Give the vulnerability filer a folder of its own, and leave its item's score to rank_todo.py

This commit is contained in:
2026-10-14 13:31:21 +02:00
committed by Lillem4n
parent 15965e7a77
commit 5a6928750c
6 changed files with 26 additions and 47 deletions
@@ -0,0 +1,143 @@
#!/usr/bin/env python3
"""File each module's known vulnerabilities as one security item atop todo.md's next release.
Usage: file_vulnerabilities.py <govulncheck -format json report> <todo.md>
Exits 1 while any vulnerability is known, filed now or before, and 0 on a clean report.
todo.md's `Next ID = N` gives the item's ID, and the release of the plan's top row, the next
release, gives its release. Its Score is 0.0 until rank_todo.py, which CI cannot run, scores
it; a security item ranks first in its release whatever it scores. The test that files into
the repository's own todo.md fails the gate when the plan's shape drifts from what this
script reads.
Scores, as todo.md's table spells them: R (risk), S (size) and C (difficulty) are 1, since
the fix is a version bump; A (value to the audience) and G (value to the goal) are 8, the goal
being README goal 1.1, installing fejkdata the way one installs other tools, which a flagged
binary works against.
"""
import json
import re
import sys
from dataclasses import dataclass, field
ROW = re.compile(r"^\| (\d+) \| ([^|]+?) \| [^|]*\| \*\*", re.M)
NEXT_ID = re.compile(r"`Next ID = (\d+)`")
SCORES = {"R": 1, "S": 1, "C": 1, "A": 8, "G": 8}
GOAL = "1.1"
@dataclass
class Module:
path: str
found_in: str
fixed: list[str] = field(default_factory=list)
unfixed: bool = False
osvs: dict[str, dict] = field(default_factory=dict)
@property
def name(self) -> str:
return "Go" if self.path == "stdlib" else f"`{self.path}`"
def version(self, version: str) -> str:
return version.removeprefix("v") if self.path == "stdlib" else version
@property
def carried(self) -> str:
return f"{self.name} {self.version(self.found_in)}"
def messages(report: str):
decoder, at = json.JSONDecoder(), 0
while at < len(report):
if report[at].isspace():
at += 1
continue
message, at = decoder.raw_decode(report, at)
yield message
def vulnerable_modules(report: str) -> list[Module]:
"""Each module govulncheck found a vulnerability in, Go (stdlib) first, then by path."""
osvs, modules = {}, {}
for message in messages(report):
if "osv" in message:
osv = message["osv"]
osvs[osv["id"]] = {"id": osv["id"], "aliases": osv.get("aliases", []), "summary": osv.get("summary", "")}
elif "finding" in message:
finding = message["finding"]
frame = finding["trace"][0]
module = modules.setdefault(frame["module"], Module(frame["module"], frame["version"]))
module.osvs[finding["osv"]] = osvs[finding["osv"]]
if "fixed_version" in finding:
module.fixed.append(finding["fixed_version"])
else:
module.unfixed = True
return sorted(modules.values(), key=lambda m: (m.path != "stdlib", m.path))
def version_key(version: str):
return [int(part) for part in re.findall(r"\d+", version)]
def tagline(module: Module) -> str:
if module.unfixed:
return f"Ship no release while fejkdata carries {module.carried}, some of whose known vulnerabilities no version fixes yet."
fixed = module.version(max(module.fixed, key=version_key))
if module.path == "stdlib":
return f"Raise Go to {fixed}, so the release binaries carry none of {module.carried}'s known vulnerabilities."
return f"Raise {module.name} to {fixed}, so fejkdata requires no version of it with a known vulnerability."
def detail(module: Module) -> str:
lines = []
for osv in sorted(module.osvs.values(), key=lambda o: o["id"]):
aliases = f" ({', '.join(osv['aliases'])})" if osv["aliases"] else ""
summary = f": {osv['summary']}" if osv["summary"] else ""
lines.append(f"- {osv['id']}{aliases}{summary}")
return f"govulncheck found these in {module.carried}:\n\n" + "\n".join(lines) + "\n"
def file_items(modules: list[Module], todo: str) -> tuple[str, list[int]]:
"""Return todo.md with an item for each module whose vulnerabilities it does not yet name, and the items' IDs."""
filed = []
for module in modules:
if all(osv in todo for osv in module.osvs):
continue
first = ROW.search(todo)
if first is None:
raise ValueError("todo.md has no item to take the next release from")
next_id = int(NEXT_ID.search(todo).group(1))
line = tagline(module)
scores = " | ".join(str(SCORES[k]) for k in "RSCAG")
row = f"| {next_id} | {first.group(2)} | security | **{line}** | {scores} | {GOAL} | 0.0 |\n"
todo = todo[: first.start()] + row + todo[first.start() :]
todo = todo.replace(f"`Next ID = {next_id}`", f"`Next ID = {next_id + 1}`", 1)
todo = todo.replace("## Details\n\n", f"## Details\n\n### {next_id}. {line}\n\n{detail(module)}\n", 1)
filed.append(next_id)
return todo, filed
def main(argv: list[str]) -> int:
report_path, todo_path = argv
with open(report_path, encoding="utf-8") as f:
modules = vulnerable_modules(f.read())
if not modules:
print("No known vulnerabilities.")
return 0
count = sum(len(m.osvs) for m in modules)
known = f"{count} known {'vulnerability' if count == 1 else 'vulnerabilities'}"
with open(todo_path, encoding="utf-8") as f:
todo, filed = file_items(modules, f.read())
if not filed:
print(f"{known}, already planned in todo.md.")
return 1
with open(todo_path, "w", encoding="utf-8") as f:
f.write(todo)
items = ", ".join(map(str, filed))
print(f"{known}; todo.md {'item ' + items + ' plans' if len(filed) == 1 else 'items ' + items + ' plan'} the fix.")
return 1
if __name__ == "__main__":
sys.exit(main(sys.argv[1:]))
@@ -0,0 +1,177 @@
import io
import json
import os
import tempfile
import unittest
from contextlib import redirect_stdout
import file_vulnerabilities
TODO = """# todo
## Scoring
`Score = -R - S/4 + 2*A + 2*G*W`
`Bar = 9`
`Next ID = 7`
| Goal | W |
|---|---|
| 1 | 1.00 |
| 1.1 | 0.90 |
## Items
| ID | Release | Exempt | Item | R | S | C | A | G | Goals | Score |
|---|---|---|---|---|---|---|---|---|---|---|
| 5 | 0.2.0 | | **Add a thing.** | 2 | 2 | 2 | 7 | 8 | 1 | 26.5 |
## Details
### 5. Add a thing.
Why.
"""
OSVS = (
{"id": "GO-2026-0002", "summary": "Denial of service in net/http", "aliases": ["GHSA-2222", "CVE-2026-2"]},
{"id": "GO-2026-0001", "summary": "Cross-site scripting in html/template", "aliases": ["CVE-2026-1"]},
{"id": "GO-2026-0003", "summary": "Panic in x/text"},
{"id": "GO-2026-0004"},
)
STDLIB = (
("GO-2026-0002", "stdlib", "v1.27.1", "v1.27.2"),
("GO-2026-0001", "stdlib", "v1.27.1", "v1.27.10"),
("GO-2026-0001", "stdlib", "v1.27.1", "v1.27.10"),
)
TEXT = (("GO-2026-0003", "golang.org/x/text", "v0.3.0", "v0.3.8"),)
GO_TAGLINE = "Raise Go to 1.27.10, so the release binaries carry none of Go 1.27.1's known vulnerabilities."
def report(*findings):
messages = [{"config": {"go_version": "go1.27.1", "scan_level": "module"}}]
messages += [{"osv": osv} for osv in OSVS]
for osv, module, version, fixed in findings:
finding = {"osv": osv, "trace": [{"module": module, "version": version}]}
if fixed:
finding["fixed_version"] = fixed
messages.append({"finding": finding})
return "\n".join(json.dumps(m, indent=2) for m in messages)
def file(todo, *findings):
return file_vulnerabilities.file_items(file_vulnerabilities.vulnerable_modules(report(*findings)), todo)
def rows(todo):
return [line for line in todo.splitlines() if "| **" in line]
class FileItems(unittest.TestCase):
def test_no_finding_files_nothing(self):
self.assertEqual(file(TODO), (TODO, []))
def test_files_one_security_item_per_module_atop_the_next_release(self):
todo, filed = file(TODO, *STDLIB)
self.assertEqual(filed, [7])
self.assertIn("`Next ID = 8`", todo)
self.assertEqual(rows(todo)[0], f"| 7 | 0.2.0 | security | **{GO_TAGLINE}** | 1 | 1 | 1 | 8 | 8 | 1.1 | 0.0 |")
self.assertTrue(rows(todo)[1].startswith("| 5 |"))
self.assertTrue(
todo.split("## Details\n\n", 1)[1].startswith(
f"### 7. {GO_TAGLINE}\n\n"
"govulncheck found these in Go 1.27.1:\n\n"
"- GO-2026-0001 (CVE-2026-1): Cross-site scripting in html/template\n"
"- GO-2026-0002 (GHSA-2222, CVE-2026-2): Denial of service in net/http\n\n"
"### 5. Add a thing."
)
)
def test_a_required_module_gets_its_own_item(self):
todo, filed = file(TODO, *STDLIB, *TEXT)
self.assertEqual(filed, [7, 8])
self.assertIn(
"| 8 | 0.2.0 | security | **Raise `golang.org/x/text` to v0.3.8, so fejkdata requires no version"
" of it with a known vulnerability.** |",
todo,
)
self.assertIn("govulncheck found these in `golang.org/x/text` v0.3.0:\n\n- GO-2026-0003: Panic in x/text\n", todo)
self.assertIn("`Next ID = 9`", todo)
def test_a_vulnerability_already_planned_is_not_filed_again(self):
planned, _ = file(TODO, *STDLIB)
self.assertEqual(file(planned, *STDLIB), (planned, []))
def test_a_module_newly_vulnerable_is_filed_beside_one_already_planned(self):
planned, _ = file(TODO, *STDLIB)
todo, filed = file(planned, *STDLIB, *TEXT)
self.assertEqual(filed, [8])
self.assertIn("| 8 | 0.2.0 | security | **Raise `golang.org/x/text`", todo)
def test_a_vulnerability_no_version_fixes_names_no_target(self):
todo, _ = file(TODO, *STDLIB, ("GO-2026-0004", "stdlib", "v1.27.1", ""))
self.assertIn("- GO-2026-0004\n", todo)
self.assertIn(
"| 7 | 0.2.0 | security | **Ship no release while fejkdata carries Go 1.27.1, some of whose known"
" vulnerabilities no version fixes yet.** |",
todo,
)
def test_a_plan_with_no_items_is_refused(self):
empty = TODO.split("| 5 |")[0].rstrip("\n") + "\n\n## Details\n"
with self.assertRaises(ValueError):
file(empty, *STDLIB)
def test_the_repositorys_plan_takes_an_item_in_its_own_shape(self):
path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "todo.md")
with open(path, encoding="utf-8") as f:
plan = f.read()
todo, filed = file(plan, *STDLIB)
header = next(line for line in plan.splitlines() if line.startswith("| ID |"))
row = rows(todo)[0]
self.assertEqual(row.count("|"), header.count("|"))
self.assertEqual(row.split(" | ")[1], rows(plan)[0].split(" | ")[1])
self.assertRegex(row, rf"^\| {filed[0]} \| .* \| 0\.0 \|$")
self.assertIn(f"`Next ID = {filed[0] + 1}`", todo)
class Main(unittest.TestCase):
def setUp(self):
cwd = os.getcwd()
tmp = tempfile.TemporaryDirectory()
self.addCleanup(tmp.cleanup)
os.chdir(tmp.name)
self.addCleanup(os.chdir, cwd)
with open("todo.md", "w", encoding="utf-8") as f:
f.write(TODO)
def run_main(self, *findings):
with open("report.json", "w", encoding="utf-8") as f:
f.write(report(*findings))
out = io.StringIO()
with redirect_stdout(out):
code = file_vulnerabilities.main(["report.json", "todo.md"])
with open("todo.md", encoding="utf-8") as f:
return code, out.getvalue(), f.read()
def test_a_clean_report_passes_and_leaves_the_plan(self):
self.assertEqual(self.run_main(), (0, "No known vulnerabilities.\n", TODO))
def test_a_finding_fails_and_files_it(self):
code, out, todo = self.run_main(*STDLIB)
self.assertEqual((code, out), (1, "2 known vulnerabilities; todo.md item 7 plans the fix.\n"))
self.assertIn("| 7 | 0.2.0 | security |", todo)
def test_findings_in_two_modules_name_both_items(self):
_, out, _ = self.run_main(*STDLIB, *TEXT)
self.assertEqual(out, "3 known vulnerabilities; todo.md items 7, 8 plan the fix.\n")
def test_one_finding_already_planned_still_fails(self):
self.run_main(*TEXT)
self.assertEqual(self.run_main(*TEXT)[:2], (1, "1 known vulnerability, already planned in todo.md.\n"))
if __name__ == "__main__":
unittest.main()