sensemaking 0.14.0 → 0.15.1

package/dist/esm/features/links.js

@@ -1,10 +1,14 @@
 import posix from 'node:path/posix';
+import { maskRegions } from '../fences.js';
 // links(src, target, target_base, dst, embed): target as written, target_base its baseKey
 // (indexed, for the incremental resolve below), dst the resolved path or NULL (a
 // queryable dead link), embed whether it's `![[x]]`/`![](x.md)` rather than `[[x]]`/`[](x.md)`
 // -- Obsidian's grain: a target both linked and embedded in the same note is two rows.
-// Wikilinks ([[target]], [[target#anchor|alias]], embeds) plus relative markdown links to .md files.
-function extract(_raw, body) {
+// Wikilinks ([[target]], [[target#anchor|alias]], embeds), internal markdown links, and
+// frontmatter values that are exactly a wikilink ("[[X]]"; mid-string and ![[...]] forms are
+// not links there -- Obsidian's rule, probe-verified).
+function extract(_raw, rawBody, _search, data) {
+    const body = /\[\[|\]\(/.test(rawBody) ? maskRegions(rawBody) : rawBody;
     const seen = new Set();
     const results = [];
     const add = (target, embed)=>{
@@ -20,12 +24,42 @@ function extract(_raw, body) {
     // matches; anchor and alias split off in code, since a class-based regex stops at the ].
     for (const m of body.matchAll(/\[\[(.*?)\]\]/g)){
         var _m_index;
-        const target = m[1].split('|')[0].split('#')[0].trim();
-        if (target) add(target, body[((_m_index = m.index) !== null && _m_index !== void 0 ? _m_index : 0) - 1] === '!');
+        const embed = body[((_m_index = m.index) !== null && _m_index !== void 0 ? _m_index : 0) - 1] === '!';
+        // [[#Heading]]: a same-note anchor link, which Obsidian resolves to a self-edge. Keep the
+        // target as written (leading #) so resolveTarget can special-case it.
+        const beforeAlias = m[1].split('|')[0].trim();
+        if (beforeAlias.startsWith('#')) {
+            if (beforeAlias.length > 1) add(beforeAlias, embed);
+            continue;
+        }
+        const target = beforeAlias.split('#')[0].trim();
+        if (target) add(target, embed);
+    }
+    // Any internal destination, not only .md-suffixed: Obsidian gives markdown links full
+    // linkpath resolution, so \](Zektor) resolves like [[Zektor]] and \](#anchor) is a
+    // self-edge. External URLs (a scheme anywhere survives the malformed double-paren case)
+    // and titled links (dest stops at whitespace) are skipped.
+    // A space in the destination is only valid ahead of a quoted title; a domain-shaped first
+    // segment (www.example.com/...) is external even without a scheme.
+    for (const m of body.matchAll(/(!)?\[[^\]]*\]\(((?:[^()\s]|\([^()\s]*\))+)(?:\s+(?:"[^)]*"|'[^)]*'))?\)/g)){
+        const dest = m[2].trim();
+        if (!dest || dest.includes('://')) continue;
+        if (/^www\./i.test(dest)) continue;
+        const target = dest.startsWith('#') ? dest : dest.split('#')[0];
+        if (target) add(target, m[1] === '!');
     }
-    for (const m of body.matchAll(/(!)?\[[^\]]*\]\(([^)]+\.md)(?:#[^)]*)?\)/g)){
-        const target = m[2].trim();
-        if (target && !/^[a-z]+:\/\//i.test(target)) add(target, m[1] === '!');
+    for (const value of Object.values(data !== null && data !== void 0 ? data : {})){
+        for (const item of Array.isArray(value) ? value : [
+            value
+        ]){
+            if (typeof item !== 'string') continue;
+            const m = /^\[\[(.*)\]\]$/.exec(item.trim());
+            if (!m) continue;
+            const inner = m[1].split('|')[0].trim();
+            const target = inner.split('#')[0].trim();
+            if (target) add(target, false);
+            else if (inner.startsWith('#') && inner.length > 1) add(inner, false);
+        }
     }
     return results;
 }
@@ -36,10 +70,11 @@ function baseKey(path) {
     return posix.basename(path).replace(/\.md$/i, '').toLowerCase();
 }
 // Obsidian-style: exact relative path (with/without .md), path relative to the linking
-// note's directory, then basename match (lexicographically first on ties).
+// note's directory, then basename match (shortest path wins on ties -- verified against a
+// cold-loaded Obsidian vault on 6+ real collision pairs; byBase's lists are pre-sorted that way).
 function resolveTarget(src, target, pathSet, byBase) {
-    var _ref;
-    var _byBase_get;
+    // [[#Heading]]: a same-note anchor, always the note itself -- never depends on other files.
+    if (target.startsWith('#')) return src;
     const clean = cleanTarget(target);
     const fromSrc = posix.normalize(posix.join(posix.dirname(src), clean));
     for (const candidate of [
@@ -50,11 +85,18 @@ function resolveTarget(src, target, pathSet, byBase) {
     ]){
         if (pathSet.has(candidate)) return candidate;
     }
-    return (_ref = (_byBase_get = byBase.get(baseKey(clean))) === null || _byBase_get === void 0 ? void 0 : _byBase_get[0]) !== null && _ref !== void 0 ? _ref : null;
+    const candidates = byBase.get(baseKey(clean));
+    if (!candidates) return null;
+    // The linking note itself wins a basename collision (Obsidian resolves to self before the
+    // shortest-path rule -- cold-load verified on the hub corpus).
+    return candidates.includes(src) ? src : candidates[0];
 }
+// Shortest path wins a basename collision; equal lengths fall back to lexicographic, our own
+// deterministic tiebreak for a case Obsidian itself leaves registration-order-dependent.
 function buildByBase(files) {
     const byBase = new Map();
-    for (const path of files.map((f)=>f.relPath).sort()){
+    const paths = files.map((f)=>f.relPath).sort((a, b)=>a.length - b.length || (a < b ? -1 : a > b ? 1 : 0));
+    for (const path of paths){
         const key = baseKey(path);
         const list = byBase.get(key);
         if (list) list.push(path);
@@ -126,9 +168,12 @@ export function linkEdges(db) {
     // embed whose exact (src, target) also exists as a link: that second row is new with the
     // embed grain and would double an edge that used to be one row. The NOT EXISTS probes the
     // primary key and fires only for embed rows; both branches still scan the table.
-    const sql = `SELECT src, dst FROM links WHERE dst IS NOT NULL AND embed = 0
+    // Self-edges (same-note anchors) are excluded here so PageRank mass is not self-recycled --
+    // Obsidian's own graph view hides self-loops too. The rows themselves stay in the table for
+    // backlinks/peek.
+    const sql = `SELECT src, dst FROM links WHERE dst IS NOT NULL AND embed = 0 AND src != dst
     UNION ALL
-    SELECT src, dst FROM links l WHERE dst IS NOT NULL AND embed = 1
+    SELECT src, dst FROM links l WHERE dst IS NOT NULL AND embed = 1 AND src != dst
       AND NOT EXISTS (SELECT 1 FROM links l0 WHERE l0.src = l.src AND l0.target = l.target AND l0.embed = 0)`;
     return db.prepare(sql).all().map((r)=>[
             r.src,
@@ -188,7 +233,10 @@ export const links = {
         // Upsert preserves dst on surviving rows, so an unchanged link resolves to the same
         // value and reports no change; only genuinely new rows start at NULL.
         const insert = db.prepare('INSERT INTO links (src, target, target_base, dst, embed) VALUES (?, ?, ?, NULL, ?) ON CONFLICT(src, target, embed) DO UPDATE SET target_base = excluded.target_base');
-        for (const { target, embed } of targets)insert.run(path, target, baseKey(cleanTarget(target)), embed ? 1 : 0);
+        // A same-note anchor's dst is always its own src, never another file's basename, so it
+        // gets no target_base -- SQL's `IN` never matches NULL, keeping it out of
+        // resolveIncremental's cross-file collectIn('target_base', ...) entirely.
+        for (const { target, embed } of targets)insert.run(path, target, target.startsWith('#') ? null : baseKey(cleanTarget(target)), embed ? 1 : 0);
     },
     afterReconcile (db, delta) {
         var _ref;

package/dist/esm/features/links.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/features/links.ts"],"sourcesContent":["import posix from 'node:path/posix';\nimport type { DatabaseSync } from 'node:sqlite';\nimport type { FileStat } from '../scan.ts';\nimport type { Feature, ReconcileDelta } from './types.ts';\n\n// links(src, target, target_base, dst, embed): target as written, target_base its baseKey\n// (indexed, for the incremental resolve below), dst the resolved path or NULL (a\n// queryable dead link), embed whether it's `![[x]]`/`![](x.md)` rather than `[[x]]`/`[](x.md)`\n// -- Obsidian's grain: a target both linked and embedded in the same note is two rows.\n\n// Wikilinks ([[target]], [[target#anchor|alias]], embeds) plus relative markdown links to .md files.\nfunction extract(_raw: string, body: string): Array<{ target: string; embed: boolean }> {\n  const seen = new Set<string>();\n  const results: Array<{ target: string; embed: boolean }> = [];\n  const add = (target: string, embed: boolean) => {\n    const key = `${target}\\0${embed ? '1' : '0'}`;\n    if (seen.has(key)) return;\n    seen.add(key);\n    results.push({ target, embed });\n  };\n  // To the first ]], as Obsidian parses it, so a heading or alias holding a lone ] still\n  // matches; anchor and alias split off in code, since a class-based regex stops at the ].\n  for (const m of body.matchAll(/\\[\\[(.*?)\\]\\]/g)) {\n    const target = m[1].split('|')[0].split('#')[0].trim();\n    if (target) add(target, body[(m.index ?? 0) - 1] === '!');\n  }\n  for (const m of body.matchAll(/(!)?\\[[^\\]]*\\]\\(([^)]+\\.md)(?:#[^)]*)?\\)/g)) {\n    const target = m[2].trim();\n    if (target && !/^[a-z]+:\\/\\//i.test(target)) add(target, m[1] === '!');\n  }\n  return results;\n}\n\nfunction cleanTarget(target: string): string {\n  return target.replace(/\\\\/g, '/').replace(/^\\.\\//, '');\n}\n\nfunction baseKey(path: string): string {\n  return posix.basename(path).replace(/\\.md$/i, '').toLowerCase();\n}\n\n// Obsidian-style: exact relative path (with/without .md), path relative to the linking\n// note's directory, then basename match (lexicographically first on ties).\nfunction resolveTarget(src: string, target: string, pathSet: Set<string>, byBase: Map<string, string[]>): string | null {\n  const clean = cleanTarget(target);\n  const fromSrc = posix.normalize(posix.join(posix.dirname(src), clean));\n  for (const candidate of [clean, `${clean}.md`, fromSrc, `${fromSrc}.md`]) {\n    if (pathSet.has(candidate)) return candidate;\n  }\n  return byBase.get(baseKey(clean))?.[0] ?? null;\n}\n\nfunction buildByBase(files: FileStat[]): Map<string, string[]> {\n  const byBase = new Map<string, string[]>();\n  for (const path of files.map((f) => f.relPath).sort()) {\n    const key = baseKey(path);\n    const list = byBase.get(key);\n    if (list) list.push(path);\n    else byBase.set(key, [path]);\n  }\n  return byBase;\n}\n\ninterface LinkRow {\n  src: string;\n  target: string;\n  dst: string | null;\n  embed: number;\n}\n\n// Applies resolveTarget to `rows`, writing only rows whose dst actually changes. The UPDATE\n// keys on (src, target): a link/embed sibling pair shares one resolution, so either row's\n// mismatch rewrites both.\nfunction resolveRows(db: DatabaseSync, rows: LinkRow[], pathSet: Set<string>, byBase: Map<string, string[]>): boolean {\n  const update = db.prepare('UPDATE links SET dst = ? WHERE src = ? AND target = ?');\n  let changed = false;\n  for (const row of rows) {\n    const dst = resolveTarget(row.src, row.target, pathSet, byBase);\n    if (dst !== row.dst) {\n      update.run(dst, row.src, row.target);\n      changed = true;\n    }\n  }\n  return changed;\n}\n\n// A new or deleted file can change any note's resolution, so re-resolve the whole table.\n// Fallback for cold builds and large deltas -- see afterReconcile's threshold.\nfunction resolveAll(db: DatabaseSync, files: FileStat[]): boolean {\n  const pathSet = new Set(files.map((f) => f.relPath));\n  const byBase = buildByBase(files);\n  const rows = db.prepare('SELECT src, target, dst, embed FROM links').all() as unknown as LinkRow[];\n  return resolveRows(db, rows, pathSet, byBase);\n}\n\n// Re-resolve only what this reconcile could have touched: re-stored sources, links whose\n// target basename matches an added/vanished file, and links whose dst just vanished. All\n// three are indexed lookups, so this never reads the whole table.\nfunction resolveIncremental(db: DatabaseSync, delta: ReconcileDelta): boolean {\n  const pathSet = new Set(delta.files.map((f) => f.relPath));\n  const byBase = buildByBase(delta.files);\n\n  const changedBasenames = new Set<string>();\n  for (const p of delta.added) changedBasenames.add(baseKey(p));\n  for (const p of delta.vanished) changedBasenames.add(baseKey(p));\n\n  // \\0 can't appear in a path or target, so the key never collides; a space can.\n  const candidates = new Map<string, LinkRow>();\n  const collect = (rows: LinkRow[]) => {\n    for (const row of rows) candidates.set(`${row.src}\\0${row.target}\\0${row.embed}`, row);\n  };\n  // Chunked so a large-but-under-threshold delta never exceeds SQLite's bound-variable\n  // limit (SQLITE_MAX_VARIABLE_NUMBER, 32766).\n  const collectIn = (column: string, keys: string[]) => {\n    for (let i = 0; i < keys.length; i += 500) {\n      const chunk = keys.slice(i, i + 500);\n      const placeholders = chunk.map(() => '?').join(', ');\n      collect(db.prepare(`SELECT src, target, dst, embed FROM links WHERE ${column} IN (${placeholders})`).all(...chunk) as unknown as LinkRow[]);\n    }\n  };\n\n  collectIn('src', delta.reparsed);\n  collectIn('target_base', [...changedBasenames]);\n  collectIn('dst', delta.vanished);\n\n  return resolveRows(db, [...candidates.values()], pathSet, byBase);\n}\n\n// Resolved edges, for rank and for search's graph expansion.\nexport function linkEdges(db: DatabaseSync): [string, string][] {\n  // One edge per row, as 0.12's grain always had it -- two written targets resolving to one\n  // dst stay two edges (a weight the fusion evals were gated on). The only exclusion is an\n  // embed whose exact (src, target) also exists as a link: that second row is new with the\n  // embed grain and would double an edge that used to be one row. The NOT EXISTS probes the\n  // primary key and fires only for embed rows; both branches still scan the table.\n  const sql = `SELECT src, dst FROM links WHERE dst IS NOT NULL AND embed = 0\n    UNION ALL\n    SELECT src, dst FROM links l WHERE dst IS NOT NULL AND embed = 1\n      AND NOT EXISTS (SELECT 1 FROM links l0 WHERE l0.src = l.src AND l0.target = l.target AND l0.embed = 0)`;\n  return (db.prepare(sql).all() as Array<{ src: string; dst: string }>).map((r) => [r.src, r.dst]);\n}\n\n// remove() has already deleted the rows by the time afterReconcile runs, so it records here\n// whether they carried edges. Keyed on the delta object, so the state dies with the reconcile.\nconst removedWithLinks = new WeakMap<ReconcileDelta, Set<string>>();\n\nfunction recordRemoved(delta: ReconcileDelta, path: string): void {\n  let set = removedWithLinks.get(delta);\n  if (!set) {\n    set = new Set();\n    removedWithLinks.set(delta, set);\n  }\n  set.add(path);\n}\n\n// remove() runs per path; the vanished list is an array, so cache the Set per delta.\nconst vanishedSets = new WeakMap<ReconcileDelta, Set<string>>();\nfunction vanishedSet(delta: ReconcileDelta): Set<string> {\n  let set = vanishedSets.get(delta);\n  if (!set) {\n    set = new Set(delta.vanished);\n    vanishedSets.set(delta, set);\n  }\n  return set;\n}\n\nexport const links: Feature = {\n  name: 'links',\n  schema(db) {\n    db.exec('CREATE TABLE IF NOT EXISTS links (src TEXT, target TEXT, target_base TEXT, dst TEXT, embed INTEGER, PRIMARY KEY (src, target, embed))');\n    db.exec('CREATE INDEX IF NOT EXISTS links_dst ON links(dst)');\n    db.exec('CREATE INDEX IF NOT EXISTS links_target_base ON links(target_base)');\n  },\n  extract,\n  // Reparsed docs are diffed in store() below instead of wiped here: deleting and\n  // re-inserting resets dst to NULL, which made every touch-only reparse read as an edge\n  // change and recompute PageRank -- measured as most of the remaining update cost.\n  remove(db, path, delta) {\n    if (!vanishedSet(delta).has(path)) return;\n    const result = db.prepare('DELETE FROM links WHERE src = ?').run(path);\n    if (Number(result.changes) > 0) recordRemoved(delta, path);\n  },\n  store(db, path, extracted, delta) {\n    const targets = extracted as Array<{ target: string; embed: boolean }>;\n    // Stale rows (target/embed pairs the new parse no longer contains) are real edge removals.\n    // \\0 can't appear in a target, so the composite key never collides.\n    let stale: ReturnType<ReturnType<DatabaseSync['prepare']>['run']>;\n    if (targets.length === 0) {\n      stale = db.prepare('DELETE FROM links WHERE src = ?').run(path);\n    } else {\n      const placeholders = targets.map(() => '?').join(', ');\n      const keys = targets.map((t) => `${t.target}\\0${t.embed ? '1' : '0'}`);\n      stale = db.prepare(`DELETE FROM links WHERE src = ? AND (target || char(0) || embed) NOT IN (${placeholders})`).run(path, ...keys);\n    }\n    if (Number(stale.changes) > 0) recordRemoved(delta, path);\n    // Upsert preserves dst on surviving rows, so an unchanged link resolves to the same\n    // value and reports no change; only genuinely new rows start at NULL.\n    const insert = db.prepare('INSERT INTO links (src, target, target_base, dst, embed) VALUES (?, ?, ?, NULL, ?) ON CONFLICT(src, target, embed) DO UPDATE SET target_base = excluded.target_base');\n    for (const { target, embed } of targets) insert.run(path, target, baseKey(cleanTarget(target)), embed ? 1 : 0);\n  },\n  afterReconcile(db, delta) {\n    // Any deleted rows that carried edges mean the edge set shrank -- whether the file\n    // vanished or was reparsed down to fewer/no links. dstChanged alone misses the\n    // reparsed-to-zero-links case (no surviving rows to re-resolve).\n    const removeHadLinks = (removedWithLinks.get(delta)?.size ?? 0) > 0;\n\n    const churn = delta.reparsed.length + delta.vanished.length;\n    // Past this share of the tree, a full pass is the only one guaranteed to match\n    // resolveTarget's ambiguity rules. A cold build clears the threshold on its own.\n    const large = delta.files.length === 0 || churn > 0.2 * delta.files.length;\n\n    const dstChanged = large ? resolveAll(db, delta.files) : resolveIncremental(db, delta);\n    delta.linksChanged = dstChanged || removeHadLinks;\n  },\n};\n"],"names":["posix","extract","_raw","body","seen","Set","results","add","target","embed","key","has","push","m","matchAll","split","trim","index","test","cleanTarget","replace","baseKey","path","basename","toLowerCase","resolveTarget","src","pathSet","byBase","clean","fromSrc","normalize","join","dirname","candidate","get","buildByBase","files","Map","map","f","relPath","sort","list","set","resolveRows","db","rows","update","prepare","changed","row","dst","run","resolveAll","all","resolveIncremental","delta","changedBasenames","p","added","vanished","candidates","collect","collectIn","column","keys","i","length","chunk","slice","placeholders","reparsed","values","linkEdges","sql","r","removedWithLinks","WeakMap","recordRemoved","vanishedSets","vanishedSet","links","name","schema","exec","remove","result","Number","changes","store","extracted","targets","stale","t","insert","afterReconcile","removeHadLinks","size","churn","large","dstChanged","linksChanged"],"mappings":"AAAA,OAAOA,WAAW,kBAAkB;AAKpC,0FAA0F;AAC1F,iFAAiF;AACjF,+FAA+F;AAC/F,uFAAuF;AAEvF,qGAAqG;AACrG,SAASC,QAAQC,IAAY,EAAEC,IAAY;IACzC,MAAMC,OAAO,IAAIC;IACjB,MAAMC,UAAqD,EAAE;IAC7D,MAAMC,MAAM,CAACC,QAAgBC;QAC3B,MAAMC,MAAM,GAAGF,OAAO,EAAE,EAAEC,QAAQ,MAAM,KAAK;QAC7C,IAAIL,KAAKO,GAAG,CAACD,MAAM;QACnBN,KAAKG,GAAG,CAACG;QACTJ,QAAQM,IAAI,CAAC;YAAEJ;YAAQC;QAAM;IAC/B;IACA,uFAAuF;IACvF,yFAAyF;IACzF,KAAK,MAAMI,KAAKV,KAAKW,QAAQ,CAAC,kBAAmB;YAEjBD;QAD9B,MAAML,SAASK,CAAC,CAAC,EAAE,CAACE,KAAK,CAAC,IAAI,CAAC,EAAE,CAACA,KAAK,CAAC,IAAI,CAAC,EAAE,CAACC,IAAI;QACpD,IAAIR,QAAQD,IAAIC,QAAQL,IAAI,CAAC,EAACU,WAAAA,EAAEI,KAAK,cAAPJ,sBAAAA,WAAW,KAAK,EAAE,KAAK;IACvD;IACA,KAAK,MAAMA,KAAKV,KAAKW,QAAQ,CAAC,6CAA8C;QAC1E,MAAMN,SAASK,CAAC,CAAC,EAAE,CAACG,IAAI;QACxB,IAAIR,UAAU,CAAC,gBAAgBU,IAAI,CAACV,SAASD,IAAIC,QAAQK,CAAC,CAAC,EAAE,KAAK;IACpE;IACA,OAAOP;AACT;AAEA,SAASa,YAAYX,MAAc;IACjC,OAAOA,OAAOY,OAAO,CAAC,OAAO,KAAKA,OAAO,CAAC,SAAS;AACrD;AAEA,SAASC,QAAQC,IAAY;IAC3B,OAAOtB,MAAMuB,QAAQ,CAACD,MAAMF,OAAO,CAAC,UAAU,IAAII,WAAW;AAC/D;AAEA,uFAAuF;AACvF,2EAA2E;AAC3E,SAASC,cAAcC,GAAW,EAAElB,MAAc,EAAEmB,OAAoB,EAAEC,MAA6B;;QAM9FA;IALP,MAAMC,QAAQV,YAAYX;IAC1B,MAAMsB,UAAU9B,MAAM+B,SAAS,CAAC/B,MAAMgC,IAAI,CAAChC,MAAMiC,OAAO,CAACP,MAAMG;IAC/D,KAAK,MAAMK,aAAa;QAACL;QAAO,GAAGA,MAAM,GAAG,CAAC;QAAEC;QAAS,GAAGA,QAAQ,GAAG,CAAC;KAAC,CAAE;QACxE,IAAIH,QAAQhB,GAAG,CAACuB,YAAY,OAAOA;IACrC;IACA,gBAAON,cAAAA,OAAOO,GAAG,CAACd,QAAQQ,qBAAnBD,kCAAAA,WAA4B,CAAC,EAAE,uCAAI;AAC5C;AAEA,SAASQ,YAAYC,KAAiB;IACpC,MAAMT,SAAS,IAAIU;IACnB,KAAK,MAAMhB,QAAQe,MAAME,GAAG,CAAC,CAACC,IAAMA,EAAEC,OAAO,EAAEC,IAAI,GAAI;QACrD,MAAMhC,MAAMW,QAAQC;QACpB,MAAMqB,OAAOf,OAAOO,GAAG,CAACzB;QACxB,IAAIiC,MAAMA,KAAK/B,IAAI,CAACU;aACfM,OAAOgB,GAAG,CAAClC,KAAK;YAACY;SAAK;IAC7B;IACA,OAAOM;AACT;AASA,4FAA4F;AAC5F,0FAA0F;AAC1F,0BAA0B;AAC1B,SAASiB,YAAYC,EAAgB,EAAEC,IAAe,EAAEpB,OAAoB,EAAEC,MAA6B;IACzG,MAAMoB,SAASF,GAAGG,OAAO,CAAC;IAC1B,IAAIC,UAAU;IACd,KAAK,MAAMC,OAAOJ,KAAM;QACtB,MAAMK,MAAM3B,cAAc0B,IAAIzB,GAAG,EAAEyB,IAAI3C,MAAM,EAAEmB,SAASC;QACxD,IAAIwB,QAAQD,IAAIC,GAAG,EAAE;YACnBJ,OAAOK,GAAG,CAACD,KAAKD,IAAIzB,GAAG,EAAEyB,IAAI3C,MAAM;YACnC0C,UAAU;QACZ;IACF;IACA,OAAOA;AACT;AAEA,yFAAyF;AACzF,+EAA+E;AAC/E,SAASI,WAAWR,EAAgB,EAAET,KAAiB;IACrD,MAAMV,UAAU,IAAItB,IAAIgC,MAAME,GAAG,CAAC,CAACC,IAAMA,EAAEC,OAAO;IAClD,MAAMb,SAASQ,YAAYC;IAC3B,MAAMU,OAAOD,GAAGG,OAAO,CAAC,6CAA6CM,GAAG;IACxE,OAAOV,YAAYC,IAAIC,MAAMpB,SAASC;AACxC;AAEA,yFAAyF;AACzF,yFAAyF;AACzF,kEAAkE;AAClE,SAAS4B,mBAAmBV,EAAgB,EAAEW,KAAqB;IACjE,MAAM9B,UAAU,IAAItB,IAAIoD,MAAMpB,KAAK,CAACE,GAAG,CAAC,CAACC,IAAMA,EAAEC,OAAO;IACxD,MAAMb,SAASQ,YAAYqB,MAAMpB,KAAK;IAEtC,MAAMqB,mBAAmB,IAAIrD;IAC7B,KAAK,MAAMsD,KAAKF,MAAMG,KAAK,CAAEF,iBAAiBnD,GAAG,CAACc,QAAQsC;IAC1D,KAAK,MAAMA,KAAKF,MAAMI,QAAQ,CAAEH,iBAAiBnD,GAAG,CAACc,QAAQsC;IAE7D,+EAA+E;IAC/E,MAAMG,aAAa,IAAIxB;IACvB,MAAMyB,UAAU,CAAChB;QACf,KAAK,MAAMI,OAAOJ,KAAMe,WAAWlB,GAAG,CAAC,GAAGO,IAAIzB,GAAG,CAAC,EAAE,EAAEyB,IAAI3C,MAAM,CAAC,EAAE,EAAE2C,IAAI1C,KAAK,EAAE,EAAE0C;IACpF;IACA,qFAAqF;IACrF,6CAA6C;IAC7C,MAAMa,YAAY,CAACC,QAAgBC;QACjC,IAAK,IAAIC,IAAI,GAAGA,IAAID,KAAKE,MAAM,EAAED,KAAK,IAAK;YACzC,MAAME,QAAQH,KAAKI,KAAK,CAACH,GAAGA,IAAI;YAChC,MAAMI,eAAeF,MAAM9B,GAAG,CAAC,IAAM,KAAKP,IAAI,CAAC;YAC/C+B,QAAQjB,GAAGG,OAAO,CAAC,CAAC,gDAAgD,EAAEgB,OAAO,KAAK,EAAEM,aAAa,CAAC,CAAC,EAAEhB,GAAG,IAAIc;QAC9G;IACF;IAEAL,UAAU,OAAOP,MAAMe,QAAQ;IAC/BR,UAAU,eAAe;WAAIN;KAAiB;IAC9CM,UAAU,OAAOP,MAAMI,QAAQ;IAE/B,OAAOhB,YAAYC,IAAI;WAAIgB,WAAWW,MAAM;KAAG,EAAE9C,SAASC;AAC5D;AAEA,6DAA6D;AAC7D,OAAO,SAAS8C,UAAU5B,EAAgB;IACxC,0FAA0F;IAC1F,yFAAyF;IACzF,yFAAyF;IACzF,0FAA0F;IAC1F,iFAAiF;IACjF,MAAM6B,MAAM,CAAC;;;4GAG6F,CAAC;IAC3G,OAAO,AAAC7B,GAAGG,OAAO,CAAC0B,KAAKpB,GAAG,GAA2ChB,GAAG,CAAC,CAACqC,IAAM;YAACA,EAAElD,GAAG;YAAEkD,EAAExB,GAAG;SAAC;AACjG;AAEA,4FAA4F;AAC5F,+FAA+F;AAC/F,MAAMyB,mBAAmB,IAAIC;AAE7B,SAASC,cAActB,KAAqB,EAAEnC,IAAY;IACxD,IAAIsB,MAAMiC,iBAAiB1C,GAAG,CAACsB;IAC/B,IAAI,CAACb,KAAK;QACRA,MAAM,IAAIvC;QACVwE,iBAAiBjC,GAAG,CAACa,OAAOb;IAC9B;IACAA,IAAIrC,GAAG,CAACe;AACV;AAEA,qFAAqF;AACrF,MAAM0D,eAAe,IAAIF;AACzB,SAASG,YAAYxB,KAAqB;IACxC,IAAIb,MAAMoC,aAAa7C,GAAG,CAACsB;IAC3B,IAAI,CAACb,KAAK;QACRA,MAAM,IAAIvC,IAAIoD,MAAMI,QAAQ;QAC5BmB,aAAapC,GAAG,CAACa,OAAOb;IAC1B;IACA,OAAOA;AACT;AAEA,OAAO,MAAMsC,QAAiB;IAC5BC,MAAM;IACNC,QAAOtC,EAAE;QACPA,GAAGuC,IAAI,CAAC;QACRvC,GAAGuC,IAAI,CAAC;QACRvC,GAAGuC,IAAI,CAAC;IACV;IACApF;IACA,gFAAgF;IAChF,uFAAuF;IACvF,kFAAkF;IAClFqF,QAAOxC,EAAE,EAAExB,IAAI,EAAEmC,KAAK;QACpB,IAAI,CAACwB,YAAYxB,OAAO9C,GAAG,CAACW,OAAO;QACnC,MAAMiE,SAASzC,GAAGG,OAAO,CAAC,mCAAmCI,GAAG,CAAC/B;QACjE,IAAIkE,OAAOD,OAAOE,OAAO,IAAI,GAAGV,cAActB,OAAOnC;IACvD;IACAoE,OAAM5C,EAAE,EAAExB,IAAI,EAAEqE,SAAS,EAAElC,KAAK;QAC9B,MAAMmC,UAAUD;QAChB,2FAA2F;QAC3F,oEAAoE;QACpE,IAAIE;QACJ,IAAID,QAAQxB,MAAM,KAAK,GAAG;YACxByB,QAAQ/C,GAAGG,OAAO,CAAC,mCAAmCI,GAAG,CAAC/B;QAC5D,OAAO;YACL,MAAMiD,eAAeqB,QAAQrD,GAAG,CAAC,IAAM,KAAKP,IAAI,CAAC;YACjD,MAAMkC,OAAO0B,QAAQrD,GAAG,CAAC,CAACuD,IAAM,GAAGA,EAAEtF,MAAM,CAAC,EAAE,EAAEsF,EAAErF,KAAK,GAAG,MAAM,KAAK;YACrEoF,QAAQ/C,GAAGG,OAAO,CAAC,CAAC,yEAAyE,EAAEsB,aAAa,CAAC,CAAC,EAAElB,GAAG,CAAC/B,SAAS4C;QAC/H;QACA,IAAIsB,OAAOK,MAAMJ,OAAO,IAAI,GAAGV,cAActB,OAAOnC;QACpD,oFAAoF;QACpF,sEAAsE;QACtE,MAAMyE,SAASjD,GAAGG,OAAO,CAAC;QAC1B,KAAK,MAAM,EAAEzC,MAAM,EAAEC,KAAK,EAAE,IAAImF,QAASG,OAAO1C,GAAG,CAAC/B,MAAMd,QAAQa,QAAQF,YAAYX,UAAUC,QAAQ,IAAI;IAC9G;IACAuF,gBAAelD,EAAE,EAAEW,KAAK;;YAIEoB;QAHxB,mFAAmF;QACnF,+EAA+E;QAC/E,iEAAiE;QACjE,MAAMoB,iBAAiB,UAACpB,wBAAAA,iBAAiB1C,GAAG,CAACsB,oBAArBoB,4CAAAA,sBAA6BqB,IAAI,uCAAI,KAAK;QAElE,MAAMC,QAAQ1C,MAAMe,QAAQ,CAACJ,MAAM,GAAGX,MAAMI,QAAQ,CAACO,MAAM;QAC3D,+EAA+E;QAC/E,iFAAiF;QACjF,MAAMgC,QAAQ3C,MAAMpB,KAAK,CAAC+B,MAAM,KAAK,KAAK+B,QAAQ,MAAM1C,MAAMpB,KAAK,CAAC+B,MAAM;QAE1E,MAAMiC,aAAaD,QAAQ9C,WAAWR,IAAIW,MAAMpB,KAAK,IAAImB,mBAAmBV,IAAIW;QAChFA,MAAM6C,YAAY,GAAGD,cAAcJ;IACrC;AACF,EAAE"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/features/links.ts"],"sourcesContent":["import posix from 'node:path/posix';\nimport type { DatabaseSync } from 'node:sqlite';\nimport { maskRegions } from '../fences.ts';\nimport type { FileStat } from '../scan.ts';\nimport type { Feature, ReconcileDelta } from './types.ts';\n\n// links(src, target, target_base, dst, embed): target as written, target_base its baseKey\n// (indexed, for the incremental resolve below), dst the resolved path or NULL (a\n// queryable dead link), embed whether it's `![[x]]`/`![](x.md)` rather than `[[x]]`/`[](x.md)`\n// -- Obsidian's grain: a target both linked and embedded in the same note is two rows.\n\n// Wikilinks ([[target]], [[target#anchor|alias]], embeds), internal markdown links, and\n// frontmatter values that are exactly a wikilink (\"[[X]]\"; mid-string and ![[...]] forms are\n// not links there -- Obsidian's rule, probe-verified).\nfunction extract(_raw: string, rawBody: string, _search?: { title: string; summary: string }, data?: Record<string, unknown>): Array<{ target: string; embed: boolean }> {\n  const body = /\\[\\[|\\]\\(/.test(rawBody) ? maskRegions(rawBody) : rawBody;\n  const seen = new Set<string>();\n  const results: Array<{ target: string; embed: boolean }> = [];\n  const add = (target: string, embed: boolean) => {\n    const key = `${target}\\0${embed ? '1' : '0'}`;\n    if (seen.has(key)) return;\n    seen.add(key);\n    results.push({ target, embed });\n  };\n  // To the first ]], as Obsidian parses it, so a heading or alias holding a lone ] still\n  // matches; anchor and alias split off in code, since a class-based regex stops at the ].\n  for (const m of body.matchAll(/\\[\\[(.*?)\\]\\]/g)) {\n    const embed = body[(m.index ?? 0) - 1] === '!';\n    // [[#Heading]]: a same-note anchor link, which Obsidian resolves to a self-edge. Keep the\n    // target as written (leading #) so resolveTarget can special-case it.\n    const beforeAlias = m[1].split('|')[0].trim();\n    if (beforeAlias.startsWith('#')) {\n      if (beforeAlias.length > 1) add(beforeAlias, embed);\n      continue;\n    }\n    const target = beforeAlias.split('#')[0].trim();\n    if (target) add(target, embed);\n  }\n  // Any internal destination, not only .md-suffixed: Obsidian gives markdown links full\n  // linkpath resolution, so \\](Zektor) resolves like [[Zektor]] and \\](#anchor) is a\n  // self-edge. External URLs (a scheme anywhere survives the malformed double-paren case)\n  // and titled links (dest stops at whitespace) are skipped.\n  // A space in the destination is only valid ahead of a quoted title; a domain-shaped first\n  // segment (www.example.com/...) is external even without a scheme.\n  for (const m of body.matchAll(/(!)?\\[[^\\]]*\\]\\(((?:[^()\\s]|\\([^()\\s]*\\))+)(?:\\s+(?:\"[^)]*\"|'[^)]*'))?\\)/g)) {\n    const dest = m[2].trim();\n    if (!dest || dest.includes('://')) continue;\n    if (/^www\\./i.test(dest)) continue;\n    const target = dest.startsWith('#') ? dest : dest.split('#')[0];\n    if (target) add(target, m[1] === '!');\n  }\n  for (const value of Object.values(data ?? {})) {\n    for (const item of Array.isArray(value) ? value : [value]) {\n      if (typeof item !== 'string') continue;\n      const m = /^\\[\\[(.*)\\]\\]$/.exec(item.trim());\n      if (!m) continue;\n      const inner = m[1].split('|')[0].trim();\n      const target = inner.split('#')[0].trim();\n      if (target) add(target, false);\n      else if (inner.startsWith('#') && inner.length > 1) add(inner, false);\n    }\n  }\n  return results;\n}\n\nfunction cleanTarget(target: string): string {\n  return target.replace(/\\\\/g, '/').replace(/^\\.\\//, '');\n}\n\nfunction baseKey(path: string): string {\n  return posix.basename(path).replace(/\\.md$/i, '').toLowerCase();\n}\n\n// Obsidian-style: exact relative path (with/without .md), path relative to the linking\n// note's directory, then basename match (shortest path wins on ties -- verified against a\n// cold-loaded Obsidian vault on 6+ real collision pairs; byBase's lists are pre-sorted that way).\nfunction resolveTarget(src: string, target: string, pathSet: Set<string>, byBase: Map<string, string[]>): string | null {\n  // [[#Heading]]: a same-note anchor, always the note itself -- never depends on other files.\n  if (target.startsWith('#')) return src;\n  const clean = cleanTarget(target);\n  const fromSrc = posix.normalize(posix.join(posix.dirname(src), clean));\n  for (const candidate of [clean, `${clean}.md`, fromSrc, `${fromSrc}.md`]) {\n    if (pathSet.has(candidate)) return candidate;\n  }\n  const candidates = byBase.get(baseKey(clean));\n  if (!candidates) return null;\n  // The linking note itself wins a basename collision (Obsidian resolves to self before the\n  // shortest-path rule -- cold-load verified on the hub corpus).\n  return candidates.includes(src) ? src : candidates[0];\n}\n\n// Shortest path wins a basename collision; equal lengths fall back to lexicographic, our own\n// deterministic tiebreak for a case Obsidian itself leaves registration-order-dependent.\nfunction buildByBase(files: FileStat[]): Map<string, string[]> {\n  const byBase = new Map<string, string[]>();\n  const paths = files.map((f) => f.relPath).sort((a, b) => a.length - b.length || (a < b ? -1 : a > b ? 1 : 0));\n  for (const path of paths) {\n    const key = baseKey(path);\n    const list = byBase.get(key);\n    if (list) list.push(path);\n    else byBase.set(key, [path]);\n  }\n  return byBase;\n}\n\ninterface LinkRow {\n  src: string;\n  target: string;\n  dst: string | null;\n  embed: number;\n}\n\n// Applies resolveTarget to `rows`, writing only rows whose dst actually changes. The UPDATE\n// keys on (src, target): a link/embed sibling pair shares one resolution, so either row's\n// mismatch rewrites both.\nfunction resolveRows(db: DatabaseSync, rows: LinkRow[], pathSet: Set<string>, byBase: Map<string, string[]>): boolean {\n  const update = db.prepare('UPDATE links SET dst = ? WHERE src = ? AND target = ?');\n  let changed = false;\n  for (const row of rows) {\n    const dst = resolveTarget(row.src, row.target, pathSet, byBase);\n    if (dst !== row.dst) {\n      update.run(dst, row.src, row.target);\n      changed = true;\n    }\n  }\n  return changed;\n}\n\n// A new or deleted file can change any note's resolution, so re-resolve the whole table.\n// Fallback for cold builds and large deltas -- see afterReconcile's threshold.\nfunction resolveAll(db: DatabaseSync, files: FileStat[]): boolean {\n  const pathSet = new Set(files.map((f) => f.relPath));\n  const byBase = buildByBase(files);\n  const rows = db.prepare('SELECT src, target, dst, embed FROM links').all() as unknown as LinkRow[];\n  return resolveRows(db, rows, pathSet, byBase);\n}\n\n// Re-resolve only what this reconcile could have touched: re-stored sources, links whose\n// target basename matches an added/vanished file, and links whose dst just vanished. All\n// three are indexed lookups, so this never reads the whole table.\nfunction resolveIncremental(db: DatabaseSync, delta: ReconcileDelta): boolean {\n  const pathSet = new Set(delta.files.map((f) => f.relPath));\n  const byBase = buildByBase(delta.files);\n\n  const changedBasenames = new Set<string>();\n  for (const p of delta.added) changedBasenames.add(baseKey(p));\n  for (const p of delta.vanished) changedBasenames.add(baseKey(p));\n\n  // \\0 can't appear in a path or target, so the key never collides; a space can.\n  const candidates = new Map<string, LinkRow>();\n  const collect = (rows: LinkRow[]) => {\n    for (const row of rows) candidates.set(`${row.src}\\0${row.target}\\0${row.embed}`, row);\n  };\n  // Chunked so a large-but-under-threshold delta never exceeds SQLite's bound-variable\n  // limit (SQLITE_MAX_VARIABLE_NUMBER, 32766).\n  const collectIn = (column: string, keys: string[]) => {\n    for (let i = 0; i < keys.length; i += 500) {\n      const chunk = keys.slice(i, i + 500);\n      const placeholders = chunk.map(() => '?').join(', ');\n      collect(db.prepare(`SELECT src, target, dst, embed FROM links WHERE ${column} IN (${placeholders})`).all(...chunk) as unknown as LinkRow[]);\n    }\n  };\n\n  collectIn('src', delta.reparsed);\n  collectIn('target_base', [...changedBasenames]);\n  collectIn('dst', delta.vanished);\n\n  return resolveRows(db, [...candidates.values()], pathSet, byBase);\n}\n\n// Resolved edges, for rank and for search's graph expansion.\nexport function linkEdges(db: DatabaseSync): [string, string][] {\n  // One edge per row, as 0.12's grain always had it -- two written targets resolving to one\n  // dst stay two edges (a weight the fusion evals were gated on). The only exclusion is an\n  // embed whose exact (src, target) also exists as a link: that second row is new with the\n  // embed grain and would double an edge that used to be one row. The NOT EXISTS probes the\n  // primary key and fires only for embed rows; both branches still scan the table.\n  // Self-edges (same-note anchors) are excluded here so PageRank mass is not self-recycled --\n  // Obsidian's own graph view hides self-loops too. The rows themselves stay in the table for\n  // backlinks/peek.\n  const sql = `SELECT src, dst FROM links WHERE dst IS NOT NULL AND embed = 0 AND src != dst\n    UNION ALL\n    SELECT src, dst FROM links l WHERE dst IS NOT NULL AND embed = 1 AND src != dst\n      AND NOT EXISTS (SELECT 1 FROM links l0 WHERE l0.src = l.src AND l0.target = l.target AND l0.embed = 0)`;\n  return (db.prepare(sql).all() as Array<{ src: string; dst: string }>).map((r) => [r.src, r.dst]);\n}\n\n// remove() has already deleted the rows by the time afterReconcile runs, so it records here\n// whether they carried edges. Keyed on the delta object, so the state dies with the reconcile.\nconst removedWithLinks = new WeakMap<ReconcileDelta, Set<string>>();\n\nfunction recordRemoved(delta: ReconcileDelta, path: string): void {\n  let set = removedWithLinks.get(delta);\n  if (!set) {\n    set = new Set();\n    removedWithLinks.set(delta, set);\n  }\n  set.add(path);\n}\n\n// remove() runs per path; the vanished list is an array, so cache the Set per delta.\nconst vanishedSets = new WeakMap<ReconcileDelta, Set<string>>();\nfunction vanishedSet(delta: ReconcileDelta): Set<string> {\n  let set = vanishedSets.get(delta);\n  if (!set) {\n    set = new Set(delta.vanished);\n    vanishedSets.set(delta, set);\n  }\n  return set;\n}\n\nexport const links: Feature = {\n  name: 'links',\n  schema(db) {\n    db.exec('CREATE TABLE IF NOT EXISTS links (src TEXT, target TEXT, target_base TEXT, dst TEXT, embed INTEGER, PRIMARY KEY (src, target, embed))');\n    db.exec('CREATE INDEX IF NOT EXISTS links_dst ON links(dst)');\n    db.exec('CREATE INDEX IF NOT EXISTS links_target_base ON links(target_base)');\n  },\n  extract,\n  // Reparsed docs are diffed in store() below instead of wiped here: deleting and\n  // re-inserting resets dst to NULL, which made every touch-only reparse read as an edge\n  // change and recompute PageRank -- measured as most of the remaining update cost.\n  remove(db, path, delta) {\n    if (!vanishedSet(delta).has(path)) return;\n    const result = db.prepare('DELETE FROM links WHERE src = ?').run(path);\n    if (Number(result.changes) > 0) recordRemoved(delta, path);\n  },\n  store(db, path, extracted, delta) {\n    const targets = extracted as Array<{ target: string; embed: boolean }>;\n    // Stale rows (target/embed pairs the new parse no longer contains) are real edge removals.\n    // \\0 can't appear in a target, so the composite key never collides.\n    let stale: ReturnType<ReturnType<DatabaseSync['prepare']>['run']>;\n    if (targets.length === 0) {\n      stale = db.prepare('DELETE FROM links WHERE src = ?').run(path);\n    } else {\n      const placeholders = targets.map(() => '?').join(', ');\n      const keys = targets.map((t) => `${t.target}\\0${t.embed ? '1' : '0'}`);\n      stale = db.prepare(`DELETE FROM links WHERE src = ? AND (target || char(0) || embed) NOT IN (${placeholders})`).run(path, ...keys);\n    }\n    if (Number(stale.changes) > 0) recordRemoved(delta, path);\n    // Upsert preserves dst on surviving rows, so an unchanged link resolves to the same\n    // value and reports no change; only genuinely new rows start at NULL.\n    const insert = db.prepare('INSERT INTO links (src, target, target_base, dst, embed) VALUES (?, ?, ?, NULL, ?) ON CONFLICT(src, target, embed) DO UPDATE SET target_base = excluded.target_base');\n    // A same-note anchor's dst is always its own src, never another file's basename, so it\n    // gets no target_base -- SQL's `IN` never matches NULL, keeping it out of\n    // resolveIncremental's cross-file collectIn('target_base', ...) entirely.\n    for (const { target, embed } of targets) insert.run(path, target, target.startsWith('#') ? null : baseKey(cleanTarget(target)), embed ? 1 : 0);\n  },\n  afterReconcile(db, delta) {\n    // Any deleted rows that carried edges mean the edge set shrank -- whether the file\n    // vanished or was reparsed down to fewer/no links. dstChanged alone misses the\n    // reparsed-to-zero-links case (no surviving rows to re-resolve).\n    const removeHadLinks = (removedWithLinks.get(delta)?.size ?? 0) > 0;\n\n    const churn = delta.reparsed.length + delta.vanished.length;\n    // Past this share of the tree, a full pass is the only one guaranteed to match\n    // resolveTarget's ambiguity rules. A cold build clears the threshold on its own.\n    const large = delta.files.length === 0 || churn > 0.2 * delta.files.length;\n\n    const dstChanged = large ? resolveAll(db, delta.files) : resolveIncremental(db, delta);\n    delta.linksChanged = dstChanged || removeHadLinks;\n  },\n};\n"],"names":["posix","maskRegions","extract","_raw","rawBody","_search","data","body","test","seen","Set","results","add","target","embed","key","has","push","m","matchAll","index","beforeAlias","split","trim","startsWith","length","dest","includes","value","Object","values","item","Array","isArray","exec","inner","cleanTarget","replace","baseKey","path","basename","toLowerCase","resolveTarget","src","pathSet","byBase","clean","fromSrc","normalize","join","dirname","candidate","candidates","get","buildByBase","files","Map","paths","map","f","relPath","sort","a","b","list","set","resolveRows","db","rows","update","prepare","changed","row","dst","run","resolveAll","all","resolveIncremental","delta","changedBasenames","p","added","vanished","collect","collectIn","column","keys","i","chunk","slice","placeholders","reparsed","linkEdges","sql","r","removedWithLinks","WeakMap","recordRemoved","vanishedSets","vanishedSet","links","name","schema","remove","result","Number","changes","store","extracted","targets","stale","t","insert","afterReconcile","removeHadLinks","size","churn","large","dstChanged","linksChanged"],"mappings":"AAAA,OAAOA,WAAW,kBAAkB;AAEpC,SAASC,WAAW,QAAQ,eAAe;AAI3C,0FAA0F;AAC1F,iFAAiF;AACjF,+FAA+F;AAC/F,uFAAuF;AAEvF,wFAAwF;AACxF,6FAA6F;AAC7F,uDAAuD;AACvD,SAASC,QAAQC,IAAY,EAAEC,OAAe,EAAEC,OAA4C,EAAEC,IAA8B;IAC1H,MAAMC,OAAO,YAAYC,IAAI,CAACJ,WAAWH,YAAYG,WAAWA;IAChE,MAAMK,OAAO,IAAIC;IACjB,MAAMC,UAAqD,EAAE;IAC7D,MAAMC,MAAM,CAACC,QAAgBC;QAC3B,MAAMC,MAAM,GAAGF,OAAO,EAAE,EAAEC,QAAQ,MAAM,KAAK;QAC7C,IAAIL,KAAKO,GAAG,CAACD,MAAM;QACnBN,KAAKG,GAAG,CAACG;QACTJ,QAAQM,IAAI,CAAC;YAAEJ;YAAQC;QAAM;IAC/B;IACA,uFAAuF;IACvF,yFAAyF;IACzF,KAAK,MAAMI,KAAKX,KAAKY,QAAQ,CAAC,kBAAmB;YAC3BD;QAApB,MAAMJ,QAAQP,IAAI,CAAC,EAACW,WAAAA,EAAEE,KAAK,cAAPF,sBAAAA,WAAW,KAAK,EAAE,KAAK;QAC3C,0FAA0F;QAC1F,sEAAsE;QACtE,MAAMG,cAAcH,CAAC,CAAC,EAAE,CAACI,KAAK,CAAC,IAAI,CAAC,EAAE,CAACC,IAAI;QAC3C,IAAIF,YAAYG,UAAU,CAAC,MAAM;YAC/B,IAAIH,YAAYI,MAAM,GAAG,GAAGb,IAAIS,aAAaP;YAC7C;QACF;QACA,MAAMD,SAASQ,YAAYC,KAAK,CAAC,IAAI,CAAC,EAAE,CAACC,IAAI;QAC7C,IAAIV,QAAQD,IAAIC,QAAQC;IAC1B;IACA,sFAAsF;IACtF,mFAAmF;IACnF,wFAAwF;IACxF,2DAA2D;IAC3D,0FAA0F;IAC1F,mEAAmE;IACnE,KAAK,MAAMI,KAAKX,KAAKY,QAAQ,CAAC,6EAA8E;QAC1G,MAAMO,OAAOR,CAAC,CAAC,EAAE,CAACK,IAAI;QACtB,IAAI,CAACG,QAAQA,KAAKC,QAAQ,CAAC,QAAQ;QACnC,IAAI,UAAUnB,IAAI,CAACkB,OAAO;QAC1B,MAAMb,SAASa,KAAKF,UAAU,CAAC,OAAOE,OAAOA,KAAKJ,KAAK,CAAC,IAAI,CAAC,EAAE;QAC/D,IAAIT,QAAQD,IAAIC,QAAQK,CAAC,CAAC,EAAE,KAAK;IACnC;IACA,KAAK,MAAMU,SAASC,OAAOC,MAAM,CAACxB,iBAAAA,kBAAAA,OAAQ,CAAC,GAAI;QAC7C,KAAK,MAAMyB,QAAQC,MAAMC,OAAO,CAACL,SAASA,QAAQ;YAACA;SAAM,CAAE;YACzD,IAAI,OAAOG,SAAS,UAAU;YAC9B,MAAMb,IAAI,iBAAiBgB,IAAI,CAACH,KAAKR,IAAI;YACzC,IAAI,CAACL,GAAG;YACR,MAAMiB,QAAQjB,CAAC,CAAC,EAAE,CAACI,KAAK,CAAC,IAAI,CAAC,EAAE,CAACC,IAAI;YACrC,MAAMV,SAASsB,MAAMb,KAAK,CAAC,IAAI,CAAC,EAAE,CAACC,IAAI;YACvC,IAAIV,QAAQD,IAAIC,QAAQ;iBACnB,IAAIsB,MAAMX,UAAU,CAAC,QAAQW,MAAMV,MAAM,GAAG,GAAGb,IAAIuB,OAAO;QACjE;IACF;IACA,OAAOxB;AACT;AAEA,SAASyB,YAAYvB,MAAc;IACjC,OAAOA,OAAOwB,OAAO,CAAC,OAAO,KAAKA,OAAO,CAAC,SAAS;AACrD;AAEA,SAASC,QAAQC,IAAY;IAC3B,OAAOvC,MAAMwC,QAAQ,CAACD,MAAMF,OAAO,CAAC,UAAU,IAAII,WAAW;AAC/D;AAEA,uFAAuF;AACvF,0FAA0F;AAC1F,kGAAkG;AAClG,SAASC,cAAcC,GAAW,EAAE9B,MAAc,EAAE+B,OAAoB,EAAEC,MAA6B;IACrG,4FAA4F;IAC5F,IAAIhC,OAAOW,UAAU,CAAC,MAAM,OAAOmB;IACnC,MAAMG,QAAQV,YAAYvB;IAC1B,MAAMkC,UAAU/C,MAAMgD,SAAS,CAAChD,MAAMiD,IAAI,CAACjD,MAAMkD,OAAO,CAACP,MAAMG;IAC/D,KAAK,MAAMK,aAAa;QAACL;QAAO,GAAGA,MAAM,GAAG,CAAC;QAAEC;QAAS,GAAGA,QAAQ,GAAG,CAAC;KAAC,CAAE;QACxE,IAAIH,QAAQ5B,GAAG,CAACmC,YAAY,OAAOA;IACrC;IACA,MAAMC,aAAaP,OAAOQ,GAAG,CAACf,QAAQQ;IACtC,IAAI,CAACM,YAAY,OAAO;IACxB,0FAA0F;IAC1F,+DAA+D;IAC/D,OAAOA,WAAWzB,QAAQ,CAACgB,OAAOA,MAAMS,UAAU,CAAC,EAAE;AACvD;AAEA,6FAA6F;AAC7F,yFAAyF;AACzF,SAASE,YAAYC,KAAiB;IACpC,MAAMV,SAAS,IAAIW;IACnB,MAAMC,QAAQF,MAAMG,GAAG,CAAC,CAACC,IAAMA,EAAEC,OAAO,EAAEC,IAAI,CAAC,CAACC,GAAGC,IAAMD,EAAErC,MAAM,GAAGsC,EAAEtC,MAAM,IAAKqC,CAAAA,IAAIC,IAAI,CAAC,IAAID,IAAIC,IAAI,IAAI,CAAA;IAC1G,KAAK,MAAMxB,QAAQkB,MAAO;QACxB,MAAM1C,MAAMuB,QAAQC;QACpB,MAAMyB,OAAOnB,OAAOQ,GAAG,CAACtC;QACxB,IAAIiD,MAAMA,KAAK/C,IAAI,CAACsB;aACfM,OAAOoB,GAAG,CAAClD,KAAK;YAACwB;SAAK;IAC7B;IACA,OAAOM;AACT;AASA,4FAA4F;AAC5F,0FAA0F;AAC1F,0BAA0B;AAC1B,SAASqB,YAAYC,EAAgB,EAAEC,IAAe,EAAExB,OAAoB,EAAEC,MAA6B;IACzG,MAAMwB,SAASF,GAAGG,OAAO,CAAC;IAC1B,IAAIC,UAAU;IACd,KAAK,MAAMC,OAAOJ,KAAM;QACtB,MAAMK,MAAM/B,cAAc8B,IAAI7B,GAAG,EAAE6B,IAAI3D,MAAM,EAAE+B,SAASC;QACxD,IAAI4B,QAAQD,IAAIC,GAAG,EAAE;YACnBJ,OAAOK,GAAG,CAACD,KAAKD,IAAI7B,GAAG,EAAE6B,IAAI3D,MAAM;YACnC0D,UAAU;QACZ;IACF;IACA,OAAOA;AACT;AAEA,yFAAyF;AACzF,+EAA+E;AAC/E,SAASI,WAAWR,EAAgB,EAAEZ,KAAiB;IACrD,MAAMX,UAAU,IAAIlC,IAAI6C,MAAMG,GAAG,CAAC,CAACC,IAAMA,EAAEC,OAAO;IAClD,MAAMf,SAASS,YAAYC;IAC3B,MAAMa,OAAOD,GAAGG,OAAO,CAAC,6CAA6CM,GAAG;IACxE,OAAOV,YAAYC,IAAIC,MAAMxB,SAASC;AACxC;AAEA,yFAAyF;AACzF,yFAAyF;AACzF,kEAAkE;AAClE,SAASgC,mBAAmBV,EAAgB,EAAEW,KAAqB;IACjE,MAAMlC,UAAU,IAAIlC,IAAIoE,MAAMvB,KAAK,CAACG,GAAG,CAAC,CAACC,IAAMA,EAAEC,OAAO;IACxD,MAAMf,SAASS,YAAYwB,MAAMvB,KAAK;IAEtC,MAAMwB,mBAAmB,IAAIrE;IAC7B,KAAK,MAAMsE,KAAKF,MAAMG,KAAK,CAAEF,iBAAiBnE,GAAG,CAAC0B,QAAQ0C;IAC1D,KAAK,MAAMA,KAAKF,MAAMI,QAAQ,CAAEH,iBAAiBnE,GAAG,CAAC0B,QAAQ0C;IAE7D,+EAA+E;IAC/E,MAAM5B,aAAa,IAAII;IACvB,MAAM2B,UAAU,CAACf;QACf,KAAK,MAAMI,OAAOJ,KAAMhB,WAAWa,GAAG,CAAC,GAAGO,IAAI7B,GAAG,CAAC,EAAE,EAAE6B,IAAI3D,MAAM,CAAC,EAAE,EAAE2D,IAAI1D,KAAK,EAAE,EAAE0D;IACpF;IACA,qFAAqF;IACrF,6CAA6C;IAC7C,MAAMY,YAAY,CAACC,QAAgBC;QACjC,IAAK,IAAIC,IAAI,GAAGA,IAAID,KAAK7D,MAAM,EAAE8D,KAAK,IAAK;YACzC,MAAMC,QAAQF,KAAKG,KAAK,CAACF,GAAGA,IAAI;YAChC,MAAMG,eAAeF,MAAM9B,GAAG,CAAC,IAAM,KAAKT,IAAI,CAAC;YAC/CkC,QAAQhB,GAAGG,OAAO,CAAC,CAAC,gDAAgD,EAAEe,OAAO,KAAK,EAAEK,aAAa,CAAC,CAAC,EAAEd,GAAG,IAAIY;QAC9G;IACF;IAEAJ,UAAU,OAAON,MAAMa,QAAQ;IAC/BP,UAAU,eAAe;WAAIL;KAAiB;IAC9CK,UAAU,OAAON,MAAMI,QAAQ;IAE/B,OAAOhB,YAAYC,IAAI;WAAIf,WAAWtB,MAAM;KAAG,EAAEc,SAASC;AAC5D;AAEA,6DAA6D;AAC7D,OAAO,SAAS+C,UAAUzB,EAAgB;IACxC,0FAA0F;IAC1F,yFAAyF;IACzF,yFAAyF;IACzF,0FAA0F;IAC1F,iFAAiF;IACjF,4FAA4F;IAC5F,4FAA4F;IAC5F,kBAAkB;IAClB,MAAM0B,MAAM,CAAC;;;4GAG6F,CAAC;IAC3G,OAAO,AAAC1B,GAAGG,OAAO,CAACuB,KAAKjB,GAAG,GAA2ClB,GAAG,CAAC,CAACoC,IAAM;YAACA,EAAEnD,GAAG;YAAEmD,EAAErB,GAAG;SAAC;AACjG;AAEA,4FAA4F;AAC5F,+FAA+F;AAC/F,MAAMsB,mBAAmB,IAAIC;AAE7B,SAASC,cAAcnB,KAAqB,EAAEvC,IAAY;IACxD,IAAI0B,MAAM8B,iBAAiB1C,GAAG,CAACyB;IAC/B,IAAI,CAACb,KAAK;QACRA,MAAM,IAAIvD;QACVqF,iBAAiB9B,GAAG,CAACa,OAAOb;IAC9B;IACAA,IAAIrD,GAAG,CAAC2B;AACV;AAEA,qFAAqF;AACrF,MAAM2D,eAAe,IAAIF;AACzB,SAASG,YAAYrB,KAAqB;IACxC,IAAIb,MAAMiC,aAAa7C,GAAG,CAACyB;IAC3B,IAAI,CAACb,KAAK;QACRA,MAAM,IAAIvD,IAAIoE,MAAMI,QAAQ;QAC5BgB,aAAajC,GAAG,CAACa,OAAOb;IAC1B;IACA,OAAOA;AACT;AAEA,OAAO,MAAMmC,QAAiB;IAC5BC,MAAM;IACNC,QAAOnC,EAAE;QACPA,GAAGjC,IAAI,CAAC;QACRiC,GAAGjC,IAAI,CAAC;QACRiC,GAAGjC,IAAI,CAAC;IACV;IACAhC;IACA,gFAAgF;IAChF,uFAAuF;IACvF,kFAAkF;IAClFqG,QAAOpC,EAAE,EAAE5B,IAAI,EAAEuC,KAAK;QACpB,IAAI,CAACqB,YAAYrB,OAAO9D,GAAG,CAACuB,OAAO;QACnC,MAAMiE,SAASrC,GAAGG,OAAO,CAAC,mCAAmCI,GAAG,CAACnC;QACjE,IAAIkE,OAAOD,OAAOE,OAAO,IAAI,GAAGT,cAAcnB,OAAOvC;IACvD;IACAoE,OAAMxC,EAAE,EAAE5B,IAAI,EAAEqE,SAAS,EAAE9B,KAAK;QAC9B,MAAM+B,UAAUD;QAChB,2FAA2F;QAC3F,oEAAoE;QACpE,IAAIE;QACJ,IAAID,QAAQpF,MAAM,KAAK,GAAG;YACxBqF,QAAQ3C,GAAGG,OAAO,CAAC,mCAAmCI,GAAG,CAACnC;QAC5D,OAAO;YACL,MAAMmD,eAAemB,QAAQnD,GAAG,CAAC,IAAM,KAAKT,IAAI,CAAC;YACjD,MAAMqC,OAAOuB,QAAQnD,GAAG,CAAC,CAACqD,IAAM,GAAGA,EAAElG,MAAM,CAAC,EAAE,EAAEkG,EAAEjG,KAAK,GAAG,MAAM,KAAK;YACrEgG,QAAQ3C,GAAGG,OAAO,CAAC,CAAC,yEAAyE,EAAEoB,aAAa,CAAC,CAAC,EAAEhB,GAAG,CAACnC,SAAS+C;QAC/H;QACA,IAAImB,OAAOK,MAAMJ,OAAO,IAAI,GAAGT,cAAcnB,OAAOvC;QACpD,oFAAoF;QACpF,sEAAsE;QACtE,MAAMyE,SAAS7C,GAAGG,OAAO,CAAC;QAC1B,uFAAuF;QACvF,0EAA0E;QAC1E,0EAA0E;QAC1E,KAAK,MAAM,EAAEzD,MAAM,EAAEC,KAAK,EAAE,IAAI+F,QAASG,OAAOtC,GAAG,CAACnC,MAAM1B,QAAQA,OAAOW,UAAU,CAAC,OAAO,OAAOc,QAAQF,YAAYvB,UAAUC,QAAQ,IAAI;IAC9I;IACAmG,gBAAe9C,EAAE,EAAEW,KAAK;;YAIEiB;QAHxB,mFAAmF;QACnF,+EAA+E;QAC/E,iEAAiE;QACjE,MAAMmB,iBAAiB,UAACnB,wBAAAA,iBAAiB1C,GAAG,CAACyB,oBAArBiB,4CAAAA,sBAA6BoB,IAAI,uCAAI,KAAK;QAElE,MAAMC,QAAQtC,MAAMa,QAAQ,CAAClE,MAAM,GAAGqD,MAAMI,QAAQ,CAACzD,MAAM;QAC3D,+EAA+E;QAC/E,iFAAiF;QACjF,MAAM4F,QAAQvC,MAAMvB,KAAK,CAAC9B,MAAM,KAAK,KAAK2F,QAAQ,MAAMtC,MAAMvB,KAAK,CAAC9B,MAAM;QAE1E,MAAM6F,aAAaD,QAAQ1C,WAAWR,IAAIW,MAAMvB,KAAK,IAAIsB,mBAAmBV,IAAIW;QAChFA,MAAMyC,YAAY,GAAGD,cAAcJ;IACrC;AACF,EAAE"}

package/dist/esm/features/sections.js

@@ -1,14 +1,12 @@
+import { fenceTracker } from '../fences.js';
 // Headings outside fenced code blocks.
 function extract(raw) {
     const lines = raw.split('\n');
     const found = [];
-    let inFence = false;
+    const fence = fenceTracker();
     for(let i = 0; i < lines.length; i++){
-        if (/^(```|~~~)/.test(lines[i])) {
-            inFence = !inFence;
-            continue;
-        }
-        if (inFence) continue;
+        if (fence.feed(lines[i])) continue;
+        if (fence.inFence) continue;
         const m = lines[i].match(/^(#{1,6}) +(.*)/);
         if (m) found.push({
             level: m[1].length,

package/dist/esm/features/sections.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/features/sections.ts"],"sourcesContent":["import type { Feature } from './types.ts';\n\n// sections(path, idx, level, heading, start_line, end_line, tokens): the heading outline,\n// 1-indexed over the raw file so a row is a direct Read range; tokens is a chars/4 estimate.\n\nexport interface Section {\n  level: number;\n  heading: string;\n  startLine: number;\n  endLine: number;\n  tokens: number;\n}\n\n// Headings outside fenced code blocks.\nfunction extract(raw: string): Section[] {\n  const lines = raw.split('\\n');\n  const found: Section[] = [];\n  let inFence = false;\n  for (let i = 0; i < lines.length; i++) {\n    if (/^(```|~~~)/.test(lines[i])) {\n      inFence = !inFence;\n      continue;\n    }\n    if (inFence) continue;\n    const m = lines[i].match(/^(#{1,6}) +(.*)/);\n    if (m) found.push({ level: m[1].length, heading: m[2].trim(), startLine: i + 1, endLine: lines.length, tokens: 0 });\n  }\n  for (let s = 0; s < found.length; s++) {\n    if (s + 1 < found.length) found[s].endLine = found[s + 1].startLine - 1;\n    const chars = lines.slice(found[s].startLine - 1, found[s].endLine).join('\\n').length;\n    found[s].tokens = Math.ceil(chars / 4);\n  }\n  return found;\n}\n\nexport const sections: Feature = {\n  name: 'sections',\n  schema(db) {\n    db.exec(`CREATE TABLE IF NOT EXISTS sections (\"path\" TEXT, idx INTEGER, level INTEGER, heading TEXT, start_line INTEGER, end_line INTEGER, tokens INTEGER, PRIMARY KEY (\"path\", idx))`);\n  },\n  extract,\n  remove(db, path) {\n    db.prepare('DELETE FROM sections WHERE \"path\" = ?').run(path);\n  },\n  store(db, path, extracted) {\n    const insert = db.prepare('INSERT INTO sections (\"path\", idx, level, heading, start_line, end_line, tokens) VALUES (?, ?, ?, ?, ?, ?, ?)');\n    (extracted as Section[]).forEach((s, idx) => insert.run(path, idx, s.level, s.heading, s.startLine, s.endLine, s.tokens));\n  },\n};\n"],"names":["extract","raw","lines","split","found","inFence","i","length","test","m","match","push","level","heading","trim","startLine","endLine","tokens","s","chars","slice","join","Math","ceil","sections","name","schema","db","exec","remove","path","prepare","run","store","extracted","insert","forEach","idx"],"mappings":"AAaA,uCAAuC;AACvC,SAASA,QAAQC,GAAW;IAC1B,MAAMC,QAAQD,IAAIE,KAAK,CAAC;IACxB,MAAMC,QAAmB,EAAE;IAC3B,IAAIC,UAAU;IACd,IAAK,IAAIC,IAAI,GAAGA,IAAIJ,MAAMK,MAAM,EAAED,IAAK;QACrC,IAAI,aAAaE,IAAI,CAACN,KAAK,CAACI,EAAE,GAAG;YAC/BD,UAAU,CAACA;YACX;QACF;QACA,IAAIA,SAAS;QACb,MAAMI,IAAIP,KAAK,CAACI,EAAE,CAACI,KAAK,CAAC;QACzB,IAAID,GAAGL,MAAMO,IAAI,CAAC;YAAEC,OAAOH,CAAC,CAAC,EAAE,CAACF,MAAM;YAAEM,SAASJ,CAAC,CAAC,EAAE,CAACK,IAAI;YAAIC,WAAWT,IAAI;YAAGU,SAASd,MAAMK,MAAM;YAAEU,QAAQ;QAAE;IACnH;IACA,IAAK,IAAIC,IAAI,GAAGA,IAAId,MAAMG,MAAM,EAAEW,IAAK;QACrC,IAAIA,IAAI,IAAId,MAAMG,MAAM,EAAEH,KAAK,CAACc,EAAE,CAACF,OAAO,GAAGZ,KAAK,CAACc,IAAI,EAAE,CAACH,SAAS,GAAG;QACtE,MAAMI,QAAQjB,MAAMkB,KAAK,CAAChB,KAAK,CAACc,EAAE,CAACH,SAAS,GAAG,GAAGX,KAAK,CAACc,EAAE,CAACF,OAAO,EAAEK,IAAI,CAAC,MAAMd,MAAM;QACrFH,KAAK,CAACc,EAAE,CAACD,MAAM,GAAGK,KAAKC,IAAI,CAACJ,QAAQ;IACtC;IACA,OAAOf;AACT;AAEA,OAAO,MAAMoB,WAAoB;IAC/BC,MAAM;IACNC,QAAOC,EAAE;QACPA,GAAGC,IAAI,CAAC,CAAC,4KAA4K,CAAC;IACxL;IACA5B;IACA6B,QAAOF,EAAE,EAAEG,IAAI;QACbH,GAAGI,OAAO,CAAC,yCAAyCC,GAAG,CAACF;IAC1D;IACAG,OAAMN,EAAE,EAAEG,IAAI,EAAEI,SAAS;QACvB,MAAMC,SAASR,GAAGI,OAAO,CAAC;QACzBG,UAAwBE,OAAO,CAAC,CAAClB,GAAGmB,MAAQF,OAAOH,GAAG,CAACF,MAAMO,KAAKnB,EAAEN,KAAK,EAAEM,EAAEL,OAAO,EAAEK,EAAEH,SAAS,EAAEG,EAAEF,OAAO,EAAEE,EAAED,MAAM;IACzH;AACF,EAAE"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/features/sections.ts"],"sourcesContent":["import { fenceTracker } from '../fences.ts';\nimport type { Feature } from './types.ts';\n\n// sections(path, idx, level, heading, start_line, end_line, tokens): the heading outline,\n// 1-indexed over the raw file so a row is a direct Read range; tokens is a chars/4 estimate.\n\nexport interface Section {\n  level: number;\n  heading: string;\n  startLine: number;\n  endLine: number;\n  tokens: number;\n}\n\n// Headings outside fenced code blocks.\nfunction extract(raw: string): Section[] {\n  const lines = raw.split('\\n');\n  const found: Section[] = [];\n  const fence = fenceTracker();\n  for (let i = 0; i < lines.length; i++) {\n    if (fence.feed(lines[i])) continue;\n    if (fence.inFence) continue;\n    const m = lines[i].match(/^(#{1,6}) +(.*)/);\n    if (m) found.push({ level: m[1].length, heading: m[2].trim(), startLine: i + 1, endLine: lines.length, tokens: 0 });\n  }\n  for (let s = 0; s < found.length; s++) {\n    if (s + 1 < found.length) found[s].endLine = found[s + 1].startLine - 1;\n    const chars = lines.slice(found[s].startLine - 1, found[s].endLine).join('\\n').length;\n    found[s].tokens = Math.ceil(chars / 4);\n  }\n  return found;\n}\n\nexport const sections: Feature = {\n  name: 'sections',\n  schema(db) {\n    db.exec(`CREATE TABLE IF NOT EXISTS sections (\"path\" TEXT, idx INTEGER, level INTEGER, heading TEXT, start_line INTEGER, end_line INTEGER, tokens INTEGER, PRIMARY KEY (\"path\", idx))`);\n  },\n  extract,\n  remove(db, path) {\n    db.prepare('DELETE FROM sections WHERE \"path\" = ?').run(path);\n  },\n  store(db, path, extracted) {\n    const insert = db.prepare('INSERT INTO sections (\"path\", idx, level, heading, start_line, end_line, tokens) VALUES (?, ?, ?, ?, ?, ?, ?)');\n    (extracted as Section[]).forEach((s, idx) => insert.run(path, idx, s.level, s.heading, s.startLine, s.endLine, s.tokens));\n  },\n};\n"],"names":["fenceTracker","extract","raw","lines","split","found","fence","i","length","feed","inFence","m","match","push","level","heading","trim","startLine","endLine","tokens","s","chars","slice","join","Math","ceil","sections","name","schema","db","exec","remove","path","prepare","run","store","extracted","insert","forEach","idx"],"mappings":"AAAA,SAASA,YAAY,QAAQ,eAAe;AAc5C,uCAAuC;AACvC,SAASC,QAAQC,GAAW;IAC1B,MAAMC,QAAQD,IAAIE,KAAK,CAAC;IACxB,MAAMC,QAAmB,EAAE;IAC3B,MAAMC,QAAQN;IACd,IAAK,IAAIO,IAAI,GAAGA,IAAIJ,MAAMK,MAAM,EAAED,IAAK;QACrC,IAAID,MAAMG,IAAI,CAACN,KAAK,CAACI,EAAE,GAAG;QAC1B,IAAID,MAAMI,OAAO,EAAE;QACnB,MAAMC,IAAIR,KAAK,CAACI,EAAE,CAACK,KAAK,CAAC;QACzB,IAAID,GAAGN,MAAMQ,IAAI,CAAC;YAAEC,OAAOH,CAAC,CAAC,EAAE,CAACH,MAAM;YAAEO,SAASJ,CAAC,CAAC,EAAE,CAACK,IAAI;YAAIC,WAAWV,IAAI;YAAGW,SAASf,MAAMK,MAAM;YAAEW,QAAQ;QAAE;IACnH;IACA,IAAK,IAAIC,IAAI,GAAGA,IAAIf,MAAMG,MAAM,EAAEY,IAAK;QACrC,IAAIA,IAAI,IAAIf,MAAMG,MAAM,EAAEH,KAAK,CAACe,EAAE,CAACF,OAAO,GAAGb,KAAK,CAACe,IAAI,EAAE,CAACH,SAAS,GAAG;QACtE,MAAMI,QAAQlB,MAAMmB,KAAK,CAACjB,KAAK,CAACe,EAAE,CAACH,SAAS,GAAG,GAAGZ,KAAK,CAACe,EAAE,CAACF,OAAO,EAAEK,IAAI,CAAC,MAAMf,MAAM;QACrFH,KAAK,CAACe,EAAE,CAACD,MAAM,GAAGK,KAAKC,IAAI,CAACJ,QAAQ;IACtC;IACA,OAAOhB;AACT;AAEA,OAAO,MAAMqB,WAAoB;IAC/BC,MAAM;IACNC,QAAOC,EAAE;QACPA,GAAGC,IAAI,CAAC,CAAC,4KAA4K,CAAC;IACxL;IACA7B;IACA8B,QAAOF,EAAE,EAAEG,IAAI;QACbH,GAAGI,OAAO,CAAC,yCAAyCC,GAAG,CAACF;IAC1D;IACAG,OAAMN,EAAE,EAAEG,IAAI,EAAEI,SAAS;QACvB,MAAMC,SAASR,GAAGI,OAAO,CAAC;QACzBG,UAAwBE,OAAO,CAAC,CAAClB,GAAGmB,MAAQF,OAAOH,GAAG,CAACF,MAAMO,KAAKnB,EAAEN,KAAK,EAAEM,EAAEL,OAAO,EAAEK,EAAEH,SAAS,EAAEG,EAAEF,OAAO,EAAEE,EAAED,MAAM;IACzH;AACF,EAAE"}

package/dist/esm/features/tags.js

@@ -1,14 +1,92 @@
+import { commentTracker, fenceTracker, maskCodeSpans } from '../fences.js';
 // tags(path, tag): Obsidian's file.tags grain -- frontmatter list/string tags plus inline
 // #tags from the prose, deduplicated, source not distinguished. Nested tags store full
 // (book/scifi); `tag = 'book' OR tag LIKE 'book/%'` is how a caller matches the parent too.
-const FENCE_RE = /^(```|~~~)/;
-const INLINE_CODE_RE = /`[^`]*`/g;
 // Obsidian treats [[#Heading]] as a same-note link, not a tag.
 const WIKILINK_RE = /\[\[.*?\]\]/g; // to the first ]], so a heading holding a lone ] still masks
 // Obsidian doesn't read tags inside HTML markup.
 const HTML_TAG_RE = /<\/?[a-zA-Z][^>]*>/g; // tag-shaped only: a comparison's `< 5` must not open a span
 // Anchors on start-of-line or a preceding whitespace/(/[ so `a#b` and URL fragments don't count.
 const INLINE_TAG_RE = /(?:^|[\s([])#([\p{L}\p{N}_/-]+)/gu;
+// A markdown link destination `](...)` -- `[text](#anchor)` is a same-page link, not a tag.
+const LINK_DEST_RE = /\]\((?:[^()]|\([^()]*\))*\)/g; // one paren-nesting level, as CommonMark destinations allow: (https://x/a_(b)#frag)
+// CommonMark's HTML-block type-6 list (fixed by the spec, not a drifting enumeration): a line
+// starting with an open or close tag of one of these, at column 0, opens a block that swallows
+// following lines -- including any #tag in them -- until a blank line closes it.
+const HTML_BLOCK_TAGS = new Set([
+    'address',
+    'article',
+    'aside',
+    'base',
+    'basefont',
+    'blockquote',
+    'body',
+    'caption',
+    'center',
+    'col',
+    'colgroup',
+    'dd',
+    'details',
+    'dialog',
+    'dir',
+    'div',
+    'dl',
+    'dt',
+    'fieldset',
+    'figcaption',
+    'figure',
+    'footer',
+    'form',
+    'frame',
+    'frameset',
+    'h1',
+    'h2',
+    'h3',
+    'h4',
+    'h5',
+    'h6',
+    'head',
+    'header',
+    'hr',
+    'html',
+    'iframe',
+    'legend',
+    'li',
+    'link',
+    'main',
+    'menu',
+    'menuitem',
+    'nav',
+    'noframes',
+    'ol',
+    'optgroup',
+    'option',
+    'p',
+    'param',
+    'search',
+    'section',
+    'summary',
+    'table',
+    'tbody',
+    'td',
+    'tfoot',
+    'th',
+    'thead',
+    'title',
+    'tr',
+    'track',
+    'ul'
+]);
+// Type-1 blocks (script/pre/style/textarea): closes on the line holding the matching end tag,
+// not on a blank line, and that line is the last one skipped.
+const HTML_PRE_TAGS = new Set([
+    'script',
+    'pre',
+    'style',
+    'textarea'
+]);
+// An opening or closing tag at column 0, tag name captured for the lookups above.
+const HTML_BLOCK_OPEN_RE = /^<\/?([a-zA-Z][a-zA-Z0-9]*)(?:[ \t]|\/?>|$)/;
 // Strips a leading # (frontmatter entries may carry one) and a trailing /; rejects an
 // all-digit result -- a tag needs at least one non-digit character.
 function normalizeTag(raw) {
@@ -31,19 +109,59 @@ function frontmatterTags(data) {
     }
     return found;
 }
-// #tag tokens outside fenced code blocks, inline code spans, wikilinks, and HTML tags.
+// #tag tokens outside fenced code blocks, inline code spans, wikilinks, HTML tags, HTML blocks,
+// <!-- --> comments, and link destinations.
 function inlineTags(body) {
     const found = [];
-    let inFence = false;
+    const fence = fenceTracker();
+    const comment = commentTracker();
+    let inHtmlBlock = false;
+    let htmlBlockClose = null; // set while inside a type-1 (script/pre/style/textarea) block
     for (const line of body.split('\n')){
-        if (FENCE_RE.test(line)) {
-            inFence = !inFence;
+        if (inHtmlBlock) {
+            // A fence-like line here is still HTML-block content -- the block wins until it closes.
+            if (htmlBlockClose) {
+                if (htmlBlockClose.test(line)) {
+                    inHtmlBlock = false;
+                    htmlBlockClose = null;
+                }
+            } else if (/^[ \t>]*$/.test(line)) {
+                inHtmlBlock = false;
+            }
             continue;
         }
-        if (inFence) continue;
-        if (!line.includes('#')) continue; // most lines; skip the regex work
-        let cleaned = line.includes('`') ? line.replace(INLINE_CODE_RE, (m)=>' '.repeat(m.length)) : line;
+        if (!comment.inComment) {
+            if (fence.feed(line)) continue;
+            if (fence.inFence) continue;
+        }
+        // Obsidian parses nothing inside a <!-- --> comment; mask its span (which may open, close,
+        // or run the whole line) before any other check sees this line's text. Skipped when there
+        // is no comment marker anywhere near this line -- most lines, so worth the branch.
+        const masked = comment.inComment || line.includes('<!--') ? comment.mask(line) : line;
+        // Indented or blockquoted HTML blocks still swallow their content in Obsidian, so the
+        // opener test runs after stripping leading whitespace and > markers.
+        const stripped = masked.replace(/^[ \t>]*/, '');
+        if (stripped[0] === '<') {
+            const openMatch = HTML_BLOCK_OPEN_RE.exec(stripped);
+            if (openMatch) {
+                const tagName = openMatch[1].toLowerCase();
+                const isClosingTag = stripped[1] === '/';
+                if (!isClosingTag && HTML_PRE_TAGS.has(tagName)) {
+                    inHtmlBlock = true;
+                    // Any of the four type-1 closers ends the block, not only the tag that opened it.
+                    htmlBlockClose = /<\/(?:script|pre|style|textarea)>/i;
+                    continue;
+                }
+                if (HTML_BLOCK_TAGS.has(tagName)) {
+                    inHtmlBlock = true;
+                    continue;
+                }
+            }
+        }
+        if (!masked.includes('#')) continue; // most lines; skip the regex work
+        let cleaned = masked.includes('`') ? maskCodeSpans(masked) : masked;
         if (cleaned.includes('[[')) cleaned = cleaned.replace(WIKILINK_RE, (m)=>' '.repeat(m.length));
+        if (cleaned.includes('](')) cleaned = cleaned.replace(LINK_DEST_RE, (m)=>' '.repeat(m.length));
         if (cleaned.includes('<')) cleaned = cleaned.replace(HTML_TAG_RE, (m)=>' '.repeat(m.length));
         for (const m of cleaned.matchAll(INLINE_TAG_RE)){
             const tag = normalizeTag(m[1]);

package/dist/esm/features/tags.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/features/tags.ts"],"sourcesContent":["import type { Feature } from './types.ts';\n\n// tags(path, tag): Obsidian's file.tags grain -- frontmatter list/string tags plus inline\n// #tags from the prose, deduplicated, source not distinguished. Nested tags store full\n// (book/scifi); `tag = 'book' OR tag LIKE 'book/%'` is how a caller matches the parent too.\n\nconst FENCE_RE = /^(```|~~~)/;\nconst INLINE_CODE_RE = /`[^`]*`/g;\n// Obsidian treats [[#Heading]] as a same-note link, not a tag.\nconst WIKILINK_RE = /\\[\\[.*?\\]\\]/g; // to the first ]], so a heading holding a lone ] still masks\n// Obsidian doesn't read tags inside HTML markup.\nconst HTML_TAG_RE = /<\\/?[a-zA-Z][^>]*>/g; // tag-shaped only: a comparison's `< 5` must not open a span\n// Anchors on start-of-line or a preceding whitespace/(/[ so `a#b` and URL fragments don't count.\nconst INLINE_TAG_RE = /(?:^|[\\s([])#([\\p{L}\\p{N}_/-]+)/gu;\n\n// Strips a leading # (frontmatter entries may carry one) and a trailing /; rejects an\n// all-digit result -- a tag needs at least one non-digit character.\nfunction normalizeTag(raw: string): string | null {\n  const stripped = raw.replace(/^#/, '').replace(/\\/+$/, '');\n  if (!stripped || /^\\d+$/.test(stripped)) return null;\n  return stripped;\n}\n\n// data.tags: a YAML list (Obsidian also accepts a bare string). Null members and non-string\n// members are skipped rather than throwing -- `tags:\\n  -` parses to [null].\nfunction frontmatterTags(data?: Record<string, unknown>): string[] {\n  const raw = data?.tags;\n  const list = Array.isArray(raw) ? raw : typeof raw === 'string' ? [raw] : [];\n  const found: string[] = [];\n  for (const item of list) {\n    if (typeof item !== 'string') continue;\n    const tag = normalizeTag(item);\n    if (tag) found.push(tag);\n  }\n  return found;\n}\n\n// #tag tokens outside fenced code blocks, inline code spans, wikilinks, and HTML tags.\nfunction inlineTags(body: string): string[] {\n  const found: string[] = [];\n  let inFence = false;\n  for (const line of body.split('\\n')) {\n    if (FENCE_RE.test(line)) {\n      inFence = !inFence;\n      continue;\n    }\n    if (inFence) continue;\n    if (!line.includes('#')) continue; // most lines; skip the regex work\n    let cleaned = line.includes('`') ? line.replace(INLINE_CODE_RE, (m) => ' '.repeat(m.length)) : line;\n    if (cleaned.includes('[[')) cleaned = cleaned.replace(WIKILINK_RE, (m) => ' '.repeat(m.length));\n    if (cleaned.includes('<')) cleaned = cleaned.replace(HTML_TAG_RE, (m) => ' '.repeat(m.length));\n    for (const m of cleaned.matchAll(INLINE_TAG_RE)) {\n      const tag = normalizeTag(m[1]);\n      if (tag) found.push(tag);\n    }\n  }\n  return found;\n}\n\nfunction extract(_raw: string, body: string, _search?: { title: string; summary: string }, data?: Record<string, unknown>): string[] {\n  return [...new Set([...frontmatterTags(data), ...inlineTags(body)])].sort();\n}\n\nexport const tags: Feature = {\n  name: 'tags',\n  schema(db) {\n    db.exec('CREATE TABLE IF NOT EXISTS tags (\"path\" TEXT, tag TEXT, PRIMARY KEY (\"path\", tag))');\n    db.exec('CREATE INDEX IF NOT EXISTS tags_tag ON tags(tag)');\n  },\n  extract,\n  // Per-file rows with nothing else to resolve, so only a vanished file needs a delete here;\n  // store() below handles a reparse's stale rows itself.\n  remove(db, path, delta) {\n    if (!delta.vanished.includes(path)) return;\n    db.prepare('DELETE FROM tags WHERE \"path\" = ?').run(path);\n  },\n  store(db, path, extracted) {\n    const found = extracted as string[];\n    if (found.length === 0) {\n      db.prepare('DELETE FROM tags WHERE \"path\" = ?').run(path);\n    } else {\n      const placeholders = found.map(() => '?').join(', ');\n      db.prepare(`DELETE FROM tags WHERE \"path\" = ? AND tag NOT IN (${placeholders})`).run(path, ...found);\n    }\n    const insert = db.prepare('INSERT OR IGNORE INTO tags (\"path\", tag) VALUES (?, ?)');\n    for (const tag of found) insert.run(path, tag);\n  },\n};\n"],"names":["FENCE_RE","INLINE_CODE_RE","WIKILINK_RE","HTML_TAG_RE","INLINE_TAG_RE","normalizeTag","raw","stripped","replace","test","frontmatterTags","data","tags","list","Array","isArray","found","item","tag","push","inlineTags","body","inFence","line","split","includes","cleaned","m","repeat","length","matchAll","extract","_raw","_search","Set","sort","name","schema","db","exec","remove","path","delta","vanished","prepare","run","store","extracted","placeholders","map","join","insert"],"mappings":"AAEA,0FAA0F;AAC1F,uFAAuF;AACvF,4FAA4F;AAE5F,MAAMA,WAAW;AACjB,MAAMC,iBAAiB;AACvB,+DAA+D;AAC/D,MAAMC,cAAc,gBAAgB,6DAA6D;AACjG,iDAAiD;AACjD,MAAMC,cAAc,uBAAuB,6DAA6D;AACxG,iGAAiG;AACjG,MAAMC,gBAAgB;AAEtB,sFAAsF;AACtF,oEAAoE;AACpE,SAASC,aAAaC,GAAW;IAC/B,MAAMC,WAAWD,IAAIE,OAAO,CAAC,MAAM,IAAIA,OAAO,CAAC,QAAQ;IACvD,IAAI,CAACD,YAAY,QAAQE,IAAI,CAACF,WAAW,OAAO;IAChD,OAAOA;AACT;AAEA,4FAA4F;AAC5F,6EAA6E;AAC7E,SAASG,gBAAgBC,IAA8B;IACrD,MAAML,MAAMK,iBAAAA,2BAAAA,KAAMC,IAAI;IACtB,MAAMC,OAAOC,MAAMC,OAAO,CAACT,OAAOA,MAAM,OAAOA,QAAQ,WAAW;QAACA;KAAI,GAAG,EAAE;IAC5E,MAAMU,QAAkB,EAAE;IAC1B,KAAK,MAAMC,QAAQJ,KAAM;QACvB,IAAI,OAAOI,SAAS,UAAU;QAC9B,MAAMC,MAAMb,aAAaY;QACzB,IAAIC,KAAKF,MAAMG,IAAI,CAACD;IACtB;IACA,OAAOF;AACT;AAEA,uFAAuF;AACvF,SAASI,WAAWC,IAAY;IAC9B,MAAML,QAAkB,EAAE;IAC1B,IAAIM,UAAU;IACd,KAAK,MAAMC,QAAQF,KAAKG,KAAK,CAAC,MAAO;QACnC,IAAIxB,SAASS,IAAI,CAACc,OAAO;YACvBD,UAAU,CAACA;YACX;QACF;QACA,IAAIA,SAAS;QACb,IAAI,CAACC,KAAKE,QAAQ,CAAC,MAAM,UAAU,kCAAkC;QACrE,IAAIC,UAAUH,KAAKE,QAAQ,CAAC,OAAOF,KAAKf,OAAO,CAACP,gBAAgB,CAAC0B,IAAM,IAAIC,MAAM,CAACD,EAAEE,MAAM,KAAKN;QAC/F,IAAIG,QAAQD,QAAQ,CAAC,OAAOC,UAAUA,QAAQlB,OAAO,CAACN,aAAa,CAACyB,IAAM,IAAIC,MAAM,CAACD,EAAEE,MAAM;QAC7F,IAAIH,QAAQD,QAAQ,CAAC,MAAMC,UAAUA,QAAQlB,OAAO,CAACL,aAAa,CAACwB,IAAM,IAAIC,MAAM,CAACD,EAAEE,MAAM;QAC5F,KAAK,MAAMF,KAAKD,QAAQI,QAAQ,CAAC1B,eAAgB;YAC/C,MAAMc,MAAMb,aAAasB,CAAC,CAAC,EAAE;YAC7B,IAAIT,KAAKF,MAAMG,IAAI,CAACD;QACtB;IACF;IACA,OAAOF;AACT;AAEA,SAASe,QAAQC,IAAY,EAAEX,IAAY,EAAEY,OAA4C,EAAEtB,IAA8B;IACvH,OAAO;WAAI,IAAIuB,IAAI;eAAIxB,gBAAgBC;eAAUS,WAAWC;SAAM;KAAE,CAACc,IAAI;AAC3E;AAEA,OAAO,MAAMvB,OAAgB;IAC3BwB,MAAM;IACNC,QAAOC,EAAE;QACPA,GAAGC,IAAI,CAAC;QACRD,GAAGC,IAAI,CAAC;IACV;IACAR;IACA,2FAA2F;IAC3F,uDAAuD;IACvDS,QAAOF,EAAE,EAAEG,IAAI,EAAEC,KAAK;QACpB,IAAI,CAACA,MAAMC,QAAQ,CAAClB,QAAQ,CAACgB,OAAO;QACpCH,GAAGM,OAAO,CAAC,qCAAqCC,GAAG,CAACJ;IACtD;IACAK,OAAMR,EAAE,EAAEG,IAAI,EAAEM,SAAS;QACvB,MAAM/B,QAAQ+B;QACd,IAAI/B,MAAMa,MAAM,KAAK,GAAG;YACtBS,GAAGM,OAAO,CAAC,qCAAqCC,GAAG,CAACJ;QACtD,OAAO;YACL,MAAMO,eAAehC,MAAMiC,GAAG,CAAC,IAAM,KAAKC,IAAI,CAAC;YAC/CZ,GAAGM,OAAO,CAAC,CAAC,kDAAkD,EAAEI,aAAa,CAAC,CAAC,EAAEH,GAAG,CAACJ,SAASzB;QAChG;QACA,MAAMmC,SAASb,GAAGM,OAAO,CAAC;QAC1B,KAAK,MAAM1B,OAAOF,MAAOmC,OAAON,GAAG,CAACJ,MAAMvB;IAC5C;AACF,EAAE"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/features/tags.ts"],"sourcesContent":["import { commentTracker, fenceTracker, maskCodeSpans } from '../fences.ts';\nimport type { Feature } from './types.ts';\n\n// tags(path, tag): Obsidian's file.tags grain -- frontmatter list/string tags plus inline\n// #tags from the prose, deduplicated, source not distinguished. Nested tags store full\n// (book/scifi); `tag = 'book' OR tag LIKE 'book/%'` is how a caller matches the parent too.\n\n// Obsidian treats [[#Heading]] as a same-note link, not a tag.\nconst WIKILINK_RE = /\\[\\[.*?\\]\\]/g; // to the first ]], so a heading holding a lone ] still masks\n// Obsidian doesn't read tags inside HTML markup.\nconst HTML_TAG_RE = /<\\/?[a-zA-Z][^>]*>/g; // tag-shaped only: a comparison's `< 5` must not open a span\n// Anchors on start-of-line or a preceding whitespace/(/[ so `a#b` and URL fragments don't count.\nconst INLINE_TAG_RE = /(?:^|[\\s([])#([\\p{L}\\p{N}_/-]+)/gu;\n// A markdown link destination `](...)` -- `[text](#anchor)` is a same-page link, not a tag.\nconst LINK_DEST_RE = /\\]\\((?:[^()]|\\([^()]*\\))*\\)/g; // one paren-nesting level, as CommonMark destinations allow: (https://x/a_(b)#frag)\n\n// CommonMark's HTML-block type-6 list (fixed by the spec, not a drifting enumeration): a line\n// starting with an open or close tag of one of these, at column 0, opens a block that swallows\n// following lines -- including any #tag in them -- until a blank line closes it.\nconst HTML_BLOCK_TAGS = new Set([\n  'address',\n  'article',\n  'aside',\n  'base',\n  'basefont',\n  'blockquote',\n  'body',\n  'caption',\n  'center',\n  'col',\n  'colgroup',\n  'dd',\n  'details',\n  'dialog',\n  'dir',\n  'div',\n  'dl',\n  'dt',\n  'fieldset',\n  'figcaption',\n  'figure',\n  'footer',\n  'form',\n  'frame',\n  'frameset',\n  'h1',\n  'h2',\n  'h3',\n  'h4',\n  'h5',\n  'h6',\n  'head',\n  'header',\n  'hr',\n  'html',\n  'iframe',\n  'legend',\n  'li',\n  'link',\n  'main',\n  'menu',\n  'menuitem',\n  'nav',\n  'noframes',\n  'ol',\n  'optgroup',\n  'option',\n  'p',\n  'param',\n  'search',\n  'section',\n  'summary',\n  'table',\n  'tbody',\n  'td',\n  'tfoot',\n  'th',\n  'thead',\n  'title',\n  'tr',\n  'track',\n  'ul',\n]);\n// Type-1 blocks (script/pre/style/textarea): closes on the line holding the matching end tag,\n// not on a blank line, and that line is the last one skipped.\nconst HTML_PRE_TAGS = new Set(['script', 'pre', 'style', 'textarea']);\n// An opening or closing tag at column 0, tag name captured for the lookups above.\nconst HTML_BLOCK_OPEN_RE = /^<\\/?([a-zA-Z][a-zA-Z0-9]*)(?:[ \\t]|\\/?>|$)/;\n\n// Strips a leading # (frontmatter entries may carry one) and a trailing /; rejects an\n// all-digit result -- a tag needs at least one non-digit character.\nfunction normalizeTag(raw: string): string | null {\n  const stripped = raw.replace(/^#/, '').replace(/\\/+$/, '');\n  if (!stripped || /^\\d+$/.test(stripped)) return null;\n  return stripped;\n}\n\n// data.tags: a YAML list (Obsidian also accepts a bare string). Null members and non-string\n// members are skipped rather than throwing -- `tags:\\n  -` parses to [null].\nfunction frontmatterTags(data?: Record<string, unknown>): string[] {\n  const raw = data?.tags;\n  const list = Array.isArray(raw) ? raw : typeof raw === 'string' ? [raw] : [];\n  const found: string[] = [];\n  for (const item of list) {\n    if (typeof item !== 'string') continue;\n    const tag = normalizeTag(item);\n    if (tag) found.push(tag);\n  }\n  return found;\n}\n\n// #tag tokens outside fenced code blocks, inline code spans, wikilinks, HTML tags, HTML blocks,\n// <!-- --> comments, and link destinations.\nfunction inlineTags(body: string): string[] {\n  const found: string[] = [];\n  const fence = fenceTracker();\n  const comment = commentTracker();\n  let inHtmlBlock = false;\n  let htmlBlockClose: RegExp | null = null; // set while inside a type-1 (script/pre/style/textarea) block\n  for (const line of body.split('\\n')) {\n    if (inHtmlBlock) {\n      // A fence-like line here is still HTML-block content -- the block wins until it closes.\n      if (htmlBlockClose) {\n        if (htmlBlockClose.test(line)) {\n          inHtmlBlock = false;\n          htmlBlockClose = null;\n        }\n      } else if (/^[ \\t>]*$/.test(line)) {\n        inHtmlBlock = false;\n      }\n      continue;\n    }\n    if (!comment.inComment) {\n      if (fence.feed(line)) continue;\n      if (fence.inFence) continue;\n    }\n    // Obsidian parses nothing inside a <!-- --> comment; mask its span (which may open, close,\n    // or run the whole line) before any other check sees this line's text. Skipped when there\n    // is no comment marker anywhere near this line -- most lines, so worth the branch.\n    const masked = comment.inComment || line.includes('<!--') ? comment.mask(line) : line;\n    // Indented or blockquoted HTML blocks still swallow their content in Obsidian, so the\n    // opener test runs after stripping leading whitespace and > markers.\n    const stripped = masked.replace(/^[ \\t>]*/, '');\n    if (stripped[0] === '<') {\n      const openMatch = HTML_BLOCK_OPEN_RE.exec(stripped);\n      if (openMatch) {\n        const tagName = openMatch[1].toLowerCase();\n        const isClosingTag = stripped[1] === '/';\n        if (!isClosingTag && HTML_PRE_TAGS.has(tagName)) {\n          inHtmlBlock = true;\n          // Any of the four type-1 closers ends the block, not only the tag that opened it.\n          htmlBlockClose = /<\\/(?:script|pre|style|textarea)>/i;\n          continue;\n        }\n        if (HTML_BLOCK_TAGS.has(tagName)) {\n          inHtmlBlock = true;\n          continue;\n        }\n      }\n    }\n    if (!masked.includes('#')) continue; // most lines; skip the regex work\n    let cleaned = masked.includes('`') ? maskCodeSpans(masked) : masked;\n    if (cleaned.includes('[[')) cleaned = cleaned.replace(WIKILINK_RE, (m) => ' '.repeat(m.length));\n    if (cleaned.includes('](')) cleaned = cleaned.replace(LINK_DEST_RE, (m) => ' '.repeat(m.length));\n    if (cleaned.includes('<')) cleaned = cleaned.replace(HTML_TAG_RE, (m) => ' '.repeat(m.length));\n    for (const m of cleaned.matchAll(INLINE_TAG_RE)) {\n      const tag = normalizeTag(m[1]);\n      if (tag) found.push(tag);\n    }\n  }\n  return found;\n}\n\nfunction extract(_raw: string, body: string, _search?: { title: string; summary: string }, data?: Record<string, unknown>): string[] {\n  return [...new Set([...frontmatterTags(data), ...inlineTags(body)])].sort();\n}\n\nexport const tags: Feature = {\n  name: 'tags',\n  schema(db) {\n    db.exec('CREATE TABLE IF NOT EXISTS tags (\"path\" TEXT, tag TEXT, PRIMARY KEY (\"path\", tag))');\n    db.exec('CREATE INDEX IF NOT EXISTS tags_tag ON tags(tag)');\n  },\n  extract,\n  // Per-file rows with nothing else to resolve, so only a vanished file needs a delete here;\n  // store() below handles a reparse's stale rows itself.\n  remove(db, path, delta) {\n    if (!delta.vanished.includes(path)) return;\n    db.prepare('DELETE FROM tags WHERE \"path\" = ?').run(path);\n  },\n  store(db, path, extracted) {\n    const found = extracted as string[];\n    if (found.length === 0) {\n      db.prepare('DELETE FROM tags WHERE \"path\" = ?').run(path);\n    } else {\n      const placeholders = found.map(() => '?').join(', ');\n      db.prepare(`DELETE FROM tags WHERE \"path\" = ? AND tag NOT IN (${placeholders})`).run(path, ...found);\n    }\n    const insert = db.prepare('INSERT OR IGNORE INTO tags (\"path\", tag) VALUES (?, ?)');\n    for (const tag of found) insert.run(path, tag);\n  },\n};\n"],"names":["commentTracker","fenceTracker","maskCodeSpans","WIKILINK_RE","HTML_TAG_RE","INLINE_TAG_RE","LINK_DEST_RE","HTML_BLOCK_TAGS","Set","HTML_PRE_TAGS","HTML_BLOCK_OPEN_RE","normalizeTag","raw","stripped","replace","test","frontmatterTags","data","tags","list","Array","isArray","found","item","tag","push","inlineTags","body","fence","comment","inHtmlBlock","htmlBlockClose","line","split","inComment","feed","inFence","masked","includes","mask","openMatch","exec","tagName","toLowerCase","isClosingTag","has","cleaned","m","repeat","length","matchAll","extract","_raw","_search","sort","name","schema","db","remove","path","delta","vanished","prepare","run","store","extracted","placeholders","map","join","insert"],"mappings":"AAAA,SAASA,cAAc,EAAEC,YAAY,EAAEC,aAAa,QAAQ,eAAe;AAG3E,0FAA0F;AAC1F,uFAAuF;AACvF,4FAA4F;AAE5F,+DAA+D;AAC/D,MAAMC,cAAc,gBAAgB,6DAA6D;AACjG,iDAAiD;AACjD,MAAMC,cAAc,uBAAuB,6DAA6D;AACxG,iGAAiG;AACjG,MAAMC,gBAAgB;AACtB,4FAA4F;AAC5F,MAAMC,eAAe,gCAAgC,oFAAoF;AAEzI,8FAA8F;AAC9F,+FAA+F;AAC/F,iFAAiF;AACjF,MAAMC,kBAAkB,IAAIC,IAAI;IAC9B;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AACD,8FAA8F;AAC9F,8DAA8D;AAC9D,MAAMC,gBAAgB,IAAID,IAAI;IAAC;IAAU;IAAO;IAAS;CAAW;AACpE,kFAAkF;AAClF,MAAME,qBAAqB;AAE3B,sFAAsF;AACtF,oEAAoE;AACpE,SAASC,aAAaC,GAAW;IAC/B,MAAMC,WAAWD,IAAIE,OAAO,CAAC,MAAM,IAAIA,OAAO,CAAC,QAAQ;IACvD,IAAI,CAACD,YAAY,QAAQE,IAAI,CAACF,WAAW,OAAO;IAChD,OAAOA;AACT;AAEA,4FAA4F;AAC5F,6EAA6E;AAC7E,SAASG,gBAAgBC,IAA8B;IACrD,MAAML,MAAMK,iBAAAA,2BAAAA,KAAMC,IAAI;IACtB,MAAMC,OAAOC,MAAMC,OAAO,CAACT,OAAOA,MAAM,OAAOA,QAAQ,WAAW;QAACA;KAAI,GAAG,EAAE;IAC5E,MAAMU,QAAkB,EAAE;IAC1B,KAAK,MAAMC,QAAQJ,KAAM;QACvB,IAAI,OAAOI,SAAS,UAAU;QAC9B,MAAMC,MAAMb,aAAaY;QACzB,IAAIC,KAAKF,MAAMG,IAAI,CAACD;IACtB;IACA,OAAOF;AACT;AAEA,gGAAgG;AAChG,4CAA4C;AAC5C,SAASI,WAAWC,IAAY;IAC9B,MAAML,QAAkB,EAAE;IAC1B,MAAMM,QAAQ3B;IACd,MAAM4B,UAAU7B;IAChB,IAAI8B,cAAc;IAClB,IAAIC,iBAAgC,MAAM,8DAA8D;IACxG,KAAK,MAAMC,QAAQL,KAAKM,KAAK,CAAC,MAAO;QACnC,IAAIH,aAAa;YACf,wFAAwF;YACxF,IAAIC,gBAAgB;gBAClB,IAAIA,eAAehB,IAAI,CAACiB,OAAO;oBAC7BF,cAAc;oBACdC,iBAAiB;gBACnB;YACF,OAAO,IAAI,YAAYhB,IAAI,CAACiB,OAAO;gBACjCF,cAAc;YAChB;YACA;QACF;QACA,IAAI,CAACD,QAAQK,SAAS,EAAE;YACtB,IAAIN,MAAMO,IAAI,CAACH,OAAO;YACtB,IAAIJ,MAAMQ,OAAO,EAAE;QACrB;QACA,2FAA2F;QAC3F,0FAA0F;QAC1F,mFAAmF;QACnF,MAAMC,SAASR,QAAQK,SAAS,IAAIF,KAAKM,QAAQ,CAAC,UAAUT,QAAQU,IAAI,CAACP,QAAQA;QACjF,sFAAsF;QACtF,qEAAqE;QACrE,MAAMnB,WAAWwB,OAAOvB,OAAO,CAAC,YAAY;QAC5C,IAAID,QAAQ,CAAC,EAAE,KAAK,KAAK;YACvB,MAAM2B,YAAY9B,mBAAmB+B,IAAI,CAAC5B;YAC1C,IAAI2B,WAAW;gBACb,MAAME,UAAUF,SAAS,CAAC,EAAE,CAACG,WAAW;gBACxC,MAAMC,eAAe/B,QAAQ,CAAC,EAAE,KAAK;gBACrC,IAAI,CAAC+B,gBAAgBnC,cAAcoC,GAAG,CAACH,UAAU;oBAC/CZ,cAAc;oBACd,kFAAkF;oBAClFC,iBAAiB;oBACjB;gBACF;gBACA,IAAIxB,gBAAgBsC,GAAG,CAACH,UAAU;oBAChCZ,cAAc;oBACd;gBACF;YACF;QACF;QACA,IAAI,CAACO,OAAOC,QAAQ,CAAC,MAAM,UAAU,kCAAkC;QACvE,IAAIQ,UAAUT,OAAOC,QAAQ,CAAC,OAAOpC,cAAcmC,UAAUA;QAC7D,IAAIS,QAAQR,QAAQ,CAAC,OAAOQ,UAAUA,QAAQhC,OAAO,CAACX,aAAa,CAAC4C,IAAM,IAAIC,MAAM,CAACD,EAAEE,MAAM;QAC7F,IAAIH,QAAQR,QAAQ,CAAC,OAAOQ,UAAUA,QAAQhC,OAAO,CAACR,cAAc,CAACyC,IAAM,IAAIC,MAAM,CAACD,EAAEE,MAAM;QAC9F,IAAIH,QAAQR,QAAQ,CAAC,MAAMQ,UAAUA,QAAQhC,OAAO,CAACV,aAAa,CAAC2C,IAAM,IAAIC,MAAM,CAACD,EAAEE,MAAM;QAC5F,KAAK,MAAMF,KAAKD,QAAQI,QAAQ,CAAC7C,eAAgB;YAC/C,MAAMmB,MAAMb,aAAaoC,CAAC,CAAC,EAAE;YAC7B,IAAIvB,KAAKF,MAAMG,IAAI,CAACD;QACtB;IACF;IACA,OAAOF;AACT;AAEA,SAAS6B,QAAQC,IAAY,EAAEzB,IAAY,EAAE0B,OAA4C,EAAEpC,IAA8B;IACvH,OAAO;WAAI,IAAIT,IAAI;eAAIQ,gBAAgBC;eAAUS,WAAWC;SAAM;KAAE,CAAC2B,IAAI;AAC3E;AAEA,OAAO,MAAMpC,OAAgB;IAC3BqC,MAAM;IACNC,QAAOC,EAAE;QACPA,GAAGhB,IAAI,CAAC;QACRgB,GAAGhB,IAAI,CAAC;IACV;IACAU;IACA,2FAA2F;IAC3F,uDAAuD;IACvDO,QAAOD,EAAE,EAAEE,IAAI,EAAEC,KAAK;QACpB,IAAI,CAACA,MAAMC,QAAQ,CAACvB,QAAQ,CAACqB,OAAO;QACpCF,GAAGK,OAAO,CAAC,qCAAqCC,GAAG,CAACJ;IACtD;IACAK,OAAMP,EAAE,EAAEE,IAAI,EAAEM,SAAS;QACvB,MAAM3C,QAAQ2C;QACd,IAAI3C,MAAM2B,MAAM,KAAK,GAAG;YACtBQ,GAAGK,OAAO,CAAC,qCAAqCC,GAAG,CAACJ;QACtD,OAAO;YACL,MAAMO,eAAe5C,MAAM6C,GAAG,CAAC,IAAM,KAAKC,IAAI,CAAC;YAC/CX,GAAGK,OAAO,CAAC,CAAC,kDAAkD,EAAEI,aAAa,CAAC,CAAC,EAAEH,GAAG,CAACJ,SAASrC;QAChG;QACA,MAAM+C,SAASZ,GAAGK,OAAO,CAAC;QAC1B,KAAK,MAAMtC,OAAOF,MAAO+C,OAAON,GAAG,CAACJ,MAAMnC;IAC5C;AACF,EAAE"}

package/dist/esm/fences.d.ts

@@ -0,0 +1,12 @@
+export interface FenceTracker {
+    feed(line: string): boolean;
+    readonly inFence: boolean;
+}
+export declare function fenceTracker(): FenceTracker;
+export interface CommentTracker {
+    mask(line: string): string;
+    readonly inComment: boolean;
+}
+export declare function commentTracker(): CommentTracker;
+export declare function maskRegions(body: string): string;
+export declare function maskCodeSpans(line: string): string;