Compare commits

..
Author SHA1 Message Date
Garry TanandClaude Fable 5 a42352081c test(extract): pin the version-arm stamp floor with an explicit regression test
The floor in extractStaleFromDB was only covered incidentally — the #1768
test's updated_at (2026-06-02) fell below the bumped VERSION_TS, but its
comment still says the date was chosen to sit ABOVE the watermark. A future
date 'repair' there would silently drop floor coverage. This test pins it
directly: pre-watermark page clears after --stale and stays cleared.
Verified fail-without-fix against master's extract.ts.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-22 11:59:09 -07:00
Garry TanandClaude Fable 5 070f4c5678 fix(extract): floor --stale stamp at LINK_EXTRACTOR_VERSION_TS so version bumps don't create permanently-stale pages
The version-ts bump in this PR exposed a latent conflict: extractStaleFromDB
stamps links_extracted_at = the page's read updated_at (#1768 µs fix), but a
page last edited BEFORE the new LINK_EXTRACTOR_VERSION_TS then lands below the
version watermark and the 'links_extracted_at < versionTs' arm re-flags it
stale on every run — an infinite re-extraction loop. Stamp
max(updated_at, versionTs); versionTs is always a past release date, so the
D4 concurrent-edit race guard still holds.

Fixes the test/extract-stale.test.ts CI failure on this branch.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-22 10:56:38 -07:00
Garry TanandClaude Fable 5 4b2df89935 fix(extract): resolve path-style wikilinks outside DIR_PATTERN + add ops/ to whitelist (#2576 bug 2)
- Add ops to DIR_PATTERN so [[ops/services/...]] wikilinks and bare
  ops/... slug refs are recognized as qualified entity references.
- queryBasenameIndex (the shared basename matcher behind
  resolveBasenameMatches, the FS resolver, and the doctor check) now
  falls back to the path tail when a path-style ref misses — before,
  normalizeBasename stripped slashes into a garbage key that could
  never hit the tail-keyed index.
- Bump LINK_EXTRACTOR_VERSION_TS so extract --stale re-sweeps
  previously-stamped pages with the new extraction logic.

Bugs 1+3 of #2576 are covered by PR #2717 (--stale nullResolver).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-21 14:33:45 -07:00
8 changed files with 70 additions and 88 deletions
-7
View File
@@ -438,13 +438,6 @@ export async function runApplyMigrations(args: string[]): Promise<void> {
const result = await m.orchestrator(orchestratorOptsFrom(cli));
if (result.status === 'failed') {
console.error(`Migration v${m.version} reported status=failed.`);
// Surface each failed phase's detail — the ledger records it, but
// the operator needs it on stderr to act (#921).
for (const p of result.phases) {
if (p.status === 'failed') {
console.error(` phase ${p.name}: ${p.detail ?? '(no detail)'}`);
}
}
// Record the attempt as 'partial' (not 'complete') so the cap counts
// it. Don't let a failed orchestrator look like it never ran.
try {
+9 -1
View File
@@ -1743,7 +1743,15 @@ async function extractStaleFromDB(
// `page.updated_at.toISOString()` — the JS Date is ms-truncated, so the
// µs-precision DB updated_at stayed strictly greater and the page never
// cleared on Postgres. Stamping the exact value makes them equal.
processedRefs.push({ slug: page.slug, source_id: page.source_id, extractedAt: page.updated_at_iso });
//
// Version-arm floor: a page last edited BEFORE LINK_EXTRACTOR_VERSION_TS
// would otherwise be stamped below the version watermark and stay
// permanently stale (`links_extracted_at < versionTs` re-fires every run).
// Stamp max(updated_at, versionTs) — versionTs is always a past release
// date, so a concurrent edit's now() still exceeds the stamp and D4 holds.
// Tie at ms precision picks updated_at_iso (its µs ≥ versionTs's .000000).
const stampTs = new Date(page.updated_at_iso) >= new Date(versionTs) ? page.updated_at_iso : versionTs;
processedRefs.push({ slug: page.slug, source_id: page.source_id, extractedAt: stampTs });
}
// Flush NON-swallowing (CDX-4): a throw here propagates out of the sweep so
+11 -15
View File
@@ -186,6 +186,17 @@ async function phaseBFenceFacts(
const localPathById = new Map<string, string | null>();
for (const s of sources) localPathById.set(s.id, s.local_path);
// Dirty-tree refusal: check every source's local_path before writing.
for (const [id, localPath] of localPathById) {
if (localPath && isLocalPathDirty(localPath)) {
return {
name: 'fence_facts',
status: 'failed',
detail: `source "${id}" has uncommitted changes in ${localPath}. Commit or stash, then re-run.`,
};
}
}
// Walk legacy rows in (source_id, entity_slug) groups for per-page
// atomic writes.
const legacy = await engine.executeRaw<LegacyFactRow>(
@@ -224,21 +235,6 @@ async function phaseBFenceFacts(
groups.set(key, list);
}
// Dirty-tree refusal: check ONLY the sources we are about to write
// into. A dirty tree in an unrelated source (or zero fenceable rows
// at all) must not block a no-op or a targeted backfill (#927).
const targetSourceIds = new Set([...groups.keys()].map(k => k.split('\0')[0]));
for (const id of targetSourceIds) {
const localPath = localPathById.get(id);
if (localPath && isLocalPathDirty(localPath)) {
return {
name: 'fence_facts',
status: 'failed',
detail: `source "${id}" has uncommitted changes in ${localPath}. Commit or stash, then re-run.`,
};
}
}
for (const [key, group] of groups) {
const [sourceId, entitySlug] = key.split('\0');
const localPath = localPathById.get(sourceId)!;
+13 -4
View File
@@ -28,7 +28,7 @@ import { ensureWellFormed } from './text-safe.ts';
* OR updated_at > links_extracted_at`. It is an ISO-8601 string (NOT a number) —
* the column is TIMESTAMPTZ and the predicate binds it as `::timestamptz`.
*/
export const LINK_EXTRACTOR_VERSION_TS = '2026-05-31T00:00:00Z';
export const LINK_EXTRACTOR_VERSION_TS = '2026-07-21T00:00:00Z';
// ─── Entity references ──────────────────────────────────────────
@@ -80,10 +80,10 @@ export type LinkResolutionType = 'qualified' | 'unqualified';
* Directory prefix whitelist. These are the top-level slug dirs the extractor
* recognizes as entity references. Upstream canonical + our extensions:
* - Gbrain canonical: people, companies, meetings, concepts, deal, civic, project, source, media, yc, projects
* - Our domain extensions: tech, finance, personal, openclaw (domain-organized wikis)
* - Our domain extensions: tech, finance, personal, openclaw, ops (domain-organized wikis)
* - Our entity prefix: entities (we kept some legacy entities/projects/ pages)
*/
const DIR_PATTERN = '(?:people|companies|meetings|concepts|deal|civic|project|projects|source|media|yc|tech|finance|personal|openclaw|entities)';
const DIR_PATTERN = '(?:people|companies|meetings|concepts|deal|civic|project|projects|source|media|yc|tech|finance|personal|openclaw|entities|ops)';
/**
* Match `[Name](path)` markdown links pointing to entity directories.
@@ -865,7 +865,16 @@ export function queryBasenameIndex(idx: Map<string, string[]>, name: string): st
if (!name || typeof name !== 'string') return [];
const trimmed = name.trim();
if (!trimmed) return [];
const hit = idx.get(trimmed) ?? idx.get(trimmed.toLowerCase()) ?? idx.get(normalizeBasename(trimmed));
let hit = idx.get(trimmed) ?? idx.get(trimmed.toLowerCase()) ?? idx.get(normalizeBasename(trimmed));
// Issue #2576 bug 2: path-style refs (`runbooks/2026-05-01-x`) from dirs
// outside DIR_PATTERN reach here, but normalizeBasename strips slashes
// into a garbage key (`runbooks2026-05-01-x`) that can never hit the
// tail-keyed index. Fall back to the path tail so qualified refs resolve
// by basename like everything else.
if (!hit && trimmed.includes('/')) {
const tail = trimmed.slice(trimmed.lastIndexOf('/') + 1).trim();
if (tail) hit = idx.get(tail) ?? idx.get(tail.toLowerCase()) ?? idx.get(normalizeBasename(tail));
}
return hit ? [...hit].sort(basenameSort) : [];
}
-13
View File
@@ -180,16 +180,3 @@ describe('runApplyMigrations exit codes (v0.36.1.x #1062)', () => {
expect(src).toMatch(/All migrations up to date[\s\S]{0,80}process\.exit\(0\)/);
});
});
// #921: a failed orchestrator must print each failed phase's detail to
// stderr — not just "reported status=failed" — so the operator can act
// without digging through the ledger.
describe('failed migration prints phase detail (#921)', () => {
test('runner loops result.phases and console.errors failed phase details', async () => {
const { readFileSync } = await import('fs');
const src = readFileSync('src/commands/apply-migrations.ts', 'utf8');
expect(src).toMatch(
/reported status=failed[\s\S]{0,400}for \(const p of result\.phases\)[\s\S]{0,200}p\.status === 'failed'[\s\S]{0,200}console\.error\([\s\S]{0,80}p\.name[\s\S]{0,80}p\.detail/,
);
});
});
+18
View File
@@ -209,6 +209,24 @@ describe('gbrain extract --stale', () => {
expect(usRows[0]?.eq).toBe(true);
});
test('version-arm floor: page edited BEFORE LINK_EXTRACTOR_VERSION_TS clears after --stale (issue #2576 bug 3)', async () => {
// A page whose updated_at predates the version watermark used to be
// stamped at its updated_at (< versionTs), so the version arm re-fired
// every run — permanently stale. The sweep now floors the stamp at
// versionTs. (The #1768 test above also covers this since the v0.42.x
// VERSION_TS bump moved its date below the watermark, but this pins the
// behavior explicitly so a date "repair" there can't drop coverage.)
await engine.putPage('people/alice', personPage('Alice'));
await engine.executeRaw(`UPDATE pages SET updated_at = '2000-01-01T00:00:00Z' WHERE slug = 'people/alice'`);
expect(await engine.countStalePagesForExtraction({ versionTs: LINK_EXTRACTOR_VERSION_TS })).toBe(1);
await runExtract(engine, ['--stale']);
// Pre-floor this stayed 1 forever (stamp < versionTs → version arm re-fires).
expect(await engine.countStalePagesForExtraction({ versionTs: LINK_EXTRACTOR_VERSION_TS })).toBe(0);
await runExtract(engine, ['--stale']);
expect(await engine.countStalePagesForExtraction({ versionTs: LINK_EXTRACTOR_VERSION_TS })).toBe(0);
});
test('CDX-4 (D2): a link-flush throw aborts the sweep and leaves pages UNSTAMPED', async () => {
await engine.putPage('people/alice', personPage('Alice'));
await engine.putPage('companies/acme', companyPage('Acme', '[Alice](people/alice) founded [Acme](companies/acme).'));
+18
View File
@@ -140,6 +140,15 @@ describe('extractEntityRefs', () => {
expect(wikiRefs[0].needsResolution).toBe(true);
});
test('recognizes ops/ qualified wikilinks (issue #2576 bug 2)', () => {
// `ops` was missing from DIR_PATTERN, so [[ops/...]] fell through to
// the generic 2c pass (needsResolution) instead of being a real ref.
const refs = extractEntityRefs('Deployed via [[ops/services/pointer-agent]].');
expect(refs.length).toBe(1);
expect(refs[0].slug).toBe('ops/services/pointer-agent');
expect(refs[0].needsResolution).toBeUndefined();
});
test('skips qualified-syntax tokens (those belong to 2a)', () => {
// [[wiki:topics/ai]] looks like 2a's qualified shape — even though
// it wouldn't satisfy DIR_PATTERN, 2c must not claim it either
@@ -1069,6 +1078,15 @@ describe('makeResolver — fallback chain', () => {
]);
});
test('resolveBasenameMatches: path-style ref falls back to the tail (issue #2576 bug 2)', async () => {
// normalizeBasename strips slashes, so `runbooks/2026-05-01-pointer-agent`
// used to normalize to a garbage key that never hit the tail-keyed index.
const engine = makeFakeEngineWithSlugs(['ops/changes/2026-05-01-pointer-agent']);
const r = makeResolver(engine);
expect(await r.resolveBasenameMatches!('runbooks/2026-05-01-pointer-agent'))
.toEqual(['ops/changes/2026-05-01-pointer-agent']);
});
test('resolveBasenameMatches: case-insensitive fallback', async () => {
const engine = makeFakeEngineWithSlugs(['companies/fast-weigh']);
const r = makeResolver(engine);
+1 -48
View File
@@ -10,11 +10,10 @@
* __setTestEngineOverride so we don't need a configured brain.
*/
import { describe, test, expect, beforeAll, afterAll, beforeEach, afterEach } from 'bun:test';
import { describe, test, expect, beforeAll, afterAll, beforeEach } from 'bun:test';
import { mkdtempSync, rmSync, existsSync, readFileSync, writeFileSync, mkdirSync } from 'node:fs';
import { tmpdir } from 'node:os';
import { join } from 'node:path';
import { execFileSync } from 'node:child_process';
import { PGLiteEngine } from '../src/core/pglite-engine.ts';
import { v0_32_2, __setTestEngineOverride, __testing } from '../src/commands/migrations/v0_32_2.ts';
@@ -239,52 +238,6 @@ describe('phaseBFenceFacts — happy path backfill', () => {
});
});
describe('phaseBFenceFacts — dirty-tree refusal scoping (#927)', () => {
let dirtyDir: string;
beforeEach(async () => {
// A second source whose local_path is a git repo with uncommitted changes.
dirtyDir = mkdtempSync(join(tmpdir(), 'mig-v0_32_2-dirty-'));
execFileSync('git', ['-C', dirtyDir, 'init', '-q']);
writeFileSync(join(dirtyDir, 'uncommitted.md'), 'dirty', 'utf-8');
// eslint-disable-next-line @typescript-eslint/no-explicit-any
await (engine as any).db.query(
`INSERT INTO sources (id, name, local_path) VALUES ('other', 'other', $1)`,
[dirtyDir],
);
});
afterEach(async () => {
// eslint-disable-next-line @typescript-eslint/no-explicit-any
await (engine as any).db.query(`DELETE FROM sources WHERE id = 'other'`);
rmSync(dirtyDir, { recursive: true, force: true });
});
test('no legacy facts at all → complete, dirty unrelated source ignored', async () => {
const r = await __testing.phaseBFenceFacts(engine, OPTS);
expect(r.status).toBe('complete');
expect(r.detail).toContain('scanned=0');
});
test('facts scoped to a clean source fence despite dirty unrelated source', async () => {
await seedLegacyFact({ entity_slug: 'people/alice', fact: 'Founded Acme' });
const r = await __testing.phaseBFenceFacts(engine, OPTS);
expect(r.status).toBe('complete');
expect(r.detail).toContain('fenced=1');
expect(existsSync(join(brainDir, 'people/alice.md'))).toBe(true);
});
test('still refuses when the TARGETED source is dirty', async () => {
await seedLegacyFact({ entity_slug: 'people/alice', fact: 'F1', source_id: 'other' });
const r = await __testing.phaseBFenceFacts(engine, OPTS);
expect(r.status).toBe('failed');
expect(r.detail).toContain('"other"');
expect(r.detail).toContain('uncommitted changes');
});
});
describe('phaseCVerify', () => {
test('returns complete when fence + DB row counts match', async () => {
await seedLegacyFact({ entity_slug: 'people/alice', fact: 'F1' });