fix(import): fall back to body H1 for title when frontmatter lacks title: (#2446)

Title precedence is now frontmatter title: > body's first ATX H1 > the
slug/filename-humanized fallback. Slug-based imports (contacts, calendar)
carry a correct # Heading but no frontmatter title; without the H1 fallback
they got junk titles humanized from the slug. The H1 scan skips h2+ and
lines inside fenced code blocks, and strips closed-ATX trailing hashes.

Takeover of #2495.

Co-authored-by: javieraldape <javieraldape@users.noreply.github.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

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, slugifyPath, pruneDir, isSyncable } from '../core/sync.ts';
+import { pathToSlug, 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
@@ -269,24 +269,14 @@ 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;
 
-  // 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 s1 = join(fileDir, targetNoExt);
+  if (allSlugs.has(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 = hit(ancestor ? join(ancestor, targetNoExt) : targetNoExt);
-    if (candidate) return candidate;
+    const candidate = ancestor ? join(ancestor, targetNoExt) : targetNoExt;
+    if (allSlugs.has(candidate)) return candidate;
   }
 
   return null;

src/core/link-extraction.ts

@@ -14,7 +14,6 @@
 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
@@ -483,29 +482,14 @@ export async function extractPageLinks(
     // pre-v0.40.8.2 behavior of dropping bare wikilinks outside
     // DIR_PATTERN.
     if (ref.needsResolution) {
-      if (typeof resolver.resolveBasenameMatches !== 'function') continue;
+      if (!opts.globalBasename || 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.
-      //
-      // 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);
-      }
+      const 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/markdown.ts

@@ -135,7 +135,16 @@ export function parseMarkdown(
   const type = coerceFrontmatterString(frontmatter.type) || (
     opts?.activePack ? inferTypeFromPack(filePath, opts.activePack) : inferType(filePath)
   );
-  const title = coerceFrontmatterString(frontmatter.title).trim() || inferTitle(filePath);
+  // #2446: title precedence is frontmatter `title:` > the body's first H1 >
+  // the slug/filename-humanized fallback. Slug-based imports (contacts,
+  // calendar) write a correct `# Heading` but no frontmatter title; without
+  // the H1 fallback they get junk titles humanized from the slug
+  // (`Contact 20170928 5 John Defalco`), which also breaks anything keyed on
+  // the title (e.g. the by-mention gazetteer's first-token bucketing).
+  const title =
+    coerceFrontmatterString(frontmatter.title).trim() ||
+    inferTitleFromBody(body) ||
+    inferTitle(filePath);
   const tags = extractTags(frontmatter);
   const slug = coerceFrontmatterString(frontmatter.slug) || inferSlug(filePath);
 
@@ -602,6 +611,31 @@ function inferTypeWithPrefixes(
   return 'concept';
 }
 
+/**
+ * #2446: derive a title from the body's first ATX H1 (`# Heading`).
+ *
+ * Returns the trimmed heading text with the leading `# ` and any decorative
+ * trailing `#` run stripped, or '' if the body has no H1. Only a SINGLE leading
+ * `#` matches — `##`+ (h2 and deeper) are skipped — and lines inside a fenced
+ * code block (```/~~~) are ignored so a `# comment` in a shell snippet can't be
+ * mistaken for the page title.
+ */
+function inferTitleFromBody(body: string): string {
+  let inFence = false;
+  for (const raw of body.split('\n')) {
+    const fence = /^\s*(`{3,}|~{3,})/.exec(raw);
+    if (fence) {
+      inFence = !inFence;
+      continue;
+    }
+    if (inFence) continue;
+    // Exactly one leading `#`, then whitespace, then the heading text.
+    const m = /^#(?!#)\s+(.+?)\s*$/.exec(raw);
+    if (m) return m[1].replace(/\s+#+\s*$/, '').trim();
+  }
+  return '';
+}
+
 function inferTitle(filePath?: string): string {
   if (!filePath) return 'Untitled';
 

test/extract-fs.test.ts

@@ -389,33 +389,6 @@ 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

@@ -9,8 +9,6 @@ import {
   parseTimelineEntries,
   isAutoLinkEnabled,
   FRONTMATTER_LINK_MAP,
-  buildBasenameIndex,
-  queryBasenameIndex,
   type SlugResolver,
 } from '../src/core/link-extraction.ts';
 import type { BrainEngine } from '../src/core/engine.ts';
@@ -426,46 +424,6 @@ 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/markdown.test.ts

@@ -343,3 +343,47 @@ describe('issue #1939 — non-string frontmatter coercion', () => {
     expect(parsed.title).toBe('A Normal Title');
   });
 });
+
+// issue #2446 — when frontmatter has no `title:`, prefer the body's first H1
+// over the slug/filename-humanized fallback. Slug-based imports (contacts,
+// calendar) carry a correct `# Heading` but no frontmatter title; humanizing
+// the slug leaks date/id tokens and loses casing (`Defalco` vs `DeFalco`).
+describe('issue #2446 — body H1 fallback for missing frontmatter title', () => {
+  test('no frontmatter title uses the body H1, not the slug-humanized junk', () => {
+    const md = '---\ntype: person\n---\n\n# John DeFalco\n\nNotes about John.\n';
+    const parsed = parseMarkdown(md, 'people/contact-20170928-5-john-defalco.md');
+    expect(parsed.title).toBe('John DeFalco');
+    // The slug-derived junk title must NOT win.
+    expect(parsed.title).not.toBe('Contact 20170928 5 John Defalco');
+  });
+
+  test('no frontmatter title and no H1 falls back to the inferred slug title', () => {
+    const md = '---\ntype: note\n---\n\njust body prose, no heading\n';
+    const parsed = parseMarkdown(md, 'people/alice-example.md');
+    expect(parsed.title).toBe('Alice Example');
+  });
+
+  test('frontmatter title wins over a body H1 (no regression)', () => {
+    const md = '---\ntitle: Frontmatter Wins\n---\n\n# Body Heading\n\nbody\n';
+    const parsed = parseMarkdown(md, 'people/some-slug.md');
+    expect(parsed.title).toBe('Frontmatter Wins');
+  });
+
+  test('h2 is not treated as the title; first real H1 is used', () => {
+    const md = '---\ntype: note\n---\n\n## Subsection First\n\n# The Real Title\n\nbody\n';
+    const parsed = parseMarkdown(md, 'notes/x.md');
+    expect(parsed.title).toBe('The Real Title');
+  });
+
+  test('a # inside a fenced code block is not mistaken for the title', () => {
+    const md = '---\ntype: note\n---\n\n```sh\n# this is a shell comment, not a heading\n```\n\n# Actual Heading\n';
+    const parsed = parseMarkdown(md, 'notes/x.md');
+    expect(parsed.title).toBe('Actual Heading');
+  });
+
+  test('trailing closing hashes are stripped from the H1', () => {
+    const md = '---\ntype: note\n---\n\n# Closed ATX Heading #\n\nbody\n';
+    const parsed = parseMarkdown(md, 'notes/x.md');
+    expect(parsed.title).toBe('Closed ATX Heading');
+  });
+});