand values expand
-// read, so the evaluator touches neither the rng nor the node tree.
-type calcNode interface {
-	eval(operands []string) float64
-}
-
-type calcNum float64 // a number literal
-type calcVar string  // a sibling-field name, before indexVars places it
-type calcIdx int     // an operand, by its position in the values expand read
-type calcNeg struct{ x calcNode }
-type calcBin struct { // a + - * / b
-	op   byte
-	l, r calcNode
-}
-
-func (n calcNum) eval([]string) float64 { return float64(n) }
-
-func (n calcIdx) eval(operands []string) float64 {
-	v, err := strconv.ParseFloat(strings.TrimSpace(operands[n]), 64)
-	if err != nil {
-		return math.NaN() // a non-numeric operand stays visible, never an error
-	}
-	return v
-}
-
-// eval on an unplaced name cannot happen: calcPrep runs indexVars over every
-// expression it compiles, so only a calcIdx reaches a render. It panics rather
-// than returning NaN, so a node kind indexVars forgets is a stack trace and not a
-// silently wrong number.
-func (n calcVar) eval([]string) float64 {
-	panic(fmt.Sprintf("fejkdata: calc operand %q was never placed", string(n)))
-}
-
-func (n calcNeg) eval(operands []string) float64 { return -n.x.eval(operands) }
-
-func (n calcBin) eval(operands []string) float64 {
-	l, r := n.l.eval(operands), n.r.eval(operands)
-	switch n.op {
-	case '+':
-		return l + r
-	case '-':
-		return l - r
-	case '*':
-		return l * r
-	default: // '/'
-		return l / r
-	}
-}
-
-// checkCalc validates a calc token at compile time: a parseable expression whose
-// operands all name existing fields, and an optional non-negative integer dp. It
-// is a builtin check (fields first), so calc dispatches through the registry like
-// every other {name(args)} function.
-func checkCalc(fields map[string]node, args []string) error {
-	if len(args) < 1 || len(args) > 2 {
-		return fmt.Errorf("calc takes an expression and an optional decimals count, got %d args", len(args))
-	}
-	expr, err := parseCalc(args[0])
-	if err != nil {
-		return fmt.Errorf("calc(%q): %w", args[0], err)
-	}
-	for _, name := range calcVars(expr) {
-		operand, ok := fields[name]
-		if !ok {
-			return fmt.Errorf("calc(%q): no field %q", args[0], name)
+// checkCalcFields runs checkOperands over each calc operand naming a field; an operand naming no
+// field is a name, which checkCalcNames checks once names resolve.
+func checkCalcFields(args []string, fields map[string]node) error {
+	return checkOperands(args[0], builtinfunc.ParsedCalc(args[0]), func(name string) []node {
+		if n, ok := fields[name]; ok {
+			return []node{n}
 		}
-		if text, never := neverNumeric(operand); never {
-			return fmt.Errorf("calc(%q): operand %q is never a number: it renders %q", args[0], name, text)
+		return nil
+	})
+}
+
+// checkCalcNames holds each calc of t reading a name to the checks checkCalcFields makes of a field.
+func checkCalcNames(label string, t *template) error {
+	for _, o := range t.compiled.ops {
+		if o.Fn != builtinfunc.CalcName || !slices.ContainsFunc(o.operands, func(a arm) bool { return a.kind == namedRead }) {
+			continue
 		}
-	}
-	if len(args) == 2 {
-		if dp, err := strconv.Atoi(args[1]); err != nil || dp < 0 || dp > maxDecimals {
-			return fmt.Errorf("calc decimals %q must be an integer in 0..%d", args[1], maxDecimals)
+		if err := checkOperands(o.Args[0], builtinfunc.ParsedCalc(o.Args[0]), operandNodes(o)); err != nil {
+			return fmt.Errorf("%s: token {%s}: %w", label, o.Body, err)
 		}
 	}
 	return nil
 }
 
-// neverNumeric reports a node no render of which is a number: fixed text that does
-// not parse, or a choice of only such items. text is one such render.
-func neverNumeric(n node) (text string, never bool) {
+// operandNodes lists every node each operand of o may render, once its template and names resolve.
+func operandNodes(o op) func(name string) []node {
+	return func(name string) []node {
+		a := o.operands[slices.IndexFunc(o.operands, func(a arm) bool { return a.head == name })]
+		if len(a.leaves) == 0 {
+			panic(invariant.Broken("calc operand %q was read before it was compiled", name))
+		}
+		return a.leaves
+	}
+}
+
+// checkOperands refuses an operand that is never a number and a division by a constant zero.
+// operand lists every node an operand may render, nil while that is unknown.
+func checkOperands(text string, c grammar.Calc, operand func(name string) []node) error {
+	for _, name := range c.Operands {
+		if rendered, never := allNeverNumeric(operand(name)); never {
+			return fmt.Errorf("calc(%q): operand %q is never a number: it renders %q", text, name, rendered)
+		}
+	}
+	if divisor, zero := constantZeroDivisor(c.Expr, operand); zero {
+		return fmt.Errorf("calc(%q) divides by %s, which is always zero", text, divisor)
+	}
+	return nil
+}
+
+// constantZeroDivisor finds a division whose right side is a constant zero: number
+// literals and fixed operands folded, anything that varies left unknown.
+func constantZeroDivisor(n grammar.CalcNode, operand func(string) []node) (string, bool) {
 	switch n := n.(type) {
-	case *template:
-		if !n.fixed || n.repeat > 1 {
-			return "", false
-		}
-		if _, err := strconv.ParseFloat(strings.TrimSpace(n.lit), 64); err != nil {
-			return n.lit, true
-		}
-	case *choice:
-		for _, it := range n.items {
-			t, itemNever := neverNumeric(it)
-			if !itemNever {
-				return "", false
+	case grammar.CalcNeg:
+		return constantZeroDivisor(n.X, operand)
+	case grammar.CalcBin:
+		if n.Operator == '/' {
+			if v, known := constantValue(n.R, operand); known && v == 0 {
+				return grammar.CalcText(n.R), true
 			}
-			text = t
 		}
-		return text, true
+		if d, zero := constantZeroDivisor(n.L, operand); zero {
+			return d, true
+		}
+		return constantZeroDivisor(n.R, operand)
 	}
 	return "", false
 }
 
-// calcPrep parses the expression and decimals once, at compile time, and places each
-// operand name at the position expand will read it into. checkCalc proved both args
-// valid, so no step here can fail; dp -1 prints the minimal form.
-func calcPrep(args []string) callFn {
-	expr, err := parseCalc(args[0])
-	if err != nil { // a nil AST would be a nil dereference per render, with no message
-		panic(fmt.Sprintf("fejkdata: calc(%q) reached prep unparsed: %v", args[0], err))
-	}
-	at := make(map[string]int)
-	for i, name := range calcVars(expr) {
-		at[name] = i
-	}
-	placed := indexVars(expr, at)
-	dp := -1
-	if len(args) == 2 {
-		dp = atoi(args[1])
-	}
-	return func(_ *session, _ string, operands []string) string {
-		return strconv.FormatFloat(placed.eval(operands), 'f', dp, 64)
-	}
-}
-
-// indexVars replaces each operand name with its position in the values expand reads.
-// Both sides take that order from calcVars, so they cannot drift.
-func indexVars(n calcNode, at map[string]int) calcNode {
+// constantValue evaluates an expression whose every operand is fixed.
+func constantValue(n grammar.CalcNode, operand func(string) []node) (float64, bool) {
 	switch n := n.(type) {
-	case calcVar:
-		i, placed := at[string(n)]
-		if !placed { // calcVars named every operand, so a miss means the two disagree
-			panic(fmt.Sprintf("fejkdata: calc operand %q is not among the names read for it", string(n)))
+	case grammar.CalcNum:
+		return float64(n), true
+	case grammar.CalcVar:
+		nodes := operand(n.Name)
+		if len(nodes) != 1 {
+			return 0, false
 		}
-		return calcIdx(i)
-	case calcNeg:
-		return calcNeg{indexVars(n.x, at)}
-	case calcBin:
-		return calcBin{n.op, indexVars(n.l, at), indexVars(n.r, at)}
-	}
-	return n
-}
-
-// calcOperands lists the sibling-field names a calc's args read. checkCalc reports
-// an expression that does not parse, so one that does not simply names nothing.
-func calcOperands(args []string) []string {
-	if len(args) == 0 {
-		return nil
-	}
-	expr, err := parseCalc(args[0])
-	if err != nil {
-		return nil
-	}
-	return calcVars(expr)
-}
-
-// calcVars lists the distinct field names an expression reads, in the order it first
-// names each. That order is the contract between expand, which reads the operands
-// into a slice, and indexVars, which places each name at its position in it.
-func calcVars(n calcNode) []string {
-	var out []string
-	seen := map[string]bool{}
-	var walk func(calcNode)
-	walk = func(n calcNode) {
-		switch n := n.(type) {
-		case calcVar:
-			if name := string(n); !seen[name] {
-				seen[name] = true
-				out = append(out, name)
-			}
-		case calcNeg:
-			walk(n.x)
-		case calcBin:
-			walk(n.l)
-			walk(n.r)
+		t, ok := nodes[0].(*template)
+		if !ok || t.repeat != 1 {
+			return 0, false
 		}
-	}
-	walk(n)
-	return out
-}
-
-// calcParser is a recursive-descent parser over the expression runes, threading
-// expr -> term -> factor for the standard * / before + - precedence.
-type calcParser struct {
-	rs  []rune
-	pos int
-}
-
-// parseCalc parses a whole expression, requiring it to consume all input.
-func parseCalc(expr string) (calcNode, error) {
-	p := &calcParser{rs: []rune(expr)}
-	if p.space(); p.pos >= len(p.rs) {
-		return nil, fmt.Errorf("empty expression")
-	}
-	n, err := p.expr()
-	if err != nil {
-		return nil, err
-	}
-	if p.space(); p.pos != len(p.rs) {
-		return nil, fmt.Errorf("unexpected %q", string(p.rs[p.pos:]))
-	}
-	return n, nil
-}
-
-func (p *calcParser) space() {
-	for p.pos < len(p.rs) && unicode.IsSpace(p.rs[p.pos]) {
-		p.pos++
-	}
-}
-
-func (p *calcParser) expr() (calcNode, error) { return p.binary(p.term, '+', '-') }
-func (p *calcParser) term() (calcNode, error) { return p.binary(p.factor, '*', '/') }
-
-// binary parses a left-associative run of next() operands joined by the given
-// operators, the one shape expr and term share.
-func (p *calcParser) binary(next func() (calcNode, error), ops ...byte) (calcNode, error) {
-	n, err := next()
-	if err != nil {
-		return nil, err
-	}
-	for {
-		p.space()
-		if p.pos >= len(p.rs) || !contains(ops, byte(p.rs[p.pos])) {
-			return n, nil
+		lit, fixed := t.fixedText()
+		if !fixed {
+			return 0, false
 		}
-		op := byte(p.rs[p.pos])
-		p.pos++
-		r, err := next()
-		if err != nil {
-			return nil, err
+		v, err := strconv.ParseFloat(strings.TrimSpace(lit), 64)
+		return v, err == nil
+	case grammar.CalcNeg:
+		v, ok := constantValue(n.X, operand)
+		return -v, ok
+	case grammar.CalcBin:
+		l, lok := constantValue(n.L, operand)
+		r, rok := constantValue(n.R, operand)
+		if !lok || !rok {
+			return 0, false
 		}
-		n = calcBin{op, n, r}
+		return builtinfunc.Arith(n.Operator, l, r), true
 	}
+	return 0, false
 }
 
-func (p *calcParser) factor() (calcNode, error) {
-	p.space()
-	if p.pos >= len(p.rs) {
-		return nil, fmt.Errorf("unexpected end of expression")
-	}
-	switch c := p.rs[p.pos]; {
-	case c == '-':
-		p.pos++
-		x, err := p.factor()
-		if err != nil {
-			return nil, err
-		}
-		return calcNeg{x}, nil
-	case c == '(':
-		p.pos++
-		n, err := p.expr()
-		if err != nil {
-			return nil, err
-		}
-		if p.space(); p.pos >= len(p.rs) || p.rs[p.pos] != ')' {
-			return nil, fmt.Errorf("missing ')'")
-		}
-		p.pos++
-		return n, nil
-	case c == '.' || c >= '0' && c <= '9':
-		return p.number()
-	case c == '_' || unicode.IsLetter(c):
-		return p.ident()
-	default:
-		return nil, fmt.Errorf("unexpected %q", string(c))
+// neverNumeric reports a node no render of which is a number: a null, a repeat of 0, fixed
+// text that does not parse, or a choice of only such items. text is one such render.
+func neverNumeric(n node) (text string, never bool) {
+	switch n := n.(type) {
+	case *nullItem:
+		return "", true
+	case *template:
+		lit, fixed := n.fixedText()
+		if n.repeat == 0 {
+			return "", true
+		}
+		if !fixed || n.repeat != 1 {
+			return "", false
+		}
+		if _, err := strconv.ParseFloat(strings.TrimSpace(lit), 64); err != nil {
+			return lit, true
+		}
+	case *choice:
+		return allNeverNumeric(n.items)
 	}
+	return "", false
 }
 
-func (p *calcParser) number() (calcNode, error) {
-	start, dot := p.pos, false
-	for p.pos < len(p.rs) {
-		if c := p.rs[p.pos]; c >= '0' && c <= '9' {
-			p.pos++
-		} else if c == '.' && !dot {
-			dot, p.pos = true, p.pos+1
-		} else {
-			break
+// allNeverNumeric reports whether no render of any of nodes is a number; false for no nodes.
+func allNeverNumeric(nodes []node) (text string, never bool) {
+	for _, n := range nodes {
+		t, nodeNever := neverNumeric(n)
+		if !nodeNever {
+			return "", false
 		}
+		text = t
 	}
-	v, err := strconv.ParseFloat(string(p.rs[start:p.pos]), 64)
-	if err != nil {
-		return nil, fmt.Errorf("bad number %q", string(p.rs[start:p.pos]))
-	}
-	return calcNum(v), nil
-}
-
-// ident reads a field name: a letter or '_', then letters, digits or '_'. A '-'
-// is always the minus operator, so a hyphenated field name can't be an operand.
-func (p *calcParser) ident() (calcNode, error) {
-	start := p.pos
-	for p.pos < len(p.rs) {
-		if c := p.rs[p.pos]; c == '_' || unicode.IsLetter(c) || unicode.IsDigit(c) {
-			p.pos++
-		} else {
-			break
-		}
-	}
-	return calcVar(string(p.rs[start:p.pos])), nil
-}
-
-func contains(bs []byte, b byte) bool {
-	for _, x := range bs {
-		if x == b {
-			return true
-		}
-	}
-	return false
+	return text, len(nodes) > 0
 }

calc_test.go

@@ -3,8 +3,11 @@ package fejkdata
 import (
 	"fmt"
 	"math"
+	"strconv"
 	"strings"
 	"testing"
+
+	"github.com/larvit/fejkdata/internal/grammar"
 )
 
 // TestCalcArithmetic pins the operators, precedence, parentheses and unary minus
@@ -36,6 +39,7 @@ func TestCalcAuto(t *testing.T) {
 		`"{calc(10 / 3)}"`: "3.3333333333333335",
 		`"{calc(6 / 2)}"`:  "3",
 		`"{calc(1 / 4)}"`:  "0.25",
+		`"{calc(0 * -1)}"`: "0",
 	}
 	for tmpl, want := range cases {
 		if got := mustRender(t, f, tmpl); got != want {
@@ -52,6 +56,7 @@ func TestCalcDecimals(t *testing.T) {
 		`"{calc(10 / 3, 2)}"`: "3.33",
 		`"{calc(10 / 3, 0)}"`: "3",
 		`"{calc(2 * 3, 2)}"`:  "6.00",
+		`"{calc(-0.001, 2)}"`: "0.00",
 	}
 	for tmpl, want := range cases {
 		if got := mustRender(t, f, tmpl); got != want {
@@ -70,6 +75,9 @@ func TestCalcFields(t *testing.T) {
 	if got := mustRender(t, engine(1), `{"format":"{calc(a - b)}","a":{"format":"{n}","n":"10"},"b":"3"}`); got != "7" {
 		t.Fatalf("calc(a - b) = %q, want 7", got)
 	}
+	if got := mustRender(t, engine(1), `{"format":"{calc(b * (a - b))}","a":"10","b":"3"}`); got != "21" {
+		t.Fatalf("calc(b * (a - b)) = %q, want 21", got)
+	}
 }
 
 // TestCalcNonNumericIsNaN pins the never-fail rule: a field that sometimes does
@@ -92,43 +100,26 @@ func TestCalcReproducible(t *testing.T) {
 	}
 }
 
-// TestTokenOperandsReadsOnlyAnOperandBuiltin pins what the helper answers for a body that
-// is not a calc, and for one whose expression does not parse: nothing, either way.
-// checkCalc is what reports a bad expression, so the callers that run after it
-// never meet one — but they must not have to depend on that order to be safe.
-func TestTokenOperandsReadsOnlyAnOperandBuiltin(t *testing.T) {
-	for _, body := range []string{"plain", "luhn()", "calc()", "calc(1 +)", "calc(()"} {
-		if got := tokenOperands(body); got != nil {
-			t.Errorf("tokenOperands(%q) = %v, want none", body, got)
+// TestParsedCallReadsOnlyAnOperandBuiltin pins the operands a parsed call carries for a
+// call that is not a calc, and for one whose expression does not parse: none, either way.
+func TestParsedCallReadsOnlyAnOperandBuiltin(t *testing.T) {
+	for _, format := range []string{"{luhn()}", "{calc()}", "{calc(1 +)}", "{calc(()}"} {
+		if toks, err := grammar.ParseFormat(format); err != nil || len(toks) != 1 || tokenReads(toks[0]) != nil {
+			t.Errorf("ParseFormat(%q) = %+v, %v, want one call reading no operand", format, toks, err)
 		}
 	}
-	if got := tokenOperands("calc(net * qty)"); len(got) != 2 {
-		t.Errorf("tokenOperands(calc(net * qty)) = %v, want both operands", got)
-	}
-}
-
-// TestCalcGuardsPanic pins the guards on calc's own invariants: checkCalc parsed
-// the expression before prep sees it, and indexVars places every name calcVars
-// read, so a break in either is reported rather than computed around.
-func TestCalcGuardsPanic(t *testing.T) {
-	for name, call := range map[string]func(){
-		"prep on an expression that does not parse": func() { calcPrep([]string{"1 +"}) },
-		"an operand name never placed":              func() { calcVar("n").eval(nil) },
-		"indexVars on a name it did not read":       func() { indexVars(calcVar("n"), map[string]int{}) },
-	} {
-		mustPanic(t, name, call)
+	if toks, err := grammar.ParseFormat("{calc(net * qty)}"); err != nil || len(toks) != 1 || len(tokenReads(toks[0])) != 2 {
+		t.Errorf("ParseFormat({calc(net * qty)}) = %+v, %v, want both operands", toks, err)
 	}
 }
 
 // --- one draw, one value: a calc operand reads the expansion's draw ---
 
-// TestCalcOperandReadsTheExpansionsDraw pins the one-draw rule for a calc operand.
-// A field the format renders and a calc reads is drawn once per expansion, so the
-// operand shown is the operand computed — the correlation the dotted-path rule
-// already gives a level (see bound_test.go), applied to a plain sibling.
-func TestCalcOperandReadsTheExpansionsDraw(t *testing.T) {
+// TestANamedCalcOperandIsTheOneShown pins how a calc computes from the value its format
+// shows: the format and the calc read one name.
+func TestANamedCalcOperandIsTheOneShown(t *testing.T) {
 	dir := writeData(t, map[string]string{
-		"inv": `{"format":"{net} x {qty} = {calc(net * qty, 2)}","net":["19.99","5.00","100.00"],"qty":["2","3","7"]}`,
+		"inv": `{"format":"{net as n}{qty as q}{n} x {q} = {calc(n * q, 2)}","net":["19.99","5.00","100.00"],"qty":["2","3","7"]}`,
 	})
 	f := newGenerator(t, dir, WithSeed(3))
 	for i := 0; i < 300; i++ {
@@ -143,39 +134,23 @@ func TestCalcOperandReadsTheExpansionsDraw(t *testing.T) {
 	}
 }
 
-// TestCalcOperandSharesOneDraw pins the reach of that hold: the draw belongs to the
-// expansion, not to the calc, so the bare token rendering the same name reads it too.
-func TestCalcOperandSharesOneDraw(t *testing.T) {
-	dir := writeData(t, map[string]string{
-		"same": `{"format":"{w} {calc(w)}","w":["1","2","3","4","5"]}`,
-	})
+// TestACalcOperandDrawsAfresh pins that a calc reading a field draws it as any {…} does.
+func TestACalcOperandDrawsAfresh(t *testing.T) {
+	dir := writeData(t, map[string]string{"same": `{"format":"{w} {calc(w)}","w":["1","2","3","4","5"]}`})
 	f := newGenerator(t, dir, WithSeed(5))
 	for i := 0; i < 200; i++ {
-		got := fake(t, f, "same")
-		if p := strings.Fields(got); len(p) != 2 || p[0] != p[1] {
-			t.Fatalf("same = %q, want one value twice", got)
-		}
-	}
-}
-
-// TestFieldNoCalcReadsDrawsEachTime guards the boundary: only a name a calc reads is
-// held, so an ordinary {w} {w} still draws twice.
-func TestFieldNoCalcReadsDrawsEachTime(t *testing.T) {
-	dir := writeData(t, map[string]string{"two": `{"format":"{w} {w}","w":["1","2","3","4","5"]}`})
-	f := newGenerator(t, dir, WithSeed(5))
-	for i := 0; i < 200; i++ {
-		if p := strings.Fields(fake(t, f, "two")); p[0] != p[1] {
+		if p := strings.Fields(fake(t, f, "same")); p[0] != p[1] {
 			return
 		}
 	}
-	t.Fatal("{w} {w} never differed in 200 draws, want two independent draws")
+	t.Fatal("{w} {calc(w)} never differed in 200 draws, want two independent draws")
 }
 
-// TestCalcHoldIsPerExpansion pins the scope of the hold: each repeat iteration is
-// its own expansion, so it draws again while staying self-consistent.
-func TestCalcHoldIsPerExpansion(t *testing.T) {
+// TestANamedCalcOperandPicksEachIteration pins that a name a repeat binds picks again on each
+// iteration, staying one value within it.
+func TestANamedCalcOperandPicksEachIteration(t *testing.T) {
 	dir := writeData(t, map[string]string{
-		"rep": `{"format":"{n}={calc(n * 1)}","repeat":8,"separator":" ","n":["2","3","4","5","6","7","8","9"]}`,
+		"rep": `{"format":"{n as m}{m}={calc(m * 1)}","repeat":8,"separator":" ","n":["2","3","4","5","6","7","8","9"]}`,
 	})
 	f := newGenerator(t, dir, WithSeed(11))
 	varied := false
@@ -196,12 +171,12 @@ func TestCalcHoldIsPerExpansion(t *testing.T) {
 	}
 }
 
-// TestCalcHoldIsPerTemplate pins that a nested template holds its own: the inner
-// {v} and its calc agree with each other, not with the outer pair.
-func TestCalcHoldIsPerTemplate(t *testing.T) {
+// TestTwoNamesPickApart pins that a nested template's name is a pick of its own: the inner
+// pair agrees with itself, not with the outer pair.
+func TestTwoNamesPickApart(t *testing.T) {
 	dir := writeData(t, map[string]string{
-		"nest": `{"format":"{v}={calc(v * 1)} {inner}","v":["2","3","4","5","6","7","8","9"],
-			"inner":{"format":"{v}={calc(v * 1)}","v":["2","3","4","5","6","7","8","9"]}}`,
+		"nest": `{"format":"{v as a}{a}={calc(a * 1)} {inner}","v":["2","3","4","5","6","7","8","9"],
+			"inner":{"format":"{v as b}{b}={calc(b * 1)}","v":["2","3","4","5","6","7","8","9"]}}`,
 	})
 	f := newGenerator(t, dir, WithSeed(13))
 	differed := false
@@ -229,7 +204,7 @@ func TestCalcOverANeverNumericOperandIsRejected(t *testing.T) {
 		`{"format":"{calc(x * 2)}","x":["a","b"]}`:       `"x"`,
 		`{"format":"{calc(x + y)}","x":"1","y":"{{2}}"}`: `"{2}"`,
 	} {
-		_, err := compile(parse(t, src))
+		_, err := resolved(t, src)
 		if err == nil || !strings.Contains(err.Error(), want) || !strings.Contains(err.Error(), "never a number") {
 			t.Errorf("compile(%s) = %v, want the operand rejected naming %s", src, err, want)
 		}
@@ -240,8 +215,136 @@ func TestCalcOverANeverNumericOperandIsRejected(t *testing.T) {
 		`{"format":"{calc(x * 2)}","x":["1","abc"]}`,
 		`{"format":"{calc(x * 2)}","x":"{digits(2)}"}`,
 	} {
-		if _, err := compile(parse(t, ok)); err != nil {
+		if _, err := resolved(t, ok); err != nil {
 			t.Errorf("compile(%s) = %v, want it accepted", ok, err)
 		}
 	}
 }
+
+func TestCalcConstantZeroDivisorIsRejected(t *testing.T) {
+	for src, want := range map[string]string{
+		`"{calc(1/0)}"`:     "divides by 0",
+		`"{calc(2/(1-1))}"`: "divides by (1 - 1)",
+		`{"format":"{calc(x/y)}","x":"1","y":"0"}`:       "divides by y",
+		`{"format":"{calc(x/(y*2))}","x":"1","y":" 0 "}`: "divides by (y * 2)",
+		`{"format":"{calc((a/0)+b)}","a":"1","b":"2"}`:   "divides by 0",
+		`{"format":"{calc(a/-0)}","a":"1"}`:              "divides by -0",
+	} {
+		if _, err := resolved(t, src); err == nil || !strings.Contains(err.Error(), want) {
+			t.Errorf("compile(%s) = %v, want the constant zero divisor rejected naming %q", src, err, want)
+		}
+	}
+	if _, err := compile(parse(t, `{"format":"{calc(a/(b*c))}","a":"1","b":"0","c":["1","2"]}`)); err != nil {
+		t.Errorf("compile(a/(b*c)) = %v, want a divisor that varies accepted", err)
+	}
+	f := engine(1)
+	for i := 0; i < 50; i++ {
+		if got := mustRender(t, f, `{"format":"{calc(x/y)}","x":"1","y":["0","1"]}`); got == "+Inf" {
+			return
+		}
+	}
+	t.Fatal("a sometimes-zero divisor never printed +Inf in 50 draws")
+}
+
+func TestCalcCompileErrors(t *testing.T) {
+	for _, bad := range []string{
+		`"{calc(1/0)}"`, // a constant zero divisor
+		`{"format":"{calc(x/y)}","x":"1","y":"0"}`, // a fixed zero divisor
+		`"{calc()}"`,        // calc needs an expression
+		`"{calc(1 +)}"`,     // dangling operator
+		`"{calc((1 + 2)}"`,  // unbalanced parenthesis
+		`"{calc(1 2)}"`,     // two operands, no operator
+		`"{calc(price)}"`,   // operand names no field
+		`"{calc(1, 2, 3)}"`, // too many args
+		`"{calc(1, x)}"`,    // decimals arg not an integer
+		`"{calc(1, -1)}"`,   // decimals negative
+	} {
+		if _, err := resolved(t, bad); err == nil {
+			t.Errorf("compile(%s) = nil error, want error", bad)
+		}
+	}
+}
+
+func TestCalcReadsAName(t *testing.T) {
+	dir := writeData(t, map[string]string{
+		"n":    `["2","3","5","7"]`,
+		"card": `"{/n as k}{k} x 2 = {calc(k * 2)}"`,
+	})
+	f := newGenerator(t, dir, WithSeed(1))
+	for i := 0; i < 50; i++ {
+		var k, double int
+		got := fake(t, f, "card")
+		if _, err := fmt.Sscanf(got, "%d x 2 = %d", &k, &double); err != nil || double != k*2 {
+			t.Fatalf("card = %q, want the calc to compute from the pick {k} prints", got)
+		}
+	}
+}
+
+func TestCalcReadsANameItReadsNowhereElse(t *testing.T) {
+	dir := writeData(t, map[string]string{
+		"n":    `["2","3"]`,
+		"card": `"{/n as k}{calc(k * 2)}"`,
+	})
+	f := newGenerator(t, dir, WithSeed(1))
+	if got := fake(t, f, "card"); got != "4" && got != "6" {
+		t.Fatalf("card = %q, want 4 or 6", got)
+	}
+}
+
+func TestCalcRefusesABadOperand(t *testing.T) {
+	for card, want := range map[string]string{
+		`"{/word as w}{w}{calc(w * 2)}"`:    `operand "w" is never a number: it renders "def"`,
+		`"{/zero as z}{z}{calc(1 / z)}"`:    `divides by z, which is always zero`,
+		`"{calc(nope * 2)}"`:                `no field "nope"`,
+		`"{/word as w}{w}{calc(nope * 2)}"`: `no field or name "nope"; the names bound here are "w"`,
+	} {
+		dir := writeData(t, map[string]string{
+			"word": `["abc","def"]`,
+			"zero": `"0"`,
+			"card": card,
+		})
+		if _, err := New(withShipped(), WithDataPath(dir), WithSeed(1)); err == nil || !strings.Contains(err.Error(), want) {
+			t.Errorf("New with card %s = %v, want an error containing %q", card, err, want)
+		}
+	}
+}
+
+func TestATypedColumnRefusesACalcOverNoDigits(t *testing.T) {
+	_, err := New(WithSeed(1), WithDataPath(writeData(t, map[string]string{
+		"row": `{"format":"","x":{"format":"{calc(d + 1)}","d":"{digits(0)}","datatype":"integer"}}`,
+	})))
+	if err == nil || !strings.Contains(err.Error(), "{digits(0)} prints nothing, which is no number") {
+		t.Fatalf("New = %v, want a calc over {digits(0)} refused in a typed column", err)
+	}
+}
+
+func TestATypedColumnProvesACalcOverAName(t *testing.T) {
+	dir := writeData(t, map[string]string{
+		"n":     `"{int(1,9)}"`,
+		"order": `{"format":"{/n as k}{k}","double":{"format":"{calc(k * 2)}","datatype":"integer"}}`,
+	})
+	f := newGenerator(t, dir, WithSeed(1))
+	r, err := f.FakeRecord("order")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if v, err := strconv.Atoi(r.Columns()[0].Value); err != nil || v < 2 || v > 18 || v%2 != 0 {
+		t.Fatalf("double = %q, want an even integer in 2..18", r.Columns()[0].Value)
+	}
+	if !strings.Contains(r.JSON(), `"double":`+r.Columns()[0].Value) {
+		t.Fatalf("JSON = %s, want double written as a number", r.JSON())
+	}
+	bad := writeData(t, map[string]string{
+		"word":  `["1","x"]`,
+		"order": `{"format":"{/word as w}{w}","double":{"format":"{calc(w * 2)}","datatype":"integer"}}`,
+	})
+	if _, err := New(withShipped(), WithDataPath(bad), WithSeed(1)); err == nil || !strings.Contains(err.Error(), `operand "w": "x" is not a number`) {
+		t.Fatalf("New = %v, want the typed column refused over an operand that is not always a number", err)
+	}
+}
+
+func TestARepeatOfZeroIsNeverANumber(t *testing.T) {
+	if _, err := resolved(t, `{"format":"{calc(x / y)}","x":"1","y":{"format":"0","repeat":0}}`); err == nil || !strings.Contains(err.Error(), `operand "y" is never a number: it renders ""`) {
+		t.Errorf("a calc over a repeat of 0 = %v, want the operand refused as never a number", err)
+	}
+}

cli/cli.go

@@ -0,0 +1,706 @@
+// Package cli is the fejkdata command: it renders the template on its stdin with fake
+// values from the modules it is given and any directories layered over them. A program
+// passing its own modules to Main is a fejkdata CLI carrying them, for example
+// os.Exit(cli.Main(append(data.Modules(), mydata)...)).
+package cli
+
+import (
+	"bufio"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"io/fs"
+	"math/rand/v2"
+	"os"
+	"path/filepath"
+	"reflect"
+	"runtime/debug"
+	"sort"
+	"strconv"
+	"strings"
+
+	"github.com/larvit/fejkdata"
+)
+
+const usage = `Usage: <template> | COMMAND [flags]
+       COMMAND --list | --help | --version
+
+COMMAND reads one template from stdin and renders it:
+
+  echo 'id: {int(1,999)}' | COMMAND
+  echo '{"format":"{x}","x":["bosse","lina"]}' | COMMAND
+  COMMAND < template.txt
+
+COMMAND <<'EOF'
+born {date(1990-01-01,2010-12-31,'2006-01-02')}
+EOF
+
+A template is a format string, or a JSON object, array, string or null. Its {…} tokens
+reach the data by reference from the root: {/path}, for any path --list prints, shipped
+or layered with --data-path. A reference names a category or a dotted path into one, a
+table's row by key or name ({/table[KEY]}), and .. steps up from a row to the row it
+links to.
+
+Under --format text, the default, what you send is what renders, and COMMAND adds
+nothing: a format string keeps every byte, so echo's newline ends each render, and
+printf '%s' sends none. -n joins renders with --separator, empty by default. A JSON
+object, array or string drops the whitespace around it, and null alone renders nothing;
+42 or true alone is a format string, printed as written. An empty or blank stdin is a
+format string like any other, and renders as sent. A quoted heredoc, <<'EOF', passes $,
+a backslash or a quote as written.
+
+With --format json, ndjson, csv or sql the template must be a record — a JSON
+template whose fields are its columns, or a lone reference such as {/path}:
+one reference and nothing else, with one newline after it allowed — and the rows are
+written as one JSON array, one JSON object per line, one CSV row (after a header), or
+one INSERT.
+
+  -d, --data-path D      a data directory to layer over the shipped data (repeatable; last wins on a clash)
+      --format F         output form: text (default), json, ndjson, csv or sql
+  -h, --help             print this help, then exit
+      --list             list the paths the data offers, each written after {/ in a template, then exit;
+                         --repeat, --separator, --format and --table do nothing here
+      --no-shipped-data  load only the --data-path directories
+  -n, --repeat N         render the template N times, 0..1048576 (default 1); 0 prints nothing,
+                         [] under --format json, or the header under --format csv
+  -s, --seed N           same seed, version and data: identical output (default: a random seed)
+      --separator S      string between repeated renders (default empty); refused with a record
+                         --format, which has its own separators
+      --table T          the INSERT target for --format sql, ignored under any other
+                         (default: a lone reference's last segment, else records)
+      --version          print the version, then exit
+
+A short flag's value attaches or follows (-n3, -n 3); short flags bundle (-hn 3);
+-- ends the flags.
+
+Exit codes: 0 success, 1 runtime error, 2 misuse.
+`
+
+type invocation struct {
+	args         []string
+	dirs         []string
+	format       string
+	help         bool
+	list         bool
+	noShipped    bool
+	repeat       int
+	seed         uint64
+	seeded       bool
+	separator    string
+	separatorSet bool
+	table        string
+	tableSet     bool
+	version      bool
+}
+
+type flagDef struct {
+	long  string
+	short string
+	value bool
+	set   func(*invocation, string) error
+}
+
+var flagDefs = []flagDef{
+	{"data-path", "d", true, func(in *invocation, v string) error { in.dirs = append(in.dirs, v); return nil }},
+	{"format", "", true, func(in *invocation, v string) error { in.format = v; return nil }},
+	{"help", "h", false, func(in *invocation, _ string) error { in.help = true; return nil }},
+	{"list", "", false, func(in *invocation, _ string) error { in.list = true; return nil }},
+	{"no-shipped-data", "", false, func(in *invocation, _ string) error { in.noShipped = true; return nil }},
+	{"repeat", "n", true, func(in *invocation, v string) error {
+		n, err := strconv.Atoi(v)
+		if err != nil || n < 0 || n > fejkdata.MaxRepeat {
+			return fmt.Errorf("--repeat needs an integer in 0..%d, got %q", fejkdata.MaxRepeat, v)
+		}
+		in.repeat = n
+		return nil
+	}},
+	{"seed", "s", true, func(in *invocation, v string) error {
+		n, err := strconv.ParseUint(v, 10, 64)
+		if err != nil {
+			return fmt.Errorf("--seed needs an unsigned integer, got %q", v)
+		}
+		in.seed, in.seeded = n, true
+		return nil
+	}},
+	{"separator", "", true, func(in *invocation, v string) error { in.separator, in.separatorSet = v, true; return nil }},
+	{"table", "", true, func(in *invocation, v string) error { in.table, in.tableSet = v, true; return nil }},
+	{"version", "", false, func(in *invocation, _ string) error { in.version = true; return nil }},
+}
+
+func flagByLong(name string) *flagDef {
+	for i := range flagDefs {
+		if flagDefs[i].long == name {
+			return &flagDefs[i]
+		}
+	}
+	return nil
+}
+
+func flagByShort(name string) *flagDef {
+	for i := range flagDefs {
+		if flagDefs[i].short == name {
+			return &flagDefs[i]
+		}
+	}
+	return nil
+}
+
+// flagArg is one flag as written: its definition and the value attached to it,
+// if any.
+type flagArg struct {
+	def    *flagDef
+	value  string
+	inline bool
+}
+
+// splitFlags reads one argv element as flags: --name or --name=value, or -abc
+// where each letter is a short flag and the first that takes a value takes the
+// rest of the element.
+func splitFlags(arg string) ([]flagArg, error) {
+	if strings.HasPrefix(arg, "--") {
+		name, value, inline := strings.Cut(arg[2:], "=")
+		def := flagByLong(name)
+		if def == nil {
+			return nil, fmt.Errorf("unknown flag --%s", name)
+		}
+		return []flagArg{{def, value, inline}}, nil
+	}
+	if name, _, _ := strings.Cut(arg[1:], "="); len(name) > 1 && flagByLong(name) != nil {
+		return nil, fmt.Errorf("unknown flag %s; use --%s", arg, name)
+	}
+	var flags []flagArg
+	letters := arg[1:]
+	for i, r := range letters {
+		letter := string(r)
+		def := flagByShort(letter)
+		if def == nil {
+			return nil, fmt.Errorf("unknown flag -%s", letter)
+		}
+		if !def.value {
+			flags = append(flags, flagArg{def: def})
+			continue
+		}
+		rest := letters[i+len(letter):]
+		if strings.HasPrefix(rest, "=") {
+			return nil, fmt.Errorf("-%s takes its value attached (-%s%s) or next (-%s %s); = belongs to --%s=%s",
+				letter, letter, rest[1:], letter, rest[1:], def.long, rest[1:])
+		}
+		return append(flags, flagArg{def, rest, rest != ""}), nil
+	}
+	return flags, nil
+}
+
+func parseArgs(argv []string) (invocation, error) {
+	in := invocation{repeat: 1, format: "text"}
+	for i := 0; i < len(argv); i++ {
+		arg := argv[i]
+		if arg == "--" {
+			in.args = append(in.args, argv[i+1:]...)
+			return in, nil
+		}
+		if len(arg) < 2 || arg[0] != '-' {
+			in.args = append(in.args, arg)
+			continue
+		}
+		flags, err := splitFlags(arg)
+		if err != nil {
+			return in, err
+		}
+		for _, fl := range flags {
+			if !fl.def.value {
+				if fl.inline {
+					return in, fmt.Errorf("--%s takes no value", fl.def.long)
+				}
+				_ = fl.def.set(&in, "")
+				continue
+			}
+			value := fl.value
+			if !fl.inline {
+				if i++; i >= len(argv) {
+					return in, fmt.Errorf("--%s needs a value", fl.def.long)
+				}
+				value = argv[i]
+			}
+			if err := fl.def.set(&in, value); err != nil {
+				return in, err
+			}
+		}
+	}
+	return in, nil
+}
+
+// recordFormat is one way to write a record out: the line each record renders,
+// the header that precedes the first one, and the open/close frame plus the
+// between-record separator a document form needs.
+type recordFormat struct {
+	header func(*fejkdata.Record) string
+	line   func(r *fejkdata.Record, table string) string
+	open   string
+	close  string
+	sep    string
+	writes string // what it writes between values, which --separator would seem to set
+}
+
+// recordFormats is every --format that writes records. json frames the records
+// as one array document; ndjson writes one object per line.
+var recordFormats = map[string]recordFormat{
+	"csv":    {header: (*fejkdata.Record).CSVHeader, line: func(r *fejkdata.Record, _ string) string { return r.CSVLine() }, sep: "\n", writes: `always separates fields with ","`},
+	"json":   {line: jsonLine, open: "[", close: "]", sep: ",\n", writes: "writes the records as one JSON array"},
+	"ndjson": {line: jsonLine, sep: "\n", writes: "writes one record per line"},
+	"sql":    {line: func(r *fejkdata.Record, table string) string { return r.SQLInsert(table) }, sep: "\n", writes: "writes one INSERT per line"},
+}
+
+func jsonLine(r *fejkdata.Record, _ string) string { return r.JSON() }
+
+func (in invocation) writesRecords() bool {
+	return in.format != "text"
+}
+
+// formatNames lists the --format values, text included, for the misuse error.
+func formatNames() string {
+	names := []string{"text"}
+	for name := range recordFormats {
+		names = append(names, name)
+	}
+	sort.Strings(names)
+	return strings.Join(names[:len(names)-1], ", ") + " or " + names[len(names)-1]
+}
+
+// checkFlags rejects a flag value or combination that cannot run.
+func (in invocation) checkFlags() error {
+	if _, ok := recordFormats[in.format]; !ok && in.format != "text" {
+		return fmt.Errorf("--format takes %s, got %q", formatNames(), in.format)
+	}
+	if in.tableSet && in.table == "" && in.format == "sql" {
+		return errors.New("--table names the INSERT target, so it cannot be empty")
+	}
+	if f, records := recordFormats[in.format]; records && in.separatorSet && !in.list {
+		return fmt.Errorf("--separator joins text renders; --format %s %s", in.format, f.writes)
+	}
+	return nil
+}
+
+func pipeHint(name string) string { return "echo 'id: {int(1,999)}' | " + name }
+
+// operandError refuses an argument that is no flag, naming the spelling that reads what
+// it most likely means: dropping it for -, a file for a file's name, else a template
+// piped in.
+func operandError(arg, name string) error {
+	unexpected := "unexpected argument '" + arg + "'"
+	if arg == "-" {
+		return fmt.Errorf("%s; stdin is read already, so drop it", unexpected)
+	}
+	if fi, err := os.Stat(arg); err == nil && fi.Mode().IsRegular() {
+		return fmt.Errorf("%s; read a file from stdin: %s < %s", unexpected, name, shellQuoted(arg))
+	}
+	ref := "{/" + strings.TrimPrefix(arg, "/") + "}"
+	_, argLone := loneReference(arg)
+	inline, err := fejkdata.IsTemplate(arg)
+	_, refLone := loneReference(ref)
+	example := pipeHint(name)
+	switch {
+	case argLone || inline:
+		example = piped(arg, name)
+	case err == nil && refLone:
+		example = piped(ref, name)
+	}
+	return fmt.Errorf("%s; the template is read from stdin: %s", unexpected, example)
+}
+
+// check rejects a flag combination or an operand that cannot run, before any data is
+// read.
+func (in invocation) check(name string) error {
+	if err := in.checkFlags(); err != nil {
+		return err
+	}
+	if len(in.args) > 0 {
+		return operandError(in.args[0], name)
+	}
+	return nil
+}
+
+func (in invocation) options(modules []fs.FS, systemSeed uint64) []fejkdata.Option {
+	if in.noShipped {
+		modules = nil
+	}
+	opts := []fejkdata.Option{fejkdata.WithDataFS(modules...)}
+	for _, dir := range in.dirs {
+		opts = append(opts, fejkdata.WithDataPath(dir))
+	}
+	seed := in.seed
+	if !in.seeded {
+		seed = systemSeed
+	}
+	return append(opts, fejkdata.WithSeed(seed))
+}
+
+// write streams the input's renders to w, repeat of them joined by the separator; a write
+// failure surfaces from Flush, bufio keeping the first one.
+func (in invocation) write(f *fejkdata.Generator, src input, w io.Writer) error {
+	if in.writesRecords() {
+		return in.writeRecords(f, src, w)
+	}
+	t, err := f.NewTemplate(src.template)
+	if err != nil {
+		return templateError(err)
+	}
+	out := bufio.NewWriter(w)
+	for i := 0; i < in.repeat; i++ {
+		if i > 0 {
+			out.WriteString(in.separator)
+		}
+		out.WriteString(t.Fake())
+	}
+	return out.Flush()
+}
+
+// recordStream builds the record drawer for the input, plus the INSERT table
+// a sql format names.
+func (in invocation) recordStream(f *fejkdata.Generator, src input) (func() (*fejkdata.Record, error), string, error) {
+	record := func() (*fejkdata.Record, error) { return f.FakeRecord(src.record) }
+	if src.record == "" {
+		t, err := f.NewRecordTemplate(src.template)
+		if err != nil {
+			return nil, "", templateError(spacedReference(src.template, err))
+		}
+		record = func() (*fejkdata.Record, error) { return t.Fake(), nil }
+	}
+	table := in.table
+	if table == "" {
+		table = defaultTable(src.record)
+	}
+	return record, table, nil
+}
+
+// writeRecords streams a record per line in the chosen format, framing a document
+// form with its open/close brackets and a header preceding the first record.
+func (in invocation) writeRecords(f *fejkdata.Generator, src input, w io.Writer) error {
+	record, table, err := in.recordStream(f, src)
+	if err != nil {
+		return err
+	}
+	format := recordFormats[in.format]
+	if in.repeat == 0 {
+		return writeNoRecord(record, format, w)
+	}
+	out := bufio.NewWriter(w)
+	if format.open != "" {
+		out.WriteString(format.open)
+		out.WriteByte('\n')
+	}
+	for i := 0; i < in.repeat; i++ {
+		r, err := record()
+		if err != nil {
+			return err
+		}
+		if i == 0 && format.header != nil {
+			out.WriteString(format.header(r))
+			out.WriteByte('\n')
+		}
+		if i > 0 {
+			out.WriteString(format.sep)
+		}
+		out.WriteString(format.line(r, table))
+	}
+	if format.close != "" {
+		out.WriteByte('\n')
+		out.WriteString(format.close)
+	}
+	out.WriteByte('\n')
+	return out.Flush()
+}
+
+// writeNoRecord writes zero records: the header where the format writes one, an empty document
+// where it frames one, else nothing. It draws one record, which names the header's columns, so a
+// lone reference to nothing fails as it would with records.
+func writeNoRecord(record func() (*fejkdata.Record, error), format recordFormat, w io.Writer) error {
+	r, err := record()
+	switch {
+	case err != nil:
+		return err
+	case format.header != nil:
+		_, err = io.WriteString(w, format.header(r)+"\n")
+	case format.open != "":
+		_, err = io.WriteString(w, format.open+format.close+"\n")
+	}
+	return err
+}
+
+// defaultTable names the INSERT target when --table is absent: the path's last
+// segment, or "records" when stdin is not a lone reference.
+func defaultTable(path string) string {
+	if path == "" {
+		return "records"
+	}
+	segments := strings.Split(withoutSelectors(path), ".")
+	return segments[len(segments)-1]
+}
+
+// withoutSelectors is path with its [selectors] cut out.
+func withoutSelectors(path string) string {
+	var names strings.Builder
+	for depth, i := 0, 0; i < len(path); i++ {
+		switch {
+		case path[i] == '[':
+			depth++
+		case path[i] == ']':
+			depth--
+		case depth == 0:
+			names.WriteByte(path[i])
+		}
+	}
+	return names.String()
+}
+
+// inputError marks a failure that is the input's own fault — nothing piped in, a template
+// that does not compile, or a stdin that cannot be read. run reports it as misuse (exit 2, with a
+// pointer to --help), unlike a lone reference to nothing under --format, which is a runtime
+// error (exit 1).
+type inputError struct{ error }
+
+func (e inputError) Unwrap() error { return e.error }
+
+// jsonSpace is the whitespace JSON allows around a value.
+const jsonSpace = " \t\r\n"
+
+// input is what stdin holds: the template, and the path a lone reference names as a
+// record.
+type input struct {
+	template string
+	record   string
+}
+
+// parseInput keeps stdin whole as the template, and names a record where stdin is a lone
+// reference once JSON's surrounding whitespace, or one final newline, is cut.
+func parseInput(raw string) input {
+	trimmed := strings.Trim(raw, jsonSpace)
+	text := raw
+	if json.Valid([]byte(trimmed)) {
+		text = trimmed
+	}
+	text, found := strings.CutSuffix(text, "\n")
+	if found {
+		text = strings.TrimSuffix(text, "\r")
+	}
+	record, _ := loneReference(text)
+	return input{template: raw, record: record}
+}
+
+// loneReference returns the path that text reads when it is one reference from the root
+// and nothing else, bare or as a JSON string.
+func loneReference(text string) (string, bool) {
+	var unquoted string
+	if strings.HasPrefix(text, `"`) && strings.HasSuffix(text, `"`) && json.Unmarshal([]byte(text), &unquoted) == nil {
+		text = unquoted
+	}
+	path, opens := strings.CutPrefix(text, "{/")
+	path, closes := strings.CutSuffix(path, "}")
+	if !opens || !closes || !isPath(path) {
+		return "", false
+	}
+	return path, true
+}
+
+// isPath reports whether a reference's body after its / is a path alone: no second /, no
+// folder sigil, no empty segment, and none of the |, ( or " as " a token reads outside a selector.
+// It copies the grammar, since the CLI reads only the library's public API.
+func isPath(body string) bool {
+	if body == "" || strings.ContainsAny(body[:1], "/.") {
+		return false
+	}
+	names := withoutSelectors(body)
+	if strings.ContainsAny(names, "|(") || strings.Contains(names, " as ") || hasEmptySegment(names) {
+		return false
+	}
+	inline, err := fejkdata.IsTemplate(body)
+	return !inline && err == nil
+}
+
+// hasEmptySegment reports whether a dotted path holds a segment with no name: two dots are a
+// step up, and only a step up may end the path.
+func hasEmptySegment(path string) bool {
+	segments := strings.Split(path, ".")
+	stepUp := false
+	for i, seg := range segments[1:] {
+		last := i == len(segments)-2
+		switch {
+		case seg != "":
+			stepUp = false
+		case stepUp != last:
+			return true
+		default:
+			stepUp = !stepUp
+		}
+	}
+	return false
+}
+
+// spacedReference adds to a record refusal that whitespace around a lone reference made
+// the template text.
+func spacedReference(template string, err error) error {
+	if trimmed := strings.TrimSpace(template); trimmed != template && errors.Is(err, fejkdata.ErrNoColumns) {
+		if path, lone := loneReference(trimmed); lone {
+			return fmt.Errorf("%w; the whitespace around {/%s} makes it text, so remove it to write the record %s", err, path, path)
+		}
+	}
+	return err
+}
+
+// piped is the command piping template into the command called name; some shells' echo
+// reads a backslash as an escape, so a template holding one goes through printf.
+func piped(template, name string) string {
+	if strings.ContainsRune(template, '\\') {
+		return "printf '%s' " + shellQuoted(template) + " | " + name
+	}
+	return "echo " + shellQuoted(template) + " | " + name
+}
+
+// shellQuoted is s as a shell reads it back: bare where it holds nothing a shell expands.
+func shellQuoted(s string) string {
+	if s != "" && !strings.ContainsAny(s, " \t\r\n'\"$`\\*?[]{}()<>|&;#~!") {
+		return s
+	}
+	return "'" + strings.ReplaceAll(s, "'", `'\''`) + "'"
+}
+
+// Main runs the command on the process's arguments and standard streams over modules, each
+// passed to [fejkdata.WithDataFS], and returns the exit code: 0 ok, 1 runtime error,
+// 2 misuse. --no-shipped-data leaves modules out.
+func Main(modules ...fs.FS) int {
+	var stdin io.Reader = os.Stdin
+	if fi, err := os.Stdin.Stat(); err == nil && fi.Mode()&os.ModeCharDevice != 0 {
+		stdin = nil
+	}
+	return run(commandName(os.Args[0]), modules, os.Args[1:], rand.Uint64(), stdin, os.Stdout, os.Stderr)
+}
+
+// commandName is the name the command runs under: the file arg0 names, less a Windows
+// .exe, or fejkdata where arg0 names none.
+func commandName(arg0 string) string {
+	name := filepath.Base(arg0)
+	if ext := filepath.Ext(name); strings.EqualFold(ext, ".exe") {
+		name = strings.TrimSuffix(name, ext)
+	}
+	if name == "." || name == "" {
+		return "fejkdata"
+	}
+	return name
+}
+
+// run is Main over the given streams, run as the command name, seeded by systemSeed where
+// no --seed is given. A nil stdin is a terminal, which holds no template.
+func run(name string, modules []fs.FS, args []string, systemSeed uint64, stdin io.Reader, stdout, stderr io.Writer) int {
+	in, err := parseArgs(args)
+	if err != nil {
+		return misuse(stderr, name, err)
+	}
+	if in.help {
+		io.WriteString(stdout, strings.ReplaceAll(usage, "COMMAND", name))
+		return 0
+	}
+	if in.version {
+		info, _ := debug.ReadBuildInfo()
+		fmt.Fprintln(stdout, versionLine(name, info))
+		return 0
+	}
+	if err := in.check(name); err != nil {
+		return misuse(stderr, name, err)
+	}
+	var src input
+	if !in.list {
+		if src, err = readStdin(stdin, name); err != nil {
+			return fail(stderr, name, err)
+		}
+	}
+	f, err := fejkdata.New(in.options(modules, systemSeed)...)
+	if err != nil {
+		return fail(stderr, name, err)
+	}
+	if in.list {
+		for _, p := range f.List() {
+			fmt.Fprintln(stdout, p)
+		}
+		return 0
+	}
+	if err := in.write(f, src, stdout); err != nil {
+		return fail(stderr, name, err)
+	}
+	return 0
+}
+
+func readStdin(stdin io.Reader, name string) (input, error) {
+	if stdin == nil {
+		return input{}, inputError{fmt.Errorf("the template comes from stdin, and nothing is piped in: %s; %s --list prints the paths a {/path} reference reads", pipeHint(name), name)}
+	}
+	raw, err := io.ReadAll(stdin)
+	if err != nil {
+		return input{}, inputError{fmt.Errorf("stdin cannot be read: %w", err)}
+	}
+	return parseInput(string(raw)), nil
+}
+
+// templateError is misuse unless the data the template reads failed to load.
+func templateError(err error) error {
+	if errors.Is(err, fejkdata.ErrLoad) {
+		return err
+	}
+	return inputError{err}
+}
+
+func fail(stderr io.Writer, name string, err error) int {
+	var te inputError
+	if errors.As(err, &te) {
+		return misuse(stderr, name, te.error)
+	}
+	report(stderr, name, err)
+	if errors.Is(err, fejkdata.ErrNoColumns) {
+		fmt.Fprintln(stderr, "wrap that JSON in single quotes, which keep a shell from expanding its braces")
+	}
+	return 1
+}
+
+func misuse(stderr io.Writer, name string, err error) int {
+	report(stderr, name, err)
+	fmt.Fprintf(stderr, "try '%s --help'\n", name)
+	return 2
+}
+
+// report writes err under the command name, in place of the library's own on each line.
+func report(stderr io.Writer, name string, err error) {
+	msg := strings.ReplaceAll(strings.TrimPrefix(err.Error(), "fejkdata: "), "\nfejkdata: ", "\n"+name+": ")
+	fmt.Fprintf(stderr, "%s: %s\n", name, msg)
+}
+
+// fejkdataModule is the path of the module holding the library; the root package's path is
+// the module's.
+var fejkdataModule = reflect.TypeOf((*fejkdata.Generator)(nil)).Elem().PkgPath()
+
+// versionLine names the fejkdata version the binary is built with. Where another module's
+// main package carries the CLI, the main module's version comes first, under the command name.
+func versionLine(name string, info *debug.BuildInfo) string {
+	if info == nil {
+		return "fejkdata devel"
+	}
+	if info.Main.Path == fejkdataModule {
+		return "fejkdata " + moduleVersion(info.Main)
+	}
+	for _, m := range info.Deps {
+		if m.Path == fejkdataModule {
+			return name + " " + moduleVersion(info.Main) + ", fejkdata " + moduleVersion(*m)
+		}
+	}
+	return name + " " + moduleVersion(info.Main) + ", fejkdata devel"
+}
+
+// moduleVersion is the version m is built at: the replacement's where a replace directive
+// names one, and "devel" where it carries none, as a local directory does.
+func moduleVersion(m debug.Module) string {
+	if m.Replace != nil {
+		m = *m.Replace
+	}
+	if m.Version == "" || m.Version == "(devel)" {
+		return "devel"
+	}
+	return m.Version
+}

cli/cli_test.go

@@ -0,0 +1,961 @@
+package cli
+
+import (
+	"bytes"
+	"encoding/csv"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io/fs"
+	"os"
+	"path/filepath"
+	"regexp"
+	"runtime/debug"
+	"strings"
+	"testing"
+	"testing/fstest"
+)
+
+// shipped holds the shipped modules as read from the checkout, each folder under ../data
+// holding a manifest, so these tests link no data package.
+var shipped = func() []fs.FS {
+	var modules []fs.FS
+	err := filepath.WalkDir(filepath.Join("..", "data"), func(path string, d fs.DirEntry, err error) error {
+		if err == nil && d.Name() == ".fejkdata.json" {
+			modules = append(modules, os.DirFS(filepath.Dir(path)))
+		}
+		return err
+	})
+	if err != nil || len(modules) == 0 {
+		panic(fmt.Sprintf("no shipped module found under ../data: %v", err))
+	}
+	return modules
+}()
+
+// svSE, enUS and misc are copies of a shipped module's files without its manifest, so a test
+// layers its categories unprefixed with -d.
+var svSE, enUS, misc string
+
+func TestMain(m *testing.M) {
+	tmp, err := os.MkdirTemp("", "fejkdata-cli-test")
+	if err != nil {
+		panic(err)
+	}
+	svSE, enUS, misc = bareCopy(tmp, "sv_SE"), bareCopy(tmp, "en_US"), bareCopy(tmp, "misc")
+	code := m.Run()
+	os.RemoveAll(tmp)
+	os.Exit(code)
+}
+
+// bareCopy copies the category and rows files of the shipped module in data/<module> into a
+// folder under tmp, and returns that folder.
+func bareCopy(tmp, module string) string {
+	src, dst := filepath.Join("..", "data", module), filepath.Join(tmp, module)
+	if err := os.Mkdir(dst, 0o755); err != nil {
+		panic(err)
+	}
+	for _, pattern := range []string{"*.json", "*.tsv"} {
+		files, err := filepath.Glob(filepath.Join(src, pattern))
+		if err != nil {
+			panic(err)
+		}
+		for _, f := range files {
+			if strings.HasPrefix(filepath.Base(f), ".") {
+				continue
+			}
+			b, err := os.ReadFile(f)
+			if err == nil {
+				err = os.WriteFile(filepath.Join(dst, filepath.Base(f)), b, 0o644)
+			}
+			if err != nil {
+				panic(err)
+			}
+		}
+	}
+	return dst
+}
+
+func runOut(stdin string, args ...string) (int, string, string) {
+	var out, errb bytes.Buffer
+	code := run("fejkdata", shipped, args, 1, strings.NewReader(stdin), &out, &errb)
+	return code, out.String(), errb.String()
+}
+
+// runTerminal runs as from a shell with nothing piped in.
+func runTerminal(args ...string) (int, string, string) {
+	var out, errb bytes.Buffer
+	code := run("fejkdata", shipped, args, 1, nil, &out, &errb)
+	return code, out.String(), errb.String()
+}
+
+func TestRunLoadsTheModulesItIsGiven(t *testing.T) {
+	module := fstest.MapFS{"pet.json": {Data: []byte(`{"format":"{name}","name":["Fido"]}`)}}
+	var out, errb bytes.Buffer
+	if code := run("fejkdata", []fs.FS{module}, nil, 1, strings.NewReader("{/pet}"), &out, &errb); code != 0 || out.String() != "Fido" {
+		t.Errorf("run over a module holding pet = %d, %q, %q, want Fido", code, out.String(), errb.String())
+	}
+	out.Reset()
+	if code := run("fejkdata", []fs.FS{module}, []string{"--list"}, 1, nil, &out, &errb); code != 0 || out.String() != "pet\npet.name\n" {
+		t.Errorf("--list over a module holding pet = %d, %q, want pet and pet.name only", code, out.String())
+	}
+}
+
+func TestRunOutputsValue(t *testing.T) {
+	code, out, errb := runOut("{/person}", "--data-path", svSE)
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	if out == "" || strings.Contains(out, "\n") {
+		t.Fatalf("output = %q, stderr=%q; want one value and nothing added", out, errb)
+	}
+}
+
+func TestRunDotPath(t *testing.T) {
+	code, full, _ := runOut("{/person}", "--seed", "7", "--data-path", svSE)
+	if code != 0 {
+		t.Fatalf("person run = %d", code)
+	}
+	code, last, errb := runOut("{/person.last}", "--seed", "7", "--data-path", svSE)
+	if code != 0 {
+		t.Fatalf("person.last run = %d, stderr=%q", code, errb)
+	}
+	if strings.TrimSpace(last) == "" {
+		t.Fatal("empty person.last output")
+	}
+	if last == full {
+		t.Errorf("person.last %q should differ from person %q", last, full)
+	}
+}
+
+func TestRunWithoutASeedTakesTheSystemSeed(t *testing.T) {
+	var drawn, seeded, errb bytes.Buffer
+	if code := run("fejkdata", shipped, nil, 42, strings.NewReader("{uuid()}"), &drawn, &errb); code != 0 {
+		t.Fatalf("run with system seed 42 = %d, stderr=%q", code, errb.String())
+	}
+	if code := run("fejkdata", shipped, []string{"--seed", "42"}, 7, strings.NewReader("{uuid()}"), &seeded, &errb); code != 0 {
+		t.Fatalf("run --seed 42 = %d, stderr=%q", code, errb.String())
+	}
+	if drawn.String() != seeded.String() {
+		t.Fatalf("run with system seed 42 = %q, run --seed 42 = %q; want the same draw", drawn.String(), seeded.String())
+	}
+}
+
+func TestRunSeedSpellings(t *testing.T) {
+	_, want, _ := runOut("{/address}", "--seed", "42", "--data-path", svSE)
+	for _, args := range [][]string{
+		{"--seed=42", "--data-path", svSE},
+		{"-s", "42", "-d", svSE},
+		{"--data-path", svSE, "--seed", "42"},
+	} {
+		code, got, errb := runOut("{/address}", args...)
+		if code != 0 {
+			t.Fatalf("run(%v) = %d, stderr=%q", args, code, errb)
+		}
+		if got != want {
+			t.Errorf("run(%v) = %q, want %q", args, got, want)
+		}
+	}
+}
+
+func TestRunShortFlagValues(t *testing.T) {
+	_, want, _ := runOut("{/address}", "--seed", "42", "--data-path", svSE)
+	for _, args := range [][]string{
+		{"-s42", "-d", svSE},
+		{"-s", "42", "-d" + svSE},
+		{"-d", svSE, "-s42"},
+	} {
+		code, got, errb := runOut("{/address}", args...)
+		if code != 0 {
+			t.Fatalf("run(%v) = %d, stderr=%q", args, code, errb)
+		}
+		if got != want {
+			t.Errorf("run(%v) = %q, want %q", args, got, want)
+		}
+	}
+	_, three, _ := runOut("{/word}\n", "-s", "1", "-n3", "-d", svSE)
+	if lines := strings.Split(strings.TrimRight(three, "\n"), "\n"); len(lines) != 3 {
+		t.Errorf("-n3 gave %d lines: %q", len(lines), three)
+	}
+	for _, c := range []struct {
+		args []string
+		want string
+	}{
+		{[]string{"-s=42", "-d", svSE}, "-s42"},
+		{[]string{"-s=42", "-d", svSE}, "--seed=42"},
+		{[]string{"-d=" + svSE}, "--data-path="},
+		{[]string{"-nd", "3", "-d", svSE}, `--repeat needs an integer in 0..1048576, got "d"`},
+		{[]string{"--seed=", "-d", svSE}, `--seed needs an unsigned integer, got ""`},
+	} {
+		code, out, errb := runOut("{/word}", c.args...)
+		if code != 2 || out != "" {
+			t.Errorf("run(%v) = %d, stdout %q, want misuse", c.args, code, out)
+		}
+		if !strings.Contains(errb, c.want) {
+			t.Errorf("run(%v) stderr = %q, want %q", c.args, errb, c.want)
+		}
+	}
+	code, out, _ := runOut("", "-hd", svSE)
+	if code != 0 || !strings.Contains(out, "Usage") {
+		t.Errorf("-hd (bundled help) = %d, %q, want usage", code, out)
+	}
+}
+
+func TestRunDoubleDashEndsFlags(t *testing.T) {
+	code, out, errb := runOut("{/person}", "--data-path", svSE, "--")
+	if code != 0 || strings.TrimSpace(out) == "" {
+		t.Fatalf("run = %d, out=%q, stderr=%q", code, out, errb)
+	}
+	code, _, errb = runOut("{/person}", "--data-path", svSE, "--", "--list")
+	if code != 2 || !strings.Contains(errb, "stdin") {
+		t.Errorf("after --, --list should be an argument: code %d, stderr %q", code, errb)
+	}
+}
+
+func TestRunSingleDashLongIsRejected(t *testing.T) {
+	for _, c := range []struct {
+		args []string
+		want string
+	}{
+		{[]string{"-seed", "42", "-d", svSE}, "use --seed"},
+		{[]string{"-seed=42", "-d", svSE}, "use --seed"},
+		{[]string{"-data-path", svSE}, "use --data-path"},
+		{[]string{"-list", "-d", svSE}, "use --list"},
+		{[]string{"--nope", "-d", svSE}, "unknown flag --nope"},
+		{[]string{"-x", "-d", svSE}, "unknown flag -x"},
+	} {
+		code, _, errb := runOut("{/person}", c.args...)
+		if code != 2 {
+			t.Errorf("run(%v) = %d, want 2", c.args, code)
+		}
+		if !strings.Contains(errb, c.want) {
+			t.Errorf("run(%v) stderr = %q, want %q", c.args, errb, c.want)
+		}
+	}
+}
+
+func TestRunFlagValues(t *testing.T) {
+	for _, c := range []struct {
+		args []string
+		want string
+	}{
+		{[]string{"--seed", "-d", svSE}, `--seed needs an unsigned integer, got "-d"`},
+		{[]string{"-d", svSE, "--seed"}, "--seed needs a value"},
+		{[]string{"--seed", "x", "-d", svSE}, "--seed"},
+		{[]string{"--list=1", "-d", svSE}, "--list takes no value"},
+		{[]string{"--repeat", "-1", "-d", svSE}, "--repeat"},
+		{[]string{"-n", "-1", "-d", svSE}, "--repeat"},
+	} {
+		code, _, errb := runOut("{/person}", c.args...)
+		if code != 2 {
+			t.Errorf("run(%v) = %d, want 2", c.args, code)
+		}
+		if !strings.Contains(errb, c.want) {
+			t.Errorf("run(%v) stderr = %q, want %q", c.args, errb, c.want)
+		}
+	}
+}
+
+func TestRunRepeat(t *testing.T) {
+	code, out, errb := runOut("{/word}\n", "--seed", "1", "--repeat", "3", "--data-path", svSE)
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	lines := strings.Split(strings.TrimRight(out, "\n"), "\n")
+	if len(lines) != 3 {
+		t.Errorf("repeat=3 gave %d lines: %q", len(lines), out)
+	}
+	_, short, _ := runOut("{/word}\n", "--seed", "1", "-n", "3", "-d", svSE)
+	if short != out {
+		t.Errorf("-n 3 = %q, want the same as --repeat 3 %q", short, out)
+	}
+}
+
+func TestRunSeparator(t *testing.T) {
+	code, out, errb := runOut("{/word}", "--repeat", "3", "--separator", ",", "--data-path", svSE)
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	if n := strings.Count(out, "\n"); n != 0 {
+		t.Errorf("want no newline, got %d: %q", n, out)
+	}
+	if !strings.Contains(out, ",") {
+		t.Errorf("values should be comma-joined: %q", out)
+	}
+}
+
+func TestRunRepeatAdvancesRNG(t *testing.T) {
+	code, out, errb := runOut("{/person}\n", "--seed", "1", "--repeat", "5", "--data-path", svSE)
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	uniq := map[string]bool{}
+	for _, l := range strings.Split(strings.TrimRight(out, "\n"), "\n") {
+		uniq[l] = true
+	}
+	if len(uniq) < 2 {
+		t.Errorf("repeat should vary output, all identical: %q", out)
+	}
+}
+
+func TestRunMisuse(t *testing.T) {
+	for _, c := range []struct {
+		stdin string
+		args  []string
+	}{
+		{"{/person}", []string{"-d", svSE, "word"}},
+		{"", []string{"-d", svSE, "--list", "person"}},
+		{"{/sv_SE.person}", []string{"--no-shipped-data"}},
+	} {
+		code, out, errb := runOut(c.stdin, c.args...)
+		if code != 2 {
+			t.Errorf("run(%q, %v) = %d, want 2", c.stdin, c.args, code)
+		}
+		if !strings.Contains(errb, "try 'fejkdata --help'") {
+			t.Errorf("run(%q, %v) stderr = %q, want a pointer to --help", c.stdin, c.args, errb)
+		}
+		if out != "" {
+			t.Errorf("run(%q, %v) stdout = %q, want nothing", c.stdin, c.args, out)
+		}
+	}
+}
+
+func TestRunTakesNoOperand(t *testing.T) {
+	file := filepath.Join(t.TempDir(), "users.tmpl")
+	if err := os.WriteFile(file, []byte("{/sv_SE.person}"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	for _, c := range []struct {
+		args []string
+		want string
+	}{
+		{[]string{"sv_SE.person"}, "echo '{/sv_SE.person}' | fejkdata"},
+		{[]string{"/en_US.address"}, "echo '{/en_US.address}' | fejkdata"},
+		{[]string{"{/en_US.person}"}, "echo '{/en_US.person}' | fejkdata"},
+		{[]string{"name: {/sv_SE.person.last}"}, "echo 'name: {/sv_SE.person.last}' | fejkdata"},
+		{[]string{"geo.US.locality[O'Fallon].name"}, `echo '{/geo.US.locality[O'\''Fallon].name}' | fejkdata`},
+		{[]string{"en_US.address"}, "echo '{/en_US.address}' | fejkdata"},
+		{[]string{`C:\temp {/sv_SE.person}`}, `printf '%s' 'C:\temp {/sv_SE.person}' | fejkdata`},
+		{[]string{"-"}, "drop it"},
+		{[]string{file}, "fejkdata < " + file},
+	} {
+		code, out, errb := runOut("", c.args...)
+		if code != 2 || out != "" || !strings.Contains(errb, "unexpected argument") || !strings.Contains(errb, c.want) {
+			t.Errorf("run(%v) = %d, %q, %q; want misuse naming %q", c.args, code, out, errb, c.want)
+		}
+	}
+}
+
+type failingReader struct{}
+
+func (failingReader) Read([]byte) (int, error) { return 0, errors.New("bad file descriptor") }
+
+func TestRunUnreadableStdinIsMisuse(t *testing.T) {
+	var out, errb bytes.Buffer
+	if code := run("fejkdata", shipped, nil, 1, failingReader{}, &out, &errb); code != 2 || !strings.Contains(errb.String(), "cannot be read") {
+		t.Errorf("run with an unreadable stdin = %d, %q; want misuse naming it", code, errb.String())
+	}
+}
+
+func TestRunTerminalStdinIsMisuse(t *testing.T) {
+	code, out, errb := runTerminal("--seed", "1")
+	if code != 2 || out != "" || !strings.Contains(errb, "nothing is piped in: echo 'id: {int(1,999)}' | fejkdata; fejkdata --list prints the paths a {/path} reference reads\n") {
+		t.Errorf("run with nothing piped = %d, %q, %q; want misuse showing a template piped to fejkdata", code, out, errb)
+	}
+	for _, args := range [][]string{{"--list"}, {"--help"}, {"--version"}} {
+		if code, out, errb := runTerminal(args...); code != 0 || out == "" {
+			t.Errorf("run(%v) with nothing piped = %d, %q, %q; want it to run, reading no stdin", args, code, out, errb)
+		}
+	}
+}
+
+func TestRunRendersExactlyWhatIsSent(t *testing.T) {
+	for stdin, want := range map[string]string{
+		"hihi":                           "hihi",
+		"hihi\n":                         "hihi\n",
+		"hihi\r\n":                       "hihi\r\n",
+		"hihi\n\n":                       "hihi\n\n",
+		"{{x}}\n":                        "{x}\n",
+		"\"a\"\n":                        "a",
+		" \t\"a\"\r\n\n":                 "a",
+		"{/misc.territory[SE].alpha2}":   "SE",
+		"{/misc.territory[SE].alpha2}\n": "SE\n",
+	} {
+		code, out, errb := runOut(stdin, "--seed", "1")
+		if code != 0 || out != want {
+			t.Errorf("run(%q) = %d, %q, %q; want %q", stdin, code, out, errb, want)
+		}
+	}
+	if code, out, errb := runOut("{\"format\":\"{x}\\n\",\"x\":[\"a\",\"b\"]}\n"); code != 0 || (out != "a\n" && out != "b\n") {
+		t.Errorf("a JSON template's format = %d, %q, %q; want its own newline only", code, out, errb)
+	}
+	if code, out, errb := runOut("{/sv_SE.word}\n", "--seed", "1", "-n", "3", "--separator", ","); code != 0 || strings.Count(out, "\n,") != 2 || !strings.HasSuffix(out, "\n") {
+		t.Errorf("echo's newline and a separator = %d, %q, %q; want each render's newline, then the separator", code, out, errb)
+	}
+}
+
+func TestRunRendersBlankStdin(t *testing.T) {
+	for _, stdin := range []string{"", "\n", " \r\n\t"} {
+		if code, out, errb := runOut(stdin); code != 0 || out != stdin {
+			t.Errorf("run(%q) = %d, %q, %q; want the blank template rendered as it is", stdin, code, out, errb)
+		}
+	}
+}
+
+func TestRunTextPrintsAsWritten(t *testing.T) {
+	for _, stdin := range []string{"sv_SE.person", "[Skåne län]", "x[1]y", `say "hi"`} {
+		code, out, errb := runOut(stdin)
+		if code != 0 || out != stdin {
+			t.Errorf("run(%q) = %d, %q, %q; want the text as written", stdin, code, out, errb)
+		}
+	}
+}
+
+func TestRunMultipleDataPaths(t *testing.T) {
+	code, out, errb := runOut("{/person}", "-d", enUS, "--data-path", svSE)
+	if code != 0 {
+		t.Fatalf("run(multi-dir) = %d, stderr=%q", code, errb)
+	}
+	if strings.TrimSpace(out) == "" {
+		t.Fatal("empty output for multi-dir run")
+	}
+}
+
+func TestRunUnknownCategoryFails(t *testing.T) {
+	code, _, errb := runOut("{/nope}", "--data-path", svSE, "--format", "csv")
+	if code != 1 {
+		t.Fatalf("run = %d, want 1", code)
+	}
+	if !strings.Contains(errb, "nope") {
+		t.Errorf("stderr %q should name the unknown category", errb)
+	}
+}
+
+func TestRunList(t *testing.T) {
+	code, out, errb := runOut("", "--data-path", svSE, "--list")
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	for _, want := range []string{"person", "person.last", "address"} {
+		if !strings.Contains(out, want) {
+			t.Errorf("list output missing %q:\n%s", want, out)
+		}
+	}
+}
+
+func TestRunHelpOnStdout(t *testing.T) {
+	for _, h := range []string{"-h", "--help"} {
+		code, out, errb := runOut("", h)
+		if code != 0 {
+			t.Errorf("%s = %d, want 0", h, code)
+		}
+		if !strings.Contains(out, "Usage") || errb != "" {
+			t.Errorf("%s: stdout = %q, stderr = %q, want usage on stdout only", h, out, errb)
+		}
+	}
+}
+
+func TestVersionLineNamesFejkdatasVersion(t *testing.T) {
+	fejkdata := func(version string, replace *debug.Module) *debug.Module {
+		return &debug.Module{Path: "github.com/larvit/fejkdata", Version: version, Replace: replace}
+	}
+	mycli := debug.Module{Path: "example.com/mycli", Version: "v9.0.0"}
+	for _, c := range []struct {
+		info *debug.BuildInfo
+		want string
+	}{
+		{&debug.BuildInfo{Main: *fejkdata("v0.1.0", nil)}, "fejkdata v0.1.0"},
+		{&debug.BuildInfo{Main: *fejkdata("(devel)", nil)}, "fejkdata devel"},
+		{&debug.BuildInfo{Main: mycli, Deps: []*debug.Module{fejkdata("v0.1.0", nil)}}, "mycli v9.0.0, fejkdata v0.1.0"},
+		{&debug.BuildInfo{Main: mycli, Deps: []*debug.Module{fejkdata("v0.1.0", &debug.Module{Path: "../fejkdata"})}}, "mycli v9.0.0, fejkdata devel"},
+		{&debug.BuildInfo{Main: mycli, Deps: []*debug.Module{fejkdata("v0.1.0", &debug.Module{Path: "example.com/fork", Version: "v0.1.1"})}}, "mycli v9.0.0, fejkdata v0.1.1"},
+		{&debug.BuildInfo{Main: mycli}, "mycli v9.0.0, fejkdata devel"},
+		{nil, "fejkdata devel"},
+	} {
+		if got := versionLine("mycli", c.info); got != c.want {
+			t.Errorf("versionLine(mycli, %+v) = %q, want %q", c.info, got, c.want)
+		}
+	}
+}
+
+func TestCommandNameIsTheFileItRunsFrom(t *testing.T) {
+	for arg0, want := range map[string]string{
+		"/usr/local/bin/mycli": "mycli",
+		"mycli.EXE":            "mycli",
+		"mycli.exe":            "mycli",
+		"my.cli":               "my.cli",
+		"":                     "fejkdata",
+	} {
+		if got := commandName(arg0); got != want {
+			t.Errorf("commandName(%q) = %q, want %q", arg0, got, want)
+		}
+	}
+}
+
+func TestRunNamesItselfByTheCommandName(t *testing.T) {
+	var out, errb bytes.Buffer
+	if code := run("mycli", shipped, []string{"--nope"}, 1, nil, &out, &errb); code != 2 || errb.String() != "mycli: unknown flag --nope\ntry 'mycli --help'\n" {
+		t.Errorf("mycli --nope = %d, %q", code, errb.String())
+	}
+	errb.Reset()
+	if code := run("mycli", shipped, nil, 1, nil, &out, &errb); code != 2 || !strings.Contains(errb.String(), "| mycli; mycli --list prints") {
+		t.Errorf("mycli with nothing piped = %d, %q, want a pipe into mycli", code, errb.String())
+	}
+	errb.Reset()
+	if code := run("mycli", []fs.FS{nil, nil}, nil, 1, strings.NewReader("x"), &out, &errb); code != 1 || strings.Count(errb.String(), "mycli: ") != 2 || strings.Contains(errb.String(), "fejkdata") {
+		t.Errorf("mycli over two nil modules = %d, %q, want each line under mycli", code, errb.String())
+	}
+	errb.Reset()
+	if code := run("mycli", shipped, []string{"-d", "/nope"}, 1, strings.NewReader("{/sv_SE.person}"), &out, &errb); code != 1 || !strings.HasPrefix(errb.String(), "mycli: ") || strings.Contains(errb.String(), "fejkdata") {
+		t.Errorf("mycli -d /nope = %d, %q, want a load error under mycli", code, errb.String())
+	}
+	if code := run("mycli", shipped, []string{"--help"}, 1, nil, &out, &errb); code != 0 || !strings.HasPrefix(out.String(), "Usage: <template> | mycli [flags]\n") || strings.Contains(out.String(), "fejkdata") {
+		t.Errorf("mycli --help = %d, %q, want mycli named throughout", code, out.String())
+	}
+}
+
+func TestRunVersion(t *testing.T) {
+	code, out, errb := runOut("", "--version")
+	if code != 0 {
+		t.Fatalf("--version = %d, stderr=%q", code, errb)
+	}
+	version, ok := strings.CutPrefix(out, "fejkdata ")
+	if !ok || strings.TrimSpace(version) == "" {
+		t.Errorf("--version = %q, want the command name and a version on stdout", out)
+	}
+}
+
+func TestRunMissingDirFails(t *testing.T) {
+	code, _, errb := runOut("{/person}", "--data-path", "../data/nope")
+	if code != 1 {
+		t.Fatalf("run = %d, want 1", code)
+	}
+	if errb == "" {
+		t.Error("want an error message on stderr")
+	}
+}
+
+func TestRunShippedDataByDefault(t *testing.T) {
+	code, out, errb := runOut("{/sv_SE.person}", "--seed", "1")
+	if code != 0 || strings.TrimSpace(out) == "" {
+		t.Fatalf("run = %d, out=%q, stderr=%q", code, out, errb)
+	}
+	code, list, _ := runOut("", "--list")
+	if code != 0 || !strings.Contains(list, "en_US.person\n") || !strings.Contains(list, "misc.uuid\n") {
+		t.Errorf("--list without --data-path = %d, %q", code, list)
+	}
+	code, _, errb = runOut("{/person}")
+	if code != 2 || !strings.Contains(errb, "person") {
+		t.Errorf("a category outside the shipped tree: code %d, stderr %q", code, errb)
+	}
+}
+
+func TestRunSelectsATableRow(t *testing.T) {
+	dir := t.TempDir()
+	for name, content := range map[string]string{
+		"region.json": `{"format":"{name}","rows":"region.tsv","key":"code","name":"name"}`,
+		"region.tsv":  "code\tname\n01\tStockholms län\n12\tSkåne län\n",
+	} {
+		if err := os.WriteFile(filepath.Join(dir, name), []byte(content), 0o644); err != nil {
+			t.Fatal(err)
+		}
+	}
+	if code, out, errb := runOut("{/region[12]}", "--no-shipped-data", "-d", dir); code != 0 || out != "Skåne län" {
+		t.Fatalf("{/region[12]} = %d, %q, stderr=%q", code, out, errb)
+	}
+	if code, out, _ := runOut("{/region[Skåne län].code}", "--no-shipped-data", "-d", dir); code != 0 || out != "12" {
+		t.Fatalf("{/region[Skåne län].code} = %d, %q", code, out)
+	}
+	if code, out, _ := runOut("{/region[12]}", "--no-shipped-data", "-d", dir, "--format", "sql"); code != 0 || out != `INSERT INTO "region" ("code", "name") VALUES ('12', 'Skåne län');`+"\n" {
+		t.Fatalf("--format sql {/region[12]} = %d, %q, want the table named without its selector", code, out)
+	}
+	if code, _, errb := runOut("{/region[99]}", "--no-shipped-data", "-d", dir); code != 2 || !strings.Contains(errb, `"99"`) {
+		t.Fatalf("{/region[99]} = %d, stderr=%q, want misuse naming the row", code, errb)
+	}
+	if code, out, _ := runOut("{/misc.territory[SE]}", "--seed", "1", "--format", "csv"); code != 0 || !strings.HasPrefix(out, "alpha2,") || !strings.Contains(out, "\nSE,SWE,") {
+		t.Fatalf("--format csv {/misc.territory[SE]} = %d, %q", code, out)
+	}
+}
+
+func TestUsageExamplesRenderWithNoData(t *testing.T) {
+	examples := regexp.MustCompile(`(?m)^  echo '([^']*)' \| COMMAND$`).FindAllStringSubmatch(usage+"\n  "+pipeHint("COMMAND")+"\n", -1)
+	if len(examples) < 3 {
+		t.Fatalf("usage and pipe hint hold %d echo examples, want the help's two and the hint's", len(examples))
+	}
+	for _, e := range examples {
+		if code, out, errb := runOut(e[1], "--no-shipped-data"); code != 0 || strings.TrimSpace(out) == "" {
+			t.Errorf("echo %q with no data: run = %d, %q, stderr %q", e[1], code, out, errb)
+		}
+	}
+	for _, token := range regexp.MustCompile(`\{/[^}]*\}`).FindAllString(usage, -1) {
+		if token != "{/path}" && token != "{/table[KEY]}" {
+			t.Errorf("usage names %s, a path a custom CLI may not carry", token)
+		}
+	}
+	if !strings.Contains(usage, "\nExit codes: 0 success, 1 runtime error, 2 misuse.\n") {
+		t.Errorf("usage does not list the exit codes")
+	}
+}
+
+func TestRunOperandMisuseBeforeLoad(t *testing.T) {
+	code, _, errb := runOut("", "--no-shipped-data", "x")
+	if code != 2 || !strings.Contains(errb, "unexpected argument") || strings.Contains(errb, "--data-path") {
+		t.Fatalf("an operand with no data = %d, %q; want the operand refused before any load", code, errb)
+	}
+}
+
+func TestRunInlineTemplate(t *testing.T) {
+	code, out, errb := runOut("name: {/sv_SE.person.last}", "--seed", "1")
+	if code != 0 || !strings.HasPrefix(out, "name: ") || strings.Contains(out, "{") {
+		t.Fatalf("inline format string = %d, %q, stderr %q", code, out, errb)
+	}
+	code, out, errb = runOut(`{"format":"name: {x}","x":["bosse","lina"]}`, "--seed", "1")
+	if code != 0 || (out != "name: bosse" && out != "name: lina") {
+		t.Fatalf("inline JSON template = %d, %q, want one name, stderr %q", code, out, errb)
+	}
+	code, out, errb = runOut(`"name: {/sv_SE.person.last}"`, "--seed", "1")
+	if code != 0 || !strings.HasPrefix(out, "name: ") || strings.Contains(out, "{") {
+		t.Fatalf("inline JSON string = %d, %q, stderr %q", code, out, errb)
+	}
+	code, out, errb = runOut("{digits(1)}\n", "--seed", "1", "-n", "2")
+	if code != 0 || len(strings.Split(strings.TrimRight(out, "\n"), "\n")) != 2 {
+		t.Fatalf("inline template with --repeat = %d, %q, stderr %q", code, out, errb)
+	}
+	code, out, errb = runOut("{/sv_SE.person} hihi", "--seed", "1")
+	if code != 0 || !strings.HasSuffix(out, " hihi") || strings.Contains(out, "{") {
+		t.Fatalf("a reference and text = %d, %q, stderr %q", code, out, errb)
+	}
+}
+
+func TestRunFormatTakesOnlyALoneReferenceAsARecord(t *testing.T) {
+	for _, stdin := range []string{"{//sv_SE.person}", "{/sv_SE.person|/misc.uuid}", "{/sv_SE.person as p}", "{/}", "{/.sv_SE.person}"} {
+		if code, out, errb := runOut(stdin, "--format", "json"); code != 2 || out != "" {
+			t.Errorf("run(%q, --format json) = %d, %q, %q; want misuse", stdin, code, out, errb)
+		}
+	}
+	if code, out, errb := runOut(`{"format":"x"}`); code != 0 || out != "x" {
+		t.Errorf(`run({"format":"x"}) = %d, %q, %q; want x`, code, out, errb)
+	}
+}
+
+func TestRunTemplateMisuse(t *testing.T) {
+	for stdin, want := range map[string]string{
+		"{bad":                  "unterminated",
+		"name: {/no.such.path}": "no entry",
+		"a } b":                 "}}",
+		"{//sv_SE.person}":      "write {/sv_SE.person}",
+	} {
+		code, out, errb := runOut(stdin, "--seed", "1")
+		if code != 2 || out != "" || !strings.Contains(errb, "try 'fejkdata --help'") || !strings.Contains(errb, want) {
+			t.Errorf("run(%q) = %d, %q, %q; want misuse naming %q and --help", stdin, code, out, errb, want)
+		}
+		if strings.Contains(errb, "fejkdata: fejkdata:") {
+			t.Errorf("run(%q) doubled the program prefix: %q", stdin, errb)
+		}
+	}
+}
+
+func TestRunNoShippedData(t *testing.T) {
+	code, list, errb := runOut("", "--no-shipped-data", "-d", misc, "--list")
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	if strings.Contains(list, "sv_SE") || !strings.Contains(list, "uuid\n") {
+		t.Errorf("--no-shipped-data --list = %q, want only the given dir", list)
+	}
+	code, out, _ := runOut("{/uuid}", "--no-shipped-data", "-d", misc, "-s", "3")
+	if code != 0 || strings.TrimSpace(out) == "" {
+		t.Errorf("run = %d, out=%q", code, out)
+	}
+	if code, out, errb := runOut("{digits(3)}", "--no-shipped-data"); code != 0 || len(out) != 3 {
+		t.Errorf("--no-shipped-data alone = %d, %q, %q, want a template reading no data rendered", code, out, errb)
+	}
+	if code, _, errb := runOut("{/sv_SE.person}", "--no-shipped-data"); code != 2 || !strings.HasSuffix(errb, "no data is loaded\ntry 'fejkdata --help'\n") {
+		t.Errorf("--no-shipped-data alone = %d, %q, want a reference to nothing refused", code, errb)
+	}
+}
+
+func TestRunRepeatOfZeroWritesNoRecord(t *testing.T) {
+	dir := recordDir(t)
+	for _, c := range []struct {
+		args []string
+		want string
+	}{
+		{nil, ""},
+		{[]string{"--format", "json"}, "[]\n"},
+		{[]string{"--format", "ndjson"}, ""},
+		{[]string{"--format", "csv"}, "first,last\n"},
+		{[]string{"--format", "sql"}, ""},
+	} {
+		if code, out, errb := runOut("{/users}", append(c.args, "-n", "0", "-d", dir)...); code != 0 || out != c.want {
+			t.Errorf("run(-n 0 %v) = %d, %q, %q, want %q", c.args, code, out, errb, c.want)
+		}
+	}
+	if code, _, _ := runOut("{/nope}", "-n", "0", "--format", "json", "-d", dir); code != 1 {
+		t.Errorf("run(-n 0) of a reference to nothing = %d, want 1", code)
+	}
+}
+
+func TestRunRepeatIsBounded(t *testing.T) {
+	for _, args := range [][]string{
+		{"--repeat", "9223372036854775807"},
+		{"--repeat", "1048577"},
+		{"-n", "99999999999"},
+	} {
+		code, out, errb := runOut("{/sv_SE.word}", args...)
+		if code != 2 || out != "" || !strings.Contains(errb, "0..1048576") {
+			t.Errorf("run(%v) = %d, %q, %q, want misuse naming the range", args, code, out, errb)
+		}
+	}
+	dir := t.TempDir()
+	if err := os.WriteFile(filepath.Join(dir, "x.json"), []byte(`"x"`), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	code, out, errb := runOut("{/x}", "--no-shipped-data", "-d", dir, "--repeat", "1048576", "--separator", "")
+	if code != 0 || len(out) != 1048576 {
+		t.Errorf("repeat at the cap = %d, %d bytes, stderr %q; want every render streamed", code, len(out), errb)
+	}
+}
+
+func TestRunIgnoresAFlagWithNoEffect(t *testing.T) {
+	dir := recordDir(t)
+	for _, c := range []struct {
+		stdin         string
+		base, ignored []string
+	}{
+		{"", []string{"--list"}, []string{"-n", "3"}},
+		{"", []string{"--list"}, []string{"--separator", ","}},
+		{"", []string{"--list"}, []string{"--format", "json"}},
+		{"", []string{"--list"}, []string{"--table", "t"}},
+		{"{/users}", nil, []string{"--table", "t"}},
+		{"{/users}", []string{"--format", "json"}, []string{"--table", "t"}},
+	} {
+		args := append([]string{"--seed", "1", "--data-path", dir}, c.base...)
+		_, want, _ := runOut(c.stdin, args...)
+		code, got, errb := runOut(c.stdin, append(args, c.ignored...)...)
+		if code != 0 || want == "" || got != want {
+			t.Errorf("run(%v + %v) = %d, %q, %q; want what %v alone prints, %q", c.base, c.ignored, code, got, errb, c.base, want)
+		}
+	}
+}
+
+func TestRunReadsABareNumberAsAFormatString(t *testing.T) {
+	if code, out, errb := runOut("42\n", "-n", "3"); code != 0 || out != "42\n42\n42\n" {
+		t.Errorf("echo 42 | fejkdata -n 3 = %d, %q, %q, want 42 on three lines", code, out, errb)
+	}
+}
+
+func TestRunRendersAnyJSONValue(t *testing.T) {
+	for stdin, want := range map[string]string{"42": "42", "42\n": "42\n", "true\n": "true\n", " 7 ": " 7 ", "null\n": "", " null ": ""} {
+		if code, out, errb := runOut(stdin); code != 0 || out != want {
+			t.Errorf("run(%q) = %d, %q, %q, want %q", stdin, code, out, errb, want)
+		}
+	}
+}
+
+func TestRunUnknownFlagIsNamedByRune(t *testing.T) {
+	code, _, errb := runOut("{/sv_SE.word}", "-ä")
+	if code != 2 || !strings.Contains(errb, "unknown flag -ä") {
+		t.Errorf("run(-ä) = %d, %q, want the flag named whole", code, errb)
+	}
+}
+
+func TestRunEmptyDataPathIsNamed(t *testing.T) {
+	code, _, errb := runOut("{/sv_SE.word}", "--data-path=")
+	if code != 1 || !strings.Contains(errb, "empty") {
+		t.Errorf("run(--data-path=) = %d, %q, want the empty path named", code, errb)
+	}
+}
+
+func recordDir(t *testing.T) string {
+	t.Helper()
+	dir := t.TempDir()
+	content := `{"format":"{first} {last}","first":["Ada","Bo"],"last":["Lovelace","Ek"]}`
+	if err := os.WriteFile(filepath.Join(dir, "users.json"), []byte(content), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	return dir
+}
+
+func TestRunRecordJSON(t *testing.T) {
+	code, out, errb := runOut("{/users}", "--seed", "1", "--format", "json", "--repeat", "2", "--data-path", recordDir(t))
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	var arr []map[string]string
+	if err := json.Unmarshal([]byte(out), &arr); err != nil {
+		t.Fatalf("json output is not one JSON array: %v\n%q", err, out)
+	}
+	if len(arr) != 2 || len(arr[0]) != 2 || arr[0]["first"] == "" || arr[0]["last"] == "" {
+		t.Fatalf("json output = %q, want an array of two records with first and last columns", out)
+	}
+}
+
+func TestRunRecordNDJSON(t *testing.T) {
+	code, out, errb := runOut("{/users}", "--seed", "1", "--format", "ndjson", "--repeat", "2", "--data-path", recordDir(t))
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	lines := strings.Split(strings.TrimRight(out, "\n"), "\n")
+	if len(lines) != 2 {
+		t.Fatalf("ndjson output = %d lines %q, want one object per line", len(lines), out)
+	}
+	for _, line := range lines {
+		var m map[string]string
+		if err := json.Unmarshal([]byte(line), &m); err != nil || len(m) != 2 {
+			t.Fatalf("ndjson line %q is not one object: %v", line, err)
+		}
+	}
+}
+
+func TestRunRecordCSVRoundTrips(t *testing.T) {
+	code, out, errb := runOut("{/users}\r\n", "--seed", "1", "--format", "csv", "--repeat", "3", "--data-path", recordDir(t))
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	lines := strings.Split(strings.TrimRight(out, "\n"), "\n")
+	if len(lines) != 4 {
+		t.Fatalf("csv output = %d lines %q, want a header and 3 rows", len(lines), out)
+	}
+	if lines[0] != "first,last" {
+		t.Fatalf("csv header = %q, want first,last", lines[0])
+	}
+	r, err := csv.NewReader(strings.NewReader(out)).ReadAll()
+	if err != nil {
+		t.Fatalf("csv output is not valid CSV: %v", err)
+	}
+	if len(r) != 4 || len(r[0]) != 2 {
+		t.Fatalf("csv parsed to %v, want 4 records of 2 fields", r)
+	}
+}
+
+func TestRunRecordSQL(t *testing.T) {
+	code, out, errb := runOut("{/users}", "--seed", "1", "--format", "sql", "--repeat", "2", "--table", "people", "--data-path", recordDir(t))
+	if code != 0 {
+		t.Fatalf("run = %d, stderr=%q", code, errb)
+	}
+	lines := strings.Split(strings.TrimRight(out, "\n"), "\n")
+	if len(lines) != 2 || !strings.HasPrefix(lines[0], `INSERT INTO "people" ("first", "last") VALUES (`) {
+		t.Fatalf("sql output = %q, want two INSERT INTO people statements", out)
+	}
+}
+
+func TestRunRecordDefaultTable(t *testing.T) {
+	code, out, errb := runOut("{/users}", "--seed", "1", "--format", "sql", "--data-path", recordDir(t))
+	if code != 0 || !strings.HasPrefix(out, `INSERT INTO "users" (`) {
+		t.Fatalf("default table = %d, %q, want the reference's last segment; stderr %q", code, out, errb)
+	}
+}
+
+func TestRunRecordOfAQuotedLoneReference(t *testing.T) {
+	code, out, errb := runOut(`"{/users}"`, "--seed", "1", "--format", "sql", "--data-path", recordDir(t))
+	if code != 0 || !strings.HasPrefix(out, `INSERT INTO "users" (`) {
+		t.Fatalf("a JSON string holding {/users} = %d, %q, %q; want the record users", code, out, errb)
+	}
+	code, out, errb = runOut("{/users} ", "--format", "csv", "--data-path", recordDir(t))
+	if code != 2 || out != "" || !strings.Contains(errb, "whitespace around {/users}") {
+		t.Errorf("{/users} and a space = %d, %q, %q; want misuse naming the whitespace", code, out, errb)
+	}
+}
+
+func TestRunRecordInlineTemplate(t *testing.T) {
+	code, out, errb := runOut(`{"format":"{x}","x":["a","b"]}`, "--seed", "1", "--format", "json")
+	if code != 0 {
+		t.Fatalf("inline record = %d, stderr=%q", code, errb)
+	}
+	var arr []map[string]string
+	if err := json.Unmarshal([]byte(out), &arr); err != nil || len(arr) != 1 || (arr[0]["x"] != "a" && arr[0]["x"] != "b") {
+		t.Fatalf("inline record json = %q, want one record with a column x (err %v)", out, err)
+	}
+}
+
+func TestRunRecordMisuse(t *testing.T) {
+	for _, c := range []struct {
+		args []string
+		want string
+	}{
+		{[]string{"--format", "yaml"}, "--format takes csv, json, ndjson, sql or text"},
+		{[]string{"--format", "sql", "--table", ""}, "--table names the INSERT target"},
+		{[]string{"--format", "csv", "--separator", ";"}, `--separator joins text renders; --format csv always separates fields with ","`},
+		{[]string{"--format", "json", "--separator", ","}, "--separator joins text renders"},
+		{[]string{"--format", "ndjson", "--separator", ","}, "--separator joins text renders"},
+		{[]string{"--format", "sql", "--separator", ","}, "--separator joins text renders"},
+	} {
+		code, out, errb := runOut("{/users}", c.args...)
+		if code != 2 || out != "" || !strings.Contains(errb, c.want) {
+			t.Errorf("run(%v) = %d, %q, %q; want misuse naming %q", c.args, code, out, errb, c.want)
+		}
+	}
+}
+
+func TestRunRecordOnABareValueIsRuntimeError(t *testing.T) {
+	dir := t.TempDir()
+	if err := os.WriteFile(filepath.Join(dir, "word.json"), []byte(`["a","b"]`), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	code, _, errb := runOut("{/word}", "--format", "json", "--data-path", dir)
+	if code != 1 {
+		t.Fatalf("record on a choice = %d, want exit 1 (runtime error), stderr %q", code, errb)
+	}
+}
+
+func TestRunRecordRejectsATopLevelRepeat(t *testing.T) {
+	dir := t.TempDir()
+	content := `{"format":"{a}-","repeat":3,"separator":"|","a":["x","y"]}`
+	if err := os.WriteFile(filepath.Join(dir, "rep.json"), []byte(content), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	code, _, errb := runOut("{/rep}", "--format", "json", "--data-path", dir)
+	if code != 1 || !strings.Contains(errb, "carries repeat 3") {
+		t.Errorf("record on a repeating path = %d, %q; want exit 1 naming the repeat", code, errb)
+	}
+	code, _, errb = runOut(content, "--format", "json")
+	if code != 2 || !strings.Contains(errb, "carries repeat 3") {
+		t.Errorf("record on a repeating inline template = %d, %q; want exit 2 naming the repeat", code, errb)
+	}
+	if code, out, _ := runOut("{/rep}", "--seed", "1", "--data-path", dir); code != 0 || strings.TrimSpace(out) != "x-|x-|x-" {
+		t.Errorf("text view = %d, %q; want the repeat still composed", code, out)
+	}
+}
+
+func TestRunNumbersFollowTheShell(t *testing.T) {
+	_, want, _ := runOut("{/sv_SE.word}", "--seed", "7", "--repeat", "3")
+	code, got, errb := runOut("{/sv_SE.word}", "--seed", "007", "--repeat", "+3")
+	if code != 0 || got != want {
+		t.Errorf("run(--seed 007 --repeat +3) = %d, %q, stderr %q; want the same as --seed 7 --repeat 3 %q", code, got, errb, want)
+	}
+}
+
+// TestRunRecordOfAFieldlessPath pins the way out of asking a category of one value
+// for columns: the record to write, and the quoting a shell needs to take it. The
+// template is well-formed and only the data cannot serve it, so the code is 1.
+func TestRunRecordOfAFieldlessPath(t *testing.T) {
+	code, out, errb := runOut("{/sv_SE.personnummer}", "--format", "csv")
+	if code != 1 {
+		t.Fatalf("run(--format csv) of {/sv_SE.personnummer} = %d, want 1", code)
+	}
+	for _, want := range []string{`{"format":"","personnummer":"{/sv_SE.personnummer}"}`, "single quotes"} {
+		if !strings.Contains(errb, want) {
+			t.Errorf("stderr = %q, want it to name %s", errb, want)
+		}
+	}
+	if out != "" {
+		t.Errorf("stdout = %q, want nothing", out)
+	}
+}
+
+func TestADataPathCategoryFailingAtFirstReachExits1(t *testing.T) {
+	dir := t.TempDir()
+	for name, data := range map[string]string{
+		".fejkdata.json": `{"index": {"bad": {"paths": [""]}}}`,
+		"bad.json":       `"{/nope}"`,
+	} {
+		if err := os.WriteFile(filepath.Join(dir, name), []byte(data), 0o644); err != nil {
+			t.Fatal(err)
+		}
+	}
+	for _, args := range [][]string{{"-d", dir}, {"-d", dir, "--format", "json"}} {
+		if code, _, errb := runOut("{/bad}", args...); code != 1 || !strings.Contains(errb, "{/nope}") {
+			t.Errorf("%v on a category failing to load = %d, %q; want exit 1 naming the reference", args, code, errb)
+		}
+	}
+}

cli/lone_test.go

@@ -0,0 +1,52 @@
+package cli
+
+import (
+	"testing"
+
+	"github.com/larvit/fejkdata/internal/grammar"
+)
+
+// grammarLoneReference is the path a format reads when the grammar parses it as one path
+// read of one reference from the root, and nothing else.
+func grammarLoneReference(text string) (string, bool) {
+	toks, err := grammar.ParseFormat(text)
+	if err != nil || len(toks) != 1 || toks[0].Kind != grammar.PathRead || len(toks[0].Arms) != 1 {
+		return "", false
+	}
+	sigil, path, err := grammar.RefShape(toks[0].Arms[0])
+	if err != nil || sigil != "/" {
+		return "", false
+	}
+	return path, true
+}
+
+func TestLoneReferenceAgreesWithTheGrammar(t *testing.T) {
+	for _, text := range []string{
+		" {/x}",
+		"x",
+		"{.x}",
+		"{/.x}",
+		"{//x}",
+		"{/a...b}",
+		"{/a..b}",
+		"{/a..}",
+		"{/a.[x]}",
+		"{/a.}",
+		"{/a|/b}",
+		"{/a}{/b}",
+		"{/misc.currency[US Dollar (Next day)]}",
+		"{/sv_SE.person}",
+		"{/t[a as b].c}",
+		"{/t[f(x)]}",
+		"{/x as p}",
+		"{/x(1)}",
+		"{/x}}",
+		"{/}",
+	} {
+		path, lone := loneReference(text)
+		want, wantLone := grammarLoneReference(text)
+		if lone != wantLone || path != want {
+			t.Errorf("loneReference(%q) = %q, %v; the grammar reads %q, %v", text, path, lone, want, wantLone)
+		}
+	}
+}

cmd/fejkdata/main.go

@@ -1,277 +1,12 @@
-// Command fejkdata prints fake values from the shipped data and any directories
-// layered over it.
-//
-//	fejkdata sv_SE.person                        # a full person
-//	fejkdata sv_SE.person.last                   # just the surname
-//	fejkdata --data-path ./mydata sv_SE.person   # layer custom data; the last dir wins
-//	fejkdata --seed 42 sv_SE.address
+// Command fejkdata renders the template on its stdin with fake values from every shipped
+// module; package cli documents it.
 package main
 
 import (
-	"errors"
-	"fmt"
-	"io"
 	"os"
-	"runtime/debug"
-	"strconv"
-	"strings"
 
-	"gitea.larvit.se/larvit/fejkdata"
+	"github.com/larvit/fejkdata/cli"
+	"github.com/larvit/fejkdata/data"
 )
 
-const usage = `Usage: fejkdata [flags] <path>
-
-  <path>                 a category, or a dotted path into one (person, person.last)
-
-  -d, --data-path D      a data directory to layer over the shipped data (repeatable; last wins on a clash)
-  -h, --help             print this help, then exit
-      --list             list the paths the data offers, then exit
-      --no-shipped-data  load only the --data-path directories
-  -n, --repeat N         render the path N times (default 1)
-  -s, --seed N           seed for reproducible output
-      --separator S      string between repeated values (default newline)
-      --version          print the version, then exit
-
-Flags may come before or after <path>; -- ends the flags. A short flag's value
-attaches or follows (-n3, -n 3); short flags bundle (-hn 3).
-`
-
-type invocation struct {
-	dirs      []string
-	help      bool
-	list      bool
-	noShipped bool
-	paths     []string
-	repeat    int
-	seed      uint64
-	seeded    bool
-	separator string
-	version   bool
-}
-
-type flagDef struct {
-	long  string
-	short string
-	value bool
-	set   func(*invocation, string) error
-}
-
-var flagDefs = []flagDef{
-	{"data-path", "d", true, func(in *invocation, v string) error { in.dirs = append(in.dirs, v); return nil }},
-	{"help", "h", false, func(in *invocation, _ string) error { in.help = true; return nil }},
-	{"list", "", false, func(in *invocation, _ string) error { in.list = true; return nil }},
-	{"no-shipped-data", "", false, func(in *invocation, _ string) error { in.noShipped = true; return nil }},
-	{"repeat", "n", true, func(in *invocation, v string) error {
-		n, err := strconv.Atoi(v)
-		if err != nil || n < 1 {
-			return fmt.Errorf("--repeat needs a positive integer, got %q", v)
-		}
-		in.repeat = n
-		return nil
-	}},
-	{"seed", "s", true, func(in *invocation, v string) error {
-		n, err := strconv.ParseUint(v, 10, 64)
-		if err != nil {
-			return fmt.Errorf("--seed needs an unsigned integer, got %q", v)
-		}
-		in.seed, in.seeded = n, true
-		return nil
-	}},
-	{"separator", "", true, func(in *invocation, v string) error { in.separator = v; return nil }},
-	{"version", "", false, func(in *invocation, _ string) error { in.version = true; return nil }},
-}
-
-func flagByLong(name string) *flagDef {
-	for i := range flagDefs {
-		if flagDefs[i].long == name {
-			return &flagDefs[i]
-		}
-	}
-	return nil
-}
-
-func flagByShort(name string) *flagDef {
-	for i := range flagDefs {
-		if flagDefs[i].short == name {
-			return &flagDefs[i]
-		}
-	}
-	return nil
-}
-
-// flagArg is one flag as written: its definition and the value attached to it,
-// if any.
-type flagArg struct {
-	def    *flagDef
-	value  string
-	inline bool
-}
-
-// splitFlags reads one argv element as flags: --name or --name=value, or -abc
-// where each letter is a short flag and the first that takes a value takes the
-// rest of the element.
-func splitFlags(arg string) ([]flagArg, error) {
-	if strings.HasPrefix(arg, "--") {
-		name, value, inline := strings.Cut(arg[2:], "=")
-		def := flagByLong(name)
-		if def == nil {
-			return nil, fmt.Errorf("unknown flag --%s", name)
-		}
-		return []flagArg{{def, value, inline}}, nil
-	}
-	if name, _, _ := strings.Cut(arg[1:], "="); len(name) > 1 && flagByLong(name) != nil {
-		return nil, fmt.Errorf("unknown flag %s; use --%s", arg, name)
-	}
-	var flags []flagArg
-	letters := arg[1:]
-	for i := 0; i < len(letters); i++ {
-		letter := letters[i : i+1]
-		def := flagByShort(letter)
-		if def == nil {
-			return nil, fmt.Errorf("unknown flag -%s", letter)
-		}
-		if !def.value {
-			flags = append(flags, flagArg{def: def})
-			continue
-		}
-		rest := letters[i+1:]
-		if strings.HasPrefix(rest, "=") {
-			return nil, fmt.Errorf("-%s takes its value attached (-%s%s) or next (-%s %s); = belongs to --%s=%s",
-				letter, letter, rest[1:], letter, rest[1:], def.long, rest[1:])
-		}
-		return append(flags, flagArg{def, rest, rest != ""}), nil
-	}
-	return flags, nil
-}
-
-func parseArgs(argv []string) (invocation, error) {
-	in := invocation{repeat: 1, separator: "\n"}
-	for i := 0; i < len(argv); i++ {
-		arg := argv[i]
-		if arg == "--" {
-			in.paths = append(in.paths, argv[i+1:]...)
-			return in, nil
-		}
-		if len(arg) < 2 || arg[0] != '-' {
-			in.paths = append(in.paths, arg)
-			continue
-		}
-		flags, err := splitFlags(arg)
-		if err != nil {
-			return in, err
-		}
-		for _, fl := range flags {
-			if !fl.def.value {
-				if fl.inline {
-					return in, fmt.Errorf("--%s takes no value", fl.def.long)
-				}
-				_ = fl.def.set(&in, "")
-				continue
-			}
-			value := fl.value
-			if !fl.inline {
-				if i++; i >= len(argv) {
-					return in, fmt.Errorf("--%s needs a value", fl.def.long)
-				}
-				value = argv[i]
-			}
-			if err := fl.def.set(&in, value); err != nil {
-				return in, err
-			}
-		}
-	}
-	return in, nil
-}
-
-// check rejects a flag combination that cannot run.
-func (in invocation) check() error {
-	if in.list && len(in.paths) > 0 {
-		return errors.New("--list takes no path")
-	}
-	if !in.list && len(in.paths) != 1 {
-		return fmt.Errorf("expected one path, got %d", len(in.paths))
-	}
-	return nil
-}
-
-func (in invocation) options() []fejkdata.Option {
-	var opts []fejkdata.Option
-	if in.noShipped {
-		opts = append(opts, fejkdata.WithoutShippedData())
-	}
-	for _, dir := range in.dirs {
-		opts = append(opts, fejkdata.WithDataPath(dir))
-	}
-	if in.seeded {
-		opts = append(opts, fejkdata.WithSeed(in.seed))
-	}
-	return opts
-}
-
-// values renders the path repeat times, joined by the separator.
-func (in invocation) values(f *fejkdata.Generator) (string, error) {
-	vals := make([]string, in.repeat)
-	for i := range vals {
-		v, err := f.Fake(in.paths[0])
-		if err != nil {
-			return "", err
-		}
-		vals[i] = v
-	}
-	return strings.Join(vals, in.separator), nil
-}
-
-func main() { os.Exit(run(os.Args[1:], os.Stdout, os.Stderr)) }
-
-// run returns the exit code: 0 ok, 1 runtime error, 2 misuse.
-func run(args []string, stdout, stderr io.Writer) int {
-	in, err := parseArgs(args)
-	if err != nil {
-		return misuse(stderr, err)
-	}
-	if in.help {
-		fmt.Fprint(stdout, usage)
-		return 0
-	}
-	if in.version {
-		fmt.Fprintln(stdout, "fejkdata "+buildVersion())
-		return 0
-	}
-	if err := in.check(); err != nil {
-		return misuse(stderr, err)
-	}
-	f, err := fejkdata.New(in.options()...)
-	if errors.Is(err, fejkdata.ErrNoData) {
-		return misuse(stderr, errors.New("--no-shipped-data needs at least one --data-path"))
-	}
-	if err != nil {
-		fmt.Fprintln(stderr, err)
-		return 1
-	}
-	if in.list {
-		for _, p := range f.List() {
-			fmt.Fprintln(stdout, p)
-		}
-		return 0
-	}
-	out, err := in.values(f)
-	if err != nil {
-		fmt.Fprintln(stderr, err)
-		return 1
-	}
-	fmt.Fprintln(stdout, out)
-	return 0
-}
-
-func misuse(stderr io.Writer, err error) int {
-	fmt.Fprintf(stderr, "fejkdata: %v\ntry 'fejkdata --help'\n", err)
-	return 2
-}
-
-// buildVersion is the module version go install stamps into the binary, or "devel".
-func buildVersion() string {
-	if info, ok := debug.ReadBuildInfo(); ok && info.Main.Version != "" {
-		return info.Main.Version
-	}
-	return "devel"
-}
+func main() { os.Exit(cli.Main(data.Modules()...)) }

cmd/fejkdata/main_test.go

@@ -2,317 +2,46 @@ package main
 
 import (
 	"bytes"
+	"os"
+	"os/exec"
 	"strings"
 	"testing"
 )
 
-const (
-	svSE = "../../data/sv_SE"
-	enUS = "../../data/en_US"
-)
-
-func runOut(args ...string) (int, string, string) {
-	var out, errb bytes.Buffer
-	code := run(args, &out, &errb)
-	return code, out.String(), errb.String()
+func TestMain(m *testing.M) {
+	if os.Getenv("FEJKDATA_RUN_MAIN") == "1" {
+		main()
+	}
+	os.Exit(m.Run())
 }
 
-func TestRunOutputsValue(t *testing.T) {
-	code, out, errb := runOut("--data-path", svSE, "person")
-	if code != 0 {
-		t.Fatalf("run = %d, stderr=%q", code, errb)
-	}
-	if strings.TrimSpace(out) == "" {
-		t.Fatalf("empty output, stderr=%q", errb)
-	}
-	if !strings.HasSuffix(out, "\n") {
-		t.Errorf("output should end with newline, got %q", out)
+func TestMainRendersTheShippedData(t *testing.T) {
+	cmd := exec.Command(os.Args[0], "--seed", "1")
+	cmd.Env = append(os.Environ(), "FEJKDATA_RUN_MAIN=1")
+	cmd.Stdin = strings.NewReader("{/misc.territory[SE].alpha2} {/en_US.person.last} {/sv_SE.address.locality}")
+	var stderr bytes.Buffer
+	cmd.Stderr = &stderr
+	out, err := cmd.Output()
+	if fields := strings.Fields(string(out)); err != nil || len(fields) < 3 || fields[0] != "SE" {
+		t.Errorf("fejkdata = %v, %q, %q, want SE followed by a last name and a locality", err, out, stderr.String())
 	}
 }
 
-func TestRunDotPath(t *testing.T) {
-	code, full, _ := runOut("--seed", "7", "--data-path", svSE, "person")
-	if code != 0 {
-		t.Fatalf("person run = %d", code)
-	}
-	code, last, errb := runOut("--seed", "7", "--data-path", svSE, "person.last")
-	if code != 0 {
-		t.Fatalf("person.last run = %d, stderr=%q", code, errb)
-	}
-	if strings.TrimSpace(last) == "" {
-		t.Fatal("empty person.last output")
-	}
-	if last == full {
-		t.Errorf("person.last %q should differ from person %q", last, full)
-	}
-}
-
-func TestRunSeedSpellings(t *testing.T) {
-	_, want, _ := runOut("--seed", "42", "--data-path", svSE, "address")
-	for _, args := range [][]string{
-		{"--seed=42", "--data-path", svSE, "address"},
-		{"-s", "42", "-d", svSE, "address"},
-		{"--data-path", svSE, "address", "--seed", "42"},
-	} {
-		code, got, errb := runOut(args...)
-		if code != 0 {
-			t.Fatalf("run(%v) = %d, stderr=%q", args, code, errb)
-		}
-		if got != want {
-			t.Errorf("run(%v) = %q, want %q", args, got, want)
+func TestMainWithoutASeedDrawsANewSeedEachRun(t *testing.T) {
+	var outs [2]string
+	for i := range outs {
+		cmd := exec.Command(os.Args[0])
+		cmd.Env = append(os.Environ(), "FEJKDATA_RUN_MAIN=1")
+		cmd.Stdin = strings.NewReader("{uuid()}")
+		var stderr bytes.Buffer
+		cmd.Stderr = &stderr
+		out, err := cmd.Output()
+		if err != nil || len(out) == 0 {
+			t.Fatalf("fejkdata without --seed = %v, %q, %q", err, out, stderr.String())
 		}
-	}
-}
-
-func TestRunShortFlagValues(t *testing.T) {
-	_, want, _ := runOut("--seed", "42", "--data-path", svSE, "address")
-	for _, args := range [][]string{
-		{"-s42", "-d", svSE, "address"},
-		{"-s", "42", "-d" + svSE, "address"},
-		{"-d", svSE, "address", "-s42"},
-	} {
-		code, got, errb := runOut(args...)
-		if code != 0 {
-			t.Fatalf("run(%v) = %d, stderr=%q", args, code, errb)
-		}
-		if got != want {
-			t.Errorf("run(%v) = %q, want %q", args, got, want)
-		}
-	}
-	_, three, _ := runOut("-s", "1", "-n3", "-d", svSE, "word")
-	if lines := strings.Split(strings.TrimRight(three, "\n"), "\n"); len(lines) != 3 {
-		t.Errorf("-n3 gave %d lines: %q", len(lines), three)
-	}
-	for _, c := range []struct {
-		args []string
-		want string
-	}{
-		{[]string{"-s=42", "-d", svSE, "address"}, "-s42"},
-		{[]string{"-s=42", "-d", svSE, "address"}, "--seed=42"},
-		{[]string{"-d=" + svSE, "address"}, "--data-path="},
-		{[]string{"-nd", "3", "-d", svSE, "word"}, `--repeat needs a positive integer, got "d"`},
-		{[]string{"--seed=", "-d", svSE, "word"}, `--seed needs an unsigned integer, got ""`},
-	} {
-		code, out, errb := runOut(c.args...)
-		if code != 2 || out != "" {
-			t.Errorf("run(%v) = %d, stdout %q, want misuse", c.args, code, out)
-		}
-		if !strings.Contains(errb, c.want) {
-			t.Errorf("run(%v) stderr = %q, want %q", c.args, errb, c.want)
-		}
-	}
-	code, out, _ := runOut("-hd", svSE)
-	if code != 0 || !strings.Contains(out, "Usage") {
-		t.Errorf("-hd (bundled help) = %d, %q, want usage", code, out)
-	}
-}
-
-func TestRunDoubleDashEndsFlags(t *testing.T) {
-	code, out, errb := runOut("--data-path", svSE, "--", "person")
-	if code != 0 || strings.TrimSpace(out) == "" {
-		t.Fatalf("run = %d, out=%q, stderr=%q", code, out, errb)
-	}
-	code, _, errb = runOut("--data-path", svSE, "--", "--list")
-	if code != 1 || !strings.Contains(errb, "--list") {
-		t.Errorf("after --, --list should be a path: code %d, stderr %q", code, errb)
-	}
-}
-
-func TestRunSingleDashLongIsRejected(t *testing.T) {
-	for _, c := range []struct {
-		args []string
-		want string
-	}{
-		{[]string{"-seed", "42", "-d", svSE, "person"}, "use --seed"},
-		{[]string{"-seed=42", "-d", svSE, "person"}, "use --seed"},
-		{[]string{"-data-path", svSE, "person"}, "use --data-path"},
-		{[]string{"-list", "-d", svSE}, "use --list"},
-		{[]string{"--nope", "-d", svSE, "person"}, "unknown flag --nope"},
-		{[]string{"-x", "-d", svSE, "person"}, "unknown flag -x"},
-	} {
-		code, _, errb := runOut(c.args...)
-		if code != 2 {
-			t.Errorf("run(%v) = %d, want 2", c.args, code)
-		}
-		if !strings.Contains(errb, c.want) {
-			t.Errorf("run(%v) stderr = %q, want %q", c.args, errb, c.want)
-		}
-	}
-}
-
-func TestRunFlagValues(t *testing.T) {
-	for _, c := range []struct {
-		args []string
-		want string
-	}{
-		{[]string{"--seed", "-d", svSE, "person"}, `--seed needs an unsigned integer, got "-d"`},
-		{[]string{"-d", svSE, "person", "--seed"}, "--seed needs a value"},
-		{[]string{"--seed", "x", "-d", svSE, "person"}, "--seed"},
-		{[]string{"--list=1", "-d", svSE}, "--list takes no value"},
-		{[]string{"--repeat", "0", "-d", svSE, "person"}, "--repeat"},
-		{[]string{"-n", "-1", "-d", svSE, "person"}, "--repeat"},
-	} {
-		code, _, errb := runOut(c.args...)
-		if code != 2 {
-			t.Errorf("run(%v) = %d, want 2", c.args, code)
-		}
-		if !strings.Contains(errb, c.want) {
-			t.Errorf("run(%v) stderr = %q, want %q", c.args, errb, c.want)
-		}
-	}
-}
-
-func TestRunRepeat(t *testing.T) {
-	code, out, errb := runOut("--seed", "1", "--repeat", "3", "--data-path", svSE, "word")
-	if code != 0 {
-		t.Fatalf("run = %d, stderr=%q", code, errb)
-	}
-	lines := strings.Split(strings.TrimRight(out, "\n"), "\n")
-	if len(lines) != 3 {
-		t.Errorf("repeat=3 gave %d lines: %q", len(lines), out)
-	}
-	_, short, _ := runOut("--seed", "1", "-n", "3", "-d", svSE, "word")
-	if short != out {
-		t.Errorf("-n 3 = %q, want the same as --repeat 3 %q", short, out)
-	}
-}
-
-func TestRunSeparator(t *testing.T) {
-	code, out, errb := runOut("--repeat", "3", "--separator", ",", "--data-path", svSE, "word")
-	if code != 0 {
-		t.Fatalf("run = %d, stderr=%q", code, errb)
-	}
-	if n := strings.Count(out, "\n"); n != 1 {
-		t.Errorf("want one trailing newline, got %d: %q", n, out)
-	}
-	if !strings.Contains(out, ",") {
-		t.Errorf("values should be comma-joined: %q", out)
-	}
-}
-
-func TestRunRepeatAdvancesRNG(t *testing.T) {
-	code, out, errb := runOut("--seed", "1", "--repeat", "5", "--data-path", svSE, "person")
-	if code != 0 {
-		t.Fatalf("run = %d, stderr=%q", code, errb)
-	}
-	uniq := map[string]bool{}
-	for _, l := range strings.Split(strings.TrimRight(out, "\n"), "\n") {
-		uniq[l] = true
-	}
-	if len(uniq) < 2 {
-		t.Errorf("repeat should vary output, all identical: %q", out)
-	}
-}
-
-func TestRunMisuse(t *testing.T) {
-	for _, args := range [][]string{{}, {"--data-path", svSE}, {"-d", svSE, "person", "word"}, {"-d", svSE, "--list", "person"}, {"--no-shipped-data", "sv_SE.person"}} {
-		code, out, errb := runOut(args...)
-		if code != 2 {
-			t.Errorf("run(%v) = %d, want 2", args, code)
-		}
-		if !strings.Contains(errb, "try 'fejkdata --help'") {
-			t.Errorf("run(%v) stderr = %q, want a pointer to --help", args, errb)
-		}
-		if out != "" {
-			t.Errorf("run(%v) stdout = %q, want nothing", args, out)
-		}
-	}
-}
-
-func TestRunMultipleDataPaths(t *testing.T) {
-	code, out, errb := runOut("-d", enUS, "--data-path", svSE, "person")
-	if code != 0 {
-		t.Fatalf("run(multi-dir) = %d, stderr=%q", code, errb)
-	}
-	if strings.TrimSpace(out) == "" {
-		t.Fatal("empty output for multi-dir run")
-	}
-}
-
-func TestRunUnknownCategoryFails(t *testing.T) {
-	code, _, errb := runOut("--data-path", svSE, "nope")
-	if code != 1 {
-		t.Fatalf("run = %d, want 1", code)
-	}
-	if !strings.Contains(errb, "nope") {
-		t.Errorf("stderr %q should name the unknown category", errb)
-	}
-}
-
-func TestRunList(t *testing.T) {
-	code, out, errb := runOut("--data-path", svSE, "--list")
-	if code != 0 {
-		t.Fatalf("run = %d, stderr=%q", code, errb)
-	}
-	for _, want := range []string{"person", "person.last", "address"} {
-		if !strings.Contains(out, want) {
-			t.Errorf("list output missing %q:\n%s", want, out)
-		}
-	}
-}
-
-func TestRunHelpOnStdout(t *testing.T) {
-	for _, h := range []string{"-h", "--help"} {
-		code, out, errb := runOut(h)
-		if code != 0 {
-			t.Errorf("%s = %d, want 0", h, code)
-		}
-		if !strings.Contains(out, "Usage") || errb != "" {
-			t.Errorf("%s: stdout = %q, stderr = %q, want usage on stdout only", h, out, errb)
-		}
-	}
-}
-
-func TestRunVersion(t *testing.T) {
-	code, out, errb := runOut("--version")
-	if code != 0 {
-		t.Fatalf("--version = %d, stderr=%q", code, errb)
-	}
-	version, ok := strings.CutPrefix(out, "fejkdata ")
-	if !ok || strings.TrimSpace(version) == "" {
-		t.Errorf("--version = %q, want the command name and a version on stdout", out)
-	}
-}
-
-func TestRunMissingDirFails(t *testing.T) {
-	code, _, errb := runOut("--data-path", "../../data/nope", "person")
-	if code != 1 {
-		t.Fatalf("run = %d, want 1", code)
-	}
-	if errb == "" {
-		t.Error("want an error message on stderr")
-	}
-}
-
-func TestRunShippedDataByDefault(t *testing.T) {
-	code, out, errb := runOut("--seed", "1", "sv_SE.person")
-	if code != 0 || strings.TrimSpace(out) == "" {
-		t.Fatalf("run = %d, out=%q, stderr=%q", code, out, errb)
-	}
-	code, list, _ := runOut("--list")
-	if code != 0 || !strings.Contains(list, "en_US.person\n") || !strings.Contains(list, "misc.uuid\n") {
-		t.Errorf("--list without --data-path = %d, %q", code, list)
-	}
-	code, _, errb = runOut("person")
-	if code != 1 || !strings.Contains(errb, "person") {
-		t.Errorf("a category outside the shipped tree: code %d, stderr %q", code, errb)
-	}
-}
-
-func TestRunNoShippedData(t *testing.T) {
-	code, list, errb := runOut("--no-shipped-data", "-d", svSE, "--list")
-	if code != 0 {
-		t.Fatalf("run = %d, stderr=%q", code, errb)
-	}
-	if strings.Contains(list, "en_US") || !strings.Contains(list, "person\n") {
-		t.Errorf("--no-shipped-data --list = %q, want only the given dir", list)
-	}
-	code, out, _ := runOut("--no-shipped-data", "-d", svSE, "-s", "3", "person")
-	if code != 0 || strings.TrimSpace(out) == "" {
-		t.Errorf("run = %d, out=%q", code, out)
-	}
-	code, _, errb = runOut("--no-shipped-data", "sv_SE.person")
-	if code != 2 || !strings.Contains(errb, "--no-shipped-data needs at least one --data-path") {
-		t.Errorf("--no-shipped-data alone = %d, %q, want misuse naming --data-path", code, errb)
+		outs[i] = string(out)
+	}
+	if outs[0] == outs[1] {
+		t.Errorf("two runs without --seed both printed %q, want a new seed each run", outs[0])
 	}
 }

compose.yaml

@@ -1,21 +1,5 @@
-# Docker-based dev workflow — no local tooling beyond Docker is required.
-#
-#   docker compose run --rm test
-#   docker compose run --rm vet
-#   docker compose run --rm build
-#
-# The full gate — vet, formatting and tests — is `docker build .`, which is what
-# CI runs.
-#
-# Build caches persist in the `gocache` volume, so re-runs are fast and the
-# host tree stays clean. Commands that rewrite source (fmt, tidy) keep your
-# ownership when run with `--user "$(id -u):$(id -g)"`.
-#
-# GO_VERSION picks the toolchain image (default: latest stable). Test the lowest
-# supported Go with: GO_VERSION=1.22 docker compose run --rm test
-
 x-go: &go
-  image: golang:${GO_VERSION:-1.26.4}
+  image: golang:${GO_VERSION:-1.27.2}
   working_dir: /app
   volumes:
     - .:/app
@@ -24,6 +8,7 @@ x-go: &go
     HOME: /cache
     GOCACHE: /cache/build
     GOMODCACHE: /cache/mod
+    REPIN: ${REPIN:-}
 
 services:
   test:
@@ -32,16 +17,28 @@ services:
 
   cover:
     <<: *go
-    command: go test -cover ./...
+    command: sh -c "go test -coverpkg=./... -coverprofile=/tmp/cover.out ./... && go tool cover -func=/tmp/cover.out | tail -1"
 
   bench:
     <<: *go
     command: go test -run XXX -bench . -benchmem ./...
 
+  generate:
+    <<: *go
+    command: go generate ./...
+
   vet:
     <<: *go
     command: go vet ./...
 
+  cyclo:
+    <<: *go
+    command: go run github.com/fzipp/gocyclo/cmd/gocyclo@v0.6.0 -over 14 -ignore _test .
+
+  vulncheck:
+    <<: *go
+    command: go run golang.org/x/vuln/cmd/govulncheck@v1.8.0 -scan module
+
   build:
     <<: *go
     command: go build ./...
@@ -54,10 +51,16 @@ services:
     <<: *go
     command: go mod tidy
 
-  # Interactive shell for ad-hoc work: docker compose run --rm dev
   dev:
     <<: *go
     command: sh
 
+  data-import:
+    image: python:3.14.8-slim
+    working_dir: /app
+    volumes:
+      - .:/app
+    entrypoint: python3
+
 volumes:
   gocache:

data-import/currency.py

@@ -0,0 +1,68 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/currency.tsv from datasets/currency-codes (PDDL) and CLDR's symbols (Unicode).
+
+    data-import/currency.py [--source URL_OR_FILE] [--symbols URL_OR_FILE ...] [--cache DIR] [--out FILE]
+
+The symbols come from the first locale file that has one, narrow symbols before wide.
+"""
+import argparse
+import csv
+import io
+import xml.etree.ElementTree as ET
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://raw.githubusercontent.com/datasets/currency-codes/main/data/codes-all.csv"
+SYMBOLS = [
+    "https://raw.githubusercontent.com/unicode-org/cldr/main/common/main/en.xml",
+    "https://raw.githubusercontent.com/unicode-org/cldr/main/common/main/root.xml",
+]
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "currency.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["code", "decimals", "name", "numeric", "symbol"]
+
+
+def symbols(xml_texts):
+    """CLDR's symbol per code: the first locale's narrow symbol, else the first locale's wide one."""
+    narrow, wide = {}, {}
+    for text in xml_texts:
+        for c in ET.fromstring(text).iter("currency"):
+            for s in c.findall("symbol"):
+                into = narrow if s.get("alt") == "narrow" else wide if s.get("alt") is None else None
+                if into is not None:
+                    into.setdefault(c.get("type"), s.text)
+    return {**wide, **narrow}
+
+
+def rows(text, symbol):
+    seen = set()
+    for r in csv.DictReader(io.StringIO(text)):
+        code = r["AlphabeticCode"]
+        if not code or code in seen or r["WithdrawalDate"] or r["Entity"].startswith("ZZ") or not r["MinorUnit"].isdigit():
+            continue
+        seen.add(code)
+        yield {
+            "code": code,
+            "decimals": r["MinorUnit"],
+            "name": r["Currency"],
+            "numeric": r["NumericCode"].zfill(3),
+            "symbol": symbol.get(code, code),
+        }
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--symbols", nargs="+", default=SYMBOLS)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    symbol = symbols(source.fetch(s, a.cache, Path(s).name).decode("utf-8") for s in a.symbols)
+    table = rows(source.fetch(a.source, a.cache, "codes-all.csv").decode("utf-8"), symbol)
+    tsv.write(a.out, COLUMNS, sorted(table, key=lambda r: r["code"]))
+
+
+if __name__ == "__main__":
+    main()

data-import/geo-se.py

@@ -0,0 +1,212 @@
+#!/usr/bin/env python3
+"""Rebuild data/geo/SE/*.tsv from SCB (CC0), GeoNames (CC BY 4.0) and Trafikverket NVDB (CC0).
+
+    TRAFIKVERKET_API_KEY=… data-import/geo-se.py [--key-file FILE] [--cache DIR] [--min-streets N] [--street-share F] [--out DIR]
+"""
+import collections
+import csv
+import io
+import json
+import os
+import re
+import sys
+import urllib.request
+import zipfile
+from pathlib import Path
+
+import geo
+import source
+import tsv
+import xlsx
+
+CODES = "https://www.scb.se/contentassets/7a89e48960f741e08918e489ea36354a/kommunlankod-2026.xlsx"
+POPULATION = "https://api.scb.se/OV0104/v1/doris/sv/ssd/START/BE/BE0101/BE0101A/BefolkningNy"
+POPULATION_QUERY = {
+    "query": [
+        {"code": "Region", "selection": {"filter": "all", "values": ["*"]}},
+        {"code": "ContentsCode", "selection": {"filter": "item", "values": ["BE0101N1"]}},
+        {"code": "Tid", "selection": {"filter": "top", "values": ["1"]}},
+    ],
+    "response": {"format": "json"},
+}
+TATORTER = "https://geodata.scb.se/geoserver/stat/wfs?service=WFS&version=2.0.0&request=GetFeature&typeNames=stat:Tatorter_2023&outputFormat=csv&propertyName=tatort,kommun,bef"
+POSTAL_CODES = "https://download.geonames.org/export/zip/SE.zip"
+NVDB = "https://api.trafikinfo.trafikverket.se/v2/data.json"
+NVDB_PAGE = 50000
+TIMEZONE = "Europe/Stockholm"
+# This script's reading of the "exceptions in big cities" in docs/research/research-geo-se.md, Postal codes → Structure.
+ONE_POSITION = {"Stockholm", "Göteborg", "Malmö"}
+# A tätort's floor: "minst 200 invånare", docs/research/research-geo-se.md, Lantmäteriet → Ortnamn.
+UNMATCHED_POPULATION = 200
+# Not a street: a road number, a ferry route, a cycle path's or road association's label, a placeholder, or two names
+# or a description in one.
+NOT_A_STREET = re.compile(r"^(Europaväg|Länsväg|Riksväg) \d|^Färjeled |\b(GC|SMF|VSF)\b|^NN$|,")
+PARTICLES = {"af", "de", "von"}
+
+
+def scb_codes(cache):
+    regions, municipalities = {}, {}
+    for cells in xlsx.rows(source.fetch(CODES, cache, "kommunlankod.xlsx", magic=b"PK")):
+        if len(cells) < 2 or not re.fullmatch(r"\d{2}|\d{4}", cells[0]):
+            continue
+        (regions if len(cells[0]) == 2 else municipalities)[cells[0]] = cells[1].strip()
+    return regions, municipalities
+
+
+def scb_population(cache):
+    body = json.dumps(POPULATION_QUERY).encode()
+    data = source.fetch(POPULATION, cache, "befolkning.json", data=body, headers={"Content-Type": "application/json"})
+    return {row["key"][0]: row["values"][0] for row in json.loads(data.decode("utf-8-sig"))["data"]}
+
+
+def scb_tatorter(cache):
+    text = source.fetch(TATORTER, cache, "tatorter.csv").decode("utf-8")
+    by_name = collections.defaultdict(list)
+    for r in csv.DictReader(io.StringIO(text)):
+        by_name[r["tatort"]].append((r["kommun"], int(r["bef"])))
+    return by_name
+
+
+def geonames(cache):
+    z = zipfile.ZipFile(io.BytesIO(source.fetch(POSTAL_CODES, cache, "SE.zip", magic=b"PK")))
+    rows = []
+    for line in z.read("SE.txt").decode("utf-8").splitlines():
+        f = line.split("\t")
+        lat, lon = (float(f[9]), float(f[10])) if f[9] and f[10] else (None, None)
+        rows.append({"code": f[1], "locality": f[2], "municipality": f[6], "lat": lat, "lon": lon})
+    return rows
+
+
+def nvdb_segments(cache, key):
+    path = cache / "nvdb-gatunamn.tsv"
+    if not path.exists():
+        with open(path.with_suffix(".part"), "w", encoding="utf-8") as out:
+            change = "0"
+            while True:
+                query = (
+                    f'<REQUEST><LOGIN authenticationkey="{key}"/>'
+                    f'<QUERY objecttype="Gatunamn" namespace="vägdata.nvdb_dk_o" schemaversion="1.0" limit="{NVDB_PAGE}" changeid="{change}">'
+                    "<FILTER><EQ name=\"Deleted\" value=\"false\"/></FILTER>"
+                    "<INCLUDE>Namn</INCLUDE><INCLUDE>Geometry.WKT-WGS84-3D</INCLUDE></QUERY></REQUEST>"
+                )
+                req = urllib.request.Request(NVDB, data=query.encode(), headers={"Content-Type": "text/xml"})
+                with urllib.request.urlopen(req, timeout=600) as r:
+                    result = json.load(r)["RESPONSE"]["RESULT"][0]
+                rows = result.get("Gatunamn", [])
+                for row in rows:
+                    m = re.match(r"LINESTRING Z \(([-\d.]+) ([-\d.]+) ", row.get("Geometry", {}).get("WKT-WGS84-3D", ""))
+                    name = " ".join(row.get("Namn", "").split())
+                    if m and name:
+                        out.write(f"{name}\t{m.group(1)}\t{m.group(2)}\n")
+                change = result["INFO"]["LASTCHANGEID"]
+                if len(rows) < NVDB_PAGE:
+                    break
+        path.with_suffix(".part").rename(path)
+    for line in path.read_text(encoding="utf-8").splitlines():
+        name, lon, lat = line.split("\t")
+        yield name, float(lat), float(lon)
+
+
+def street_delivery(name, codes):
+    """The codes delivered to a street. A postort's own prefix is three digits where one three-digit prefix starts at
+    least half its codes, else two. Outside ONE_POSITION's cities, the digit after that prefix is 0 or 1 for boxes and 8
+    for reply mail; inside them, a second digit 0 marks boxes. The box and reply-mail digits are per
+    docs/research/research-geo-se.md, Postal codes → Structure."""
+    if name in ONE_POSITION:
+        return [c for c in codes if c[1] != "0"]
+    largest = collections.Counter(c[:3] for c in codes).most_common(1)[0][1]
+    length = 3 if largest * 2 >= len(codes) else 2
+    return [c for c in codes if c[length] not in "018"]
+
+
+def municipality_of(name, rows, tatorter, municipalities):
+    named = [code for code, n in municipalities.items() if n == name]
+    if named:
+        return named[0], "kommun"
+    voted = collections.Counter(r["municipality"] for r in rows if r["municipality"] in municipalities)
+    matches = tatorter.get(name, [])
+    if matches:
+        in_vote = [m for m in matches if voted and m[0] == voted.most_common(1)[0][0]]
+        return max(in_vote or matches, key=lambda m: m[1])[0], "tatort"
+    if voted:
+        return voted.most_common(1)[0][0], "codes"
+    return None, "unplaced"
+
+
+def population_of(name, municipality, tatorter, municipalities, population):
+    matches = [m for m in tatorter.get(name, []) if m[0] == municipality]
+    if matches:
+        return str(max(m[1] for m in matches))
+    return population[municipality] if municipalities[municipality] == name else str(UNMATCHED_POPULATION)
+
+
+def well_cased(name):
+    """Whether each word of name is capitalised: GeoNames also spells some postorter, such as STOCKHOLM and örebro, in
+    capitals or lower case."""
+    return all(part[:1].isupper() and (len(part) == 1 or not part.isupper()) for part in re.split(r"[ -]", name))
+
+
+def localities(codes, tatorter, municipalities, population):
+    """Each postort with its municipality, weight, centroid and street-delivery codes."""
+    by_locality = collections.defaultdict(list)
+    for r in codes:
+        by_locality[r["locality"]].append(r)
+    out, how = {}, collections.Counter()
+    for name, rows in by_locality.items():
+        municipality, method = municipality_of(name, rows, tatorter, municipalities)
+        how[method] += 1
+        kept = street_delivery(name, [r["code"].replace(" ", "") for r in rows])
+        point = geo.centroid(rows)
+        if municipality is None or not kept or point is None or not well_cased(name):
+            continue
+        lat, lon = point
+        out[name] = {"name": name, "municipality": municipality, "population": population_of(name, municipality, tatorter, municipalities, population), "lat": f"{lat:.4f}", "lon": f"{lon:.4f}", "codes": kept}
+    print(f"municipality by {dict(how)}; {len(by_locality) - len(out)} postorter dropped before streets", file=sys.stderr)
+    return out
+
+
+def street_name(name):
+    # A word of four capitals or fewer is an acronym or a Roman numeral: IOGT-vägen, Karl XII gatan.
+    opening = name[0].isupper() or name.split()[0] in PARTICLES
+    return opening and not NOT_A_STREET.search(name) and not any(len(w) >= 5 and w.isupper() for w in name.split())
+
+
+def streets(segments, codes, localities, share, minimum):
+    """The names with most segments per locality, each segment at its nearest code centroid."""
+    nearest = geo.Nearest((r["lat"], r["lon"], r["locality"]) for r in codes if r["lat"] is not None and r["locality"] in localities)
+    count = collections.Counter()
+    for name, lat, lon in segments:
+        if street_name(name):
+            count[(nearest.find(lat, lon), name)] += 1
+    return geo.top_streets(count, share, minimum, "segments")
+
+
+def main():
+    a = geo.parser(__doc__, "SE", 1.0, key_file={"help": "file holding the Trafikverket API key; TRAFIKVERKET_API_KEY otherwise"}).parse_args()
+    key = Path(a.key_file).read_text().strip() if a.key_file else os.environ.get("TRAFIKVERKET_API_KEY")
+    if not key:
+        sys.exit("set TRAFIKVERKET_API_KEY or pass --key-file")
+    cache, out = geo.directories(a)
+
+    region_names, municipality_names = scb_codes(cache)
+    population = scb_population(cache)
+    codes = geonames(cache)
+    places = localities(codes, scb_tatorter(cache), municipality_names, population)
+    named = streets(nvdb_segments(cache, key), codes, places, a.street_share, a.min_streets)
+    regions = {c: {"code": c, "name": n, "population": population[c], "timezone": TIMEZONE} for c, n in region_names.items()}
+    municipalities = {c: {"code": c, "name": n, "region": c[:2], "population": population[c]} for c, n in municipality_names.items()}
+    kept, dropped = geo.fence(regions, municipalities, places, named, {k: l["codes"] for k, l in places.items()})
+    geo.log_dropped_localities(dropped["locality"])
+    if dropped["municipality"] or dropped["region"]:
+        sys.exit(f"municipalities without a locality: {sorted(dropped['municipality'])}; regions without a municipality: {sorted(dropped['region'])}")
+    places = kept["locality"]
+
+    tsv.write(out / "region.tsv", ["code", "name", "population", "timezone"], [r for _, r in sorted(kept["region"].items())])
+    tsv.write(out / "municipality.tsv", ["code", "name", "region", "population"], [m for _, m in sorted(kept["municipality"].items())])
+    tsv.write(out / "locality.tsv", ["name", "municipality", "population", "lat", "lon"], [l for _, l in sorted(places.items())])
+    tsv.write(out / "postal-code.tsv", ["code", "locality"], sorted(({"code": f"{c[:3]} {c[3:]}", "locality": l["name"]} for l in places.values() for c in l["codes"]), key=lambda r: r["code"]))
+    tsv.write(out / "street.tsv", ["name", "locality", "segments"], [s for locality in sorted(places) for s in named[locality]])
+
+
+if __name__ == "__main__":
+    main()

data-import/geo-us.py

@@ -0,0 +1,237 @@
+#!/usr/bin/env python3
+"""Rebuild data/geo/US/*.tsv from the Census Bureau's Gazetteer, population estimates, ZCTA relationships and TIGER/Line files (public domain), and Wikidata's state capitals (CC0).
+
+    data-import/geo-us.py [--cache DIR] [--min-streets N] [--place-share F] [--street-share F] [--out DIR]
+"""
+import collections
+import concurrent.futures
+import csv
+import io
+import math
+import random
+import re
+import struct
+import sys
+import urllib.parse
+import zipfile
+
+import geo
+import source
+import tsv
+
+GAZETTEER = "https://www2.census.gov/geo/docs/maps-data/data/gazetteer/2026_Gazetteer/2026_Gaz_{}_national.zip"
+POPULATION = "https://www2.census.gov/programs-surveys/popest/datasets/2020-2025/{}"
+STATES = POPULATION.format("state/totals/NST-EST2025-ALLDATA.csv")
+COUNTIES = POPULATION.format("counties/totals/co-est2025-alldata.csv")
+PLACES = POPULATION.format("cities/totals/sub-est2025.csv")
+ZCTA_PLACE = "https://www2.census.gov/geo/docs/maps-data/data/rel2020/zcta520/tab20_zcta520_place20_natl.txt"
+TIGER = "https://www2.census.gov/geo/tiger/TIGER2025/{0}/tl_2025_{1}_{2}.zip"
+ESTIMATE = "POPESTIMATE2025"
+CDP = "57"
+# Each state's capital from Wikidata, with the capital's GNIS ids, which the Gazetteer carries as ANSICODE.
+CAPITALS = "https://query.wikidata.org/sparql?" + urllib.parse.urlencode({"query": """
+SELECT ?iso ?capitalLabel ?gnis WHERE {
+  ?state wdt:P31 wd:Q35657; wdt:P300 ?iso; wdt:P36 ?capital.
+  OPTIONAL { ?capital wdt:P590 ?gnis. }
+  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
+}"""})
+SHIPPED_POPULATION = 100000
+HIGHWAY = re.compile(r"\b(I- |Hwy |Highway |Loop |Rte |Route |Rd )\d")
+SUFFIX = re.compile(r" (city and borough|city|town|village|borough|municipality|comunidad|zona urbana|metropolitan government|metro government|consolidated government|unified government|urban county|corporation|plantation)( \(balance\))?$")
+# Places whose Census name is not the postal city an address carries.
+NAMES = {"0665042": "Ventura", "1303440": "Athens", "1304204": "Augusta", "1349008": "Macon", "1608830": "Boise", "2146027": "Lexington", "2148006": "Louisville", "3011397": "Butte", "4732742": "Hartsville", "4744382": "Lynchburg", "4752006": "Nashville"}
+TIMEZONES = {
+    "AK": "America/Anchorage", "AL": "America/Chicago", "AR": "America/Chicago", "AZ": "America/Phoenix",
+    "CA": "America/Los_Angeles", "CO": "America/Denver", "CT": "America/New_York", "DC": "America/New_York",
+    "DE": "America/New_York", "FL": "America/New_York", "GA": "America/New_York", "HI": "Pacific/Honolulu",
+    "IA": "America/Chicago", "ID": "America/Boise", "IL": "America/Chicago", "IN": "America/Indiana/Indianapolis",
+    "KS": "America/Chicago", "KY": "America/New_York", "LA": "America/Chicago", "MA": "America/New_York",
+    "MD": "America/New_York", "ME": "America/New_York", "MI": "America/Detroit", "MN": "America/Chicago",
+    "MO": "America/Chicago", "MS": "America/Chicago", "MT": "America/Denver", "NC": "America/New_York",
+    "ND": "America/Chicago", "NE": "America/Chicago", "NH": "America/New_York", "NJ": "America/New_York",
+    "NM": "America/Denver", "NV": "America/Los_Angeles", "NY": "America/New_York", "OH": "America/New_York",
+    "OK": "America/Chicago", "OR": "America/Los_Angeles", "PA": "America/New_York", "PR": "America/Puerto_Rico",
+    "RI": "America/New_York", "SC": "America/New_York", "SD": "America/Chicago", "TN": "America/Chicago",
+    "TX": "America/Chicago", "UT": "America/Denver", "VA": "America/New_York", "VT": "America/New_York",
+    "WA": "America/Los_Angeles", "WI": "America/Chicago", "WV": "America/New_York", "WY": "America/Denver",
+}
+
+
+def text(data):
+    try:
+        return data.decode("utf-8-sig")
+    except UnicodeDecodeError:
+        return data.decode("latin-1")
+
+
+def gazetteer(cache, kind):
+    z = zipfile.ZipFile(io.BytesIO(source.fetch(GAZETTEER.format(kind), cache, f"gaz_{kind}.zip", magic=b"PK")))
+    rows = text(z.read(z.namelist()[0])).splitlines()
+    header = [h.strip() for h in rows[0].split("|")]
+    return [dict(zip(header, (c.strip() for c in row.split("|")))) for row in rows[1:]]
+
+
+def csv_rows(cache, url, name):
+    return list(csv.DictReader(io.StringIO(text(source.fetch(url, cache, name)))))
+
+
+def dbf_rows(data, wanted):
+    """The records of a dBASE file, the wanted fields only."""
+    count, header_len, record_len = struct.unpack("<xxxxIHH", data[:12])
+    fields, pos = [], 32
+    while data[pos] != 0x0D:
+        name = data[pos:pos + 11].split(b"\0")[0].decode()
+        fields.append((name, data[pos + 16]))
+        pos += 32
+    pos = header_len
+    for _ in range(count):
+        record = data[pos:pos + record_len]
+        pos += record_len
+        if record[:1] == b"*":
+            continue
+        row, at = {}, 1
+        for name, length in fields:
+            if name in wanted:
+                row[name] = record[at:at + length].decode("utf-8", "replace").strip()
+            at += length
+        yield row
+
+
+def tiger_zip(cache, kind, county):
+    return source.fetch(TIGER.format(kind.upper(), county, kind), cache, f"tl_{county}_{kind}.zip", magic=b"PK")
+
+
+def tiger(cache, kind, county, wanted):
+    z = zipfile.ZipFile(io.BytesIO(tiger_zip(cache, kind, county)))
+    return dbf_rows(z.read(f"tl_2025_{county}_{kind}.dbf"), wanted)
+
+
+def place_name(geoid, name):
+    if geoid in NAMES:
+        return NAMES[geoid]
+    stripped = SUFFIX.sub("", name)
+    if stripped == name:
+        print(f"{geoid}: no suffix stripped from {name!r}", file=sys.stderr)
+    return stripped
+
+
+def localities(cache, counties):
+    """Each incorporated place with its county, population and centroid."""
+    place_population, county_part = {}, collections.defaultdict(list)
+    for r in csv_rows(cache, PLACES, "sub-est2025.csv"):
+        if r["SUMLEV"] == "162":
+            place_population[r["STATE"] + r["PLACE"]] = int(r[ESTIMATE])
+        elif r["SUMLEV"] == "157":
+            county_part[r["STATE"] + r["PLACE"]].append((int(r[ESTIMATE]), r["STATE"] + r["COUNTY"]))
+    out = {}
+    for r in gazetteer(cache, "place"):
+        geoid, population = r["GEOID"], place_population.get(r["GEOID"], 0)
+        if r["FUNCSTAT"] not in ("A", "B", "F", "N") or r["LSAD"] == CDP or not county_part.get(geoid):
+            continue
+        county = max(county_part[geoid])[1]
+        if county not in counties:
+            print(f"{geoid} {r['NAME']}: county {county} unknown, dropped", file=sys.stderr)
+            continue
+        out[geoid] = {"code": geoid, "gnis": int(r["ANSICODE"]), "name": place_name(geoid, r["NAME"]), "municipality": county, "population": population, "lat": r["INTPTLAT"], "lon": r["INTPTLONG"]}
+    return out
+
+
+def capitals(cache, places, counties):
+    """The codes of the places that are a state's capital, matched by any of its GNIS ids, else by a name no other place in the state carries."""
+    by_gnis = {p["gnis"]: code for code, p in places.items()}
+    named = collections.defaultdict(list)
+    for code, p in places.items():
+        named[(counties[p["municipality"]]["region"], p["name"])].append(code)
+    by_name = {key: codes[0] for key, codes in named.items() if len(codes) == 1}
+    gnis_of = collections.defaultdict(set)
+    for r in csv.DictReader(io.StringIO(text(source.fetch(CAPITALS, cache, "capitals.csv", magic=b"iso,", headers={"Accept": "text/csv"})))):
+        gnis_of[(r["iso"].removeprefix("US-"), r["capitalLabel"])].add(int(r["gnis"] or 0))
+    out = set()
+    for (state, name), ids in sorted(gnis_of.items()):
+        code = next((by_gnis[i] for i in sorted(ids) if i in by_gnis), None) or by_name.get((state, name))
+        if code:
+            out.add(code)
+        else:
+            print(f"{state} {name}: capital matches no incorporated place by GNIS id or by a name unique in its state", file=sys.stderr)
+    return out
+
+
+def drawn(places, share):
+    """The share of the places with people, each drawn with odds by its population: Efraimidis and Spirakis's weighted
+    sampling without replacement. Each place's random number is seeded by its code, so a place added to or dropped from
+    the sources moves only the places near the cut."""
+    peopled = [code for code, p in places.items() if p["population"] > 0]
+    key = {code: math.log(1 - random.Random(code).random()) / places[code]["population"] for code in peopled}
+    return {code: places[code] for code in sorted(peopled, key=lambda c: (key[c], c), reverse=True)[:round(len(peopled) * share)]}
+
+
+def zcta_localities(cache, localities):
+    """Each ZCTA mapped to the incorporated place holding most of its land inside incorporated places, where localities holds that place."""
+    parts = {}
+    for r in csv.DictReader(io.StringIO(text(source.fetch(ZCTA_PLACE, cache, "zcta-place.txt"))), delimiter="|"):
+        if r["GEOID_ZCTA5_20"] and r["GEOID_PLACE_20"] and not r["NAMELSAD_PLACE_20"].endswith(" CDP"):
+            parts.setdefault(r["GEOID_ZCTA5_20"], []).append((int(r["AREALAND_PART"]), r["GEOID_PLACE_20"]))
+    largest = {zcta: max(p)[1] for zcta, p in parts.items()}
+    return {zcta: place for zcta, place in largest.items() if place in localities}
+
+
+def streets(cache, counties, locality_of_zcta, share, minimum):
+    """The names with most TIGER address ranges per place, and the address ranges per ZCTA."""
+    with concurrent.futures.ThreadPoolExecutor(3) as pool:
+        list(pool.map(lambda c: (tiger_zip(cache, "addr", c), tiger_zip(cache, "featnames", c)), counties))
+    addresses, count = collections.Counter(), collections.Counter()
+    for county in counties:
+        zips = collections.defaultdict(set)
+        for r in tiger(cache, "addr", county, {"TLID", "ZIP"}):
+            if r["ZIP"] in locality_of_zcta:
+                zips[r["TLID"]].add(r["ZIP"])
+                addresses[r["ZIP"]] += 1
+        for r in tiger(cache, "featnames", county, {"TLID", "FULLNAME", "PAFLAG"}):
+            if r["PAFLAG"] == "P" and r["FULLNAME"] and not HIGHWAY.search(r["FULLNAME"]):
+                for z in zips.get(r["TLID"], ()):
+                    count[(locality_of_zcta[z], r["FULLNAME"])] += 1
+    return geo.top_streets(count, share, minimum, "addresses"), addresses
+
+
+def main():
+    a = geo.parser(__doc__, "US", 0.25, place_share={"type": float, "default": 0.1}).parse_args()
+    cache, out = geo.directories(a)
+
+    state_population = {r["STATE"]: r[ESTIMATE] for r in csv_rows(cache, STATES, "nst-est2025.csv") if r["SUMLEV"] == "040"}
+    regions = {r["USPS"]: {"abbr": r["USPS"], "code": r["GEOID"], "name": r["NAME"], "population": state_population[r["GEOID"]], "timezone": TIMEZONES[r["USPS"]]} for r in gazetteer(cache, "state")}
+    county_population = {r["STATE"] + r["COUNTY"]: r[ESTIMATE] for r in csv_rows(cache, COUNTIES, "co-est2025.csv") if r["SUMLEV"] == "050"}
+    counties, unestimated = {}, collections.Counter()
+    for r in gazetteer(cache, "counties"):
+        if r["GEOID"] not in county_population:
+            unestimated[r["USPS"]] += 1
+            continue
+        counties[r["GEOID"]] = {"code": r["GEOID"], "name": r["NAME"], "region": r["USPS"], "population": county_population[r["GEOID"]]}
+    print(f"counties without a population estimate, dropped: {dict(unestimated)}", file=sys.stderr)
+
+    places = localities(cache, counties)
+    owner_of_zcta = zcta_localities(cache, places)
+    capital_codes = capitals(cache, places, counties)
+    owners = {p: places[p] for p in set(owner_of_zcta.values())}
+    for code in sorted(capital_codes - set(owners)):
+        print(f"{code} {places[code]['name']}: capital owns no ZCTA, dropped", file=sys.stderr)
+    places = drawn(owners, a.place_share) | {code: p for code, p in owners.items() if code in capital_codes or p["population"] >= SHIPPED_POPULATION}
+    locality_of_zcta = {zcta: p for zcta, p in owner_of_zcta.items() if p in places}
+    named, addresses = streets(cache, sorted({l["municipality"] for l in places.values()}), locality_of_zcta, a.street_share, a.min_streets)
+    zctas = collections.defaultdict(list)
+    for zcta, locality in sorted(locality_of_zcta.items()):
+        if addresses[zcta]:
+            zctas[locality].append(zcta)
+    kept, dropped = geo.fence(regions, counties, places, named, zctas)
+    geo.log_dropped_localities(dropped["locality"])
+    print(f"counties without a place, dropped: {len(dropped['municipality'])}; states without a county, dropped: {sorted(dropped['region'])}", file=sys.stderr)
+    places = kept["locality"]
+
+    tsv.write(out / "region.tsv", ["abbr", "code", "name", "population", "timezone"], [r for _, r in sorted(kept["region"].items())])
+    tsv.write(out / "municipality.tsv", ["code", "name", "region", "population"], [c for _, c in sorted(kept["municipality"].items())])
+    tsv.write(out / "locality.tsv", ["code", "name", "municipality", "population", "lat", "lon"], [l for _, l in sorted(places.items())])
+    tsv.write(out / "postal-code.tsv", ["code", "locality", "addresses"], sorted(({"code": z, "locality": l, "addresses": addresses[z]} for l in places for z in zctas[l]), key=lambda r: r["code"]))
+    tsv.write(out / "street.tsv", ["name", "locality", "addresses"], [s for locality in sorted(places) for s in named[locality]])
+
+
+if __name__ == "__main__":
+    main()

data-import/geo.py

@@ -0,0 +1,98 @@
+"""The steps a country's geo import may share with another's."""
+import argparse
+import collections
+import math
+import sys
+from pathlib import Path
+
+DATA = Path(__file__).resolve().parent.parent / "data" / "geo"
+CACHE = Path(__file__).resolve().parent / "cache"
+
+
+def parser(doc, country, street_share, **own):
+    """A parser for the options every country's import takes, plus the script's own: each keyword names one option, with
+    "_" standing for "-"."""
+    options = {"cache": {"default": str(CACHE)}, "out": {"default": str(DATA / country)}, "min_streets": {"type": int, "default": 10}, "street_share": {"type": float, "default": street_share}, **own}
+    p = argparse.ArgumentParser(description=doc.splitlines()[0])
+    for name in sorted(options):
+        p.add_argument("--" + name.replace("_", "-"), **options[name])
+    return p
+
+
+def directories(args):
+    """The cache and out directories of the parsed options, both created."""
+    cache, out = Path(args.cache), Path(args.out)
+    cache.mkdir(parents=True, exist_ok=True)
+    out.mkdir(parents=True, exist_ok=True)
+    return cache, out
+
+
+def centroid(rows):
+    """The mean lat and lon of the rows with a point, or None where no row has one."""
+    points = [r for r in rows if r["lat"] is not None]
+    if not points:
+        return None
+    return sum(r["lat"] for r in points) / len(points), sum(r["lon"] for r in points) / len(points)
+
+
+class Nearest:
+    """Nearest point by an equirectangular distance, over a degree grid."""
+
+    def __init__(self, points, cell=0.05):
+        self.cell = cell
+        self.grid = collections.defaultdict(list)
+        for lat, lon, value in points:
+            self.grid[(int(lat // cell), int(lon // cell))].append((lat, lon, value))
+
+    def find(self, lat, lon):
+        ci, cj = int(lat // self.cell), int(lon // self.cell)
+        best, best_d = None, math.inf
+        ring = 0
+        while ring < 400:
+            for i in range(ci - ring, ci + ring + 1):
+                for j in range(cj - ring, cj + ring + 1):
+                    if max(abs(i - ci), abs(j - cj)) != ring:
+                        continue
+                    for plat, plon, value in self.grid.get((i, j), ()):
+                        d = (plat - lat) ** 2 + ((plon - lon) * math.cos(math.radians(lat))) ** 2
+                        if d < best_d:
+                            best, best_d = value, d
+            if best is not None and math.sqrt(best_d) < ring * self.cell * math.cos(math.radians(lat)):
+                return best
+            ring += 1
+        return best
+
+
+def top_streets(count, share, minimum, column):
+    """count maps (locality, street name) to a count. Returns, per locality, rows for its most-counted names, share of
+    them rounded up and at least minimum, ties broken by name; each row holds its count under column."""
+    of = collections.defaultdict(list)
+    for (locality, name), n in count.items():
+        of[locality].append((n, name))
+    return {locality: [{"name": name, "locality": locality, column: n} for n, name in sorted(named, key=lambda s: (-s[0], s[1]))[:max(minimum, math.ceil(len(named) * share))]] for locality, named in of.items()}
+
+
+def fence(regions, municipalities, localities, streets, postal_codes):
+    """Meet the loader's fence that every parent row has a child row. Keeps each locality with entries in both streets
+    and postal_codes, then each municipality and region left with a child.
+
+    regions and municipalities map a code to a row; a municipality row names its region under "region". localities,
+    streets and postal_codes map a locality's key to its row, its streets and its postal codes. A locality row holds
+    name, municipality, population, lat and lon, plus its country's own keys.
+
+    Returns kept, then dropped: each maps "locality", "municipality" and "region" to that table's rows."""
+    kept = {"locality": {k: l for k, l in localities.items() if streets.get(k) and postal_codes.get(k)}}
+    municipality_codes = {l["municipality"] for l in kept["locality"].values()}
+    kept["municipality"] = {c: m for c, m in municipalities.items() if c in municipality_codes}
+    region_codes = {m["region"] for m in kept["municipality"].values()}
+    kept["region"] = {c: r for c, r in regions.items() if c in region_codes}
+    every = {"locality": localities, "municipality": municipalities, "region": regions}
+    dropped = {table: {k: row for k, row in rows.items() if k not in kept[table]} for table, rows in every.items()}
+    return kept, dropped
+
+
+def log_dropped_localities(rows):
+    """Logs each locality row as dropped, by its key, plus its name where that differs."""
+    for key, row in rows.items():
+        label = key if key == row["name"] else f"{key} {row['name']}"
+        print(f"{label}: no street or no postal code, dropped", file=sys.stderr)

data-import/httpmethod.py

@@ -0,0 +1,59 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/httpmethod.tsv from the IANA HTTP Method Registry.
+
+    data-import/httpmethod.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
+
+A method ships when the register cites RFC 9110 section 9.3 or RFC 5789 for it, which
+is the eight HTTP defines and PATCH; the register's other entries do not. The
+register's `yes` and `no` ship as `true` and `false`, which a consumer's boolean reads.
+"""
+import argparse
+import csv
+import io
+import re
+import sys
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://www.iana.org/assignments/http-methods/methods.csv"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "httpmethod.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["idempotent", "method", "safe"]
+CORE = re.compile(r"\[RFC9110, Section 9\.3\.|\[RFC5789, Section 2\]")
+BOOLEAN = {"yes": "true", "no": "false"}
+SELECTOR = '[]{}"|'  # mirrors inSelector in internal/grammar/path.go
+EXPECTED = 9
+
+
+def rows(text):
+    for r in csv.DictReader(io.StringIO(text)):
+        method = (r["Method Name"] or "").strip()
+        if not CORE.search(r["Reference"] or ""):
+            continue
+        if set(method) & set(SELECTOR):
+            sys.exit(f"{method}: a method holding one of {SELECTOR} is one a selector cannot spell")
+        answers = {}
+        for column in ("Safe", "Idempotent"):
+            answer = (r[column] or "").strip()
+            if answer not in BOOLEAN:
+                sys.exit(f"{method}: {column} {answer!r} is neither yes nor no")
+            answers[column] = BOOLEAN[answer]
+        yield {"idempotent": answers["Idempotent"], "method": method, "safe": answers["Safe"]}
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = sorted(rows(source.fetch(a.source, a.cache, "http-methods.csv").decode("utf-8")), key=lambda r: r["method"])
+    if len(table) != EXPECTED:
+        sys.exit(f"{len(table)} methods cite RFC 9110 section 9.3 or RFC 5789, not {EXPECTED}; the register's Reference column has moved")
+    tsv.write(a.out, COLUMNS, table)
+
+
+if __name__ == "__main__":
+    main()

data-import/httpstatus.py

@@ -0,0 +1,40 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/httpstatus.tsv from the IANA HTTP Status Code Registry.
+
+    data-import/httpstatus.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
+
+What ships is the codes in use: a range, an unassigned code and a qualified reason are all skipped.
+"""
+import argparse
+import csv
+import io
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://www.iana.org/assignments/http-status-codes/http-status-codes-1.csv"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "httpstatus.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["code", "reason"]
+
+
+def rows(text):
+    for r in csv.DictReader(io.StringIO(text)):
+        code, reason = r["Value"].strip(), r["Description"].strip()
+        if code.isdigit() and reason and "(" not in reason and reason != "Unassigned":
+            yield {"code": code, "reason": reason}
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = rows(source.fetch(a.source, a.cache, "http-status-codes-1.csv").decode("utf-8"))
+    tsv.write(a.out, COLUMNS, sorted(table, key=lambda r: int(r["code"])))
+
+
+if __name__ == "__main__":
+    main()

data-import/language.py

@@ -0,0 +1,45 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/language.tsv from datasets/language-codes (PDDL), the LoC ISO 639-2 register.
+
+    data-import/language.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
+
+Only the 639-2 entries carrying a 639-1 code ship, since the alpha-2 code is the key.
+"""
+import argparse
+import csv
+import io
+import re
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://raw.githubusercontent.com/datasets/language-codes/main/data/language-codes-full.csv"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "language.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["code", "code3", "name"]
+
+
+def name(english):
+    """The register lists every synonym; the first, without its qualifier, is the one a selector spells."""
+    return re.sub(r"\s*\([^)]*\)", "", english.split(";")[0]).strip()
+
+
+def rows(text):
+    for r in csv.DictReader(io.StringIO(text)):
+        if r["alpha2"]:
+            yield {"code": r["alpha2"], "code3": r["alpha3-t"] or r["alpha3-b"], "name": name(r["English"])}
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = rows(source.fetch(a.source, a.cache, "language-codes-full.csv").decode("utf-8"))
+    tsv.write(a.out, COLUMNS, sorted(table, key=lambda r: r["code"]))
+
+
+if __name__ == "__main__":
+    main()

data-import/loglevel.py

@@ -0,0 +1,68 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/loglevel.tsv from RFC 5424 and POSIX <syslog.h>.
+
+    data-import/loglevel.py [--source URL_OR_FILE] [--posix URL_OR_FILE] [--cache DIR] [--out FILE]
+
+Every severity in Table 2 ships, paired to POSIX's keywords by the order each source
+lists them, which holds only because a keyword prefixes the severity RFC 5424 numbers
+alike.
+"""
+import argparse
+import html
+import re
+import sys
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://www.rfc-editor.org/rfc/rfc5424.txt"
+POSIX = "https://pubs.opengroup.org/onlinepubs/9799919799/basedefs/syslog.h.html"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "loglevel.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["code", "keyword", "severity"]
+TABLE = re.compile(r"Table 1\.\s+Syslog Message Facilities(.*?)Table 2\.\s+Syslog Message Severities", re.S)
+ROW = re.compile(r"^ +(\d) +([A-Z][a-z]+): ", re.M)
+SECTION = re.compile(r"severity level portion of the.*?macros:(.*?)The following shall be declared as functions", re.S)
+MACRO = re.compile(r"\bLOG_([A-Z]+)\b")
+EXPECTED = 8
+
+
+def severities(text):
+    table = TABLE.search(text)
+    if not table:
+        sys.exit("the RFC holds no text between the facility and the severity caption; its tables have moved")
+    return ROW.findall(table.group(1))
+
+
+def keywords(page):
+    section = SECTION.search(html.unescape(re.sub(r"<[^>]+>", "", page)))
+    if not section:
+        sys.exit("POSIX spells no severity list between its own caption and the function declarations; the page has moved")
+    return [m.lower() for m in MACRO.findall(section.group(1))]
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--posix", default=POSIX)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = sorted(severities(source.fetch(a.source, a.cache, "rfc5424.txt").decode("utf-8")))
+    codes = [code for code, _ in table]
+    if codes != [str(i) for i in range(EXPECTED)]:
+        sys.exit(f"the severity table numbers {codes}, not 0 through {EXPECTED - 1}")
+    words = keywords(source.fetch(a.posix, a.cache, "posix-syslog.h.html").decode("utf-8"))
+    if len(words) != EXPECTED:
+        sys.exit(f"POSIX lists {len(words)} severity macros, not {EXPECTED}: {words}")
+    rows = []
+    for (code, severity), keyword in zip(table, words):
+        if not severity.lower().startswith(keyword):
+            sys.exit(f"severity {code} is {severity} in the RFC and {keyword} in POSIX, which is no prefix of it")
+        rows.append({"code": code, "keyword": keyword, "severity": severity})
+    tsv.write(a.out, COLUMNS, rows)
+
+
+if __name__ == "__main__":
+    main()

data-import/mimetype.py

@@ -0,0 +1,39 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/mimetype.tsv from mime-db (MIT), the IANA media type registry with extensions.
+
+    data-import/mimetype.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
+
+Only types mime-db records as IANA-registered, with a filename extension, ship.
+"""
+import argparse
+import json
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://raw.githubusercontent.com/jshttp/mime-db/master/db.json"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "mimetype.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["ext", "type"]
+FIXUPS = {"application/mp4": "mp4s", "audio/mpeg": "mp3", "audio/ogg": "ogg", "video/quicktime": "mov"}
+
+
+def rows(db):
+    for mimetype, entry in db.items():
+        if entry.get("source") == "iana" and entry.get("extensions"):
+            yield {"ext": "." + FIXUPS.get(mimetype, entry["extensions"][0]), "type": mimetype}
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = rows(json.loads(source.fetch(a.source, a.cache, "mime-db.json").decode("utf-8")))
+    tsv.write(a.out, COLUMNS, sorted(table, key=lambda r: r["type"]))
+
+
+if __name__ == "__main__":
+    main()

data-import/names-se.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env python3
+"""Rebuild data/sv_SE/first-name.tsv and last-name.tsv from SCB's whole-population name counts (CC0, "Källa: SCB").
+
+    data-import/names-se.py [--source URL_OR_FILE] [--first N] [--last N] [--cache DIR] [--out DIR]
+
+A tilltalsnamn goes under each sex that carries it, so a name both sexes carry has two rows.
+"""
+import argparse
+import re
+from pathlib import Path
+
+import source
+import tsv
+import xlsx
+
+SOURCE = "https://www.scb.se/contentassets/9fe7dbb460994c72b835163dbc491ef9/namn-med-minst-tva-barare-31-december-2022.xlsx"
+OUT = Path(__file__).resolve().parent.parent / "data" / "sv_SE"
+CACHE = Path(__file__).resolve().parent / "cache"
+SHEETS = {"f": "Tilltalsnamn kvinnor", "m": "Tilltalsnamn män"}
+SURNAMES = "Efternamn"
+NAME = re.compile(r"^[^\W\d_]{2,}([ '-][^\W\d_]{2,})*$")
+PARTICLES = {"af", "av", "da", "de", "del", "den", "der", "di", "dos", "du", "la", "le", "van", "von"}
+
+
+def cased(name):
+    """SCB's uppercase name as it is written: each part capitalised, a particle before another part lowercased."""
+    parts = name.lower().split(" ")
+    return " ".join(w if w in PARTICLES and len(parts) > 1 else w.title() for w in parts)
+
+
+def counted(rows, top):
+    """The top names of a sheet by bearers, initials and single letters dropped."""
+    names = [(cells[0], int(cells[1])) for cells in rows if len(cells) >= 2 and cells[1].isdigit() and NAME.match(cells[0])]
+    return sorted(names, key=lambda n: (-n[1], n[0]))[:top]
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--first", type=int, default=2000, help="names per sex")
+    p.add_argument("--last", type=int, default=5000)
+    p.add_argument("--out", default=str(OUT))
+    p.add_argument("--source", default=SOURCE)
+    a = p.parse_args()
+    data = source.fetch(a.source, a.cache, "scb-namn-2022.xlsx", magic=b"PK")
+    first = [{"name": cased(n), "sex": sex, "count": c} for sex, sheet in SHEETS.items() for n, c in counted(xlsx.rows(data, sheet), a.first)]
+    last = [{"name": cased(n), "count": c} for n, c in counted(xlsx.rows(data, SURNAMES), a.last)]
+    out = Path(a.out)
+    tsv.write(out / "first-name.tsv", ["name", "sex", "count"], sorted(first, key=lambda r: (r["name"], r["sex"])))
+    tsv.write(out / "last-name.tsv", ["name", "count"], sorted(last, key=lambda r: r["name"]))
+
+
+if __name__ == "__main__":
+    main()

data-import/names-us.py

@@ -0,0 +1,79 @@
+#!/usr/bin/env python3
+"""Rebuild data/en_US/first-name.tsv and last-name.tsv from SSA baby names and the Census 2010 surnames (public domain).
+
+    data-import/names-us.py [--names URL_OR_FILE] [--surnames URL_OR_FILE] [--from-year YEAR] [--first N] [--last N] [--cache DIR] [--out DIR]
+
+A given name is counted over the births of --from-year and later, under each sex it was given to, so a
+name both sexes carry has two rows. --names takes SSA's names.zip or a merged name,sex,count,year file;
+--surnames takes the Census names.zip or its Names_2010Census.csv.
+"""
+import argparse
+import collections
+import csv
+import io
+import re
+import zipfile
+from pathlib import Path
+
+import source
+import tsv
+
+NAMES = "https://raw.githubusercontent.com/hackerb9/ssa-baby-names/master/alldata.txt"
+SURNAMES = "https://www2.census.gov/topics/genealogy/2010surnames/names.zip"
+OUT = Path(__file__).resolve().parent.parent / "data" / "en_US"
+CACHE = Path(__file__).resolve().parent / "cache"
+MC = re.compile(r"^Mc([a-z])")
+PLACEHOLDERS = {"Baby", "Female", "Infant", "Male", "Notnamed", "Unknown", "Unnamed"}
+
+
+def csv_or_zip_rows(data, member):
+    """The rows of a CSV, or of every member of a zip named like member, name,sex,count[,year]."""
+    if data.startswith(b"PK"):
+        z = zipfile.ZipFile(io.BytesIO(data))
+        for n in sorted(z.namelist()):
+            if re.fullmatch(member, n):
+                year = re.sub(r"\D", "", n)
+                for r in csv.reader(io.StringIO(z.read(n).decode("utf-8-sig"))):
+                    yield r + [year] if year else r
+        return
+    yield from csv.reader(io.StringIO(data.decode("utf-8-sig")))
+
+
+def given(data, from_year):
+    counts = collections.Counter()
+    for r in csv_or_zip_rows(data, r"yob\d{4}\.txt"):
+        if len(r) >= 4 and r[3].isdigit() and int(r[3]) >= from_year and r[0] not in PLACEHOLDERS:
+            counts[(r[0], r[1].lower())] += int(r[2])
+    return counts
+
+
+def surname(name):
+    """A Census uppercase surname as it is written: capitalised, and Mc before a capital."""
+    return MC.sub(lambda m: "Mc" + m.group(1).upper(), name.title())
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--first", type=int, default=2000, help="names per sex")
+    p.add_argument("--from-year", type=int, default=1930)
+    p.add_argument("--last", type=int, default=5000)
+    p.add_argument("--names", default=NAMES)
+    p.add_argument("--out", default=str(OUT))
+    p.add_argument("--surnames", default=SURNAMES)
+    a = p.parse_args()
+    counts = given(source.fetch(a.names, a.cache, "ssa-names.txt"), a.from_year)
+    first = []
+    for sex in ("f", "m"):
+        top = sorted(((n, c) for (n, s), c in counts.items() if s == sex), key=lambda n: (-n[1], n[0]))[:a.first]
+        first += [{"name": n, "sex": sex, "count": c} for n, c in top]
+    rows = csv_or_zip_rows(source.fetch(a.surnames, a.cache, "census-surnames-2010.zip"), r"Names_2010Census\.csv")
+    last = [{"name": surname(r[0]), "count": int(r[2])} for r in rows if len(r) >= 3 and r[2].isdigit() and r[0].isalpha()]
+    last = sorted(last, key=lambda r: (-r["count"], r["name"]))[:a.last]
+    out = Path(a.out)
+    tsv.write(out / "first-name.tsv", ["name", "sex", "count"], sorted(first, key=lambda r: (r["name"], r["sex"])))
+    tsv.write(out / "last-name.tsv", ["name", "count"], sorted(last, key=lambda r: r["name"]))
+
+
+if __name__ == "__main__":
+    main()

data-import/port.py

@@ -0,0 +1,55 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/port.tsv from the IANA Service Name and Transport Protocol Port Number Registry.
+
+    data-import/port.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
+
+A row needs a service name and a numeric TCP port. A port the registry lists more than
+once keeps the first service it describes, or the first of them where it describes none,
+so a number selects one row.
+"""
+import argparse
+import csv
+import io
+import sys
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://www.iana.org/assignments/service-names-port-numbers/service-names-port-numbers.csv"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "port.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["number", "service"]
+FLOOR = 4000
+
+
+def rows(text):
+    best, described = {}, set()
+    for r in csv.DictReader(io.StringIO(text)):
+        service, number = (r["Service Name"] or "").strip(), (r["Port Number"] or "").strip()
+        transport, description = (r["Transport Protocol"] or "").strip(), (r["Description"] or "").strip()
+        if transport != "tcp" or not service or not number.isdigit():
+            continue
+        if not 1 <= int(number) <= 65535:
+            sys.exit(f"{service}: port {number} is outside 1-65535")
+        if number not in best or (description and number not in described):
+            best[number] = {"number": number, "service": service}
+        if description:
+            described.add(number)
+    return sorted(best.values(), key=lambda r: int(r["number"]))
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = rows(source.fetch(a.source, a.cache, "service-names-port-numbers.csv").decode("utf-8"))
+    if len(table) < FLOOR:
+        sys.exit(f"only {len(table)} ports named a TCP service; the registry's columns have moved")
+    tsv.write(a.out, COLUMNS, table)
+
+
+if __name__ == "__main__":
+    main()

data-import/protocol.py

@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/protocol.tsv from the IANA Protocol Numbers registry.
+
+    data-import/protocol.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
+
+A row needs a number and a keyword, which drops the seven numbers assigned to a class of
+protocols rather than to one; a keyword the registry reserves or marks deprecated goes
+too. Where the registry names no protocol beside the keyword, the keyword is the name.
+"""
+import argparse
+import csv
+import io
+import sys
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://www.iana.org/assignments/protocol-numbers/protocol-numbers-1.csv"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "protocol.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["keyword", "name", "number"]
+SKIP = ("Reserved", "deprecated")
+SELECTOR = '[]{}"|'  # mirrors inSelector in internal/grammar/path.go
+
+
+def rows(text):
+    for r in csv.DictReader(io.StringIO(text)):
+        number, keyword = (r["Decimal"] or "").strip(), (r["Keyword"] or "").strip()
+        if not number.isdigit() or not keyword or any(s in keyword for s in SKIP):
+            continue
+        if not 0 <= int(number) <= 255:
+            sys.exit(f"{keyword}: protocol number {number} is outside 0-255")
+        yield {"keyword": keyword, "name": " ".join((r["Protocol"] or "").split()) or keyword, "number": number}
+
+
+def refuse_an_unselectable_row(table):
+    """What the loader refuses at New, and a name that names two rows rather than one."""
+    keywords, names = {}, {}
+    for r in table:
+        for column in ("keyword", "name"):
+            if set(r[column]) & set(SELECTOR):
+                sys.exit(f"{r['keyword']}: {column} {r[column]!r} holds one of {SELECTOR}, which a selector cannot spell")
+        if r["keyword"] in keywords:
+            sys.exit(f"{r['keyword']}: repeats the keyword of protocol {keywords[r['keyword']]}")
+        keywords[r["keyword"]] = r["number"]
+    for r in table:
+        if r["name"] != r["keyword"] and r["name"] in keywords:
+            sys.exit(f"{r['keyword']}: name {r['name']!r} is another protocol's keyword")
+        if r["name"] in names:
+            sys.exit(f"{r['keyword']}: name {r['name']!r} repeats {names[r['name']]}'s, so it selects neither")
+        names[r["name"]] = r["keyword"]
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = sorted(rows(source.fetch(a.source, a.cache, "protocol-numbers-1.csv").decode("utf-8")), key=lambda r: int(r["number"]))
+    refuse_an_unselectable_row(table)
+    tsv.write(a.out, COLUMNS, table)
+
+
+if __name__ == "__main__":
+    main()

data-import/source.py

@@ -0,0 +1,29 @@
+"""A source fetched once into the cache."""
+import re
+import sys
+import time
+import urllib.request
+from pathlib import Path
+
+
+def fetch(source, cache, name, magic=b"", data=None, headers=None):
+    """The bytes of a URL, downloaded into cache/name once, or of a local file."""
+    if not re.match(r"^https?://", source):
+        return Path(source).read_bytes()
+    path = Path(cache) / name
+    if path.exists():
+        return path.read_bytes()
+    for attempt in range(1, 6):
+        req = urllib.request.Request(source, data=data, headers={"User-Agent": "fejkdata-data-import (https://github.com/larvit/fejkdata)", **(headers or {})})
+        try:
+            with urllib.request.urlopen(req, timeout=600) as r:
+                body = r.read()
+        except OSError:
+            body = b""
+        if body and body.startswith(magic) and b"Request Rejected" not in body[:512]:
+            path.parent.mkdir(parents=True, exist_ok=True)
+            path.write_bytes(body)
+            return body
+        if attempt < 5:
+            time.sleep(10 * attempt)
+    sys.exit(f"{source}: no valid download in 5 attempts")

data-import/territory.py

@@ -0,0 +1,85 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/territory.tsv from datasets/country-codes (PDDL).
+
+    data-import/territory.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
+"""
+import argparse
+import csv
+import io
+import re
+import sys
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://raw.githubusercontent.com/datasets/country-codes/main/data/country-codes.csv"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "territory.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["alpha2", "alpha3", "calling-code", "capital", "country", "currency", "flag", "languages", "name", "numeric", "tld"]
+SOVEREIGN = re.compile(r"^(?:Part of|Territor(?:y|ies) of|Crown dependency of|Commonwealth of|Associated with) ([A-Z]{2})$")
+STANDALONE = {"In contention", "International"}
+# Gaps in the source, keyed by alpha2.
+FIXUPS = {"TR": {"currency": "TRY"}}
+
+
+def flag(alpha2):
+    return "".join(chr(0x1F1E6 + ord(c) - ord("A")) for c in alpha2)
+
+
+def calling_code(dial):
+    first = re.sub(r"[^0-9-]", "", dial.split(",")[0])
+    return first if re.match(r"^\d+-\d{3}$", first) else first.split("-")[0]
+
+
+def languages(field):
+    return ",".join(p for p in field.split(",") if p)
+
+
+def country(alpha2, independent):
+    """The sovereign state the register records; a territory it records none for stands alone."""
+    m = SOVEREIGN.match(independent)
+    if m:
+        return m.group(1)
+    if independent == "Yes" or independent in STANDALONE:
+        return alpha2
+    sys.exit(f"{alpha2}: is_independent {independent!r} names no sovereign this script can read")
+
+
+def rows(text):
+    for r in csv.DictReader(io.StringIO(text)):
+        alpha2 = r["ISO3166-1-Alpha-2"]
+        row = {
+            "alpha2": alpha2,
+            "alpha3": r["ISO3166-1-Alpha-3"],
+            "calling-code": calling_code(r["Dial"]),
+            "capital": r["Capital"],
+            "country": country(alpha2, r["is_independent"]),
+            "currency": r["ISO4217-currency_alphabetic_code"].split(",")[0],
+            "flag": flag(alpha2),
+            "languages": languages(r["Languages"]),
+            "name": r["CLDR display name"] or r["official_name_en"],
+            "numeric": r["ISO3166-1-numeric"].zfill(3),
+            "tld": r["TLD"],
+        }
+        row.update(FIXUPS.get(alpha2, {}))
+        if all(row[c] for c in ("alpha2", "alpha3", "calling-code", "capital", "currency", "name", "tld")):
+            yield row
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = sorted(rows(source.fetch(a.source, a.cache, "country-codes.csv").decode("utf-8")), key=lambda r: r["alpha2"])
+    shipped = {r["alpha2"] for r in table}
+    for r in table:
+        if r["country"] not in shipped:
+            sys.exit(f"{r['alpha2']}: country {r['country']} is no shipped territory; a sovereign the set drops leaves its territories unreadable")
+    tsv.write(a.out, COLUMNS, table)
+
+
+if __name__ == "__main__":
+    main()

data-import/timezone.py

@@ -0,0 +1,138 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/timezone.tsv from the IANA tzdb tarball (public domain) and GeoNames (CC BY 4.0).
+
+    data-import/timezone.py [--source URL_OR_FILE] [--cities URL_OR_FILE] [--cache DIR] [--out FILE] [--territories FILE]
+
+The offset is the zone's standard offset, the first field of its Zone rule's last
+continuation line, with a Link resolved to its target. The weight is the population
+GeoNames records in the zone's cities of 15,000 or more, never below that threshold,
+so a drawn zone is one people live in.
+"""
+import argparse
+import csv
+import io
+import re
+import sys
+import tarfile
+import zipfile
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://data.iana.org/time-zones/tzdata-latest.tar.gz"
+CITIES = "https://download.geonames.org/export/dump/cities15000.zip"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "timezone.tsv"
+TERRITORIES = Path(__file__).resolve().parent.parent / "data" / "misc" / "territory.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["offset", "territory", "weight", "zone"]
+REGIONS = ["africa", "antarctica", "asia", "australasia", "backward", "etcetera", "europe", "northamerica", "southamerica"]
+# GeoNames' own cutoff, so a zone it records no city in weighs less than one would.
+FLOOR = 15000
+
+
+def member(tar, name):
+    try:
+        f = tar.extractfile(name)
+    except KeyError:
+        f = None
+    if f is None:
+        sys.exit(f"{name}: the tarball no longer holds it as a file; the tzdb layout has moved")
+    return f.read().decode("utf-8")
+
+
+def offsets(tar):
+    """Every zone's standard offset, and every link's target."""
+    std, links = {}, {}
+    for name in REGIONS:
+        zone = None
+        for raw in member(tar, name).splitlines():
+            line = raw.split("#")[0].rstrip()
+            if not line.strip():
+                continue
+            if line.startswith("Zone"):
+                f = line.split()
+                zone, std[f[1]] = f[1], f[2]
+            elif line.startswith("Link"):
+                f = line.split()
+                links[f[2]], zone = f[1], None
+            elif line[0] in " \t" and zone:
+                std[zone] = line.split()[0]
+            else:
+                zone = None
+    return std, links
+
+
+def resolve(zone, std, links):
+    for _ in range(10):
+        if zone in std:
+            return std[zone]
+        if zone not in links:
+            return None
+        zone = links[zone]
+    return None
+
+
+def utc_offset(raw):
+    """±HH:MM from a tzdb STDOFF field, which writes the minutes and seconds only when it has them."""
+    m = re.match(r"^(-)?(\d{1,2})(?::(\d{2})(?::(\d{2}))?)?$", raw)
+    if not m or (m.group(4) or "00") != "00":
+        return None
+    return f"{'-' if m.group(1) else '+'}{int(m.group(2)):02d}:{m.group(3) or '00'}"
+
+
+def population(body):
+    """The population GeoNames records per timezone, summed over its cities."""
+    with zipfile.ZipFile(io.BytesIO(body)) as z:
+        text = z.read("cities15000.txt").decode("utf-8")
+    per = {}
+    for city in csv.reader(io.StringIO(text), delimiter="\t", quoting=csv.QUOTE_NONE):
+        if len(city) > 17 and city[17] and city[14].isdigit():
+            per[city[17]] = per.get(city[17], 0) + int(city[14])
+    return per
+
+
+def rows(tar, territories, pop):
+    std, links = offsets(tar)
+    for line in member(tar, "zone.tab").splitlines():
+        if line.startswith("#") or not line.strip():
+            continue
+        fields = line.split("\t")
+        if len(fields) < 3:
+            sys.exit(f"zone.tab: {line!r} has {len(fields)} fields; a row names a territory, a location and a zone")
+        territory, zone = fields[0], fields[2]
+        if territory not in territories:
+            continue
+        raw = resolve(zone, std, links)
+        if raw is None:
+            sys.exit(f"{zone}: the tarball gives it no Zone rule and no Link to one")
+        offset = utc_offset(raw)
+        if offset is None:
+            sys.exit(f"{zone}: standard offset {raw!r} is not a ±HH:MM the table can spell")
+        yield {"offset": offset, "territory": territory, "weight": max(pop.get(zone, 0), FLOOR), "zone": zone}
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    p.add_argument("--cities", default=CITIES)
+    p.add_argument("--territories", default=str(TERRITORIES))
+    a = p.parse_args()
+    shipped = {r["alpha2"] for r in csv.DictReader(io.StringIO(Path(a.territories).read_text(encoding="utf-8")), delimiter="\t")}
+    pop = population(source.fetch(a.cities, a.cache, "cities15000.zip", magic=b"PK"))
+    body = source.fetch(a.source, a.cache, "tzdata-latest.tar.gz", magic=b"\x1f\x8b")
+    with tarfile.open(fileobj=io.BytesIO(body)) as tar:
+        table = sorted(rows(tar, shipped, pop), key=lambda r: r["zone"])
+        print(f"tzdb {tar.extractfile('version').read().decode('utf-8').strip()}", file=sys.stderr)
+    linked = {r["territory"] for r in table}
+    if missing := shipped - linked:
+        sys.exit(f"no zone for {sorted(missing)}; a parent row without a child is a load error")
+    if (floored := sum(1 for r in table if r["weight"] == FLOOR)) > len(table) // 2:
+        sys.exit(f"{floored} of {len(table)} zones fell to the floor; GeoNames' population or timezone column has moved")
+    tsv.write(a.out, COLUMNS, table)
+
+
+if __name__ == "__main__":
+    main()

data-import/tld.py

@@ -0,0 +1,68 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/tld.tsv from the IANA Root Zone Database.
+
+    data-import/tld.py [--source URL_OR_FILE] [--zone URL_OR_FILE] [--cache DIR] [--out FILE]
+
+A TLD ships when the database names a manager for it; one it marks "Not assigned" is a
+delegation the root zone no longer holds, and the zone file is fetched to prove the two
+agree. The key is the A-label the root zone holds; `unicode` is the form the database
+displays, less the bidi marks it wraps a right-to-left label in, and proved to punycode
+back to the key.
+"""
+import argparse
+import html
+import re
+import sys
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://www.iana.org/domains/root/db"
+ZONE = "https://data.iana.org/TLD/tlds-alpha-by-domain.txt"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "tld.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["tld", "type", "unicode"]
+TYPES = ("country-code", "generic", "generic-restricted", "infrastructure", "sponsored")
+BIDI = {0x200E: None, 0x200F: None}
+ROW = re.compile(r'<a href="/domains/root/db/([^"]+)\.html">([^<]+)</a>.*?</td>\s*<td>([^<]*)</td>\s*<td>(.*?)</td>', re.S)
+
+
+def rows(page):
+    for a_label, shown, kind, manager in ROW.findall(page):
+        if "Not assigned" in manager:
+            continue
+        kind = kind.strip()
+        if kind not in TYPES:
+            sys.exit(f"{a_label}: the database types it {kind!r}, which is none of {TYPES}")
+        yield {"tld": "." + a_label, "type": kind, "unicode": html.unescape(shown).strip().translate(BIDI)}
+
+
+def prove_each_unicode_form_is_its_key(table):
+    """The displayed form is the key's own spelling, which punycode encodes back to."""
+    for r in table:
+        label = r["unicode"][1:]
+        encoded = label if label.isascii() else "xn--" + label.encode("punycode").decode("ascii")
+        if "." + encoded != r["tld"]:
+            sys.exit(f"{r['tld']}: the database displays {r['unicode']!r}, which is no spelling of it")
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--zone", default=ZONE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    table = sorted(rows(source.fetch(a.source, a.cache, "root-db.html").decode("utf-8")), key=lambda r: r["tld"])
+    prove_each_unicode_form_is_its_key(table)
+    listing = source.fetch(a.zone, a.cache, "tlds-alpha-by-domain.txt").decode("utf-8")
+    delegated = {"." + line.strip().lower() for line in listing.splitlines() if line.strip() and not line.startswith("#")}
+    shipped = {r["tld"] for r in table}
+    if len(table) != len(delegated) or shipped != delegated:
+        sys.exit(f"the database assigns {len(table)} rows where the root zone holds {len(delegated)} TLDs: {sorted(shipped ^ delegated)[:10]}")
+    tsv.write(a.out, COLUMNS, table)
+
+
+if __name__ == "__main__":
+    main()

data-import/tsv.py

@@ -0,0 +1,18 @@
+"""A table written as the loader admits it."""
+import re
+import sys
+from pathlib import Path
+
+
+def write(path, columns, rows):
+    """Write the rows as a TSV of a header and at least one row, each cell stripped, non-empty and free of tabs, newlines and braces."""
+    lines = ["\t".join(columns)]
+    for row in rows:
+        cells = [str(row[c]).strip() for c in columns]
+        if not all(cells) or any(re.search(r"[\t\n{}]", c) for c in cells):
+            raise ValueError(f"{path}: a cell is empty or holds a tab, newline or brace: {row}")
+        lines.append("\t".join(cells))
+    if len(lines) < 2:
+        raise ValueError(f"{path}: has no rows below its header; a table holds at least one row")
+    Path(path).write_text("\n".join(lines) + "\n", encoding="utf-8")
+    print(f"{path}: {len(lines) - 1} rows", file=sys.stderr)

data-import/useragent.py

@@ -0,0 +1,64 @@
+#!/usr/bin/env python3
+"""Rebuild data/misc/useragent.tsv from microlinkhq/top-user-agents (MIT).
+
+    data-import/useragent.py [--source URL] [--cache DIR] [--out FILE]
+
+The desktop and mobile lists partition the whole, so `device` is read rather than
+parsed. A row ships when the string names a browser and an operating system both.
+"""
+import argparse
+import json
+import re
+import sys
+from pathlib import Path
+
+import source
+import tsv
+
+SOURCE = "https://raw.githubusercontent.com/microlinkhq/top-user-agents/master/src"
+OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "useragent.tsv"
+CACHE = Path(__file__).resolve().parent / "cache"
+COLUMNS = ["browser", "device", "os", "ua"]
+# First match wins: a Chromium fork this list does not name ships as Chrome.
+BROWSERS = [("Edge", r"Edg(A|iOS)?/"), ("Opera", r"OPR/"), ("Samsung Internet", r"SamsungBrowser/"),
+            ("Chrome", r"(Chrome|CriOS)/"), ("Firefox", r"(Firefox|FxiOS)/"), ("Safari", r"Version/[\d.]+ .*Safari")]
+SYSTEMS = [("iOS", r"iPhone|iPad|CPU OS "), ("Android", r"Android"), ("ChromeOS", r"CrOS"),
+           ("Windows", r"Windows NT"), ("macOS", r"Macintosh|Mac OS X"), ("Linux", r"X11.*Linux|Ubuntu")]
+
+
+def named(table, ua):
+    for name, pattern in table:
+        if re.search(pattern, ua):
+            return name
+    return ""
+
+
+def rows(desktop, mobile):
+    seen, kept = set(), []
+    for device, uas in (("desktop", desktop), ("mobile", mobile)):
+        for ua in uas:
+            browser, os = named(BROWSERS, ua), named(SYSTEMS, ua)
+            if not browser or not os:
+                print(f"dropped, naming no {'browser' if not browser else 'operating system'}: {ua}", file=sys.stderr)
+                continue
+            if ua in seen:
+                sys.exit(f"{ua}: listed twice, which would draw it twice as often")
+            seen.add(ua)
+            kept.append({"browser": browser, "device": device, "os": os, "ua": ua})
+    if len(kept) < 0.8 * (len(desktop) + len(mobile)):
+        sys.exit(f"only {len(kept)} of {len(desktop) + len(mobile)} strings named both; the tokens have moved")
+    return kept
+
+
+def main():
+    p = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    p.add_argument("--cache", default=str(CACHE))
+    p.add_argument("--source", default=SOURCE)
+    p.add_argument("--out", default=str(OUT))
+    a = p.parse_args()
+    lists = [json.loads(source.fetch(f"{a.source}/{n}.json", a.cache, f"top-user-agents-{n}.json")) for n in ("desktop", "mobile")]
+    tsv.write(a.out, COLUMNS, sorted(rows(*lists), key=lambda r: (r["browser"], r["os"], r["ua"])))
+
+
+if __name__ == "__main__":
+    main()

data-import/xlsx.py

@@ -0,0 +1,21 @@
+"""The rows of an xlsx sheet."""
+import io
+import xml.etree.ElementTree as ET
+import zipfile
+
+NS = {"m": "http://schemas.openxmlformats.org/spreadsheetml/2006/main", "r": "http://schemas.openxmlformats.org/officeDocument/2006/relationships"}
+
+
+def rows(data, sheet=None):
+    """The rows of an xlsx sheet named sheet, the first sheet by default, each a list of cell texts."""
+    z = zipfile.ZipFile(io.BytesIO(data))
+    strings = ["".join(t.text or "" for t in si.iter("{%s}t" % NS["m"])) for si in ET.fromstring(z.read("xl/sharedStrings.xml")).findall("m:si", NS)]
+    rels = {r.get("Id"): r.get("Target") for r in ET.fromstring(z.read("xl/_rels/workbook.xml.rels"))}
+    sheets = {s.get("name"): rels[s.get("{%s}id" % NS["r"])] for s in ET.fromstring(z.read("xl/workbook.xml")).iter("{%s}sheet" % NS["m"])}
+    target = sheets[sheet] if sheet else next(iter(sheets.values()))
+    for row in ET.fromstring(z.read("xl/" + target)).findall(".//m:row", NS):
+        cells = []
+        for c in row.findall("m:c", NS):
+            v = c.find("m:v", NS)
+            cells.append("" if v is None else strings[int(v.text)] if c.get("t") == "s" else v.text)
+        yield cells

data.go

@@ -1,141 +1,175 @@
 package fejkdata
 
 import (
-	"encoding/json"
-	"fmt"
-	"io/fs"
-	"os"
-	"path"
 	"strings"
+
+	"github.com/larvit/fejkdata/internal/datafiles"
 )
 
-// dataSource is one tree to load: an fs.FS and the directory in it to start from.
-// label prefixes file names in errors; path, when set, is a directory on disk that
-// must exist.
-type dataSource struct {
-	fsys  fs.FS
-	label string
-	path  string
-	root  string
-}
-
-func (s dataSource) name(p string) string {
-	if s.label == "" {
-		return p
-	}
-	return path.Join(s.label, p)
-}
-
-// loadData loads every source into one namespace tree and returns its root
-// children. A directory becomes a group; each *.json file in it compiles to a node
-// keyed by its base name (address.json -> "address"); each subdirectory becomes a
-// nested group, so folders turn into dot-path segments. Sources merge left to right:
-// matching groups merge by their children, and any other clash is won by the last
-// source loaded. Once merged, linkRefs binds every {..path} reference against the
-// final tree.
-func loadData(sources []dataSource) (map[string]node, error) {
-	root := map[string]node{}
+// loadSources merges sources in order, the last winning a clash. A source whose manifest
+// carries an index leaves its categories unloaded; every other source loads here, with
+// the unloaded categories it needs.
+// docs/decisions.md#a-source-whose-manifest-carries-an-index-loads-each-category-on-first-reach-any-other-loads-in-new-with-what-it-reads
+func loadSources(sources []datafiles.Source) (folder, error) {
+	root := folder{children: map[string]node{}}
 	for _, src := range sources {
-		if src.path != "" {
-			info, err := os.Stat(src.path)
-			if err != nil {
-				return nil, fmt.Errorf("%s: no such directory", src.path)
-			}
-			if !info.IsDir() {
-				return nil, fmt.Errorf("%s is not a directory", src.path)
-			}
-		}
-		dir := src.root
-		if dir == "" {
-			dir = "."
-		}
-		g, err := loadDir(src, dir)
+		m, err := src.Manifest()
 		if err != nil {
-			return nil, err
+			return folder{}, err
+		}
+		if err := addSource(&root, src, m); err != nil {
+			return folder{}, err
 		}
-		mergeChildren(root, g.children)
 	}
-	if len(root) == 0 {
-		return nil, fmt.Errorf("no .json data found")
-	}
-	if err := linkRefs(root); err != nil {
-		return nil, err
-	}
-	if err := checkNoCycles(root); err != nil {
-		return nil, err
-	}
-	if err := checkRepeatReach(root); err != nil {
-		return nil, err
-	}
-	if err := checkBoundLevelsHeld(root); err != nil {
-		return nil, err
-	}
-	return root, nil
+	return root, loadReached(&root, categorySites(&root), nil)
 }
 
-// loadDir compiles one directory into a group. fs.ReadDir yields entries sorted
-// by name, so the tree is built deterministically. Empty subdirectories (no JSON
-// anywhere under them) are skipped rather than added as empty namespaces.
-func loadDir(src dataSource, dir string) (*group, error) {
-	entries, err := fs.ReadDir(src.fsys, dir)
-	if err != nil {
-		return nil, fmt.Errorf("%s: %w", src.name(dir), err)
+func addSource(root *folder, src datafiles.Source, m datafiles.Manifest) error {
+	src = src.Mounted(m.Namespace)
+	if m.Index != nil {
+		placeIndex(root, src, m.Index)
+		return nil
 	}
-	g := &group{children: map[string]node{}}
-	for _, e := range entries {
-		if strings.HasPrefix(e.Name(), ".") { // hidden: a checkout or an editor's file, never data
-			continue
-		}
-		full := path.Join(dir, e.Name())
-		if e.IsDir() {
-			child, err := loadDir(src, full)
-			if err != nil {
-				return nil, err
-			}
-			if len(child.children) == 0 {
-				continue
-			}
-			if err := checkName(e.Name()); err != nil {
-				return nil, fmt.Errorf("%s: folder %w", src.name(full), err)
-			}
-			g.children[e.Name()] = child
-			continue
-		}
-		if !strings.HasSuffix(e.Name(), ".json") {
-			continue
-		}
-		name := strings.TrimSuffix(e.Name(), ".json")
-		if err := checkName(name); err != nil {
-			return nil, fmt.Errorf("%s: category %w", src.name(full), err)
-		}
-		b, err := fs.ReadFile(src.fsys, full)
-		if err != nil {
-			return nil, fmt.Errorf("%s: %w", src.name(full), err)
-		}
-		var raw any
-		if err := json.Unmarshal(b, &raw); err != nil {
-			return nil, fmt.Errorf("%s: %w", src.name(full), err)
-		}
-		n, err := compile(raw)
-		if err != nil {
-			return nil, fmt.Errorf("%s: %w", src.name(full), err)
-		}
-		g.children[name] = n
+	g := &folder{children: map[string]node{}}
+	if err := src.Walk(compileInto(func(dir []string) *folder { return madeFolder(g, dir) })); err != nil {
+		return err
 	}
-	return g, nil
+	mergeFolder(root, g)
+	return nil
 }
 
-// mergeChildren overlays src onto dst. Two groups under the same key merge
-// recursively (so locales/categories from several paths combine); every other
-// key is replaced, making the last-loaded directory win on a conflict.
-func mergeChildren(dst, src map[string]node) {
-	for k, v := range src {
-		if dg, ok := dst[k].(*group); ok {
-			if sg, ok := v.(*group); ok {
-				mergeChildren(dg.children, sg.children)
+func compileInto(place func(dir []string) *folder) func(datafiles.Category) error {
+	return func(c datafiles.Category) error {
+		n, err := compileCategory(c)
+		if err != nil {
+			return err
+		}
+		place(c.Folders).put(c.Name, n)
+		return nil
+	}
+}
+
+// madeFolder is the folder that dir names under root, made where it is missing or a category.
+func madeFolder(root *folder, dir []string) *folder {
+	g := root
+	for _, seg := range dir {
+		sub, isFolder := g.children[seg].(*folder)
+		if !isFolder {
+			sub = &folder{children: map[string]node{}}
+			g.put(seg, sub)
+		}
+		g = sub
+	}
+	return g
+}
+
+func mergeFolder(dst, src *folder) {
+	for k, v := range src.children {
+		if dg, ok := dst.children[k].(*folder); ok {
+			if sg, ok := v.(*folder); ok {
+				mergeFolder(dg, sg)
 				continue
 			}
 		}
-		dst[k] = v
+		dst.put(k, v)
 	}
 }
+
+// nodeSet is the set of nodes one validation pass covers: one load's categories, or a
+// single inline node.
+type nodeSet func(fn func(label string, n node) error) error
+
+// categorySite is a loaded category, where it sits and its own dot path.
+type categorySite struct {
+	categoryAt
+	n    node
+	path string
+}
+
+// categorySites lists every category under root, folders and names in sorted order.
+func categorySites(root *folder) []categorySite {
+	var out []categorySite
+	var walk func(dir []string, g *folder)
+	walk = func(dir []string, g *folder) {
+		for _, name := range sortedNames(g.children) {
+			if sub, isFolder := g.children[name].(*folder); isFolder {
+				walk(append(dir[:len(dir):len(dir)], name), sub)
+				continue
+			}
+			out = append(out, siteIn(dir, g, name))
+		}
+	}
+	walk(nil, root)
+	return out
+}
+
+func siteIn(dir []string, in *folder, name string) categorySite {
+	return categorySite{categoryAt: categoryAt{dir: dir, in: in, name: name}, n: in.children[name], path: categoryPath(dir, name)}
+}
+
+func categoryPath(dir []string, name string) string { return join(strings.Join(dir, "."), name) }
+
+func siteNodes(sites []categorySite) nodeSet {
+	return func(fn func(label string, n node) error) error {
+		for _, s := range sites {
+			if err := eachNode(s.n, s.path, fn); err != nil {
+				return err
+			}
+		}
+		return nil
+	}
+}
+
+func inlineNodes(n node, label string) nodeSet {
+	return func(fn func(label string, m node) error) error { return eachNode(n, label, fn) }
+}
+
+// pipeline is the one load pipeline, run over one nodeSet; its fields are where
+// entry points differ.
+type pipeline struct {
+	nodes   nodeSet
+	resolve func() error
+	// Set by a caller that proves the columns against their Go types itself.
+	typedByGo bool
+}
+
+// categoryPipeline is the pipeline over the categories of one load, their references
+// resolving against root.
+func categoryPipeline(sites []categorySite, root map[string]node) pipeline {
+	return pipeline{
+		nodes: siteNodes(sites),
+		resolve: func() error {
+			if err := linkTables(sites); err != nil {
+				return err
+			}
+			return resolveCategoryTemplates(sites, root)
+		},
+	}
+}
+
+func (p pipeline) run() error {
+	if err := p.resolve(); err != nil {
+		return err
+	}
+	if err := checkNoCycles(p.nodes); err != nil {
+		return err
+	}
+	settleRecords(p.nodes)
+	if !p.typedByGo {
+		if err := checkColumns(p.nodes); err != nil {
+			return err
+		}
+	}
+	return checkNodeFences(p.nodes)
+}
+
+// checkNodeFences runs the per-node fences over s, each over every node before the next,
+// so which of several broken nodes is reported does not depend on the walk. Its walks
+// recurse unguarded, so pipeline.run refuses cycles first.
+func checkNodeFences(s nodeSet) error {
+	mem := renderCounts{}
+	if err := s(func(label string, n node) error { return repeatCheck(label, n, mem) }); err != nil {
+		return err
+	}
+	return s((&valueProof{}).checkDatatype)
+}

data/en_US/.fejkdata.json

@@ -0,0 +1,27 @@
+{
+	"index": {
+		"en_US.address": {"paths":["","locality","postal-code","region","street","street-number"]},
+		"en_US.color": {"paths":[""]},
+		"en_US.company": {"paths":["","base","suffix"]},
+		"en_US.date": {"paths":[""]},
+		"en_US.email": {"paths":["","domain","local","local.n"]},
+		"en_US.first-name": {"parent":"sex","paths":["","count","name","sex"]},
+		"en_US.ip": {"paths":[""]},
+		"en_US.itin": {"paths":["","area","group","serial"]},
+		"en_US.last-name": {"paths":["","count","name"]},
+		"en_US.person": {"paths":["","first","last","prefix","sex"]},
+		"en_US.phone": {"paths":["","area","exchange","line"]},
+		"en_US.phone-exchange": {"paths":[""]},
+		"en_US.price": {"paths":["","amt","cents"]},
+		"en_US.sentence": {"paths":["","adj","noun","prep","verb"]},
+		"en_US.sex": {"paths":["","code","first-name","first-name.count","first-name.name","first-name.sex","name","title","title.name","title.sex","title.share"]},
+		"en_US.ssn": {"paths":["","area","group","serial"]},
+		"en_US.time": {"paths":[""]},
+		"en_US.title": {"parent":"sex","paths":["","name","sex","share"]},
+		"en_US.url": {"paths":["","host","path"]},
+		"en_US.username": {"paths":["","n"]},
+		"en_US.version": {"paths":["","n","pre","suffix"]},
+		"en_US.word": {"paths":[""]}
+	},
+	"namespace": "en_US"
+}

data/en_US/address.json

@@ -1,17 +1,8 @@
 {
-  "format": "{street-number} {street}\n{locality}, {region} {postal-code}",
-  "street": {
-    "format": "{name} {suffix}",
-    "name": ["Adams", "Ashby", "Aspen", "Bay", "Birch", "Bridge", "Cedar", "Chestnut", "Church", "Clark", "Cypress", "Dogwood", "Elm", "Forest", "Franklin", "Garden", "Grove", "Hawthorn", "Hickory", "Highland", "Jackson", "Jefferson", "Juniper", "Lake", "Laurel", "Liberty", "Lincoln", "Madison", "Magnolia", "Maple", "Market", "Meadow", "Mill", "Oak", "Park", "Pine", "Poplar", "Prospect", "Ridge", "River", "Spruce", "Sunset", "Sycamore", "Union", "Walnut", "Washington", "Willow", "Wilson"],
-    "suffix": ["Avenue", "Boulevard", "Circle", "Court", "Drive", "Lane", "Place", "Road", "Street", "Terrace", "Trail", "Way"]
-  },
-  "street-number": [
-    "{int(10,99)}",
-    { "format": "{int(100,999)}", "weight": 2 },
-    { "format": "{int(1000,9999)}", "weight": 0.5 },
-    { "format": "{int(10,99)}{int(100,999)}", "weight": 0.3 }
-  ],
-  "locality": ["Albany", "Atlanta", "Austin", "Baltimore", "Boston", "Charlotte", "Chicago", "Cincinnati", "Cleveland", "Columbus", "Dallas", "Denver", "Detroit", "El Paso", "Fort Worth", "Fresno", "Houston", "Indianapolis", "Jacksonville", "Kansas City", "Las Vegas", "Long Beach", "Los Angeles", "Memphis", "Mesa", "Miami", "Milwaukee", "Minneapolis", "Nashville", "New Orleans", "Oakland", "Oklahoma City", "Omaha", "Orlando", "Philadelphia", "Phoenix", "Pittsburgh", "Portland", "Raleigh", "Sacramento", "San Antonio", "San Diego", "San Jose", "Seattle", "St. Louis", "Tampa", "Tucson", "Tulsa"],
-  "region": ["AL", "AZ", "CA", "CO", "CT", "FL", "GA", "IL", "IN", "KY", "LA", "MA", "MD", "MI", "MN", "MO", "NC", "NJ", "NV", "NY", "OH", "OK", "OR", "PA", "TN", "TX", "VA", "WA", "WI"],
-  "postal-code": ["{int(10000,99999)}", { "format": "{int(10000,99999)}-{digits(4)}", "weight": 0.3 }]
+  "format": "{/geo.US.address as a}{a}",
+  "street": "{a.street}",
+  "street-number": "{a.street-number}",
+  "locality": "{a.locality}",
+  "region": "{a.region}",
+  "postal-code": "{a.postal-code}"
 }

data/en_US/date.json

@@ -1,13 +1 @@
-{
-  "format": "{month}/{day}/{year}",
-  "month": ["01", "02", "03", "04", "05", "06", "07", "08", "09", "10", "11", "12"],
-  "day": ["01", "02", "03", "04", "05", "06", "07", "08", "09", "10", "11", "12", "13", "14", "15", "16", "17", "18", "19", "20", "21", "22", "23", "24", "25", "26", "27", "28"],
-  "year": [
-    "197{digits(1)}",
-    "198{digits(1)}",
-    { "format": "199{digits(1)}", "weight": 2 },
-    { "format": "200{digits(1)}", "weight": 3 },
-    { "format": "201{digits(1)}", "weight": 3 },
-    { "format": "202{digits(1)}", "weight": 2 }
-  ]
-}
+"{date(1970-01-01,2029-12-31,'01/02/2006')}"

data/en_US/first-name.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "first-name.tsv", "name": "name", "parent": "sex", "weight": "count" }

data/en_US/fs.go

@@ -0,0 +1,19 @@
+// Package enus is fejkdata's en_US locale module, whose categories are en_US.*; its address reads
+// geo.US, which package geous holds.
+package enus
+
+import (
+	"embed"
+	"io/fs"
+
+	geous "github.com/larvit/fejkdata/data/geo/US"
+)
+
+// FS holds the locale alone, its categories and its manifest; pass Modules unless your own
+// data stands in for geo.US.
+//
+//go:embed *.json *.tsv .fejkdata.json
+var FS embed.FS
+
+// Modules returns FS and every module it reads, to pass to fejkdata.WithDataFS.
+func Modules() []fs.FS { return append([]fs.FS{FS}, geous.Modules()...) }

data/en_US/fs_test.go

@@ -0,0 +1,22 @@
+package enus_test
+
+import (
+	"io/fs"
+	"testing"
+
+	"github.com/larvit/fejkdata"
+	enus "github.com/larvit/fejkdata/data/en_US"
+)
+
+func TestFSLoadsByItsManifest(t *testing.T) {
+	if _, err := fs.Stat(enus.FS, ".fejkdata.json"); err != nil {
+		t.Fatalf("FS holds no manifest: %v", err)
+	}
+	f, err := fejkdata.New(fejkdata.WithDataFS(enus.Modules()...), fejkdata.WithSeed(1))
+	if err != nil {
+		t.Fatalf("New(WithDataFS(Modules()...)) = %v", err)
+	}
+	if len(f.List()) == 0 {
+		t.Error("List() is empty, want the module's categories")
+	}
+}

data/en_US/itin.json

@@ -0,0 +1,16 @@
+{
+  "format": "{area}-{group}-{serial}",
+  "area": "9{digits(2)}",
+  "group": [
+    { "format": "{int(50,65)}", "weight": 16 },
+    { "format": "{int(70,88)}", "weight": 19 },
+    { "format": "{int(90,92)}", "weight": 3 },
+    { "format": "{int(94,99)}", "weight": 6 }
+  ],
+  "serial": [
+    { "format": "000{int(1,9)}", "weight": 9 },
+    { "format": "00{int(10,99)}", "weight": 90 },
+    { "format": "0{int(100,999)}", "weight": 900 },
+    { "format": "{int(1000,9999)}", "weight": 9000 }
+  ]
+}

data/en_US/last-name.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "last-name.tsv", "key": "name", "weight": "count" }

data/en_US/person.json

@@ -1,7 +1,7 @@
 {
-  "format": "{prefix}{femalefirst|malefirst} {last}",
-  "femalefirst": ["Abigail", "Addison", "Amelia", "Aria", "Aubrey", "Audrey", "Aurora", "Ava", "Bella", "Brooklyn", "Camila", "Caroline", "Charlotte", "Chloe", "Claire", "Eleanor", "Elizabeth", "Ella", "Ellie", "Emily", "Emma", "Evelyn", "Gianna", "Grace", "Hannah", "Harper", "Hazel", "Isabella", "Layla", "Leah", "Lillian", "Lily", "Lucy", "Luna", "Madison", "Maya", "Mia", "Mila", "Naomi", "Natalie", "Nora", "Olivia", "Paisley", "Penelope", "Riley", "Savannah", "Scarlett", "Sofia", "Sophia", "Stella", "Victoria", "Violet", "Zoe"],
-  "malefirst": ["Aiden", "Alexander", "Andrew", "Anthony", "Asher", "Benjamin", "Caleb", "Carter", "Charles", "Christopher", "Daniel", "David", "Dylan", "Elijah", "Ethan", "Ezra", "Gabriel", "Grayson", "Henry", "Isaac", "Jack", "Jackson", "Jacob", "James", "Jayden", "John", "Joseph", "Joshua", "Julian", "Levi", "Liam", "Lincoln", "Logan", "Lucas", "Luke", "Mason", "Mateo", "Matthew", "Michael", "Nathan", "Noah", "Oliver", "Owen", "Samuel", "Sebastian", "Theodore", "Thomas", "William", "Wyatt"],
-  "last": ["Adams", "Allen", "Anderson", "Bailey", "Baker", "Bell", "Bennett", "Brooks", "Brown", "Campbell", "Carter", "Clark", "Collins", "Cook", "Cooper", "Cox", "Davis", "Edwards", "Evans", "Flores", "Foster", "Garcia", "Gonzalez", "Gray", "Green", "Hall", "Harris", "Hayes", "Hernandez", "Hill", "Howard", "Hughes", "Jackson", "James", "Jenkins", "Johnson", "Jones", "Kelly", "King", "Lee", "Lewis", "Long", "Lopez", "Martin", "Martinez", "Miller", "Mitchell", "Moore", "Morgan", "Morris", "Murphy", "Nelson", "Parker", "Perez", "Perry", "Peterson", "Phillips", "Powell", "Price", "Ramirez", "Reed", "Richardson", "Rivera", "Roberts", "Robinson", "Rodriguez", "Rogers", "Ross", "Russell", "Sanchez", "Sanders", "Scott", "Smith", "Stewart", "Sullivan", "Taylor", "Thomas", "Thompson", "Torres", "Turner", "Walker", "Ward", "Watson", "White", "Williams", "Wilson", "Wood", "Wright", "Young"],
-  "prefix": ["", { "format": "{title} ", "title": ["Dr", "Miss", "Mr", "Mrs", "Ms", "Mx", "Prof"], "weight": 0.1 }]
+  "format": "{prefix}{first} {last}",
+  "first": "{.first-name as n}{n.name}",
+  "last": "{.last-name.name}",
+  "prefix": [null, { "format": "{n..sex.title.name} ", "weight": 0.1 }],
+  "sex": "{n..sex.name}"
 }

data/en_US/phone-exchange.json

@@ -0,0 +1,17 @@
+[
+  { "format": "{int(2,9)}0{digits(1)}", "weight": 80 },
+  { "format": "{int(2,9)}1{int(2,9)}", "weight": 64 },
+  { "format": "{int(2,9)}10", "weight": 8 },
+  { "format": "{int(2,4)}{int(2,9)}{digits(1)}", "weight": 240 },
+  { "format": "{int(6,8)}{int(2,9)}{digits(1)}", "weight": 240 },
+  { "format": "5{int(2,4)}{digits(1)}", "weight": 30 },
+  { "format": "5{int(6,9)}{digits(1)}", "weight": 40 },
+  { "format": "55{int(0,4)}", "weight": 5 },
+  { "format": "55{int(6,9)}", "weight": 4 },
+  { "format": "9{int(2,4)}{digits(1)}", "weight": 30 },
+  { "format": "9{int(8,9)}{digits(1)}", "weight": 20 },
+  { "format": "95{int(1,7)}", "weight": 7 },
+  { "format": "96{digits(1)}", "weight": 10 },
+  { "format": "97{int(0,5)}", "weight": 6 },
+  { "format": "97{int(7,9)}", "weight": 3 }
+]

data/en_US/phone.json

@@ -1,15 +1,15 @@
 [
   {
-    "format": "({area}) {exch}-{line}",
+    "format": "({area}) {exchange}-{line}",
     "weight": 2,
     "area": ["202", "212", "213", "305", "312", "404", "415", "469", "503", "512", "602", "617", "646", "702", "718", "720", "737", "773", "786", "808", "917"],
-    "exch": "{int(100,999)}",
+    "exchange": "{.phone-exchange}",
     "line": "{digits(4)}"
   },
   {
-    "format": "{area}-{exch}-{line}",
+    "format": "{area}-{exchange}-{line}",
     "area": ["202", "212", "213", "305", "312", "404", "415", "469", "503", "512", "602", "617", "646", "702", "718", "720", "737", "773", "786", "808", "917"],
-    "exch": "{int(100,999)}",
+    "exchange": "{.phone-exchange}",
     "line": "{digits(4)}"
   }
 ]

data/en_US/sex.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "sex.tsv", "key": "code", "name": "name" }

data/en_US/sex.tsv

@@ -0,0 +1,3 @@
+code	name
+f	female
+m	male

data/en_US/ssn.json

@@ -1 +1,19 @@
-"{int(100,999)}-{digits(2)}-{digits(4)}"
+{
+  "format": "{area}-{group}-{serial}",
+  "area": [
+    { "format": "00{int(1,9)}", "weight": 9 },
+    { "format": "0{int(10,99)}", "weight": 90 },
+    { "format": "{int(100,665)}", "weight": 566 },
+    { "format": "{int(667,899)}", "weight": 233 }
+  ],
+  "group": [
+    { "format": "0{int(1,9)}", "weight": 9 },
+    { "format": "{int(10,99)}", "weight": 90 }
+  ],
+  "serial": [
+    { "format": "000{int(1,9)}", "weight": 9 },
+    { "format": "00{int(10,99)}", "weight": 90 },
+    { "format": "0{int(100,999)}", "weight": 900 },
+    { "format": "{int(1000,9999)}", "weight": 9000 }
+  ]
+}

data/en_US/time.json

@@ -1,6 +1 @@
-{
-  "format": "{hour}:{minute} {ampm}",
-  "hour": ["1", "2", "3", "4", "5", "6", "7", "8", "9", "10", "11", "12"],
-  "minute": { "format": "{t}{digits(1)}", "t": ["0", "1", "2", "3", "4", "5"] },
-  "ampm": ["AM", "PM"]
-}
+"{time('3:04 PM')}"

data/en_US/title.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "title.tsv", "name": "name", "parent": "sex", "weight": "share" }

data/en_US/title.tsv

@@ -0,0 +1,11 @@
+name	sex	share
+Dr	f	6
+Dr	m	6
+Miss	f	4
+Mr	m	60
+Mrs	f	22
+Ms	f	30
+Mx	f	1
+Mx	m	1
+Prof	f	3
+Prof	m	3

data/geo/SE/.fejkdata.json

@@ -0,0 +1,11 @@
+{
+	"index": {
+		"geo.SE.address": {"paths":["","locality","postal-code","street","street-number"]},
+		"geo.SE.locality": {"parent":"municipality","paths":["","lat","lon","municipality","name","population","postal-code","postal-code.code","postal-code.locality","street","street.locality","street.name","street.segments"]},
+		"geo.SE.municipality": {"parent":"region","paths":["","code","locality","locality.lat","locality.lon","locality.municipality","locality.name","locality.population","locality.postal-code","locality.postal-code.code","locality.postal-code.locality","locality.street","locality.street.locality","locality.street.name","locality.street.segments","name","population","region"]},
+		"geo.SE.postal-code": {"parent":"locality","paths":["","code","locality"]},
+		"geo.SE.region": {"paths":["","code","municipality","municipality.code","municipality.locality","municipality.locality.lat","municipality.locality.lon","municipality.locality.municipality","municipality.locality.name","municipality.locality.population","municipality.locality.postal-code","municipality.locality.postal-code.code","municipality.locality.postal-code.locality","municipality.locality.street","municipality.locality.street.locality","municipality.locality.street.name","municipality.locality.street.segments","municipality.name","municipality.population","municipality.region","name","population","timezone"]},
+		"geo.SE.street": {"parent":"locality","paths":["","locality","name","segments"]}
+	},
+	"namespace": "geo.SE"
+}

data/geo/SE/address.json

@@ -0,0 +1,14 @@
+{
+  "format": "{street} {street-number}\n{postal-code} {locality}",
+  "street": "{.locality as l}{l.street.name}",
+  "street-number": [
+    "{int(1,9)}",
+    "{int(10,99)}",
+    { "format": "{int(100,999)}", "weight": 0.2 },
+    { "format": "{int(1,9)}{upper(1)}", "weight": 0.2 },
+    { "format": "{int(10,99)}{upper(1)}", "weight": 0.1 },
+    { "format": "{int(100,999)}{upper(1)}", "weight": 0.05 }
+  ],
+  "postal-code": "{l.postal-code.code}",
+  "locality": "{l.name}"
+}

data/geo/SE/fs.go

@@ -0,0 +1,15 @@
+// Package geose is fejkdata's Swedish places module, whose categories are geo.SE.*.
+package geose
+
+import (
+	"embed"
+	"io/fs"
+)
+
+// FS holds the module's categories and its manifest.
+//
+//go:embed *.json *.tsv .fejkdata.json
+var FS embed.FS
+
+// Modules returns FS, which reads no other module, to pass to fejkdata.WithDataFS.
+func Modules() []fs.FS { return []fs.FS{FS} }

data/geo/SE/fs_test.go

@@ -0,0 +1,22 @@
+package geose_test
+
+import (
+	"io/fs"
+	"testing"
+
+	"github.com/larvit/fejkdata"
+	geose "github.com/larvit/fejkdata/data/geo/SE"
+)
+
+func TestFSLoadsByItsManifest(t *testing.T) {
+	if _, err := fs.Stat(geose.FS, ".fejkdata.json"); err != nil {
+		t.Fatalf("FS holds no manifest: %v", err)
+	}
+	f, err := fejkdata.New(fejkdata.WithDataFS(geose.Modules()...), fejkdata.WithSeed(1))
+	if err != nil {
+		t.Fatalf("New(WithDataFS(Modules()...)) = %v", err)
+	}
+	if len(f.List()) == 0 {
+		t.Error("List() is empty, want the module's categories")
+	}
+}

data/geo/SE/locality.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "locality.tsv", "key": "name", "parent": "municipality", "weight": "population" }

data/geo/SE/municipality.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "municipality.tsv", "key": "code", "name": "name", "parent": "region", "weight": "population" }

data/geo/SE/municipality.tsv

@@ -0,0 +1,291 @@
+code	name	region	population
+0114	Upplands Väsby	01	50323
+0115	Vallentuna	01	35119
+0117	Österåker	01	49787
+0120	Värmdö	01	46635
+0123	Järfälla	01	88950
+0125	Ekerö	01	28910
+0126	Huddinge	01	114304
+0127	Botkyrka	01	95905
+0128	Salem	01	17507
+0136	Haninge	01	100895
+0138	Tyresö	01	49179
+0139	Upplands-Bro	01	32868
+0140	Nykvarn	01	12342
+0160	Täby	01	77744
+0162	Danderyd	01	32425
+0163	Sollentuna	01	77624
+0180	Stockholm	01	995574
+0181	Södertälje	01	102911
+0182	Nacka	01	112112
+0183	Sundbyberg	01	56274
+0184	Solna	01	85789
+0186	Lidingö	01	48377
+0187	Vaxholm	01	11822
+0188	Norrtälje	01	66585
+0191	Sigtuna	01	52767
+0192	Nynäshamn	01	30579
+0305	Håbo	03	22973
+0319	Älvkarleby	03	9552
+0330	Knivsta	03	21193
+0331	Heby	03	14345
+0360	Tierp	03	21104
+0380	Uppsala	03	248016
+0381	Enköping	03	48591
+0382	Östhammar	03	22138
+0428	Vingåker	04	8750
+0461	Gnesta	04	11458
+0480	Nyköping	04	58344
+0481	Oxelösund	04	12031
+0482	Flen	04	15362
+0483	Katrineholm	04	34154
+0484	Eskilstuna	04	107203
+0486	Strängnäs	04	39313
+0488	Trosa	04	14927
+0509	Ödeshög	05	5237
+0512	Ydre	05	3626
+0513	Kinda	05	9957
+0560	Boxholm	05	5517
+0561	Åtvidaberg	05	11467
+0562	Finspång	05	21623
+0563	Valdemarsvik	05	7525
+0580	Linköping	05	168035
+0581	Norrköping	05	144980
+0582	Söderköping	05	14789
+0583	Motala	05	43505
+0584	Vadstena	05	7490
+0586	Mjölby	05	28695
+0604	Aneby	06	6797
+0617	Gnosjö	06	9131
+0642	Mullsjö	06	7594
+0643	Habo	06	13456
+0662	Gislaved	06	28936
+0665	Vaggeryd	06	14825
+0680	Jönköping	06	147654
+0682	Nässjö	06	31587
+0683	Värnamo	06	34542
+0684	Sävsjö	06	11563
+0685	Vetlanda	06	27528
+0686	Eksjö	06	17792
+0687	Tranås	06	18604
+0760	Uppvidinge	07	9061
+0761	Lessebo	07	8289
+0763	Tingsryd	07	11966
+0764	Alvesta	07	19830
+0765	Älmhult	07	17653
+0767	Markaryd	07	9938
+0780	Växjö	07	98334
+0781	Ljungby	07	28280
+0821	Högsby	08	5321
+0834	Torsås	08	6984
+0840	Mörbylånga	08	16224
+0860	Hultsfred	08	13673
+0861	Mönsterås	08	13069
+0862	Emmaboda	08	9006
+0880	Kalmar	08	72704
+0881	Nybro	08	19951
+0882	Oskarshamn	08	26923
+0883	Västervik	08	36447
+0884	Vimmerby	08	15384
+0885	Borgholm	08	10666
+0980	Gotland	09	60971
+1060	Olofström	10	13000
+1080	Karlskrona	10	66301
+1081	Ronneby	10	28741
+1082	Karlshamn	10	31751
+1083	Sölvesborg	10	17430
+1214	Svalöv	12	14543
+1230	Staffanstorp	12	27303
+1231	Burlöv	12	20101
+1233	Vellinge	12	37816
+1256	Östra Göinge	12	13978
+1257	Örkelljunga	12	10277
+1260	Bjuv	12	15985
+1261	Kävlinge	12	32477
+1262	Lomma	12	24715
+1263	Svedala	12	23581
+1264	Skurup	12	17099
+1265	Sjöbo	12	19337
+1266	Hörby	12	15562
+1267	Höör	12	17518
+1270	Tomelilla	12	13639
+1272	Bromölla	12	12470
+1273	Osby	12	12947
+1275	Perstorp	12	7235
+1276	Klippan	12	17714
+1277	Åstorp	12	16449
+1278	Båstad	12	16026
+1280	Malmö	12	365644
+1281	Lund	12	131590
+1282	Landskrona	12	47309
+1283	Helsingborg	12	152091
+1284	Höganäs	12	28430
+1285	Eslöv	12	34922
+1286	Ystad	12	32106
+1287	Trelleborg	12	47269
+1290	Kristianstad	12	86379
+1291	Simrishamn	12	18890
+1292	Ängelholm	12	45110
+1293	Hässleholm	12	52114
+1315	Hylte	13	10196
+1380	Halmstad	13	106084
+1381	Laholm	13	26595
+1382	Falkenberg	13	47337
+1383	Varberg	13	69070
+1384	Kungsbacka	13	85792
+1401	Härryda	14	40003
+1402	Partille	14	41060
+1407	Öckerö	14	12771
+1415	Stenungsund	14	27851
+1419	Tjörn	14	16092
+1421	Orust	14	15352
+1427	Sotenäs	14	9104
+1430	Munkedal	14	10354
+1435	Tanum	14	12773
+1438	Dals-Ed	14	4606
+1439	Färgelanda	14	6376
+1440	Ale	14	32576
+1441	Lerum	14	43570
+1442	Vårgårda	14	12474
+1443	Bollebygd	14	9802
+1444	Grästorp	14	5555
+1445	Essunga	14	5560
+1446	Karlsborg	14	7023
+1447	Gullspång	14	5031
+1452	Tranemo	14	11839
+1460	Bengtsfors	14	9076
+1461	Mellerud	14	9052
+1462	Lilla Edet	14	14442
+1463	Mark	14	35155
+1465	Svenljunga	14	10747
+1466	Herrljunga	14	9497
+1470	Vara	14	16088
+1471	Götene	14	13286
+1472	Tibro	14	11338
+1473	Töreboda	14	9043
+1480	Göteborg	14	608993
+1481	Mölndal	14	71420
+1482	Kungälv	14	50313
+1484	Lysekil	14	13907
+1485	Uddevalla	14	57010
+1486	Strömstad	14	13482
+1487	Vänersborg	14	40041
+1488	Trollhättan	14	59003
+1489	Alingsås	14	42722
+1490	Borås	14	114872
+1491	Ulricehamn	14	24985
+1492	Åmål	14	11906
+1493	Mariestad	14	24583
+1494	Lidköping	14	40425
+1495	Skara	14	18707
+1496	Skövde	14	57995
+1497	Hjo	14	9350
+1498	Tidaholm	14	12805
+1499	Falköping	14	32806
+1715	Kil	17	12061
+1730	Eda	17	8412
+1737	Torsby	17	11321
+1760	Storfors	17	3788
+1761	Hammarö	17	16992
+1762	Munkfors	17	3625
+1763	Forshaga	17	11520
+1764	Grums	17	9004
+1765	Årjäng	17	9825
+1766	Sunne	17	13356
+1780	Karlstad	17	98084
+1781	Kristinehamn	17	23756
+1782	Filipstad	17	9776
+1783	Hagfors	17	11418
+1784	Arvika	17	25547
+1785	Säffle	17	14899
+1814	Lekeberg	18	8606
+1860	Laxå	18	5423
+1861	Hallsberg	18	16120
+1862	Degerfors	18	9278
+1863	Hällefors	18	6321
+1864	Ljusnarsberg	18	4369
+1880	Örebro	18	160140
+1881	Kumla	18	22681
+1882	Askersund	18	11477
+1883	Karlskoga	18	30180
+1884	Nora	18	10639
+1885	Lindesberg	18	23141
+1904	Skinnskatteberg	19	4256
+1907	Surahammar	19	9845
+1960	Kungsör	19	8694
+1961	Hallstahammar	19	16653
+1962	Norberg	19	5452
+1980	Västerås	19	160634
+1981	Sala	19	22843
+1982	Fagersta	19	13072
+1983	Köping	19	25729
+1984	Arboga	19	13980
+2021	Vansbro	20	6752
+2023	Malung-Sälen	20	10254
+2026	Gagnef	20	10384
+2029	Leksand	20	16137
+2031	Rättvik	20	10998
+2034	Orsa	20	6851
+2039	Älvdalen	20	6882
+2061	Smedjebacken	20	10823
+2062	Mora	20	20540
+2080	Falun	20	59945
+2081	Borlänge	20	51425
+2082	Säter	20	11223
+2083	Hedemora	20	15281
+2084	Avesta	20	22417
+2085	Ludvika	20	26634
+2101	Ockelbo	21	5715
+2104	Hofors	21	9281
+2121	Ovanåker	21	11341
+2132	Nordanstig	21	9262
+2161	Ljusdal	21	18445
+2180	Gävle	21	103838
+2181	Sandviken	21	38360
+2182	Söderhamn	21	24545
+2183	Bollnäs	21	26243
+2184	Hudiksvall	21	37528
+2260	Ånge	22	9044
+2262	Timrå	22	17521
+2280	Härnösand	22	24515
+2281	Sundsvall	22	99048
+2282	Kramfors	22	17491
+2283	Sollefteå	22	18396
+2284	Örnsköldsvik	22	55443
+2303	Ragunda	23	5146
+2305	Bräcke	23	6035
+2309	Krokom	23	15680
+2313	Strömsund	23	11023
+2321	Åre	23	12693
+2326	Berg	23	7108
+2361	Härjedalen	23	10175
+2380	Östersund	23	64979
+2401	Nordmaling	24	6942
+2403	Bjurholm	24	2359
+2404	Vindeln	24	5417
+2409	Robertsfors	24	6690
+2417	Norsjö	24	3968
+2418	Malå	24	2962
+2421	Storuman	24	5577
+2422	Sorsele	24	2357
+2425	Dorotea	24	2294
+2460	Vännäs	24	9132
+2462	Vilhelmina	24	6229
+2463	Åsele	24	2694
+2480	Umeå	24	134249
+2481	Lycksele	24	12118
+2482	Skellefteå	24	78150
+2505	Arvidsjaur	25	6089
+2506	Arjeplog	25	2599
+2510	Jokkmokk	25	4701
+2513	Överkalix	25	3201
+2514	Kalix	25	15391
+2518	Övertorneå	25	4057
+2521	Pajala	25	5857
+2523	Gällivare	25	17233
+2560	Älvsbyn	25	7774
+2580	Luleå	25	79645
+2581	Piteå	25	42447
+2582	Boden	25	28049
+2583	Haparanda	25	9151
+2584	Kiruna	25	22426

data/geo/SE/postal-code.json

@@ -0,0 +1 @@
+{ "format": "{code}", "rows": "postal-code.tsv", "key": "code", "parent": "locality" }

data/geo/SE/region.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "region.tsv", "key": "code", "name": "name", "weight": "population" }

data/geo/SE/region.tsv

@@ -0,0 +1,22 @@
+code	name	population	timezone
+01	Stockholms län	2473307	Europe/Stockholm
+03	Uppsala län	407912	Europe/Stockholm
+04	Södermanlands län	301542	Europe/Stockholm
+05	Östergötlands län	472446	Europe/Stockholm
+06	Jönköpings län	370009	Europe/Stockholm
+07	Kronobergs län	203351	Europe/Stockholm
+08	Kalmar län	246352	Europe/Stockholm
+09	Gotlands län	60971	Europe/Stockholm
+10	Blekinge län	157223	Europe/Stockholm
+12	Skåne län	1428626	Europe/Stockholm
+13	Hallands län	345074	Europe/Stockholm
+14	Västra Götalands län	1772821	Europe/Stockholm
+17	Värmlands län	283384	Europe/Stockholm
+18	Örebro län	308375	Europe/Stockholm
+19	Västmanlands län	281158	Europe/Stockholm
+20	Dalarnas län	286546	Europe/Stockholm
+21	Gävleborgs län	284558	Europe/Stockholm
+22	Västernorrlands län	241458	Europe/Stockholm
+23	Jämtlands län	132839	Europe/Stockholm
+24	Västerbottens län	281138	Europe/Stockholm
+25	Norrbottens län	248620	Europe/Stockholm

data/geo/SE/street.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "street.tsv", "parent": "locality", "weight": "segments" }

data/geo/US/.fejkdata.json

@@ -0,0 +1,11 @@
+{
+	"index": {
+		"geo.US.address": {"paths":["","locality","postal-code","region","street","street-number"]},
+		"geo.US.locality": {"parent":"municipality","paths":["","code","lat","lon","municipality","name","population","postal-code","postal-code.addresses","postal-code.code","postal-code.locality","street","street.addresses","street.locality","street.name"]},
+		"geo.US.municipality": {"parent":"region","paths":["","code","locality","locality.code","locality.lat","locality.lon","locality.municipality","locality.name","locality.population","locality.postal-code","locality.postal-code.addresses","locality.postal-code.code","locality.postal-code.locality","locality.street","locality.street.addresses","locality.street.locality","locality.street.name","name","population","region"]},
+		"geo.US.postal-code": {"parent":"locality","paths":["","addresses","code","locality"]},
+		"geo.US.region": {"paths":["","abbr","code","municipality","municipality.code","municipality.locality","municipality.locality.code","municipality.locality.lat","municipality.locality.lon","municipality.locality.municipality","municipality.locality.name","municipality.locality.population","municipality.locality.postal-code","municipality.locality.postal-code.addresses","municipality.locality.postal-code.code","municipality.locality.postal-code.locality","municipality.locality.street","municipality.locality.street.addresses","municipality.locality.street.locality","municipality.locality.street.name","municipality.name","municipality.population","municipality.region","name","population","timezone"]},
+		"geo.US.street": {"parent":"locality","paths":["","addresses","locality","name"]}
+	},
+	"namespace": "geo.US"
+}

data/geo/US/address.json

@@ -0,0 +1,13 @@
+{
+  "format": "{street-number} {street}\n{locality}, {region} {postal-code}",
+  "street": "{.locality as l}{l.street.name}",
+  "street-number": [
+    "{int(10,99)}",
+    { "format": "{int(100,999)}", "weight": 2 },
+    { "format": "{int(1000,9999)}", "weight": 0.5 },
+    { "format": "{int(10,99)}{int(100,999)}", "weight": 0.3 }
+  ],
+  "locality": "{l.name}",
+  "region": "{l..municipality..region.abbr}",
+  "postal-code": "{l.postal-code.code}"
+}

data/geo/US/fs.go

@@ -0,0 +1,15 @@
+// Package geous is fejkdata's US places module, whose categories are geo.US.*.
+package geous
+
+import (
+	"embed"
+	"io/fs"
+)
+
+// FS holds the module's categories and its manifest.
+//
+//go:embed *.json *.tsv .fejkdata.json
+var FS embed.FS
+
+// Modules returns FS, which reads no other module, to pass to fejkdata.WithDataFS.
+func Modules() []fs.FS { return []fs.FS{FS} }

data/geo/US/fs_test.go

@@ -0,0 +1,22 @@
+package geous_test
+
+import (
+	"io/fs"
+	"testing"
+
+	"github.com/larvit/fejkdata"
+	geous "github.com/larvit/fejkdata/data/geo/US"
+)
+
+func TestFSLoadsByItsManifest(t *testing.T) {
+	if _, err := fs.Stat(geous.FS, ".fejkdata.json"); err != nil {
+		t.Fatalf("FS holds no manifest: %v", err)
+	}
+	f, err := fejkdata.New(fejkdata.WithDataFS(geous.Modules()...), fejkdata.WithSeed(1))
+	if err != nil {
+		t.Fatalf("New(WithDataFS(Modules()...)) = %v", err)
+	}
+	if len(f.List()) == 0 {
+		t.Error("List() is empty, want the module's categories")
+	}
+}

data/geo/US/locality.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "locality.tsv", "key": "code", "name": "name", "parent": "municipality", "weight": "population" }

data/geo/US/municipality.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "municipality.tsv", "key": "code", "name": "name", "parent": "region", "weight": "population" }

data/geo/US/municipality.tsv

@@ -0,0 +1,887 @@
+code	name	region	population
+01005	Barbour County	AL	24607
+01015	Calhoun County	AL	115834
+01021	Chilton County	AL	47708
+01043	Cullman County	AL	94009
+01045	Dale County	AL	49912
+01049	DeKalb County	AL	74085
+01051	Elmore County	AL	91577
+01055	Etowah County	AL	103886
+01065	Hale County	AL	15194
+01069	Houston County	AL	110318
+01071	Jackson County	AL	54281
+01073	Jefferson County	AL	665742
+01077	Lauderdale County	AL	97135
+01081	Lee County	AL	189881
+01089	Madison County	AL	433516
+01095	Marshall County	AL	103537
+01097	Mobile County	AL	411658
+01101	Montgomery County	AL	225891
+01103	Morgan County	AL	126483
+01109	Pike County	AL	33688
+01111	Randolph County	AL	23163
+01113	Russell County	AL	58898
+01115	St. Clair County	AL	98206
+01117	Shelby County	AL	238552
+01121	Talladega County	AL	81603
+01125	Tuscaloosa County	AL	241368
+01127	Walker County	AL	65140
+02020	Anchorage Municipality	AK	287155
+02110	Juneau City and Borough	AK	31609
+02185	North Slope Borough	AK	10582
+04003	Cochise County	AZ	126332
+04005	Coconino County	AZ	144368
+04013	Maricopa County	AZ	4689558
+04015	Mohave County	AZ	228102
+04017	Navajo County	AZ	109946
+04019	Pima County	AZ	1074685
+04021	Pinal County	AZ	539380
+04025	Yavapai County	AZ	252552
+04027	Yuma County	AZ	224449
+05005	Baxter County	AR	42982
+05007	Benton County	AR	332554
+05019	Clark County	AR	20938
+05031	Craighead County	AR	116957
+05035	Crittenden County	AR	46210
+05039	Dallas County	AR	6011
+05045	Faulkner County	AR	133979
+05071	Johnson County	AR	26225
+05083	Logan County	AR	21633
+05085	Lonoke County	AR	76664
+05089	Marion County	AR	17593
+05091	Miller County	AR	42357
+05111	Poinsett County	AR	22483
+05113	Polk County	AR	19486
+05119	Pulaski County	AR	404611
+05125	Saline County	AR	133288
+05131	Sebastian County	AR	130641
+05143	Washington County	AR	271213
+05145	White County	AR	80085
+06001	Alameda County	CA	1636630
+06007	Butte County	CA	209211
+06009	Calaveras County	CA	46605
+06013	Contra Costa County	CA	1170070
+06017	El Dorado County	CA	192323
+06019	Fresno County	CA	1035456
+06023	Humboldt County	CA	131647
+06025	Imperial County	CA	181411
+06029	Kern County	CA	927068
+06031	Kings County	CA	154327
+06035	Lassen County	CA	28117
+06037	Los Angeles County	CA	9694934
+06039	Madera County	CA	167927
+06045	Mendocino County	CA	88122
+06047	Merced County	CA	297260
+06053	Monterey County	CA	433729
+06055	Napa County	CA	132949
+06057	Nevada County	CA	101911
+06059	Orange County	CA	3149507
+06061	Placer County	CA	442081
+06065	Riverside County	CA	2544916
+06067	Sacramento County	CA	1618460
+06071	San Bernardino County	CA	2224091
+06073	San Diego County	CA	3282248
+06075	San Francisco County	CA	826079
+06077	San Joaquin County	CA	823815
+06079	San Luis Obispo County	CA	282367
+06081	San Mateo County	CA	743568
+06083	Santa Barbara County	CA	442065
+06085	Santa Clara County	CA	1914391
+06087	Santa Cruz County	CA	258852
+06089	Shasta County	CA	181648
+06095	Solano County	CA	455376
+06097	Sonoma County	CA	486444
+06099	Stanislaus County	CA	557719
+06103	Tehama County	CA	64665
+06107	Tulare County	CA	485146
+06111	Ventura County	CA	830851
+06113	Yolo County	CA	224410
+06115	Yuba County	CA	88691
+08001	Adams County	CO	554668
+08005	Arapahoe County	CO	673820
+08011	Bent County	CO	5716
+08013	Boulder County	CO	328560
+08029	Delta County	CO	32178
+08031	Denver County	CO	740613
+08033	Dolores County	CO	2466
+08035	Douglas County	CO	399396
+08037	Eagle County	CO	54291
+08041	El Paso County	CO	757040
+08043	Fremont County	CO	50039
+08045	Garfield County	CO	63474
+08049	Grand County	CO	16496
+08051	Gunnison County	CO	17439
+08059	Jefferson County	CO	580451
+08067	La Plata County	CO	56898
+08069	Larimer County	CO	377292
+08089	Otero County	CO	17683
+08091	Ouray County	CO	5265
+08101	Pueblo County	CO	169277
+08105	Rio Grande County	CO	11087
+08107	Routt County	CO	25329
+08123	Weld County	CO	378426
+09110	Capitol Planning Region	CT	994115
+09120	Greater Bridgeport Planning Region	CT	337697
+09140	Naugatuck Valley Planning Region	CT	463349
+09150	Northeastern Connecticut Planning Region	CT	98096
+09170	South Central Connecticut Planning Region	CT	578741
+09180	Southeastern Connecticut Planning Region	CT	284015
+09190	Western Connecticut Planning Region	CT	640482
+10001	Kent County	DE	194786
+10003	New Castle County	DE	588026
+10005	Sussex County	DE	277140
+11001	District of Columbia	DC	693645
+12001	Alachua County	FL	290028
+12005	Bay County	FL	204479
+12009	Brevard County	FL	663982
+12011	Broward County	FL	2013317
+12017	Citrus County	FL	171666
+12031	Duval County	FL	1062963
+12033	Escambia County	FL	333834
+12035	Flagler County	FL	140360
+12045	Gulf County	FL	15943
+12057	Hillsborough County	FL	1574115
+12069	Lake County	FL	456068
+12071	Lee County	FL	875607
+12073	Leon County	FL	299048
+12081	Manatee County	FL	468200
+12085	Martin County	FL	166272
+12086	Miami-Dade County	FL	2802029
+12087	Monroe County	FL	80406
+12089	Nassau County	FL	106879
+12091	Okaloosa County	FL	221810
+12095	Orange County	FL	1528002
+12097	Osceola County	FL	481718
+12099	Palm Beach County	FL	1575726
+12101	Pasco County	FL	674516
+12103	Pinellas County	FL	948563
+12105	Polk County	FL	874790
+12107	Putnam County	FL	77734
+12109	St. Johns County	FL	346328
+12111	St. Lucie County	FL	402449
+12115	Sarasota County	FL	479958
+12117	Seminole County	FL	491884
+12125	Union County	FL	16250
+12127	Volusia County	FL	606573
+13021	Bibb County	GA	157556
+13029	Bryan County	GA	52062
+13031	Bulloch County	GA	86949
+13039	Camden County	GA	60143
+13045	Carroll County	GA	131036
+13051	Chatham County	GA	311855
+13057	Cherokee County	GA	299273
+13059	Clarke County	GA	129921
+13063	Clayton County	GA	297471
+13067	Cobb County	GA	793345
+13073	Columbia County	GA	169189
+13077	Coweta County	GA	160240
+13089	DeKalb County	GA	774394
+13095	Dougherty County	GA	82616
+13113	Fayette County	GA	125156
+13121	Fulton County	GA	1098791
+13133	Greene County	GA	21367
+13135	Gwinnett County	GA	1018099
+13137	Habersham County	GA	50416
+13139	Hall County	GA	226568
+13151	Henry County	GA	264922
+13153	Houston County	GA	178214
+13157	Jackson County	GA	99265
+13179	Liberty County	GA	70313
+13185	Lowndes County	GA	122867
+13215	Muscogee County	GA	202171
+13217	Newton County	GA	125583
+13237	Putnam County	GA	23722
+13245	Richmond County	GA	206559
+13247	Rockdale County	GA	98416
+13255	Spalding County	GA	70775
+13277	Tift County	GA	42254
+13297	Walton County	GA	112696
+13299	Ware County	GA	36906
+16001	Ada County	ID	546141
+16019	Bonneville County	ID	135771
+16027	Canyon County	ID	275123
+16047	Gooding County	ID	16446
+16057	Latah County	ID	41842
+16083	Twin Falls County	ID	97539
+17001	Adams County	IL	64267
+17013	Calhoun County	IL	4300
+17019	Champaign County	IL	209972
+17029	Coles County	IL	46607
+17031	Cook County	IL	5194625
+17037	DeKalb County	IL	101835
+17043	DuPage County	IL	934298
+17045	Edgar County	IL	16323
+17051	Fayette County	IL	20790
+17053	Ford County	IL	13216
+17055	Franklin County	IL	36916
+17063	Grundy County	IL	54052
+17071	Henderson County	IL	5984
+17073	Henry County	IL	48095
+17077	Jackson County	IL	52501
+17085	Jo Daviess County	IL	21594
+17089	Kane County	IL	525757
+17093	Kendall County	IL	145470
+17097	Lake County	IL	719339
+17103	Lee County	IL	32914
+17111	McHenry County	IL	317751
+17113	McLean County	IL	171419
+17119	Madison County	IL	263110
+17139	Moultrie County	IL	14371
+17143	Peoria County	IL	178553
+17157	Randolph County	IL	29921
+17161	Rock Island County	IL	141869
+17163	St. Clair County	IL	250708
+17167	Sangamon County	IL	194170
+17169	Schuyler County	IL	6657
+17177	Stephenson County	IL	43220
+17195	Whiteside County	IL	54121
+17197	Will County	IL	712253
+17199	Williamson County	IL	67225
+17201	Winnebago County	IL	283674
+18001	Adams County	IN	36650
+18003	Allen County	IN	402329
+18005	Bartholomew County	IN	85729
+18011	Boone County	IN	80689
+18019	Clark County	IN	130451
+18023	Clinton County	IN	33322
+18027	Daviess County	IN	34209
+18029	Dearborn County	IN	51609
+18031	Decatur County	IN	26576
+18033	DeKalb County	IN	44535
+18035	Delaware County	IN	113106
+18037	Dubois County	IN	44016
+18039	Elkhart County	IN	208774
+18043	Floyd County	IN	82153
+18045	Fountain County	IN	17013
+18049	Fulton County	IN	20091
+18053	Grant County	IN	66524
+18057	Hamilton County	IN	387036
+18059	Hancock County	IN	90969
+18063	Hendricks County	IN	193510
+18069	Huntington County	IN	37224
+18081	Johnson County	IN	174262
+18089	Lake County	IN	504612
+18091	LaPorte County	IN	111294
+18095	Madison County	IN	135088
+18097	Marion County	IN	992196
+18105	Monroe County	IN	143345
+18107	Montgomery County	IN	38954
+18109	Morgan County	IN	74967
+18127	Porter County	IN	176049
+18141	St. Joseph County	IN	272861
+18157	Tippecanoe County	IN	190456
+18163	Vanderburgh County	IN	181995
+18167	Vigo County	IN	106512
+18169	Wabash County	IN	30713
+18177	Wayne County	IN	66169
+19009	Audubon County	IA	5509
+19027	Carroll County	IA	20303
+19055	Delaware County	IA	17674
+19061	Dubuque County	IA	99381
+19083	Hardin County	IA	16455
+19095	Iowa County	IA	16504
+19103	Johnson County	IA	160044
+19113	Linn County	IA	232028
+19123	Mahaska County	IA	21880
+19127	Marshall County	IA	39890
+19153	Polk County	IA	516546
+19155	Pottawattamie County	IA	92996
+19163	Scott County	IA	175259
+19169	Story County	IA	101291
+19179	Wapello County	IA	35210
+19193	Woodbury County	IA	106649
+20015	Butler County	KS	69484
+20035	Cowley County	KS	34382
+20041	Dickinson County	KS	18637
+20045	Douglas County	KS	120920
+20057	Ford County	KS	33993
+20061	Geary County	KS	36338
+20091	Johnson County	KS	636906
+20099	Labette County	KS	19685
+20103	Leavenworth County	KS	84590
+20107	Linn County	KS	9950
+20111	Lyon County	KS	32266
+20113	McPherson County	KS	30064
+20121	Miami County	KS	36240
+20161	Riley County	KS	72598
+20169	Saline County	KS	53377
+20173	Sedgwick County	KS	538433
+20175	Seward County	KS	21073
+20177	Shawnee County	KS	178607
+20209	Wyandotte County	KS	170597
+21009	Barren County	KY	45641
+21015	Boone County	KY	145316
+21021	Boyle County	KY	31632
+21029	Bullitt County	KY	86454
+21035	Calloway County	KY	39421
+21037	Campbell County	KY	95441
+21047	Christian County	KY	70115
+21049	Clark County	KY	37888
+21059	Daviess County	KY	104898
+21067	Fayette County	KY	329751
+21073	Franklin County	KY	52649
+21085	Grayson County	KY	27406
+21093	Hardin County	KY	113482
+21101	Henderson County	KY	44255
+21111	Jefferson County	KY	795222
+21117	Kenton County	KY	175779
+21145	McCracken County	KY	67553
+21151	Madison County	KY	101696
+21177	Muhlenberg County	KY	30634
+21183	Ohio County	KY	23934
+21207	Russell County	KY	18608
+21209	Scott County	KY	62262
+21211	Shelby County	KY	51243
+21227	Warren County	KY	149375
+22009	Avoyelles Parish	LA	37941
+22017	Caddo Parish	LA	224226
+22019	Calcasieu Parish	LA	208466
+22025	Catahoula Parish	LA	8223
+22033	East Baton Rouge Parish	LA	456180
+22049	Jackson Parish	LA	15834
+22055	Lafayette Parish	LA	257949
+22057	Lafourche Parish	LA	95505
+22061	Lincoln Parish	LA	48360
+22071	Orleans Parish	LA	362154
+22073	Ouachita Parish	LA	158542
+22077	Pointe Coupee Parish	LA	19520
+22097	St. Landry Parish	LA	80765
+22103	St. Tammany Parish	LA	279108
+22105	Tangipahoa Parish	LA	141346
+22117	Washington Parish	LA	44937
+23001	Androscoggin County	ME	116487
+23011	Kennebec County	ME	129067
+23019	Penobscot County	ME	157967
+23031	York County	ME	222434
+24001	Allegany County	MD	66848
+24003	Anne Arundel County	MD	603380
+24015	Cecil County	MD	107131
+24021	Frederick County	MD	302883
+24031	Montgomery County	MD	1074582
+24033	Prince George's County	MD	970374
+24043	Washington County	MD	157731
+24510	Baltimore city	MD	569997
+25001	Barnstable County	MA	233539
+25005	Bristol County	MA	593640
+25009	Essex County	MA	826653
+25013	Hampden County	MA	464338
+25015	Hampshire County	MA	164065
+25017	Middlesex County	MA	1669979
+25021	Norfolk County	MA	739749
+25023	Plymouth County	MA	546829
+25025	Suffolk County	MA	791891
+25027	Worcester County	MA	888502
+26003	Alger County	MI	8641
+26021	Berrien County	MI	152444
+26037	Clinton County	MI	80432
+26065	Ingham County	MI	289709
+26077	Kalamazoo County	MI	263795
+26081	Kent County	MI	675232
+26099	Macomb County	MI	886221
+26121	Muskegon County	MI	177901
+26125	Oakland County	MI	1288337
+26139	Ottawa County	MI	308459
+26161	Washtenaw County	MI	370214
+26163	Wayne County	MI	1769038
+27003	Anoka County	MN	381605
+27013	Blue Earth County	MN	70634
+27017	Carlton County	MN	36711
+27035	Crow Wing County	MN	69132
+27037	Dakota County	MN	457710
+27053	Hennepin County	MN	1284784
+27061	Itasca County	MN	45404
+27067	Kandiyohi County	MN	44720
+27071	Koochiching County	MN	11587
+27083	Lyon County	MN	25684
+27109	Olmsted County	MN	166731
+27111	Otter Tail County	MN	61041
+27115	Pine County	MN	30472
+27123	Ramsey County	MN	541623
+27127	Redwood County	MN	15265
+27131	Rice County	MN	69939
+27137	St. Louis County	MN	200518
+27139	Scott County	MN	159017
+27141	Sherburne County	MN	104194
+27143	Sibley County	MN	15356
+27145	Stearns County	MN	164110
+27147	Steele County	MN	37464
+27163	Washington County	MN	286895
+27165	Watonwan County	MN	11459
+27167	Wilkin County	MN	6288
+27171	Wright County	MN	157559
+28003	Alcorn County	MS	34569
+28033	DeSoto County	MS	197918
+28035	Forrest County	MS	79034
+28045	Hancock County	MS	46873
+28047	Harrison County	MS	217136
+28049	Hinds County	MS	211888
+28059	Jackson County	MS	147666
+28071	Lafayette County	MS	59597
+28073	Lamar County	MS	67403
+28089	Madison County	MS	116298
+28095	Monroe County	MS	33318
+28113	Pike County	MS	38712
+28117	Prentiss County	MS	25284
+28133	Sunflower County	MS	22893
+28151	Washington County	MS	40446
+28159	Winston County	MS	17473
+29001	Adair County	MO	25684
+29003	Andrew County	MO	18241
+29019	Boone County	MO	191746
+29031	Cape Girardeau County	MO	83999
+29037	Cass County	MO	115859
+29051	Cole County	MO	77908
+29069	Dunklin County	MO	26855
+29077	Greene County	MO	309286
+29095	Jackson County	MO	732994
+29097	Jasper County	MO	127428
+29099	Jefferson County	MO	233132
+29105	Laclede County	MO	36943
+29115	Linn County	MO	11815
+29123	Madison County	MO	12858
+29127	Marion County	MO	28525
+29155	Pemiscot County	MO	14465
+29161	Phelps County	MO	45247
+29167	Polk County	MO	33604
+29169	Pulaski County	MO	54442
+29175	Randolph County	MO	24513
+29183	St. Charles County	MO	426499
+29187	St. Francois County	MO	67809
+29189	St. Louis County	MO	990911
+29195	Saline County	MO	23265
+29510	St. Louis city	MO	278144
+30013	Cascade County	MT	85029
+30017	Custer County	MT	12034
+30031	Gallatin County	MT	128740
+30049	Lewis and Clark County	MT	75331
+30057	Madison County	MT	10026
+30063	Missoula County	MT	123513
+30081	Ravalli County	MT	48582
+30111	Yellowstone County	MT	172692
+31001	Adams County	NE	31071
+31023	Butler County	NE	8507
+31053	Dodge County	NE	38057
+31055	Douglas County	NE	606460
+31067	Gage County	NE	21711
+31093	Howard County	NE	6538
+31109	Lancaster County	NE	334049
+31111	Lincoln County	NE	33303
+31147	Richardson County	NE	7530
+31151	Saline County	NE	14996
+31181	Webster County	NE	3336
+31185	York County	NE	14564
+32003	Clark County	NV	2407226
+32031	Washoe County	NV	509386
+32510	Carson City	NV	58571
+33007	Coos County	NH	31343
+33011	Hillsborough County	NH	433415
+33013	Merrimack County	NH	158078
+33015	Rockingham County	NH	324077
+34001	Atlantic County	NJ	278657
+34003	Bergen County	NJ	977026
+34007	Camden County	NJ	535799
+34009	Cape May County	NJ	93390
+34011	Cumberland County	NJ	157148
+34013	Essex County	NJ	896379
+34015	Gloucester County	NJ	312638
+34017	Hudson County	NJ	735033
+34021	Mercer County	NJ	399289
+34023	Middlesex County	NJ	883335
+34025	Monmouth County	NJ	651035
+34027	Morris County	NJ	524251
+34029	Ocean County	NJ	673746
+34031	Passaic County	NJ	531624
+34035	Somerset County	NJ	356486
+34037	Sussex County	NJ	148063
+34039	Union County	NJ	601863
+35001	Bernalillo County	NM	667601
+35005	Chaves County	NM	63364
+35009	Curry County	NM	46655
+35013	Doña Ana County	NM	229091
+35029	Luna County	NM	25407
+35035	Otero County	NM	70368
+35041	Roosevelt County	NM	18615
+35043	Sandoval County	NM	159565
+35045	San Juan County	NM	120340
+35047	San Miguel County	NM	26260
+35049	Santa Fe County	NM	156907
+35057	Torrance County	NM	16100
+35061	Valencia County	NM	82013
+36001	Albany County	NY	321225
+36007	Broome County	NY	195736
+36011	Cayuga County	NY	74365
+36013	Chautauqua County	NY	124126
+36015	Chemung County	NY	80415
+36027	Dutchess County	NY	300708
+36029	Erie County	NY	946741
+36035	Fulton County	NY	52216
+36039	Greene County	NY	47238
+36045	Jefferson County	NY	111540
+36047	Kings County	NY	2653963
+36051	Livingston County	NY	61438
+36053	Madison County	NY	67120
+36055	Monroe County	NY	750506
+36057	Montgomery County	NY	50046
+36059	Nassau County	NY	1398939
+36063	Niagara County	NY	208912
+36065	Oneida County	NY	226392
+36067	Onondaga County	NY	466584
+36071	Orange County	NY	417669
+36073	Orleans County	NY	39825
+36093	Schenectady County	NY	162581
+36101	Steuben County	NY	91855
+36103	Suffolk County	NY	1546090
+36105	Sullivan County	NY	80586
+36107	Tioga County	NY	47453
+36111	Ulster County	NY	183330
+36119	Westchester County	NY	1015743
+37013	Beaufort County	NC	44670
+37017	Bladen County	NC	30007
+37019	Brunswick County	NC	174702
+37021	Buncombe County	NC	277417
+37025	Cabarrus County	NC	249725
+37027	Caldwell County	NC	81105
+37031	Carteret County	NC	70469
+37035	Catawba County	NC	170172
+37045	Cleveland County	NC	103325
+37049	Craven County	NC	105025
+37051	Cumberland County	NC	338473
+37057	Davidson County	NC	180182
+37059	Davie County	NC	45855
+37063	Durham County	NC	347240
+37067	Forsyth County	NC	401718
+37071	Gaston County	NC	246558
+37081	Guilford County	NC	562234
+37089	Henderson County	NC	122375
+37097	Iredell County	NC	211798
+37101	Johnston County	NC	256448
+37107	Lenoir County	NC	55837
+37119	Mecklenburg County	NC	1233383
+37123	Montgomery County	NC	26403
+37127	Nash County	NC	99365
+37129	New Hanover County	NC	245959
+37133	Onslow County	NC	217175
+37135	Orange County	NC	152498
+37137	Pamlico County	NC	12758
+37153	Richmond County	NC	42041
+37159	Rowan County	NC	155096
+37161	Rutherford County	NC	65745
+37165	Scotland County	NC	33322
+37179	Union County	NC	267674
+37183	Wake County	NC	1257235
+37191	Wayne County	NC	122278
+37195	Wilson County	NC	81150
+38009	Bottineau County	ND	6284
+38015	Burleigh County	ND	103251
+38017	Cass County	ND	201794
+38035	Grand Forks County	ND	74501
+38101	Ward County	ND	68233
+38105	Williams County	ND	41767
+39003	Allen County	OH	100881
+39009	Athens County	OH	63197
+39017	Butler County	OH	400128
+39023	Clark County	OH	135340
+39029	Columbiana County	OH	99377
+39035	Cuyahoga County	OH	1232925
+39041	Delaware County	OH	242032
+39043	Erie County	OH	73296
+39045	Fairfield County	OH	169752
+39047	Fayette County	OH	28552
+39049	Franklin County	OH	1361536
+39051	Fulton County	OH	41853
+39061	Hamilton County	OH	838418
+39063	Hancock County	OH	75034
+39065	Hardin County	OH	30072
+39075	Holmes County	OH	44970
+39077	Huron County	OH	58250
+39079	Jackson County	OH	32817
+39083	Knox County	OH	64002
+39085	Lake County	OH	232217
+39089	Licking County	OH	185564
+39093	Lorain County	OH	323219
+39095	Lucas County	OH	423347
+39099	Mahoning County	OH	224706
+39101	Marion County	OH	65115
+39103	Medina County	OH	185025
+39113	Montgomery County	OH	539598
+39131	Pike County	OH	27061
+39133	Portage County	OH	163404
+39135	Preble County	OH	40664
+39139	Richland County	OH	124893
+39149	Shelby County	OH	48065
+39151	Stark County	OH	373771
+39153	Summit County	OH	538376
+39155	Trumbull County	OH	198972
+39159	Union County	OH	73446
+39165	Warren County	OH	257181
+39169	Wayne County	OH	116758
+39171	Williams County	OH	36549
+39173	Wood County	OH	134176
+40001	Adair County	OK	19828
+40017	Canadian County	OK	187189
+40027	Cleveland County	OK	303973
+40031	Comanche County	OK	122158
+40037	Creek County	OK	74967
+40039	Custer County	OK	28175
+40047	Garfield County	OK	61779
+40087	McClain County	OK	49647
+40093	Major County	OK	7536
+40101	Muskogee County	OK	66708
+40109	Oklahoma County	OK	822125
+40119	Payne County	OK	83889
+40121	Pittsburg County	OK	43320
+40125	Pottawatomie County	OK	75102
+40143	Tulsa County	OK	698782
+40145	Wagoner County	OK	93263
+41005	Clackamas County	OR	426280
+41009	Columbia County	OR	54091
+41017	Deschutes County	OR	213072
+41029	Jackson County	OR	221795
+41033	Josephine County	OR	87867
+41035	Klamath County	OR	70274
+41039	Lane County	OR	381584
+41041	Lincoln County	OR	50636
+41047	Marion County	OR	355777
+41049	Morrow County	OR	12602
+41051	Multnomah County	OR	795391
+41067	Washington County	OR	611708
+41071	Yamhill County	OR	110024
+42003	Allegheny County	PA	1225035
+42011	Berks County	PA	440072
+42017	Bucks County	PA	647828
+42019	Butler County	PA	200169
+42021	Cambria County	PA	128968
+42035	Clinton County	PA	37870
+42037	Columbia County	PA	66193
+42043	Dauphin County	PA	293351
+42045	Delaware County	PA	580937
+42049	Erie County	PA	265832
+42055	Franklin County	PA	160652
+42061	Huntingdon County	PA	43001
+42063	Indiana County	PA	82878
+42069	Lackawanna County	PA	216502
+42071	Lancaster County	PA	563159
+42073	Lawrence County	PA	83911
+42075	Lebanon County	PA	146380
+42077	Lehigh County	PA	384383
+42079	Luzerne County	PA	332126
+42081	Lycoming County	PA	112587
+42087	Mifflin County	PA	46127
+42091	Montgomery County	PA	877643
+42095	Northampton County	PA	324411
+42097	Northumberland County	PA	89920
+42101	Philadelphia County	PA	1574281
+42107	Schuylkill County	PA	145085
+42111	Somerset County	PA	72004
+42119	Union County	PA	42313
+42121	Venango County	PA	49346
+42127	Wayne County	PA	51703
+42129	Westmoreland County	PA	349324
+44005	Newport County	RI	83051
+44007	Providence County	RI	678179
+45003	Aiken County	SC	181515
+45005	Allendale County	SC	7355
+45007	Anderson County	SC	219930
+45011	Barnwell County	SC	20653
+45015	Berkeley County	SC	274666
+45019	Charleston County	SC	436200
+45023	Chester County	SC	33001
+45041	Florence County	SC	138504
+45045	Greenville County	SC	583125
+45051	Horry County	SC	427551
+45059	Laurens County	SC	71848
+45077	Pickens County	SC	139198
+45079	Richland County	SC	434956
+45083	Spartanburg County	SC	380857
+46005	Beadle County	SD	19624
+46065	Hughes County	SD	17570
+46081	Lawrence County	SD	29459
+46099	Minnehaha County	SD	212691
+46103	Pennington County	SD	116792
+46127	Union County	SD	17402
+47009	Blount County	TN	143820
+47011	Bradley County	TN	115465
+47017	Carroll County	TN	29200
+47021	Cheatham County	TN	42778
+47037	Davidson County	TN	745904
+47063	Hamblen County	TN	68843
+47065	Hamilton County	TN	390833
+47071	Hardin County	TN	27983
+47073	Hawkins County	TN	59559
+47083	Houston County	TN	8479
+47093	Knox County	TN	511453
+47113	Madison County	TN	100790
+47125	Montgomery County	TN	249935
+47127	Moore County	TN	6920
+47141	Putnam County	TN	86612
+47145	Roane County	TN	57395
+47149	Rutherford County	TN	386352
+47155	Sevier County	TN	101342
+47157	Shelby County	TN	910226
+47163	Sullivan County	TN	163759
+47165	Sumner County	TN	215538
+47167	Tipton County	TN	62287
+47177	Warren County	TN	43717
+47179	Washington County	TN	141199
+47187	Williamson County	TN	272061
+47189	Wilson County	TN	175033
+48005	Angelina County	TX	88154
+48023	Baylor County	TX	3472
+48027	Bell County	TX	402248
+48029	Bexar County	TX	2160088
+48039	Brazoria County	TX	419080
+48041	Brazos County	TX	249088
+48053	Burnet County	TX	57015
+48057	Calhoun County	TX	19599
+48059	Callahan County	TX	14491
+48061	Cameron County	TX	433946
+48073	Cherokee County	TX	53337
+48085	Collin County	TX	1297179
+48089	Colorado County	TX	21439
+48091	Comal County	TX	209166
+48111	Dallam County	TX	7508
+48113	Dallas County	TX	2661397
+48121	Denton County	TX	1069346
+48135	Ector County	TX	173801
+48139	Ellis County	TX	240867
+48141	El Paso County	TX	877858
+48149	Fayette County	TX	25459
+48157	Fort Bend County	TX	975191
+48165	Gaines County	TX	23956
+48167	Galveston County	TX	372207
+48171	Gillespie County	TX	28527
+48177	Gonzales County	TX	20159
+48181	Grayson County	TX	153613
+48183	Gregg County	TX	126095
+48187	Guadalupe County	TX	201111
+48193	Hamilton County	TX	8680
+48201	Harris County	TX	5045026
+48209	Hays County	TX	304390
+48215	Hidalgo County	TX	921549
+48231	Hunt County	TX	123336
+48245	Jefferson County	TX	254321
+48251	Johnson County	TX	218048
+48257	Kaufman County	TX	209235
+48273	Kleberg County	TX	30315
+48277	Lamar County	TX	51503
+48291	Liberty County	TX	121364
+48293	Limestone County	TX	22849
+48303	Lubbock County	TX	328906
+48307	McCulloch County	TX	7472
+48309	McLennan County	TX	272020
+48325	Medina County	TX	56881
+48329	Midland County	TX	187855
+48331	Milam County	TX	26258
+48339	Montgomery County	TX	781194
+48349	Navarro County	TX	57181
+48355	Nueces County	TX	352992
+48361	Orange County	TX	86266
+48381	Randall County	TX	152351
+48397	Rockwall County	TX	140738
+48403	Sabine County	TX	10012
+48411	San Saba County	TX	5787
+48417	Shackelford County	TX	3197
+48423	Smith County	TX	252549
+48439	Tarrant County	TX	2248466
+48441	Taylor County	TX	150077
+48451	Tom Green County	TX	120602
+48453	Travis County	TX	1389670
+48465	Val Verde County	TX	47835
+48479	Webb County	TX	281224
+48485	Wichita County	TX	129555
+48491	Williamson County	TX	752827
+48497	Wise County	TX	83778
+49005	Cache County	UT	145000
+49007	Carbon County	UT	20680
+49011	Davis County	UT	381227
+49021	Iron County	UT	67141
+49023	Juab County	UT	13567
+49027	Millard County	UT	14057
+49035	Salt Lake County	UT	1220916
+49037	San Juan County	UT	14542
+49039	Sanpete County	UT	31453
+49041	Sevier County	UT	22827
+49049	Utah County	UT	759859
+49053	Washington County	UT	213670
+49057	Weber County	UT	278174
+50007	Chittenden County	VT	169115
+50023	Washington County	VT	59652
+51035	Carroll County	VA	29299
+51093	Isle of Wight County	VA	41321
+51107	Loudoun County	VA	449749
+51135	Nottoway County	VA	15796
+51183	Sussex County	VA	10755
+51195	Wise County	VA	34824
+51510	Alexandria city	VA	160662
+51540	Charlottesville city	VA	44388
+51550	Chesapeake city	VA	255332
+51600	Fairfax city	VA	26772
+51610	Falls Church city	VA	15159
+51650	Hampton city	VA	137315
+51670	Hopewell city	VA	23261
+51680	Lynchburg city	VA	81347
+51700	Newport News city	VA	183230
+51710	Norfolk city	VA	231013
+51730	Petersburg city	VA	33734
+51740	Portsmouth city	VA	96777
+51760	Richmond city	VA	237257
+51800	Suffolk city	VA	104699
+51810	Virginia Beach city	VA	453737
+51840	Winchester city	VA	28272
+53005	Benton County	WA	221722
+53011	Clark County	WA	532119
+53015	Cowlitz County	WA	114885
+53021	Franklin County	WA	102612
+53025	Grant County	WA	105727
+53033	King County	WA	2344939
+53035	Kitsap County	WA	283374
+53041	Lewis County	WA	88062
+53053	Pierce County	WA	946288
+53061	Snohomish County	WA	870656
+53063	Spokane County	WA	558344
+53065	Stevens County	WA	49668
+53067	Thurston County	WA	304261
+53071	Walla Walla County	WA	62361
+53073	Whatcom County	WA	236392
+53077	Yakima County	WA	259185
+54011	Cabell County	WV	91183
+54039	Kanawha County	WV	172381
+54045	Logan County	WV	30375
+54047	McDowell County	WV	16878
+54059	Mingo County	WV	21455
+54061	Monongalia County	WV	107991
+54069	Ohio County	WV	40496
+54083	Randolph County	WV	26882
+54107	Wood County	WV	82385
+55009	Brown County	WI	275803
+55013	Burnett County	WI	17223
+55017	Chippewa County	WI	67614
+55025	Dane County	WI	590375
+55027	Dodge County	WI	89097
+55035	Eau Claire County	WI	109033
+55049	Iowa County	WI	23951
+55055	Jefferson County	WI	86505
+55059	Kenosha County	WI	168448
+55073	Marathon County	WI	139432
+55079	Milwaukee County	WI	924216
+55087	Outagamie County	WI	195894
+55089	Ozaukee County	WI	94346
+55093	Pierce County	WI	43519
+55095	Polk County	WI	45803
+55101	Racine County	WI	198919
+55105	Rock County	WI	166472
+55115	Shawano County	WI	41560
+55121	Trempealeau County	WI	31171
+55127	Walworth County	WI	106127
+55131	Washington County	WI	139238
+55133	Waukesha County	WI	417210
+55139	Winnebago County	WI	174218
+56001	Albany County	WY	38558
+56005	Campbell County	WY	48145
+56007	Carbon County	WY	14013
+56021	Laramie County	WY	102938
+56025	Natrona County	WY	80526
+56029	Park County	WY	31171
+56037	Sweetwater County	WY	41267
+56039	Teton County	WY	23333
+56041	Uinta County	WY	20728

data/geo/US/postal-code.json

@@ -0,0 +1 @@
+{ "format": "{code}", "rows": "postal-code.tsv", "key": "code", "parent": "locality", "weight": "addresses" }

data/geo/US/region.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "region.tsv", "key": "abbr", "name": "name", "weight": "population" }

data/geo/US/region.tsv

@@ -0,0 +1,51 @@
+abbr	code	name	population	timezone
+AK	02	Alaska	737270	America/Anchorage
+AL	01	Alabama	5193088	America/Chicago
+AR	05	Arkansas	3114791	America/Chicago
+AZ	04	Arizona	7623818	America/Phoenix
+CA	06	California	39355309	America/Los_Angeles
+CO	08	Colorado	6012561	America/Denver
+CT	09	Connecticut	3688496	America/New_York
+DC	11	District of Columbia	693645	America/New_York
+DE	10	Delaware	1059952	America/New_York
+FL	12	Florida	23462518	America/New_York
+GA	13	Georgia	11302748	America/New_York
+IA	19	Iowa	3238387	America/Chicago
+ID	16	Idaho	2029733	America/Boise
+IL	17	Illinois	12719141	America/Chicago
+IN	18	Indiana	6973333	America/Indiana/Indianapolis
+KS	20	Kansas	2977220	America/Chicago
+KY	21	Kentucky	4606864	America/New_York
+LA	22	Louisiana	4618189	America/Chicago
+MA	25	Massachusetts	7154084	America/New_York
+MD	24	Maryland	6265347	America/New_York
+ME	23	Maine	1414874	America/New_York
+MI	26	Michigan	10127884	America/Detroit
+MN	27	Minnesota	5830405	America/Chicago
+MO	29	Missouri	6270541	America/Chicago
+MS	28	Mississippi	2954160	America/Chicago
+MT	30	Montana	1144694	America/Denver
+NC	37	North Carolina	11197968	America/New_York
+ND	38	North Dakota	799358	America/Chicago
+NE	31	Nebraska	2018006	America/Chicago
+NH	33	New Hampshire	1415342	America/New_York
+NJ	34	New Jersey	9548215	America/New_York
+NM	35	New Mexico	2125498	America/Denver
+NV	32	Nevada	3282188	America/Los_Angeles
+NY	36	New York	20002427	America/New_York
+OH	39	Ohio	11900510	America/New_York
+OK	40	Oklahoma	4123288	America/Chicago
+OR	41	Oregon	4273586	America/Los_Angeles
+PA	42	Pennsylvania	13059432	America/New_York
+RI	44	Rhode Island	1114521	America/New_York
+SC	45	South Carolina	5570274	America/New_York
+SD	46	South Dakota	935094	America/Chicago
+TN	47	Tennessee	7315076	America/Chicago
+TX	48	Texas	31709821	America/Chicago
+UT	49	Utah	3538904	America/Denver
+VA	51	Virginia	8880107	America/New_York
+VT	50	Vermont	644663	America/New_York
+WA	53	Washington	8001020	America/Los_Angeles
+WI	55	Wisconsin	5972787	America/Chicago
+WV	54	West Virginia	1766147	America/New_York
+WY	56	Wyoming	588753	America/Denver

data/geo/US/street.json

@@ -0,0 +1 @@
+{ "format": "{name}", "rows": "street.tsv", "parent": "locality", "weight": "addresses" }

data/misc/.fejkdata.json

@@ -0,0 +1,25 @@
+{
+	"index": {
+		"misc.car": {"paths":["","make","model"]},
+		"misc.coordinate": {"paths":["","lat","lon"]},
+		"misc.creditcard": {"paths":["","d"]},
+		"misc.currency": {"paths":["","code","decimals","name","numeric","symbol"]},
+		"misc.datetime": {"paths":[""]},
+		"misc.emoji": {"paths":[""]},
+		"misc.httpmethod": {"paths":["","idempotent","method","safe"]},
+		"misc.httpstatus": {"paths":["","code","reason"]},
+		"misc.language": {"paths":["","code","code3","name"]},
+		"misc.loglevel": {"paths":["","code","keyword","severity"]},
+		"misc.mac": {"paths":[""]},
+		"misc.mimetype": {"paths":["","ext","type"]},
+		"misc.objectid": {"paths":[""]},
+		"misc.port": {"paths":["","number","service"]},
+		"misc.protocol": {"paths":["","keyword","name","number"]},
+		"misc.territory": {"paths":["","alpha2","alpha3","calling-code","capital","country","currency","flag","languages","name","numeric","timezone","timezone.offset","timezone.territory","timezone.weight","timezone.zone","tld"]},
+		"misc.timezone": {"parent":"territory","paths":["","offset","territory","weight","zone"]},
+		"misc.tld": {"paths":["","tld","type","unicode"]},
+		"misc.useragent": {"paths":["","browser","device","os","ua"]},
+		"misc.uuid": {"paths":["","variant"]}
+	},
+	"namespace": "misc"
+}

data/misc/car.json

@@ -1,10 +1 @@
-[
-  { "format": "{maker} {model}", "maker": "BMW", "model": ["3 Series", "5 Series", "X3", "X5"] },
-  { "format": "{maker} {model}", "maker": "Ford", "model": ["Fiesta", "Focus", "Mustang", "Explorer"] },
-  { "format": "{maker} {model}", "maker": "Honda", "model": ["Civic", "Accord", "CR-V", "Jazz"] },
-  { "format": "{maker} {model}", "maker": "Mercedes-Benz", "model": ["A-Class", "C-Class", "E-Class", "GLC"] },
-  { "format": "{maker} {model}", "maker": "Tesla", "model": ["Model 3", "Model S", "Model X", "Model Y"] },
-  { "format": "{maker} {model}", "maker": "Toyota", "model": ["Corolla", "Camry", "RAV4", "Yaris"] },
-  { "format": "{maker} {model}", "maker": "Volkswagen", "model": ["Golf", "Passat", "Polo", "Tiguan"] },
-  { "format": "{maker} {model}", "maker": "Volvo", "model": ["XC40", "XC60", "XC90", "V60"] }
-]
+{ "format": "{make} {model}", "rows": "car.tsv" }

data/misc/car.tsv

@@ -0,0 +1,33 @@
+make	model
+BMW	3 Series
+BMW	5 Series
+BMW	X3
+BMW	X5
+Ford	Explorer
+Ford	Fiesta
+Ford	Focus
+Ford	Mustang
+Honda	Accord
+Honda	CR-V
+Honda	Civic
+Honda	Jazz
+Mercedes-Benz	A-Class
+Mercedes-Benz	C-Class
+Mercedes-Benz	E-Class
+Mercedes-Benz	GLC
+Tesla	Model 3
+Tesla	Model S
+Tesla	Model X
+Tesla	Model Y
+Toyota	Camry
+Toyota	Corolla
+Toyota	RAV4
+Toyota	Yaris
+Volkswagen	Golf
+Volkswagen	Passat
+Volkswagen	Polo
+Volkswagen	Tiguan
+Volvo	V60
+Volvo	XC40
+Volvo	XC60
+Volvo	XC90

data/misc/country.json

@@ -1,22 +0,0 @@
-[
-  { "format": "{name}", "name": "Australia", "alpha2": "AU", "alpha3": "AUS" },
-  { "format": "{name}", "name": "Brazil", "alpha2": "BR", "alpha3": "BRA" },
-  { "format": "{name}", "name": "Canada", "alpha2": "CA", "alpha3": "CAN" },
-  { "format": "{name}", "name": "China", "alpha2": "CN", "alpha3": "CHN" },
-  { "format": "{name}", "name": "Denmark", "alpha2": "DK", "alpha3": "DNK" },
-  { "format": "{name}", "name": "Finland", "alpha2": "FI", "alpha3": "FIN" },
-  { "format": "{name}", "name": "France", "alpha2": "FR", "alpha3": "FRA" },
-  { "format": "{name}", "name": "Germany", "alpha2": "DE", "alpha3": "DEU" },
-  { "format": "{name}", "name": "India", "alpha2": "IN", "alpha3": "IND" },
-  { "format": "{name}", "name": "Italy", "alpha2": "IT", "alpha3": "ITA" },
-  { "format": "{name}", "name": "Japan", "alpha2": "JP", "alpha3": "JPN" },
-  { "format": "{name}", "name": "Mexico", "alpha2": "MX", "alpha3": "MEX" },
-  { "format": "{name}", "name": "Netherlands", "alpha2": "NL", "alpha3": "NLD" },
-  { "format": "{name}", "name": "Norway", "alpha2": "NO", "alpha3": "NOR" },
-  { "format": "{name}", "name": "Poland", "alpha2": "PL", "alpha3": "POL" },
-  { "format": "{name}", "name": "Spain", "alpha2": "ES", "alpha3": "ESP" },
-  { "format": "{name}", "name": "Sweden", "alpha2": "SE", "alpha3": "SWE" },
-  { "format": "{name}", "name": "Switzerland", "alpha2": "CH", "alpha3": "CHE" },
-  { "format": "{name}", "name": "United Kingdom", "alpha2": "GB", "alpha3": "GBR" },
-  { "format": "{name}", "name": "United States", "alpha2": "US", "alpha3": "USA" }
-]

data/misc/currency.json

@@ -1,18 +1 @@
-[
-  { "format": "{code}", "code": "AUD", "name": "Australian Dollar", "symbol": "$" },
-  { "format": "{code}", "code": "BRL", "name": "Brazilian Real", "symbol": "R$" },
-  { "format": "{code}", "code": "CAD", "name": "Canadian Dollar", "symbol": "$" },
-  { "format": "{code}", "code": "CHF", "name": "Swiss Franc", "symbol": "CHF" },
-  { "format": "{code}", "code": "CNY", "name": "Chinese Yuan", "symbol": "¥" },
-  { "format": "{code}", "code": "DKK", "name": "Danish Krone", "symbol": "kr" },
-  { "format": "{code}", "code": "EUR", "name": "Euro", "symbol": "€" },
-  { "format": "{code}", "code": "GBP", "name": "Pound Sterling", "symbol": "£" },
-  { "format": "{code}", "code": "INR", "name": "Indian Rupee", "symbol": "₹" },
-  { "format": "{code}", "code": "JPY", "name": "Japanese Yen", "symbol": "¥" },
-  { "format": "{code}", "code": "MXN", "name": "Mexican Peso", "symbol": "$" },
-  { "format": "{code}", "code": "NOK", "name": "Norwegian Krone", "symbol": "kr" },
-  { "format": "{code}", "code": "PLN", "name": "Polish Zloty", "symbol": "zł" },
-  { "format": "{code}", "code": "SEK", "name": "Swedish Krona", "symbol": "kr" },
-  { "format": "{code}", "code": "USD", "name": "US Dollar", "symbol": "$" },
-  { "format": "{code}", "code": "ZAR", "name": "South African Rand", "symbol": "R" }
-]
+{ "format": "{code}", "rows": "currency.tsv", "key": "code", "name": "name" }

data/misc/currency.tsv

@@ -0,0 +1,166 @@
+code	decimals	name	numeric	symbol
+AED	2	UAE Dirham	784	AED
+AFN	2	Afghani	971	؋
+ALL	2	Lek	008	ALL
+AMD	2	Armenian Dram	051	֏
+AOA	2	Kwanza	973	Kz
+ARS	2	Argentine Peso	032	$
+AUD	2	Australian Dollar	036	$
+AWG	2	Aruban Florin	533	AWG
+AZN	2	Azerbaijan Manat	944	₼
+BAM	2	Convertible Mark	977	KM
+BBD	2	Barbados Dollar	052	$
+BDT	2	Taka	050	৳
+BHD	3	Bahraini Dinar	048	BHD
+BIF	0	Burundi Franc	108	BIF
+BMD	2	Bermudian Dollar	060	$
+BND	2	Brunei Dollar	096	$
+BOB	2	Boliviano	068	Bs
+BOV	2	Mvdol	984	BOV
+BRL	2	Brazilian Real	986	R$
+BSD	2	Bahamian Dollar	044	$
+BTN	2	Ngultrum	064	BTN
+BWP	2	Pula	072	P
+BYN	2	Belarusian Ruble	933	BYN
+BZD	2	Belize Dollar	084	$
+CAD	2	Canadian Dollar	124	$
+CDF	2	Congolese Franc	976	CDF
+CHE	2	WIR Euro	947	CHE
+CHF	2	Swiss Franc	756	Fr.
+CHW	2	WIR Franc	948	CHW
+CLF	4	Unidad de Fomento	990	CLF
+CLP	0	Chilean Peso	152	$
+CNY	2	Yuan Renminbi	156	¥
+COP	2	Colombian Peso	170	$
+COU	2	Unidad de Valor Real	970	COU
+CRC	2	Costa Rican Colon	188	₡
+CUP	2	Cuban Peso	192	$
+CVE	2	Cabo Verde Escudo	132	CVE
+CZK	2	Czech Koruna	203	Kč
+DJF	0	Djibouti Franc	262	DJF
+DKK	2	Danish Krone	208	kr
+DOP	2	Dominican Peso	214	$
+DZD	2	Algerian Dinar	012	DZD
+EGP	2	Egyptian Pound	818	E£
+ERN	2	Nakfa	232	ERN
+ETB	2	Ethiopian Birr	230	ETB
+EUR	2	Euro	978	€
+FJD	2	Fiji Dollar	242	$
+FKP	2	Falkland Islands Pound	238	£
+GBP	2	Pound Sterling	826	£
+GEL	2	Lari	981	₾
+GHS	2	Ghana Cedi	936	GH₵
+GIP	2	Gibraltar Pound	292	£
+GMD	2	Dalasi	270	GMD
+GNF	0	Guinean Franc	324	FG
+GTQ	2	Quetzal	320	Q
+GYD	2	Guyana Dollar	328	$
+HKD	2	Hong Kong Dollar	344	$
+HNL	2	Lempira	340	L
+HTG	2	Gourde	332	HTG
+HUF	2	Forint	348	Ft
+IDR	2	Rupiah	360	Rp
+ILS	2	New Israeli Sheqel	376	₪
+INR	2	Indian Rupee	356	₹
+IQD	3	Iraqi Dinar	368	IQD
+IRR	2	Iranian Rial	364	IRR
+ISK	0	Iceland Krona	352	kr
+JMD	2	Jamaican Dollar	388	$
+JOD	3	Jordanian Dinar	400	JOD
+JPY	0	Yen	392	¥
+KES	2	Kenyan Shilling	404	KES
+KGS	2	Som	417	⃀
+KHR	2	Riel	116	៛
+KMF	0	Comorian Franc	174	CF
+KPW	2	North Korean Won	408	₩
+KRW	0	Won	410	₩
+KWD	3	Kuwaiti Dinar	414	KWD
+KYD	2	Cayman Islands Dollar	136	$
+KZT	2	Tenge	398	₸
+LAK	2	Lao Kip	418	₭
+LBP	2	Lebanese Pound	422	L£
+LKR	2	Sri Lanka Rupee	144	Rs
+LRD	2	Liberian Dollar	430	$
+LSL	2	Loti	426	LSL
+LYD	3	Libyan Dinar	434	LYD
+MAD	2	Moroccan Dirham	504	MAD
+MDL	2	Moldovan Leu	498	MDL
+MGA	2	Malagasy Ariary	969	Ar
+MKD	2	Denar	807	MKD
+MMK	2	Kyat	104	K
+MNT	2	Tugrik	496	₮
+MOP	2	Pataca	446	MOP
+MRU	2	Ouguiya	929	MRU
+MUR	2	Mauritius Rupee	480	Rs
+MVR	2	Rufiyaa	462	MVR
+MWK	2	Malawi Kwacha	454	MWK
+MXN	2	Mexican Peso	484	$
+MXV	2	Mexican Unidad de Inversion (UDI)	979	MXV
+MYR	2	Malaysian Ringgit	458	RM
+MZN	2	Mozambique Metical	943	MZN
+NAD	2	Namibia Dollar	516	$
+NGN	2	Naira	566	₦
+NIO	2	Cordoba Oro	558	C$
+NOK	2	Norwegian Krone	578	kr
+NPR	2	Nepalese Rupee	524	Rs
+NZD	2	New Zealand Dollar	554	$
+OMR	3	Rial Omani	512	OMR
+PAB	2	Balboa	590	PAB
+PEN	2	Sol	604	PEN
+PGK	2	Kina	598	PGK
+PHP	2	Philippine Peso	608	₱
+PKR	2	Pakistan Rupee	586	Rs
+PLN	2	Zloty	985	zł
+PYG	0	Guarani	600	₲
+QAR	2	Qatari Rial	634	QAR
+RON	2	Romanian Leu	946	lei
+RSD	2	Serbian Dinar	941	RSD
+RUB	2	Russian Ruble	643	₽
+RWF	0	Rwanda Franc	646	RF
+SAR	2	Saudi Riyal	682	SAR
+SBD	2	Solomon Islands Dollar	090	$
+SCR	2	Seychelles Rupee	690	SCR
+SDG	2	Sudanese Pound	938	SDG
+SEK	2	Swedish Krona	752	kr
+SGD	2	Singapore Dollar	702	$
+SHP	2	Saint Helena Pound	654	£
+SLE	2	Leone	925	SLE
+SOS	2	Somali Shilling	706	SOS
+SRD	2	Surinam Dollar	968	$
+SSP	2	South Sudanese Pound	728	£
+STN	2	Dobra	930	Db
+SVC	2	El Salvador Colon	222	SVC
+SYP	2	Syrian Pound	760	£
+SZL	2	Lilangeni	748	SZL
+THB	2	Baht	764	฿
+TJS	2	Somoni	972	TJS
+TMT	2	Turkmenistan New Manat	934	TMT
+TND	3	Tunisian Dinar	788	TND
+TOP	2	Pa’anga	776	T$
+TRY	2	Turkish Lira	949	₺
+TTD	2	Trinidad and Tobago Dollar	780	$
+TWD	2	New Taiwan Dollar	901	$
+TZS	2	Tanzanian Shilling	834	TZS
+UAH	2	Hryvnia	980	₴
+UGX	0	Uganda Shilling	800	UGX
+USD	2	US Dollar	840	$
+USN	2	US Dollar (Next day)	997	USN
+UYI	0	Uruguay Peso en Unidades Indexadas (UI)	940	UYI
+UYU	2	Peso Uruguayo	858	$
+UYW	4	Unidad Previsional	927	UYW
+UZS	2	Uzbekistan Sum	860	UZS
+VED	2	Bolívar Soberano	926	VED
+VES	2	Bolívar Soberano	928	VES
+VND	0	Dong	704	₫
+VUV	0	Vatu	548	VUV
+WST	2	Tala	882	WST
+XAD	2	Arab Accounting Dinar	396	XAD
+XAF	0	CFA Franc BEAC	950	FCFA
+XCD	2	East Caribbean Dollar	951	$
+XCG	2	Caribbean Guilder	532	Cg.
+XOF	0	CFA Franc BCEAO	952	F CFA
+XPF	0	CFP Franc	953	CFPF
+YER	2	Yemeni Rial	886	YER
+ZAR	2	Rand	710	R
+ZMW	2	Zambian Kwacha	967	ZK
+ZWG	2	Zimbabwe Gold	924	ZWG

data/misc/datetime.json

@@ -0,0 +1 @@
+"{date(2000-01-01,2029-12-31,'2006-01-02T15:04:05Z')}"

data/misc/fs.go

@@ -0,0 +1,15 @@
+// Package misc is fejkdata's locale-neutral module, whose categories are misc.*.
+package misc
+
+import (
+	"embed"
+	"io/fs"
+)
+
+// FS holds the module's categories and its manifest.
+//
+//go:embed *.json *.tsv .fejkdata.json
+var FS embed.FS
+
+// Modules returns FS, which reads no other module, to pass to fejkdata.WithDataFS.
+func Modules() []fs.FS { return []fs.FS{FS} }

data/misc/fs_test.go

@@ -0,0 +1,22 @@
+package misc_test
+
+import (
+	"io/fs"
+	"testing"
+
+	"github.com/larvit/fejkdata"
+	"github.com/larvit/fejkdata/data/misc"
+)
+
+func TestFSLoadsByItsManifest(t *testing.T) {
+	if _, err := fs.Stat(misc.FS, ".fejkdata.json"); err != nil {
+		t.Fatalf("FS holds no manifest: %v", err)
+	}
+	f, err := fejkdata.New(fejkdata.WithDataFS(misc.Modules()...), fejkdata.WithSeed(1))
+	if err != nil {
+		t.Fatalf("New(WithDataFS(Modules()...)) = %v", err)
+	}
+	if len(f.List()) == 0 {
+		t.Error("List() is empty, want the module's categories")
+	}
+}

data/misc/httpmethod.json

@@ -0,0 +1 @@
+{ "format": "{method}", "rows": "httpmethod.tsv", "key": "method" }