fix(extract): resolve cross-directory wikilinks via sync-consistent slugification (#1964)

Raw Obsidian wikilink paths ([[llm-wiki/entities/AI 3.0]]) never matched
sync-slugified slugs (llm-wiki/entities/ai-3.0) on either extraction path.

- FS path: resolveSlug now slugifies each candidate (slugifyPath, the same
  function sync uses) when the raw path-equality check misses, in both the
  direct join and the ancestor walk. Dir-qualified wikilinks resolve as
  typed edges without the global_basename flag.
- DB path: extractPageLinks resolves dir-qualified generic wikilinks by
  slugifying the path and matching by exact slug or path suffix against
  the basename index — regardless of global_basename (the flag guards
  bare-name cross-dir ambiguity, which doesn't apply to dir-qualified
  paths). A same-tail page in a different directory can no longer be the
  only candidate and is filtered out.

Fixes #1964

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

src/cli.ts

@@ -24,7 +24,6 @@ import type { GBrainConfig } from './core/config.ts';
 import type { AIGatewayConfig } from './core/ai/types.ts';
 import type { BrainEngine } from './core/engine.ts';
 import { operations, OperationError } from './core/operations.ts';
-import { resolveSourceIdEngineFree } from './core/source-resolver.ts';
 import { formatVolunteeredPage } from './core/context/volunteer.ts';
 import type { Operation, OperationContext } from './core/operations.ts';
 import { shouldForceExitAfterMain, finishCliTeardown, flushThenExit, currentExitCode, setCliExitVerdict } from './core/cli-force-exit.ts';
@@ -383,15 +382,6 @@ async function main() {
     if (op.localOnly) {
       refuseThinClient(command, cfgPre!.remote_mcp!.mcp_url);
     }
-    // #2098: the local path resolves --source / GBRAIN_SOURCE / .gbrain-source
-    // inside makeContext (ctx.sourceId), which this route never reaches — so
-    // scope must be mapped onto the op's source_id wire param before the call.
-    try {
-      applyThinClientSourceScope(op, params);
-    } catch (e: unknown) {
-      console.error(e instanceof Error ? e.message : String(e));
-      process.exit(1);
-    }
     await runThinClientRouted(op, params, cfgPre!, cliOpts);
     return;
   }
@@ -812,60 +802,6 @@ export function parseOpArgs(op: Operation, args: string[]): Record<string, unkno
   return params;
 }
 
-/**
- * #2098: thin-client source scoping. Locally, --source / GBRAIN_SOURCE /
- * .gbrain-source resolve to ctx.sourceId in makeContext; the thin-client
- * route short-circuits before that, so `gbrain query --source X` against a
- * remote brain silently searched unscoped. This runs the engine-free tiers
- * (flag → env → dotfile; the DB-backed tiers can't run without an engine —
- * the server's grant scoping covers the rest) and maps the result onto the
- * op's `source_id` wire param.
- *
- * Ops that declare their OWN `source` param (facts add, etc.) are left
- * untouched — their --source is an op param, not scope. An explicit --source
- * on an op with no source_id wire param throws (loud beats silent drop);
- * ambient env/dotfile scope with nowhere to send it is ignored, matching the
- * pre-fix behavior for non-scopeable ops. Exported for tests.
- */
-// Ops whose `source_id` wire param is NOT read-scope semantics: get_skill's
-// source_id flips the lookup from host catalog to brain-resident-pack
-// (getResidentSkillDetail). Ambient env/dotfile scope must never leak into
-// these; an explicit --source-id still passes through untouched above.
-const NON_SCOPE_SOURCE_ID_OPS = new Set(['get_skill']);
-
-export function applyThinClientSourceScope(
-  op: Operation,
-  params: Record<string, unknown>,
-  cwd?: string,
-): void {
-  if ('source' in op.params) return; // the op owns --source; not a scope flag
-  const explicit = typeof params.source === 'string' && params.source.length > 0
-    ? (params.source as string)
-    : null;
-  delete params.source; // never a wire param on these ops — don't leak it
-  // Explicit per-call scope already on the wire wins over ambient tiers.
-  if (params.source_id !== undefined || params.all_sources === true) {
-    if (explicit) {
-      throw new Error('Pass either --source or --source-id/--all-sources, not both.');
-    }
-    return;
-  }
-  const resolved = resolveSourceIdEngineFree(explicit, cwd);
-  if (!resolved) return;
-  if (!('source_id' in op.params) || NON_SCOPE_SOURCE_ID_OPS.has(op.name)) {
-    if (explicit) {
-      const hint = NON_SCOPE_SOURCE_ID_OPS.has(op.name)
-        ? `(its source_id parameter is not a scope filter; pass --source-id explicitly if you mean it)`
-        : `(the remote op has no source_id parameter; the server scopes it to your grant)`;
-      throw new Error(
-        `gbrain ${op.cliHints?.name || op.name} does not accept --source on a thin-client install ${hint}.`,
-      );
-    }
-    return; // ambient env/dotfile scope with nowhere to send it
-  }
-  params.source_id = resolved;
-}
-
 async function makeContext(engine: BrainEngine, params: Record<string, unknown>): Promise<OperationContext> {
   // v0.31.8 (D11): resolve sourceId via the canonical 6-tier chain. Honors
   // --source / GBRAIN_SOURCE / .gbrain-source / path-match / brain default /

src/commands/extract.ts

@@ -43,7 +43,7 @@ import {
 } from '../core/link-extraction.ts';
 import { createProgress } from '../core/progress.ts';
 import { getCliOptions, cliOptsToProgressOptions } from '../core/cli-options.ts';
-import { pathToSlug, pruneDir, isSyncable } from '../core/sync.ts';
+import { pathToSlug, slugifyPath, pruneDir, isSyncable } from '../core/sync.ts';
 // v0.41.18.0: withRetry + isRetryableConnError + WithRetryOpts moved to
 // src/core/retry.ts as the canonical primitive. Engine methods
 // (addLinksBatch/addTimelineEntriesBatch/upsertChunks) now self-retry via
@@ -269,14 +269,24 @@ export function extractMarkdownLinks(content: string): { name: string; relTarget
 export function resolveSlug(fileDir: string, relTarget: string, allSlugs: Set<string>): string | null {
   const targetNoExt = relTarget.endsWith('.md') ? relTarget.slice(0, -3) : relTarget;
 
-  const s1 = join(fileDir, targetNoExt);
-  if (allSlugs.has(s1)) return s1;
+  // Issue #1964: wikilinks carry raw Obsidian paths (`[[llm-wiki/entities/AI 3.0]]`)
+  // but allSlugs holds sync-slugified slugs (`llm-wiki/entities/ai-3.0`). Try the
+  // raw candidate first (back-compat), then the sync-consistent slugified form.
+  const hit = (candidate: string): string | null => {
+    if (allSlugs.has(candidate)) return candidate;
+    const slugified = slugifyPath(candidate);
+    if (slugified !== candidate && allSlugs.has(slugified)) return slugified;
+    return null;
+  };
+
+  const s1 = hit(join(fileDir, targetNoExt));
+  if (s1) return s1;
 
   const parts = fileDir.split('/').filter(Boolean);
   for (let strip = 1; strip <= parts.length; strip++) {
     const ancestor = parts.slice(0, parts.length - strip).join('/');
-    const candidate = ancestor ? join(ancestor, targetNoExt) : targetNoExt;
-    if (allSlugs.has(candidate)) return candidate;
+    const candidate = hit(ancestor ? join(ancestor, targetNoExt) : targetNoExt);
+    if (candidate) return candidate;
   }
 
   return null;

src/core/link-extraction.ts

@@ -14,6 +14,7 @@
 import type { BrainEngine } from './engine.ts';
 import type { PageType } from './types.ts';
 import { ensureWellFormed } from './text-safe.ts';
+import { slugifyPath } from './sync.ts';
 
 /**
  * v0.42.7 — link-extraction version stamp. Bump this ISO timestamp whenever the
@@ -482,14 +483,29 @@ export async function extractPageLinks(
     // pre-v0.40.8.2 behavior of dropping bare wikilinks outside
     // DIR_PATTERN.
     if (ref.needsResolution) {
-      if (!opts.globalBasename || typeof resolver.resolveBasenameMatches !== 'function') {
-        continue;
-      }
+      if (typeof resolver.resolveBasenameMatches !== 'function') continue;
       // Issue #972 (codex): resolve by the wikilink TARGET (ref.slug — the
       // text inside `[[...]]` before any `|`), NOT the display alias
       // (ref.name = match[2]). `[[struktura|the project]]` must resolve
       // `struktura`, not "the project". The display text is for context only.
-      const matches = await resolver.resolveBasenameMatches(ref.slug);
+      //
+      // Issue #1964: a dir-qualified wikilink (`[[llm-wiki/entities/AI 3.0]]`)
+      // carries a raw Obsidian path while page slugs are sync-slugified
+      // (`llm-wiki/entities/ai-3.0`). Slugify the path the same way sync does,
+      // then match by exact slug or path-suffix (wiki-root-relative authoring).
+      // This runs regardless of global_basename — it's dir-qualified, so the
+      // cross-dir false-positive risk the flag guards against doesn't apply.
+      // Mirrors the FS path's resolveSlug ancestor walk. Bare `[[name]]`
+      // wikilinks still require the global_basename flag.
+      let matches: string[] = [];
+      const slugified = ref.slug.includes('/') ? slugifyPath(ref.slug) : '';
+      if (slugified.includes('/')) {
+        const tail = slugified.slice(slugified.lastIndexOf('/') + 1);
+        matches = (await resolver.resolveBasenameMatches(tail))
+          .filter(m => m === slugified || m.endsWith(`/${slugified}`));
+      } else if (opts.globalBasename) {
+        matches = await resolver.resolveBasenameMatches(ref.slug);
+      }
       if (matches.length === 0) continue;
       const idx = content.indexOf(ref.slug);
       const context = idx >= 0 ? excerpt(content, idx, 240) : ref.name;

src/core/source-resolver.ts

@@ -160,33 +160,6 @@ export async function resolveSourceId(
   return 'default';
 }
 
-/**
- * Engine-free tiers (1-3) of the resolution chain: explicit flag →
- * GBRAIN_SOURCE env → .gbrain-source dotfile walk. Used by the thin-client
- * CLI path (#2098), which has no local engine to run tiers 4-6 or
- * assertSourceExists against — the remote server enforces existence + grant.
- * Returns null when no engine-free tier fires.
- */
-export function resolveSourceIdEngineFree(
-  explicit: string | null | undefined,
-  cwd: string = process.cwd(),
-): string | null {
-  if (explicit) {
-    if (!SOURCE_ID_RE.test(explicit)) {
-      throw new Error(`Invalid --source value "${explicit}". Must match [a-z0-9-]{1,32}.`);
-    }
-    return explicit;
-  }
-  const env = process.env.GBRAIN_SOURCE;
-  if (env && env.length > 0) {
-    if (!SOURCE_ID_RE.test(env)) {
-      throw new Error(`Invalid GBRAIN_SOURCE value "${env}". Must match [a-z0-9-]{1,32}.`);
-    }
-    return env;
-  }
-  return readDotfileWalk(cwd);
-}
-
 /**
  * Returns the id of the SINGLE registered non-default source with a
  * local_path, when exactly one such row exists. Returns null when:

test/extract-fs.test.ts

@@ -389,6 +389,33 @@ describe('resolveSlugAll', () => {
   });
 });
 
+// ─── issue #1964: cross-directory wikilinks — slug/path mismatch ──────────
+
+describe('issue #1964: raw Obsidian wikilink paths resolve to sync-slugified slugs', () => {
+  test('resolveSlug slugifies the candidate (sync-consistent), no flag needed', () => {
+    const all = new Set(['llm-wiki/entities/ai-3.0']);
+    // Wikilink literal `[[llm-wiki/entities/AI 3.0]]` — spaces + uppercase.
+    expect(resolveSlug('llm-wiki/notes', 'llm-wiki/entities/AI 3.0.md', all))
+      .toBe('llm-wiki/entities/ai-3.0');
+  });
+
+  test('resolveSlug slugifies raw (unslugified) fileDir too', () => {
+    const all = new Set(['llm-wiki/entities/ai-3.0']);
+    // fileDir comes from dirname(relPath) — the raw on-disk directory.
+    expect(resolveSlug('LLM Wiki/Notes', 'entities/AI 3.0.md', all))
+      .toBe('llm-wiki/entities/ai-3.0');
+  });
+
+  test('extractLinksFromFile resolves cross-directory wikilink with flag OFF as a typed edge', async () => {
+    const allSlugs = new Set(['llm-wiki/entities/ai-3.0', 'llm-wiki/notes/roadmap']);
+    const content = '---\ntitle: Roadmap\ntype: concept\n---\n\nSee [[llm-wiki/entities/AI 3.0]].\n';
+    const links = await extractLinksFromFile(content, 'llm-wiki/notes/roadmap.md', allSlugs);
+    expect(links.map(l => l.to_slug)).toEqual(['llm-wiki/entities/ai-3.0']);
+    // Dir-qualified path resolution is exact, NOT the basename fallback.
+    expect(links[0].link_type).not.toBe('wikilink_basename');
+  });
+});
+
 describe('issue #972 repro: bare wikilinks resolve when flag is on', () => {
   // End-to-end: reproduces the issue's exact repro inside a tempdir +
   // PGLite, then asserts edge count under both flag states.

test/link-extraction.test.ts

@@ -9,6 +9,8 @@ import {
   parseTimelineEntries,
   isAutoLinkEnabled,
   FRONTMATTER_LINK_MAP,
+  buildBasenameIndex,
+  queryBasenameIndex,
   type SlugResolver,
 } from '../src/core/link-extraction.ts';
 import type { BrainEngine } from '../src/core/engine.ts';
@@ -424,6 +426,46 @@ describe('extractPageLinks', () => {
     expect(strk!.linkType).toBe('wikilink_basename');
   });
 
+  // ─── issue #1964: dir-qualified wikilinks with raw Obsidian paths ────────
+
+  test('#1964: dir-qualified wikilink resolves via sync-consistent slugification (flag OFF)', async () => {
+    // `[[llm-wiki/entities/AI 3.0]]` is a raw Obsidian path; the page slug
+    // is the sync-slugified `llm-wiki/entities/ai-3.0`. Must resolve WITHOUT
+    // global_basename (it's dir-qualified) and must NOT leak to a same-tail
+    // page in a different directory.
+    const resolver: SlugResolver = {
+      resolve: async () => null,
+      resolveBasenameMatches: async (name) =>
+        name === 'ai-3.0' ? ['other/ai-3.0', 'llm-wiki/entities/ai-3.0'] : [],
+    };
+    const { candidates } = await extractPageLinks(
+      'llm-wiki/notes/roadmap',
+      'See [[llm-wiki/entities/AI 3.0]] for the model.',
+      {}, 'concept', resolver,
+      // opts.globalBasename omitted (= false) — path is dir-qualified
+    );
+    expect(candidates.map(c => c.targetSlug)).toEqual(['llm-wiki/entities/ai-3.0']);
+    expect(candidates[0].linkType).toBe('wikilink_basename');
+    expect(candidates[0].linkSource).toBe('wikilink-resolved');
+  });
+
+  test('#1964: path-suffix match resolves wiki-root-relative paths against a real index', async () => {
+    // Author writes `[[llm-wiki/entities/AI 3.0]]` but the brain nests the
+    // wiki under a vault dir. Suffix match rescues it; queried through the
+    // REAL basename index so the tail-key lookup is exercised end to end.
+    const idx = buildBasenameIndex(['vault/llm-wiki/entities/ai-3.0', 'people/ai-3.0']);
+    const resolver: SlugResolver = {
+      resolve: async () => null,
+      resolveBasenameMatches: async (name) => queryBasenameIndex(idx, name),
+    };
+    const { candidates } = await extractPageLinks(
+      'vault/llm-wiki/notes/roadmap',
+      'See [[llm-wiki/entities/AI 3.0]].',
+      {}, 'concept', resolver,
+    );
+    expect(candidates.map(c => c.targetSlug)).toEqual(['vault/llm-wiki/entities/ai-3.0']);
+  });
+
   test('opts.skipFrontmatter suppresses the frontmatter pass', async () => {
     // Real resolver shape that WOULD resolve frontmatter source: too,
     // but skipFrontmatter blocks the path entirely.

test/thin-client-source-scope.test.ts

@@ -1,142 +0,0 @@
-/**
- * #2098: thin-client routing dropped --source / GBRAIN_SOURCE / .gbrain-source.
- *
- * The local CLI path resolves source scope in makeContext (ctx.sourceId); the
- * thin-client route short-circuits before that and sent params verbatim, so
- * `gbrain query --source X` against a remote brain silently searched unscoped
- * (the server op ignores the unknown `source` key).
- *
- * applyThinClientSourceScope runs the engine-free tiers (flag → env → dotfile)
- * and maps the result onto the op's `source_id` wire param. These tests fail
- * without the fix (params.source_id stays undefined / params.source leaks).
- */
-
-import { describe, test, expect } from 'bun:test';
-import { mkdtempSync, rmSync, writeFileSync } from 'fs';
-import { join } from 'path';
-import { tmpdir } from 'os';
-import { applyThinClientSourceScope, parseOpArgs } from '../src/cli.ts';
-import { operationsByName } from '../src/core/operations.ts';
-import { withEnv } from './helpers/with-env.ts';
-
-const queryOp = operationsByName.query;
-
-describe('applyThinClientSourceScope (#2098)', () => {
-  test('--source maps onto the query op wire param source_id', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const params = parseOpArgs(queryOp, ['find things', '--source', 'wiki']);
-      expect(params.source).toBe('wiki'); // pre-fix state: wrong key
-      applyThinClientSourceScope(queryOp, params, '/');
-      expect(params.source_id).toBe('wiki');
-      expect('source' in params).toBe(false); // never leaks the unknown key
-    });
-  });
-
-  test('GBRAIN_SOURCE env tier fires when no flag is passed', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'gstack' }, () => {
-      const params = parseOpArgs(queryOp, ['find things']);
-      applyThinClientSourceScope(queryOp, params, '/');
-      expect(params.source_id).toBe('gstack');
-    });
-  });
-
-  test('.gbrain-source dotfile tier fires when flag and env are absent', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const tmp = mkdtempSync(join(tmpdir(), 'gbrain-thin-scope-'));
-      try {
-        writeFileSync(join(tmp, '.gbrain-source'), 'essays\n');
-        const params = parseOpArgs(queryOp, ['find things']);
-        applyThinClientSourceScope(queryOp, params, tmp);
-        expect(params.source_id).toBe('essays');
-      } finally {
-        rmSync(tmp, { recursive: true, force: true });
-      }
-    });
-  });
-
-  test('explicit --source-id on the wire wins over ambient env scope', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'gstack' }, () => {
-      const params = parseOpArgs(queryOp, ['find things', '--source-id', 'wiki']);
-      applyThinClientSourceScope(queryOp, params, '/');
-      expect(params.source_id).toBe('wiki');
-    });
-  });
-
-  test('--source together with --source-id is rejected loudly', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const params = parseOpArgs(queryOp, ['q', '--source', 'a', '--source-id', 'b']);
-      expect(() => applyThinClientSourceScope(queryOp, params, '/')).toThrow(/not both/);
-    });
-  });
-
-  test('invalid --source value is rejected loudly', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const params = parseOpArgs(queryOp, ['q', '--source', 'Bad_Value!']);
-      expect(() => applyThinClientSourceScope(queryOp, params, '/')).toThrow(/Invalid --source/);
-    });
-  });
-
-  test('--source on an op with no source_id wire param errors instead of silently dropping', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const op = operationsByName.add_tag;
-      expect('source_id' in op.params).toBe(false);
-      const params = { slug: 'x', tag: 'y', source: 'wiki' };
-      expect(() => applyThinClientSourceScope(op, params, '/')).toThrow(/--source/);
-    });
-  });
-
-  test('ambient env scope on an op with no source_id wire param is ignored (no throw)', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'wiki' }, () => {
-      const op = operationsByName.add_tag;
-      const params: Record<string, unknown> = { slug: 'x', tag: 'y' };
-      applyThinClientSourceScope(op, params, '/');
-      expect(params.source_id).toBeUndefined();
-    });
-  });
-
-  test('ops that declare their OWN source param are left untouched', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const op = operationsByName.put_raw_data;
-      expect('source' in op.params).toBe(true);
-      const params: Record<string, unknown> = { slug: 'x', source: 'crustdata', data: {} };
-      applyThinClientSourceScope(op, params, '/');
-      expect(params.source).toBe('crustdata');
-      expect(params.source_id).toBeUndefined();
-    });
-  });
-
-  test('get_skill: ambient scope never leaks into its non-scope source_id param', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'wiki' }, () => {
-      const op = operationsByName.get_skill;
-      expect('source_id' in op.params).toBe(true); // has the param, but it is a mode switch
-      const params: Record<string, unknown> = { name: 'ingest' };
-      applyThinClientSourceScope(op, params, '/');
-      expect(params.source_id).toBeUndefined(); // would flip host catalog → brain-pack lookup
-    });
-  });
-
-  test('get_skill: explicit --source errors instead of masquerading as --source-id', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const op = operationsByName.get_skill;
-      const params: Record<string, unknown> = { name: 'ingest', source: 'wiki' };
-      expect(() => applyThinClientSourceScope(op, params, '/')).toThrow(/--source-id/);
-    });
-  });
-
-  test('get_skill: explicit --source-id passes through untouched', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'gstack' }, () => {
-      const op = operationsByName.get_skill;
-      const params: Record<string, unknown> = { name: 'ingest', source_id: 'wiki' };
-      applyThinClientSourceScope(op, params, '/');
-      expect(params.source_id).toBe('wiki');
-    });
-  });
-
-  test('no scope from any tier leaves params unchanged', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const params = parseOpArgs(queryOp, ['find things']);
-      applyThinClientSourceScope(queryOp, params, '/');
-      expect(params.source_id).toBeUndefined();
-    });
-  });
-});