fix(#2128): address shared-seam review regressions (#557, action-scan, comment-strip)

Final convergence review found the shared <tag> seam introduced 3 behavior
regressions; fixed all + locked with tests:

- #557 REGRESSION: stripTaggedBlocks's attribute-tolerance stripped `<details open>`
  (the ACTIVE-milestone marker) that the old `<details>`-only regex preserved. The
  seam now takes `allowAttributes` (default false) — details/decisions strip is
  attr-INTOLERANT (preserves `<details open>`); only `<task type="…">` opts in.
  Regression test added to roadmap-parser + markdown-sectionizer suites.
- verify.cts actionZones (negative-grep-echo security scan): reverted to a bounded
  to-first-close scan `<action>([\s\S]{0,20000}?)</action>` so a grep-echo trick
  can't hide behind an unterminated inner <action> (the seam's stop-at-next-open
  would drop it). ReDoS-safe via the cap.
- check-command-router HTML-comment strip: `(?:-->|$)` fallback wiped to EOF
  (fail-closed spurious gate block) — replaced with stop-at-next-open so an
  unclosed `<!--` leaves downstream tags intact.
- Updated the extractTaggedBlocks nested-tag tests to the new (stop-at-next-open)
  behavior: `<x><x>inner</x></x>` -> ['inner'].

All vectors still linear; every fix verified in-process.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
Tom Boucher
2026-07-10 11:12:45 -04:00
parent 01b691fae8
commit 8e4ebb49e4
6 changed files with 91 additions and 34 deletions

View File

@@ -206,7 +206,15 @@ function scanNegativeGrepCommentEcho(content: string): { errors: string[]; warni
// while a prose echo on the same line is still caught.
const cmdSpanRe =
/grep(?:\s+-{1,2}[A-Za-z][A-Za-z-]*)+\s+(?:'[^']*'|"[^"]*"|[^\s'"|>&;]+)[^\n]*?(?:==|-eq|=)\s*0\b/g;
const actionZones = extractTaggedBlocks(text, 'action');
// Security scan: must see the FULL text up to the first </action> — including a
// malformed inner <action> — so a grep-echo-0 trick cannot hide behind a
// deliberately-unclosed tag. Use a bounded to-first-close scan (ReDoS-safe via
// the {0,20000} cap, #2128), NOT the stop-at-next-open extractTaggedBlocks seam
// (which would drop the span before an unterminated inner <action>).
const actionZones: string[] = [];
const actionRe = /<action>([\s\S]{0,20000}?)<\/action>/g;
let acm: RegExpExecArray | null;
while ((acm = actionRe.exec(text)) !== null) actionZones.push(acm[1]);
const scannableActionText = actionZones.map((zone) => zone.replace(cmdSpanRe, ' ')).join('\n');
// 3. Per shell SEGMENT (split lines on && / ||) extract count-grep literals and
@@ -363,7 +371,7 @@ function scanFileWideNegativeGateConflict(content: string): { warnings: string[]
reqText: string; // <action>+<acceptance_criteria> text (requirement side)
}
const tasks: TaskInfo[] = [];
for (const tc of extractTaggedBlocks(text, 'task')) {
for (const tc of extractTaggedBlocks(text, 'task', true)) {
// Extract task name.
const namem = extractTaggedBlocks(tc, 'name');
const name = namem.length ? namem[0].trim() : 'unnamed';
@@ -565,7 +573,7 @@ function cmdVerifyPlanStructure(cwd: string, filePath: string, raw: boolean): vo
}
const tasks: Record<string, unknown>[] = [];
for (const taskContent of extractTaggedBlocks(content, 'task')) {
for (const taskContent of extractTaggedBlocks(content, 'task', true)) {
const nameArr = extractTaggedBlocks(taskContent, 'name');
const taskName = nameArr.length ? nameArr[0].trim() : 'unnamed';
const hasFiles = /<files>/.test(taskContent);