corpus/README.md

@@ -1,8 +1,15 @@
 # The corpus
 
-`<name>.json` is an ADF document; `<name>.md` is the markdown `adfToMarkdown` must emit for it,
-byte for byte including the trailing newline, and that `markdownToAdf` must read back to that same
-document (AGENTS.md §2). Directories mirror `spec/flavour.md`'s sections.
+One directory per contract kind:
 
-The JSON is editor-normal — empty `attrs`, `marks` and `content` as the absent key, adjacent
-identical-mark text nodes merged — two-space indent, keys sorted.
+- `round-trip/` — `<name>.json` + `<name>.md`: the markdown `adfToMarkdown` must emit for that
+  document, byte for byte, and that `markdownToAdf` must read back to it (AGENTS.md §2). Grouped
+  by node family: `commonmark-subset/`, `block-nodes/`, `inline-nodes/`, `opaque-carry/`,
+  `combinations/`.
+- `normalization/` — `<name>.md` + `<name>.json`: markdown input, and the document
+  `markdownToAdf` must build from it. One-way; the markdown is not canonical.
+- `errors/` — `<name>.md` + `<name>.error`: markdown input that must not convert.
+- `real-payloads/` — `<name>.json`: sanitized live ADF, round-tripped ADF→markdown→ADF. No
+  expected markdown.
+
+JSON is editor-normal (AGENTS.md §2), two-space indent, keys sorted.

corpus/canonical-form/ordered-list.json

@@ -1,53 +0,0 @@
-{
-  "content": [
-    {
-      "content": [
-        {
-          "content": [
-            {
-              "content": [
-                {
-                  "text": "Bolt M8",
-                  "type": "text"
-                }
-              ],
-              "type": "paragraph"
-            }
-          ],
-          "type": "listItem"
-        },
-        {
-          "content": [
-            {
-              "content": [
-                {
-                  "text": "Nut M8",
-                  "type": "text"
-                }
-              ],
-              "type": "paragraph"
-            }
-          ],
-          "type": "listItem"
-        },
-        {
-          "content": [
-            {
-              "content": [
-                {
-                  "text": "Washer M8",
-                  "type": "text"
-                }
-              ],
-              "type": "paragraph"
-            }
-          ],
-          "type": "listItem"
-        }
-      ],
-      "type": "orderedList"
-    }
-  ],
-  "type": "doc",
-  "version": 1
-}

corpus/canonical-form/ordered-list.md

@@ -1,3 +0,0 @@
-1. Bolt M8
-2. Nut M8
-3. Washer M8

corpus/round-trip/commonmark-subset/code-block-backtick-run-mid-line.json

@@ -0,0 +1,18 @@
+{
+  "content": [
+    {
+      "attrs": {
+        "language": "bash"
+      },
+      "content": [
+        {
+          "text": "echo \"```\" >> notes.md",
+          "type": "text"
+        }
+      ],
+      "type": "codeBlock"
+    }
+  ],
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/commonmark-subset/code-block-backtick-run-mid-line.md

@@ -0,0 +1,3 @@
+````bash
+echo "```" >> notes.md
+````

corpus/round-trip/commonmark-subset/code-block-backtick-run.json

@@ -1,6 +1,9 @@
 {
   "content": [
     {
+      "attrs": {
+        "language": "markdown"
+      },
       "content": [
         {
           "text": "```\ncode\n```",

corpus/round-trip/commonmark-subset/code-block-backtick-run.md

@@ -1,4 +1,4 @@
-````
+````markdown
 ```
 code
 ```

corpus/round-trip/commonmark-subset/document-empty.json

@@ -0,0 +1,4 @@
+{
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/commonmark-subset/text-escaping.json

@@ -35,6 +35,15 @@
         }
       ],
       "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "text": "*not emphasis*",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
     }
   ],
   "type": "doc",

corpus/round-trip/commonmark-subset/text-escaping.md

@@ -5,3 +5,5 @@
 2 * 3 * 4 = 24
 
 snake_case_name
+
+\*not emphasis*

spec/flavour.md

@@ -23,7 +23,8 @@ normalizes to it through the round-trip.
 - Blockquotes prefix lines with `> `; a blank line inside a blockquote is a bare `>`.
 - ATX headings (`#` … `######`); setext input normalizes to ATX.
 - Code fences ``` with the node's language as info string, the fence lengthened past any backtick
-  run in the content; indented-code input normalizes to fences.
+  run in the content — the longest run anywhere plus one, counting mid-line runs no closing fence
+  could match; indented-code input normalizes to fences.
 - Code spans: a backtick string one longer than the longest backtick run in the text, the text
   padded with one space on each side where it begins or ends with a backtick, or begins and ends
   with a space without being all spaces. The content is literal — inline parsing does not see
@@ -37,8 +38,11 @@ normalizes to it through the round-trip.
   CommonMark autolink (absolute URI).
 - Paragraphs on one line — no soft wrapping; a soft line break in input becomes a single space.
 - Entity references in input decode to their characters; output backslash-escapes only where text
-  would otherwise parse as syntax.
-- Blocks separated by one blank line, no trailing whitespace, single trailing newline.
+  would otherwise parse as syntax: escape the leading delimiter of a construct that would otherwise
+  open, re-scan from there, and repeat — with the opener literal the closer parses as text, so
+  `*not emphasis*` is `\*not emphasis*`, one backslash.
+- Blocks separated by one blank line, no trailing whitespace, single trailing newline; a document
+  with no blocks is the empty string.
 
 ## Directives
 

todo.md

@@ -20,11 +20,18 @@ detail is settled at its own milestone.
       escape-based, never literal, since pipe cells trim and pad. At `mediaInline`, check real
       payloads for external-URL support — if it exists, revisit the media section's
       mid-text-image error and its "no slot" ground.
-- [ ] **1d — Corpus start** (§10): checked-in ADF ↔ canonical-markdown fixture pairs per spec'd
-      node, in `corpus/`, one directory per `spec/flavour.md` section.
-  - [ ] **1d1 — Canonical form**: the plain-CommonMark subset — blockquote, bulletList,
-        codeBlock, heading, orderedList, paragraph, rule, listItem, hardBreak, text, code spans,
-        and the `code`, `em`, `link`, `strike` and `strong` marks.
+- [ ] **1d — Corpus start** (§10): checked-in fixtures per spec'd node, in `corpus/`, one
+      directory per contract kind (`corpus/README.md`).
+      **Blocked on the maintainer** (§15), not to be guessed: Canonical form has no totality
+      guard — a `paragraph`, `heading`, `blockquote` or `codeBlock` carrying a `localId`, or a
+      `blockquote` carrying marks, has no spelling that keeps it, and picking one (directive
+      sections for the six CommonMark block nodes, or the opaque carry) is a permanent format
+      decision (§8). Its three collision sites stay out of the corpus until then: an `orderedList`
+      starting at 1, a `codeBlock` whose info string is empty, and `media` with an empty `alt` —
+      each a choice between the absent attribute and the empty value.
+  - [ ] **1d1 — The CommonMark subset**: blockquote, bulletList, codeBlock, heading, orderedList,
+        paragraph, rule, listItem, hardBreak, text, code spans, and the `code`, `em`, `link`,
+        `strike` and `strong` marks — one mark per text node; nesting is 1d5's.
   - [ ] **1d2 — Block nodes**: panel, expand/nestedExpand, the media family and the CommonMark
         image shape, both table forms, task and decision lists, layout, extensions, syncBlock —
         with the reserved `marks` attribute and the fence lengths nesting forces.
@@ -38,14 +45,22 @@ detail is settled at its own milestone.
         combining nodes rather than isolating one.
   - [ ] **1d6 — Input normalization**: one-way markdown→ADF fixtures, not pairs — setext
         headings, indented code, loose lists, `*`/`+` bullets, entity references, soft wraps.
-  - [ ] **1d7 — Error input**: also one-way, a markdown input per named error. Waits on milestone
-        3 naming them; 1d's pairs are valid documents only.
-- [ ] **2 — `adfToMarkdown`.** First real code — decide here where §10's coverage check lives.
+  - [ ] **1d7 — Error input**: also one-way, a markdown input per named error, asserting only
+        that conversion fails — malformed directives, the image gap, a claimed pipe-table line
+        that does not parse, the content slot, raw HTML with no mapping. Which error each returns
+        is pinned at milestone 3, where they are named.
+- [ ] **2 — `adfToMarkdown`.** First real code — decide here where §10's coverage check lives, and
+      gate that every `corpus/**/*.json` re-serializes to itself under the library's own canonical
+      serializer: one implementation, keys sorted, two spellings — two-space indent for the corpus
+      files and the block carry's body, compact for the inline carry.
 - [ ] **3 — `markdownToAdf`.** The CommonMark parser is the largest single component. The raw-HTML
       element mapping is empty until milestone 6, so at `0.1.0` every raw-HTML construct in input
-      is an error result.
+      is an error result. The CommonMark spec suite runs against it from here (§10), and
+      `corpus/errors/` gains the error each fixture must return (1d7).
 - [ ] **4 — Round-trip property tests** over the corpus, both ways — the thing that proves 2 and
-      3. Generators emit editor-normal ADF (§2).
+      3. Generators emit editor-normal ADF (§2). Real sanitized ADF from live Atlassian APIs lands
+      here too (§10), in `corpus/real-payloads/`: an ADF→markdown→ADF check with no expected
+      markdown, the payloads supplied by the maintainer.
 - [ ] **5 — Release pipeline, ship `0.1.0`.** Publish-on-version-change (§9), `NPM_TOKEN` secret,
       the repo made public first (§6). `0.1.0` is the markdown round-trip: both markdown
       directions, the types, `isAdfDocument`. The build lands here: a build tsconfig emitting JS