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
7 changed files with 63 additions and 68 deletions
+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
+1 -5
View File
@@ -22,7 +22,6 @@
*/
import { resolveRecipe } from './model-resolver.ts';
import { listRecipes } from './recipes/index.ts';
import { AIConfigError } from './errors.ts';
export interface ProviderCapabilities {
@@ -78,10 +77,7 @@ export function getProviderCapabilities(modelString: string): ProviderCapabiliti
if (!chat) {
throw new AIConfigError(
`Provider "${recipe.id}" does not offer a chat touchpoint.`,
// Computed from the registry so the hint can't drift into listing
// chat-less providers (the pre-fix list falsely included embedding-only
// recipes, sending users in circles — #1157).
`Known providers with chat: ${listRecipes().filter(r => r.touchpoints.chat).map(r => r.id).join(', ')}. Pick one for models.tier.subagent.`,
`Known providers with chat: openai, anthropic, google, openrouter, litellm-proxy, deepseek, groq, together, azure-openai, dashscope, minimax, zhipu, ollama, llama-server. Pick one for models.tier.subagent.`,
);
}
+4 -19
View File
@@ -1,10 +1,9 @@
import type { Recipe } from '../types.ts';
/**
* Zhipu AI (智谱AI) BigModel Open Platform. OpenAI-compatible /embeddings and
* /chat/completions endpoints at open.bigmodel.cn. Hosts embedding-2 (1024d),
* embedding-3 (Matryoshka up to 2048d), and the GLM chat family (glm-5.1 etc.)
* with native tool calling — usable for models.tier.subagent (#1157).
* Zhipu AI (智谱AI) BigModel Open Platform. OpenAI-compatible /embeddings
* endpoint at open.bigmodel.cn. Hosts embedding-2 (1024d) and embedding-3
* (Matryoshka up to 2048d).
*
* embedding-3 at 2048 dims exceeds pgvector's HNSW cap of 2000 — those
* brains fall back to exact vector scans (see
@@ -26,20 +25,6 @@ export const zhipu: Recipe = {
setup_url: 'https://open.bigmodel.cn/',
},
touchpoints: {
chat: {
// Informational list (openai-compat tier: assertTouchpoint doesn't
// enforce it), so newer GLM ids pass without a recipe edit.
models: ['glm-5.1', 'glm-4.6', 'glm-4.5'],
supports_tools: true,
// gbrain-side stable tool ids (v0.38 D11) decoupled the loop from
// Anthropic response formats; GLM tool calling is stable through the
// OpenAI-compat path, same as deepseek/groq.
supports_subagent_loop: true,
// Anthropic-style cache_control markers are not honored on the
// OpenAI-compat path — the loop runs hot (degraded:no_caching warn).
supports_prompt_cache: false,
max_context_tokens: 128000,
},
embedding: {
models: ['embedding-3', 'embedding-2'],
default_dims: 1024,
@@ -51,5 +36,5 @@ export const zhipu: Recipe = {
},
},
setup_hint:
'Get an API key at https://open.bigmodel.cn/, then `export ZHIPUAI_API_KEY=...`. Chat/subagent: use `zhipu:glm-5.1`.',
'Get an API key at https://open.bigmodel.cn/, then `export ZHIPUAI_API_KEY=...`',
};
+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) : [];
}
-39
View File
@@ -69,45 +69,6 @@ describe('recipe: zhipu', () => {
expect(sql.toLowerCase()).toContain('hnsw');
});
test('chat touchpoint declares GLM models with tool + subagent-loop support (#1157)', () => {
const r = getRecipe('zhipu')!;
expect(r.touchpoints.chat).toBeDefined();
expect(r.touchpoints.chat!.models).toContain('glm-5.1');
expect(r.touchpoints.chat!.supports_tools).toBe(true);
expect(r.touchpoints.chat!.supports_subagent_loop).toBe(true);
expect(r.touchpoints.chat!.supports_prompt_cache).toBe(false);
});
test('zhipu:glm-5.1 passes the subagent capability gate (degraded:no_caching, not refused)', async () => {
// Pre-fix: getProviderCapabilities threw "does not offer a chat touchpoint"
// and classifyCapabilities returned 'unknown' → subagent submit refused.
const { getProviderCapabilities, classifyCapabilities } =
await import('../../src/core/ai/capabilities.ts');
const caps = getProviderCapabilities('zhipu:glm-5.1');
expect(caps.supportsToolCalling).toBe(true);
expect(classifyCapabilities('zhipu:glm-5.1')).toBe('degraded:no_caching');
});
test('no-chat-touchpoint error hint lists only providers that actually have chat', async () => {
// The hint is computed from the registry; every provider it names must
// really carry a chat touchpoint (pre-fix it hardcoded zhipu/dashscope/
// minimax, all embedding-only at the time).
const { getProviderCapabilities } = await import('../../src/core/ai/capabilities.ts');
const { listRecipes } = await import('../../src/core/ai/recipes/index.ts');
let hint = '';
try {
getProviderCapabilities('voyage:voyage-3');
throw new Error('expected AIConfigError for embedding-only provider');
} catch (e) {
hint = (e as { fix?: string }).fix ?? String(e);
}
const listed = hint.match(/chat: ([^.]+)\./)?.[1]?.split(', ') ?? [];
expect(listed.length).toBeGreaterThan(0);
const withChat = new Set(listRecipes().filter(r => r.touchpoints.chat).map(r => r.id));
for (const id of listed) expect(withChat.has(id)).toBe(true);
expect(listed).toContain('zhipu');
});
test('dimsProviderOptions threads dimensions for embedding-3 (Matryoshka)', async () => {
// Codex finding #1: Zhipu embedding-3 is Matryoshka 256-2048. Without
// `dimensions` on the wire, user-selected non-default dims are
+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);