sensemaking 0.15.4 → 0.16.0

package/dist/cjs/commands/related.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/commands/related.ts"],"sourcesContent":["import type { DatabaseSync } from 'node:sqlite';\nimport type { ResolvedConfig, SearchOverrides } from '../config/index.ts';\nimport { embedEnabled, resolveSearch } from '../config/index.ts';\nimport { SenseError } from '../errors.ts';\nimport { embedPending, hasEmbedding, modelPresent, similarNotes } from '../features/embed.ts';\nimport { resolveNote } from './peek.ts';\nimport { scopedPaths, scopeHasEmbeddings } from './scope.ts';\n\n// Notes most similar by cosine, excluding self and everything already linked either way.\n// Its own command, not a peek section: a full embeddings scan is ~480ms at 26k notes.\nexport async function relatedNotes(db: DatabaseSync, cfg: ResolvedConfig, pathArg: string, overrides: SearchOverrides, k: number): Promise<Array<{ path: string; similarity: number }>> {\n  const paths = (db.prepare('SELECT \"path\" FROM frontmatter').all() as Array<{ path: string }>).map((r) => r.path);\n  const path = resolveNote(paths, pathArg);\n\n  const outbound = (db.prepare('SELECT DISTINCT dst FROM links WHERE src = ? AND dst IS NOT NULL AND dst != src').all(path) as Array<{ dst: string }>).map((r) => r.dst);\n  const backlinks = (db.prepare('SELECT DISTINCT src FROM links WHERE dst = ? AND src != dst').all(path) as Array<{ src: string }>).map((r) => r.src);\n  const exclude = new Set([path, ...outbound, ...backlinks]);\n\n  // Vectors are the only signal `related` has, so every way of not having them is an error\n  // naming the cause. An empty table then means one thing: nothing near in meaning that this\n  // note does not already link to, which is a real answer.\n  const effective = resolveSearch(cfg, overrides);\n  if (!embedEnabled(cfg)) {\n    throw new SenseError('EMBED_DISABLED', 'related ranks notes by meaning, and this tree has no embedding model; add an \"embed\" block naming one to sense.config.json, then run `sense download` (search works without it, on words and links)');\n  }\n  // search gates on the same flag (see wantsVectors above); reading it here too keeps\n  // `semantic: false` meaning one thing. Without this, an overlapping semantic-on preset's\n  // vectors would answer for a scope that declined them.\n  if (!effective.semantic) {\n    throw new SenseError('PRESET_NOT_SEMANTIC', `preset \"${effective.presetName}\" sets \"semantic\": false, so this scope has no vectors and related has no other signal; search it instead (words and links), or set semantic back on for that preset`);\n  }\n  if (!modelPresent(cfg)) {\n    throw new SenseError('EMBED_MODEL_MISSING', 'related ranks notes by meaning, so it needs the embedding model, which is not downloaded; run `sense download` (search still works without it, on words and links)');\n  }\n  const allowed = scopedPaths(db, cfg, overrides);\n  // Top up pending rows before the seed check, or a fresh index reports every note as\n  // having no indexed text until some search has run.\n  await embedPending(db, cfg, cfg.baseDir);\n  if (!hasEmbedding(db, path)) {\n    throw new SenseError('NOTE_NOT_EMBEDDED', `${path} has no indexed text to compare -- a note that is frontmatter only, or empty, has nothing to rank by meaning`);\n  }\n  if (!scopeHasEmbeddings(db, cfg, allowed)) return [];\n  return similarNotes(db, cfg, path, { exclude, allowed, k });\n}\n"],"names":["relatedNotes","db","cfg","pathArg","overrides","k","paths","path","outbound","backlinks","exclude","effective","allowed","prepare","all","map","r","resolveNote","dst","src","Set","resolveSearch","embedEnabled","SenseError","semantic","presetName","modelPresent","scopedPaths","embedPending","baseDir","hasEmbedding","scopeHasEmbeddings","similarNotes"],"mappings":";;;;+BAUsBA;;;eAAAA;;;uBARsB;wBACjB;uBAC4C;sBAC3C;uBACoB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAIzC,SAAeA,aAAaC,EAAgB,EAAEC,GAAmB,EAAEC,OAAe,EAAEC,SAA0B,EAAEC,CAAS;;YACxHC,OACAC,MAEAC,UACAC,WACAC,SAKAC,WAaAC;;;;oBAvBAN,QAAQ,AAACL,GAAGY,OAAO,CAAC,kCAAkCC,GAAG,GAA+BC,GAAG,CAAC,SAACC;+BAAMA,EAAET,IAAI;;oBACzGA,OAAOU,IAAAA,mBAAW,EAACX,OAAOH;oBAE1BK,WAAW,AAACP,GAAGY,OAAO,CAAC,mFAAmFC,GAAG,CAACP,MAAiCQ,GAAG,CAAC,SAACC;+BAAMA,EAAEE,GAAG;;oBAC/JT,YAAY,AAACR,GAAGY,OAAO,CAAC,+DAA+DC,GAAG,CAACP,MAAiCQ,GAAG,CAAC,SAACC;+BAAMA,EAAEG,GAAG;;oBAC5IT,UAAU,IAAIU,IAAI;wBAACb;sBAAD,OAAO,qBAAGC,WAAU,qBAAGC;oBAE/C,yFAAyF;oBACzF,2FAA2F;oBAC3F,yDAAyD;oBACnDE,YAAYU,IAAAA,sBAAa,EAACnB,KAAKE;oBACrC,IAAI,CAACkB,IAAAA,qBAAY,EAACpB,MAAM;wBACtB,MAAM,IAAIqB,oBAAU,CAAC,kBAAkB;oBACzC;oBACA,oFAAoF;oBACpF,yFAAyF;oBACzF,uDAAuD;oBACvD,IAAI,CAACZ,UAAUa,QAAQ,EAAE;wBACvB,MAAM,IAAID,oBAAU,CAAC,uBAAuB,AAAC,WAA+B,OAArBZ,UAAUc,UAAU,EAAC;oBAC9E;oBACA,IAAI,CAACC,IAAAA,qBAAY,EAACxB,MAAM;wBACtB,MAAM,IAAIqB,oBAAU,CAAC,uBAAuB;oBAC9C;oBACMX,UAAUe,IAAAA,oBAAW,EAAC1B,IAAIC,KAAKE;oBACrC,oFAAoF;oBACpF,oDAAoD;oBACpD;;wBAAMwB,IAAAA,qBAAY,EAAC3B,IAAIC,KAAKA,IAAI2B,OAAO;;;oBAAvC;oBACA,IAAI,CAACC,IAAAA,qBAAY,EAAC7B,IAAIM,OAAO;wBAC3B,MAAM,IAAIgB,oBAAU,CAAC,qBAAqB,AAAC,GAAO,OAALhB,MAAK;oBACpD;oBACA,IAAI,CAACwB,IAAAA,2BAAkB,EAAC9B,IAAIC,KAAKU,UAAU;;;;oBAC3C;;wBAAOoB,IAAAA,qBAAY,EAAC/B,IAAIC,KAAKK,MAAM;4BAAEG,SAAAA;4BAASE,SAAAA;4BAASP,GAAAA;wBAAE;;;;IAC3D"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/commands/related.ts"],"sourcesContent":["import type { DatabaseSync } from 'node:sqlite';\nimport type { ResolvedConfig, SearchOverrides } from '../config/index.ts';\nimport { embedConfig, embedEnabled, resolveSearch } from '../config/index.ts';\nimport { embedPending, hasEmbedding, similarNotes } from '../embed/query.ts';\nimport { localModelMissing, MODEL_FILENAMES } from '../embed/store.ts';\nimport { SenseError } from '../errors.ts';\nimport { resolveNote } from './peek.ts';\nimport { scopedPaths, scopeHasEmbeddings } from './scope.ts';\n\n// Notes most similar by cosine, excluding self and everything already linked either way.\n// Its own command, not a peek section: a full embeddings scan is ~480ms at 26k notes.\nexport async function relatedNotes(db: DatabaseSync, cfg: ResolvedConfig, pathArg: string, overrides: SearchOverrides, k: number): Promise<Array<{ path: string; similarity: number }>> {\n  const paths = (db.prepare('SELECT \"path\" FROM frontmatter').all() as Array<{ path: string }>).map((r) => r.path);\n  const path = resolveNote(paths, pathArg);\n\n  const outbound = (db.prepare('SELECT DISTINCT dst FROM links WHERE src = ? AND dst IS NOT NULL AND dst != src').all(path) as Array<{ dst: string }>).map((r) => r.dst);\n  const backlinks = (db.prepare('SELECT DISTINCT src FROM links WHERE dst = ? AND src != dst').all(path) as Array<{ src: string }>).map((r) => r.src);\n  const exclude = new Set([path, ...outbound, ...backlinks]);\n\n  // Vectors are the only signal `related` has, so every way of not having them is an error\n  // naming the cause. An empty table then means one thing: nothing near in meaning that this\n  // note does not already link to, which is a real answer.\n  const effective = resolveSearch(cfg, overrides);\n  if (!embedEnabled(cfg)) {\n    throw new SenseError('EMBED_DISABLED', 'related ranks notes by meaning, and this tree has no embedding model; add an \"embed\" block naming one to sense.config.json (a Hugging Face id fetches automatically at first use; `sense download` prefetches it) -- search works without it, on words and links');\n  }\n  // search gates on the same signal (see wantsVectors above); reading it here too keeps a\n  // preset's declared signals meaning one thing. Without this, an overlapping vectors-on\n  // preset's vectors would answer for a scope that declined them.\n  if (effective.signals.vectors === undefined) {\n    throw new SenseError('PRESET_NOT_SEMANTIC', `preset \"${effective.presetName}\" has no \"vectors\" signal, so this scope has no vectors and related has no other signal; search it instead (words and links), or add \"vectors\" to that preset's signals`);\n  }\n  // A downloadable HF id proceeds -- embedPending's getProvider call fetches it lazily on\n  // consent. Only a local path with missing files can never fetch itself.\n  const e = embedConfig(cfg); // embedEnabled(cfg) above guarantees this is set\n  if (localModelMissing(e)) {\n    throw new SenseError('EMBED_MODEL_MISSING', `related ranks notes by meaning, so it needs the embedding model, but the local model path \"${e.model}\" is missing ${MODEL_FILENAMES}; point embed.model at a directory containing them (search still works without it, on words and links)`);\n  }\n  const allowed = scopedPaths(db, cfg, overrides);\n  // Top up pending rows before the seed check, or a fresh index reports every note as\n  // having no indexed text until some search has run.\n  await embedPending(db, cfg, cfg.baseDir);\n  if (!hasEmbedding(db, path)) {\n    throw new SenseError('NOTE_NOT_EMBEDDED', `${path} has no indexed text to compare -- a note that is frontmatter only, or empty, has nothing to rank by meaning`);\n  }\n  if (!scopeHasEmbeddings(db, cfg, allowed)) return [];\n  return similarNotes(db, cfg, path, { exclude, allowed, k });\n}\n"],"names":["relatedNotes","db","cfg","pathArg","overrides","k","paths","path","outbound","backlinks","exclude","effective","e","allowed","prepare","all","map","r","resolveNote","dst","src","Set","resolveSearch","embedEnabled","SenseError","signals","vectors","undefined","presetName","embedConfig","localModelMissing","MODEL_FILENAMES","model","scopedPaths","embedPending","baseDir","hasEmbedding","scopeHasEmbeddings","similarNotes"],"mappings":";;;;+BAWsBA;;;eAAAA;;;uBATmC;uBACA;uBACN;wBACxB;sBACC;uBACoB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAIzC,SAAeA,aAAaC,EAAgB,EAAEC,GAAmB,EAAEC,OAAe,EAAEC,SAA0B,EAAEC,CAAS;;YACxHC,OACAC,MAEAC,UACAC,WACAC,SAKAC,WAYAC,GAIAC;;;;oBA1BAP,QAAQ,AAACL,GAAGa,OAAO,CAAC,kCAAkCC,GAAG,GAA+BC,GAAG,CAAC,SAACC;+BAAMA,EAAEV,IAAI;;oBACzGA,OAAOW,IAAAA,mBAAW,EAACZ,OAAOH;oBAE1BK,WAAW,AAACP,GAAGa,OAAO,CAAC,mFAAmFC,GAAG,CAACR,MAAiCS,GAAG,CAAC,SAACC;+BAAMA,EAAEE,GAAG;;oBAC/JV,YAAY,AAACR,GAAGa,OAAO,CAAC,+DAA+DC,GAAG,CAACR,MAAiCS,GAAG,CAAC,SAACC;+BAAMA,EAAEG,GAAG;;oBAC5IV,UAAU,IAAIW,IAAI;wBAACd;sBAAD,OAAO,qBAAGC,WAAU,qBAAGC;oBAE/C,yFAAyF;oBACzF,2FAA2F;oBAC3F,yDAAyD;oBACnDE,YAAYW,IAAAA,sBAAa,EAACpB,KAAKE;oBACrC,IAAI,CAACmB,IAAAA,qBAAY,EAACrB,MAAM;wBACtB,MAAM,IAAIsB,oBAAU,CAAC,kBAAkB;oBACzC;oBACA,wFAAwF;oBACxF,uFAAuF;oBACvF,gEAAgE;oBAChE,IAAIb,UAAUc,OAAO,CAACC,OAAO,KAAKC,WAAW;wBAC3C,MAAM,IAAIH,oBAAU,CAAC,uBAAuB,AAAC,WAA+B,OAArBb,UAAUiB,UAAU,EAAC;oBAC9E;oBACA,wFAAwF;oBACxF,wEAAwE;oBAClEhB,IAAIiB,IAAAA,oBAAW,EAAC3B,MAAM,iDAAiD;oBAC7E,IAAI4B,IAAAA,0BAAiB,EAAClB,IAAI;wBACxB,MAAM,IAAIY,oBAAU,CAAC,uBAAuB,AAAC,8FAAoHO,OAAvBnB,EAAEoB,KAAK,EAAC,iBAA+B,OAAhBD,wBAAe,EAAC;oBACnL;oBACMlB,UAAUoB,IAAAA,oBAAW,EAAChC,IAAIC,KAAKE;oBACrC,oFAAoF;oBACpF,oDAAoD;oBACpD;;wBAAM8B,IAAAA,qBAAY,EAACjC,IAAIC,KAAKA,IAAIiC,OAAO;;;oBAAvC;oBACA,IAAI,CAACC,IAAAA,qBAAY,EAACnC,IAAIM,OAAO;wBAC3B,MAAM,IAAIiB,oBAAU,CAAC,qBAAqB,AAAC,GAAO,OAALjB,MAAK;oBACpD;oBACA,IAAI,CAAC8B,IAAAA,2BAAkB,EAACpC,IAAIC,KAAKW,UAAU;;;;oBAC3C;;wBAAOyB,IAAAA,qBAAY,EAACrC,IAAIC,KAAKK,MAAM;4BAAEG,SAAAA;4BAASG,SAAAA;4BAASR,GAAAA;wBAAE;;;;IAC3D"}

package/dist/cjs/commands/search.js

@@ -11,13 +11,12 @@ Object.defineProperty(exports, "search", {
 var _nodefs = require("node:fs");
 var _nodepath = require("node:path");
 var _indexts = require("../config/index.js");
+var _storets = require("../embed/store.js");
 var _errorsts = require("../errors.js");
-var _embedts = require("../features/embed.js");
-var _indexts1 = require("../features/index.js");
-var _graphts = require("../graph.js");
 var _searcherrorts = require("../search-error.js");
 var _segmentts = require("../segment.js");
 var _scopets = require("./scope.js");
+var _signalsts = require("./signals.js");
 function _array_like_to_array(arr, len) {
     if (len == null || len > arr.length) len = arr.length;
     for(var i = 0, arr2 = new Array(len); i < len; i++)arr2[i] = arr[i];
@@ -26,9 +25,6 @@ function _array_like_to_array(arr, len) {
 function _array_with_holes(arr) {
     if (Array.isArray(arr)) return arr;
 }
-function _array_without_holes(arr) {
-    if (Array.isArray(arr)) return _array_like_to_array(arr);
-}
 function asyncGeneratorStep(gen, resolve, reject, _next, _throw, key, arg) {
     try {
         var info = gen[key](arg);
@@ -55,11 +51,6 @@ function _async_to_generator(fn) {
         });
     };
 }
-function _iterable_to_array(iter) {
-    if (typeof Symbol !== "undefined" && iter[Symbol.iterator] != null || iter["@@iterator"] != null) {
-        return Array.from(iter);
-    }
-}
 function _iterable_to_array_limit(arr, i) {
     var _i = arr == null ? null : typeof Symbol !== "undefined" && arr[Symbol.iterator] || arr["@@iterator"];
     if (_i == null) return;
@@ -87,15 +78,9 @@ function _iterable_to_array_limit(arr, i) {
 function _non_iterable_rest() {
     throw new TypeError("Invalid attempt to destructure non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method.");
 }
-function _non_iterable_spread() {
-    throw new TypeError("Invalid attempt to spread non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method.");
-}
 function _sliced_to_array(arr, i) {
     return _array_with_holes(arr) || _iterable_to_array_limit(arr, i) || _unsupported_iterable_to_array(arr, i) || _non_iterable_rest();
 }
-function _to_consumable_array(arr) {
-    return _array_without_holes(arr) || _iterable_to_array(arr) || _unsupported_iterable_to_array(arr) || _non_iterable_spread();
-}
 function _ts_generator(thisArg, body) {
     var f, y, t, _ = {
         label: 0,
@@ -203,13 +188,8 @@ function _unsupported_iterable_to_array(o, minLen) {
     if (n === "Map" || n === "Set") return Array.from(n);
     if (n === "Arguments" || /^(?:Ui|I)nt(?:8|16|32)(?:Clamped)?Array$/.test(n)) return _array_like_to_array(o, minLen);
 }
-// Mirrors the main columns onto the sidecars, so a title match found through title_seg ranks
-// like a title match found through title, not like a body match.
-var WEIGHTED_BM25 = 'bm25(content, 10.0, 5.0, 1.0, 0, 10.0, 5.0, 1.0)';
-var RRF_K = 60;
 // snippet() re-tokenizes each candidate doc, superlinearly: ~10s for one 1MB doc
 // (BENCHMARKING.md). Past this bound, rows get the linear JS excerpt instead.
-var SNIPPET_BOUND = 16384;
 var EXCERPT_WINDOW = 160;
 // Bare terms from an FTS5 query string: strips operators/quoting so the oversized-doc excerpt
 // scan matches the same words the query matched on, not FTS5 syntax.
@@ -346,14 +326,14 @@ function lineRangeFor(db, path, line) {
 }
 function search(_0, _1, _2) {
     return _async_to_generator(function(db, cfg, terms) {
-        var opts, _hits_get, _chunkLines_get, _chunkSimilarity_get, effective, k, allPaths, scopePaths, scopeActive, allowedPaths, fetch, wantsVectors, semanticEnabled, scope, whereJoin, whereCond, scopeCond, matchSql, query, matchRows, hits, oversized, candidates, edges, linked, seeds, ranked, chunkLines, chunkSimilarity, vec, insert, _iteratorNormalCompletion, _didIteratorError, _iteratorError, _iterator, _step, _step_value, path, c, where, similarityCol, rows, bareTerms, _iteratorNormalCompletion1, _didIteratorError1, _iteratorError1, _iterator1, _step1, row, text, _computeExcerpt, excerpt, offset;
+        var opts, _hits_get, _chunkLines_get, _chunkSimilarity_get, effective, k, signals, allPaths, scopePaths, scopeActive, allowedPaths, fetch, wantsVectors, e, semanticEnabled, scope, whereJoin, whereCond, scopeCond, query, candidates, matchRows, hits, oversized, chunkLines, chunkSimilarity, ref, insert, _iteratorNormalCompletion, _didIteratorError, _iteratorError, _iterator, _step, _step_value, path, c, where, similarityCol, rows, bareTerms, _iteratorNormalCompletion1, _didIteratorError1, _iteratorError1, _iterator1, _step1, row, text, _computeExcerpt, excerpt, offset;
         var _arguments = arguments;
         return _ts_generator(this, function(_state) {
             switch(_state.label){
                 case 0:
                     opts = _arguments.length > 3 && _arguments[3] !== void 0 ? _arguments[3] : {};
                     effective = (0, _indexts.resolveSearch)(cfg, opts);
-                    k = effective.k;
+                    k = effective.k, signals = effective.signals;
                     allPaths = db.prepare('SELECT "path" FROM frontmatter').all().map(function(r) {
                         return r.path;
                     });
@@ -365,11 +345,14 @@ function search(_0, _1, _2) {
                         throw (0, _searcherrorts.searchError)(err, terms, effective.where);
                     }
                     fetch = Math.max(k * 3, 30);
-                    // Asking for vectors without a model is a misconfiguration, not a mode: degrading would make
-                    // the same search answer differently before and after a download. semantic:false never asks.
-                    wantsVectors = effective.semantic && (0, _indexts.anyPresetEmbeds)(cfg);
-                    if (wantsVectors && !(0, _embedts.modelPresent)(cfg)) {
-                        throw new _errorsts.SenseError('EMBED_MODEL_MISSING', 'preset "'.concat(effective.presetName, '" searches with vectors, but the embedding model is not available; run `sense download`, or set "semantic": false on that preset to search on words and links'));
+                    // A downloadable HF id proceeds -- getProvider fetches it lazily on consent. Only a
+                    // local path with missing files errors here, since nothing will ever fetch it for itself.
+                    wantsVectors = signals.vectors !== undefined;
+                    if (wantsVectors) {
+                        e = (0, _indexts.embedConfig)(cfg); // validate.ts guarantees this is set whenever "vectors" is declared
+                        if ((0, _storets.localModelMissing)(e)) {
+                            throw new _errorsts.SenseError('EMBED_MODEL_MISSING', 'preset "'.concat(effective.presetName, '" searches with vectors, but the local model path "').concat(e.model, '" is missing ').concat(_storets.MODEL_FILENAMES, '; point embed.model at a directory containing them, or drop "vectors" from that preset\'s signals to search without them'));
+                        }
                     }
                     semanticEnabled = wantsVectors && (0, _scopets.scopeHasEmbeddings)(db, cfg, allowedPaths);
                     // Terms pass to MATCH as written, with one transform: a run of text whose language marks no
@@ -391,17 +374,17 @@ function search(_0, _1, _2) {
                     // thousands of paths, past SQLITE_MAX_VARIABLE_NUMBER on older builds.
                     if (scopeActive) (0, _scopets.materializeScope)(db, '_search_scope', scopePaths);
                     scopeCond = scopeActive ? 'AND content.path IN (SELECT "path" FROM _search_scope)' : '';
-                    // Docs past SNIPPET_BOUND skip snippet() entirely -- SQLite
-                    // short-circuits the untaken CASE branch, so it's never invoked on them.
-                    // Column 2 (text), not -1 (best column): -1 could surface a _seg sidecar as the excerpt,
-                    // which is machine-spaced and not what the author wrote. A row that matches only in a
-                    // sidecar gets an unhighlighted text excerpt below instead -- a stated cost.
-                    matchSql = "SELECT content.path AS path, CASE WHEN length(content.text) <= ".concat(SNIPPET_BOUND, " THEN snippet(content, 2, '\xab', '\xbb', '…', 10) ELSE NULL END AS hit FROM content ").concat(whereJoin, " WHERE content MATCH ? ").concat(whereCond, " ").concat(scopeCond, " ORDER BY ").concat(WEIGHTED_BM25, " LIMIT ").concat(fetch);
                     query = (0, _indexts.contentTokenize)(cfg) === undefined ? (0, _segmentts.segmentMatch)(terms) : terms;
-                    try {
-                        matchRows = db.prepare(matchSql).all(query);
-                    } catch (err) {
-                        throw (0, _searcherrorts.searchError)(err, terms, scope);
+                    candidates = new Map();
+                    matchRows = [];
+                    // A query that rewrites to nothing (e.g. only unspaced-script punctuation) has no
+                    // searchable words: zero lexical rows, and the other signals still compose.
+                    if (signals.words !== undefined && query.trim() !== '') {
+                        try {
+                            matchRows = (0, _signalsts.wordsCandidates)(db, candidates, query, whereJoin, whereCond, scopeCond, fetch, signals.words);
+                        } catch (err) {
+                            throw (0, _searcherrorts.searchError)(err, terms, scope);
+                        }
                     }
                     hits = new Map(matchRows.map(function(r) {
                         return [
@@ -416,46 +399,7 @@ function search(_0, _1, _2) {
                     }).map(function(r) {
                         return r.path;
                     }));
-                    candidates = new Map();
-                    matchRows.forEach(function(r, i) {
-                        candidates.set(r.path, {
-                            score: 1 / (RRF_K + i),
-                            via: 'match'
-                        });
-                    });
-                    edges = (0, _indexts.featureEnabled)(cfg, 'links') && matchRows.length > 0 ? (0, _indexts1.linkEdges)(db) : [];
-                    if (edges.length > 0) {
-                        // Gates the label only, not the score: restart mass ranks every seed without an incident
-                        // edge, but dropping it from the score reweights fusion (FEVER hit@10 0.997 -> 0.907).
-                        linked = new Set(edges.flat());
-                        seeds = new Map(matchRows.map(function(r, i) {
-                            return [
-                                r.path,
-                                1 / (i + 1)
-                            ];
-                        }));
-                        ranked = _to_consumable_array((0, _graphts.personalizedRank)(allPaths, edges, seeds)).filter(function(param) {
-                            var _param = _sliced_to_array(param, 2), path = _param[0], score = _param[1];
-                            return score > 1e-9 && allowedPaths.has(path);
-                        }).sort(function(a, b) {
-                            return b[1] - a[1];
-                        }).slice(0, fetch);
-                        ranked.forEach(function(param, i) {
-                            var _param = _sliced_to_array(param, 1), path = _param[0];
-                            var existing = candidates.get(path);
-                            if (existing) {
-                                existing.score += 1 / (RRF_K + i);
-                                if (linked.has(path)) existing.via = 'match+link';
-                            } else {
-                                candidates.set(path, {
-                                    score: 1 / (RRF_K + i),
-                                    via: 'link'
-                                });
-                            }
-                        });
-                    }
-                    // Vector expansion, invoked only: a third RRF list at the swept flat-region constants
-                    // (weight 1, pool = fetch). Each row carries its best chunk's line range.
+                    if (signals.links !== undefined) (0, _signalsts.linksCandidates)(db, candidates, matchRows, allPaths, allowedPaths, fetch, signals.links);
                     chunkLines = new Map();
                     chunkSimilarity = new Map();
                     if (!semanticEnabled) return [
@@ -464,25 +408,10 @@ function search(_0, _1, _2) {
                     ];
                     return [
                         4,
-                        (0, _embedts.semanticCandidates)(db, cfg, terms, fetch, allowedPaths)
+                        (0, _signalsts.vectorsCandidates)(db, cfg, candidates, terms, fetch, allowedPaths, signals.vectors)
                     ];
                 case 1:
-                    vec = _state.sent();
-                    vec.forEach(function(param, i) {
-                        var path = param.path, lines = param.lines, similarity = param.similarity;
-                        chunkLines.set(path, lines);
-                        chunkSimilarity.set(path, similarity);
-                        var existing = candidates.get(path);
-                        if (existing) {
-                            existing.score += 1 / (RRF_K + i);
-                            existing.via = "".concat(existing.via, "+vector");
-                        } else {
-                            candidates.set(path, {
-                                score: 1 / (RRF_K + i),
-                                via: 'vector'
-                            });
-                        }
-                    });
+                    ref = _state.sent(), chunkLines = ref.chunkLines, chunkSimilarity = ref.chunkSimilarity, ref;
                     _state.label = 2;
                 case 2:
                     db.exec('DROP TABLE IF EXISTS _search');

package/dist/cjs/commands/search.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/commands/search.ts"],"sourcesContent":["import { readFileSync } from 'node:fs';\nimport { join } from 'node:path';\nimport type { DatabaseSync } from 'node:sqlite';\nimport type { ResolvedConfig } from '../config/index.ts';\nimport { anyPresetEmbeds, contentTokenize, featureEnabled, resolveSearch } from '../config/index.ts';\nimport { SenseError } from '../errors.ts';\nimport { modelPresent, semanticCandidates } from '../features/embed.ts';\nimport { linkEdges } from '../features/index.ts';\nimport { personalizedRank } from '../graph.ts';\nimport type { Row } from '../output.ts';\nimport { searchError } from '../search-error.ts';\nimport { segmentMatch } from '../segment.ts';\nimport { materializeScope, narrowByWhere, rawScope, scopeHasEmbeddings } from './scope.ts';\n\n// Mirrors the main columns onto the sidecars, so a title match found through title_seg ranks\n// like a title match found through title, not like a body match.\nconst WEIGHTED_BM25 = 'bm25(content, 10.0, 5.0, 1.0, 0, 10.0, 5.0, 1.0)';\nconst RRF_K = 60;\n\n// snippet() re-tokenizes each candidate doc, superlinearly: ~10s for one 1MB doc\n// (BENCHMARKING.md). Past this bound, rows get the linear JS excerpt instead.\nconst SNIPPET_BOUND = 16_384;\nconst EXCERPT_WINDOW = 160;\n\n// Bare terms from an FTS5 query string: strips operators/quoting so the oversized-doc excerpt\n// scan matches the same words the query matched on, not FTS5 syntax.\nfunction extractBareTerms(query: string): string[] {\n  const cleaned = query\n    .replace(/\"/g, ' ')\n    .replace(/[()*]/g, ' ')\n    .replace(/\\b(AND|OR|NOT|NEAR)\\b(\\/\\d+)?/gi, ' ');\n  return cleaned\n    .split(/\\s+/)\n    .map((tok) => tok.replace(/^[A-Za-z_]\\w*:/, '')) // column filter, e.g. title:term\n    .map((tok) => tok.toLowerCase().trim())\n    .filter((tok) => tok.length > 0);\n}\n\nfunction findOccurrences(haystackLower: string, terms: string[]): Array<{ start: number; end: number; term: string }> {\n  const occ: Array<{ start: number; end: number; term: string }> = [];\n  for (const term of terms) {\n    let idx = 0;\n    for (;;) {\n      const found = haystackLower.indexOf(term, idx);\n      if (found === -1) break;\n      occ.push({ start: found, end: found + term.length, term });\n      idx = found + term.length;\n    }\n  }\n  // Longest span first at equal start, so \"tests\" beats its substring \"test\" and the\n  // whole word gets highlighted; the emit loop then absorbs the shorter overlap.\n  return occ.sort((a, b) => a.start - b.start || b.end - a.end);\n}\n\n// Slides a window over the sorted occurrences, same semantics as FTS5's own best-window\n// pick: most distinct terms, ties broken by most total hits.\nfunction bestWindowStart(occ: Array<{ start: number; end: number; term: string }>, windowSize: number): number {\n  let best = occ[0].start;\n  let bestDistinct = 0;\n  let bestCount = 0;\n  for (let i = 0; i < occ.length; i++) {\n    const winEnd = occ[i].start + windowSize;\n    const seen = new Set<string>();\n    let count = 0;\n    for (let j = i; j < occ.length && occ[j].start < winEnd; j++) {\n      seen.add(occ[j].term);\n      count++;\n    }\n    if (seen.size > bestDistinct || (seen.size === bestDistinct && count > bestCount)) {\n      best = occ[i].start;\n      bestDistinct = seen.size;\n      bestCount = count;\n    }\n  }\n  return best;\n}\n\n// snippet()-shaped excerpt: matched terms wrapped in «», … at cut edges. Linear in doc\n// length, only run for rows actually returned (<= k), unlike snippet()'s per-row cost.\nfunction computeExcerpt(text: string, terms: string[]): { excerpt: string; offset: number } {\n  const occ = findOccurrences(text.toLowerCase(), terms);\n  if (occ.length === 0) {\n    // This is a raw substring scan, not porter-stemmed: a doc matched only through a\n    // stemmed variant (query \"negotiate\" vs doc \"negotiating\") finds no occurrence here.\n    // Fall back to the doc's start, unmarked, rather than claim a match that isn't there.\n    const end = Math.min(text.length, EXCERPT_WINDOW);\n    return { excerpt: `${text.slice(0, end).replace(/\\s+/g, ' ').trim()}${end < text.length ? '…' : ''}`, offset: 0 };\n  }\n  const start = bestWindowStart(occ, EXCERPT_WINDOW);\n  const end = Math.min(text.length, start + EXCERPT_WINDOW);\n  let out = '';\n  let cursor = start;\n  for (const o of occ) {\n    if (o.start < start || o.end > end) continue;\n    // Occurrences of duplicate or substring-overlapping terms (\"test tests\") can overlap;\n    // emitting each would duplicate document text. Keep the first, absorb the rest.\n    if (o.start < cursor) continue;\n    out += `${text.slice(cursor, o.start)}«${text.slice(o.start, o.end)}»`;\n    cursor = o.end;\n  }\n  out += text.slice(cursor, end);\n  const prefix = start > 0 ? '…' : '';\n  const suffix = end < text.length ? '…' : '';\n  return { excerpt: `${prefix}${out.replace(/\\s+/g, ' ')}${suffix}`, offset: start };\n}\n\nfunction lineNumberAt(text: string, offset: number): number {\n  let line = 1;\n  for (let i = 0; i < offset; i++) if (text.charCodeAt(i) === 10) line++;\n  return line;\n}\n\nfunction lineRangeFor(db: DatabaseSync, path: string, line: number): string | null {\n  const row = db.prepare('SELECT start_line, end_line FROM sections WHERE \"path\" = ? AND start_line <= ? AND end_line >= ? ORDER BY start_line DESC LIMIT 1').get(path, line, line) as { start_line: number; end_line: number } | undefined;\n  return row ? `L${row.start_line}-${row.end_line}` : null;\n}\n\nexport interface SearchOptions {\n  k?: number;\n  where?: string; // SQL fragment against frontmatter alias `f`, e.g. \"f.status = 'active'\"\n  preset?: string; // named preset; unknown name throws listing declared presets, undefined -> \"default\"\n  include?: string[]; // ad hoc scope override (repeatable --include); independent of exclude\n  exclude?: string[]; // ad hoc scope override (repeatable --exclude); independent of include\n  noExclude?: boolean; // --no-exclude: drop the preset's exclude for this command\n}\n\n// BM25 + link expansion + vectors, fused by reciprocal rank; `via` names the signal per row.\n// `opts` arrives already resolved (config.ts:resolveSearch).\nexport async function search(db: DatabaseSync, cfg: ResolvedConfig, terms: string, opts: SearchOptions = {}): Promise<Row[]> {\n  const effective = resolveSearch(cfg, opts);\n  const { k } = effective;\n\n  const allPaths = (db.prepare('SELECT \"path\" FROM frontmatter').all() as Array<{ path: string }>).map((r) => r.path);\n  const scopePaths = rawScope(db, cfg, opts, allPaths);\n  const scopeActive = scopePaths.size < allPaths.length;\n  // The set every candidate pool must be filtered to before truncation: scope narrowed by\n  // --where, the same composition scopedPaths() gives the other commands. Runs the same where\n  // fragment matchSql does, so a bad column gets the same attributed error either way.\n  let allowedPaths: Set<string>;\n  try {\n    allowedPaths = narrowByWhere(db, scopePaths, effective.where);\n  } catch (err) {\n    throw searchError(err as Error, terms, effective.where);\n  }\n  const fetch = Math.max(k * 3, 30);\n\n  // Asking for vectors without a model is a misconfiguration, not a mode: degrading would make\n  // the same search answer differently before and after a download. semantic:false never asks.\n  const wantsVectors = effective.semantic && anyPresetEmbeds(cfg);\n  if (wantsVectors && !modelPresent(cfg)) {\n    throw new SenseError('EMBED_MODEL_MISSING', `preset \"${effective.presetName}\" searches with vectors, but the embedding model is not available; run \\`sense download\\`, or set \"semantic\": false on that preset to search on words and links`);\n  }\n  const semanticEnabled = wantsVectors && scopeHasEmbeddings(db, cfg, allowedPaths);\n\n  // Terms pass to MATCH as written, with one transform: a run of text whose language marks no\n  // word boundaries becomes a quoted grapheme phrase (and a title:/summary:/text: qualifier\n  // ahead of it retargets its _seg column), matching how it was indexed. Text that already\n  // carries boundaries comes through untouched, so this is a no-op for the languages that\n  // never needed it. Gated on the same predicate that decided whether the sidecars were\n  // populated (contentTokenize(cfg) === undefined), so index and query can't disagree. Invalid\n  // syntax still errors rather than being rewritten.\n  // --where\n  // applies inside the candidate query (a post-filter would drop matches ranked past the pool)\n  // and again on the final select, for link-derived rows.\n  const scope = effective.where;\n  const whereJoin = scope ? `JOIN frontmatter f ON f.\"path\" = content.path` : '';\n  const whereCond = scope ? `AND (${scope})` : '';\n  // A scope narrower than the whole index must filter the candidate pool before LIMIT, not\n  // after -- otherwise scoped notes ranking below the global top-`fetch` never reach the\n  // filter. A join against a temp table, not a bound parameter list: real scopes run to\n  // thousands of paths, past SQLITE_MAX_VARIABLE_NUMBER on older builds.\n  if (scopeActive) materializeScope(db, '_search_scope', scopePaths);\n  const scopeCond = scopeActive ? `AND content.path IN (SELECT \"path\" FROM _search_scope)` : '';\n  // Docs past SNIPPET_BOUND skip snippet() entirely -- SQLite\n  // short-circuits the untaken CASE branch, so it's never invoked on them.\n  // Column 2 (text), not -1 (best column): -1 could surface a _seg sidecar as the excerpt,\n  // which is machine-spaced and not what the author wrote. A row that matches only in a\n  // sidecar gets an unhighlighted text excerpt below instead -- a stated cost.\n  const matchSql = `SELECT content.path AS path, CASE WHEN length(content.text) <= ${SNIPPET_BOUND} THEN snippet(content, 2, '«', '»', '…', 10) ELSE NULL END AS hit FROM content ${whereJoin} WHERE content MATCH ? ${whereCond} ${scopeCond} ORDER BY ${WEIGHTED_BM25} LIMIT ${fetch}`;\n  const query = contentTokenize(cfg) === undefined ? segmentMatch(terms) : terms;\n  let matchRows: Array<{ path: string; hit: string | null }>;\n  try {\n    matchRows = db.prepare(matchSql).all(query) as Array<{ path: string; hit: string | null }>;\n  } catch (err) {\n    throw searchError(err as Error, terms, scope);\n  }\n\n  const hits = new Map(matchRows.map((r) => [r.path, r.hit]));\n  // hit === null here means the bound suppressed snippet(), not \"no match\" -- distinguish\n  // from via='link' rows (never in matchRows, so absent from this set) below.\n  const oversized = new Set(matchRows.filter((r) => r.hit === null).map((r) => r.path));\n  const candidates = new Map<string, { score: number; via: string }>();\n  matchRows.forEach((r, i) => {\n    candidates.set(r.path, { score: 1 / (RRF_K + i), via: 'match' });\n  });\n\n  const edges = featureEnabled(cfg, 'links') && matchRows.length > 0 ? linkEdges(db) : [];\n  if (edges.length > 0) {\n    // Gates the label only, not the score: restart mass ranks every seed without an incident\n    // edge, but dropping it from the score reweights fusion (FEVER hit@10 0.997 -> 0.907).\n    const linked = new Set(edges.flat());\n    const seeds = new Map(matchRows.map((r, i) => [r.path, 1 / (i + 1)]));\n    const ranked = [...personalizedRank(allPaths, edges, seeds)]\n      .filter(([path, score]) => score > 1e-9 && allowedPaths.has(path))\n      .sort((a, b) => b[1] - a[1])\n      .slice(0, fetch);\n    ranked.forEach(([path], i) => {\n      const existing = candidates.get(path);\n      if (existing) {\n        existing.score += 1 / (RRF_K + i);\n        if (linked.has(path)) existing.via = 'match+link';\n      } else {\n        candidates.set(path, { score: 1 / (RRF_K + i), via: 'link' });\n      }\n    });\n  }\n\n  // Vector expansion, invoked only: a third RRF list at the swept flat-region constants\n  // (weight 1, pool = fetch). Each row carries its best chunk's line range.\n  const chunkLines = new Map<string, string>();\n  const chunkSimilarity = new Map<string, number>();\n  if (semanticEnabled) {\n    const vec = await semanticCandidates(db, cfg, terms, fetch, allowedPaths);\n    vec.forEach(({ path, lines, similarity }, i) => {\n      chunkLines.set(path, lines);\n      chunkSimilarity.set(path, similarity);\n      const existing = candidates.get(path);\n      if (existing) {\n        existing.score += 1 / (RRF_K + i);\n        existing.via = `${existing.via}+vector`;\n      } else {\n        candidates.set(path, { score: 1 / (RRF_K + i), via: 'vector' });\n      }\n    });\n  }\n\n  db.exec('DROP TABLE IF EXISTS _search');\n  db.exec('CREATE TEMP TABLE _search (\"path\" TEXT PRIMARY KEY, score REAL, via TEXT, hit TEXT, lines TEXT, similarity REAL)');\n  const insert = db.prepare('INSERT INTO _search (\"path\", score, via, hit, lines, similarity) VALUES (?, ?, ?, ?, ?, ?)');\n  for (const [path, c] of candidates) insert.run(path, c.score, c.via, hits.get(path) ?? null, chunkLines.get(path) ?? null, chunkSimilarity.get(path) ?? null);\n\n  // All three candidate paths (match, link, vector) already filtered to scope+where before\n  // reaching _search; this reapplies --where anyway since the join to frontmatter is already\n  // needed for the path column.\n  const where = scope ? `WHERE (${scope})` : '';\n  // lines is always present now: semantic rows carry their chunk's range, oversized-doc\n  // lexical rows gain one below, everything else stays null. similarity stays semantic-only.\n  const similarityCol = semanticEnabled ? ', _search.similarity' : '';\n  const rows = db\n    .prepare(\n      `SELECT f.\"path\" AS path, content.title, content.summary, _search.hit, _search.via, round(_search.score, 4) AS score, _search.lines${similarityCol}\n       FROM _search JOIN frontmatter f ON f.\"path\" = _search.\"path\" JOIN content ON content.path = _search.\"path\"\n       ${where} ORDER BY _search.score DESC LIMIT ?`\n    )\n    .all(k) as Row[];\n\n  if (oversized.size > 0) {\n    const bareTerms = extractBareTerms(terms);\n    for (const row of rows) {\n      if (row.hit !== null || !oversized.has(row.path as string)) continue;\n      let text: string;\n      try {\n        text = readFileSync(join(cfg.baseDir, row.path as string), 'utf8');\n      } catch {\n        continue; // vanished since the match; leave hit/lines null rather than throw\n      }\n      const { excerpt, offset } = computeExcerpt(text, bareTerms);\n      row.hit = excerpt;\n      if (row.lines == null) row.lines = featureEnabled(cfg, 'sections') ? lineRangeFor(db, row.path as string, lineNumberAt(text, offset)) : null;\n    }\n  }\n\n  return rows;\n}\n"],"names":["search","WEIGHTED_BM25","RRF_K","SNIPPET_BOUND","EXCERPT_WINDOW","extractBareTerms","query","cleaned","replace","split","map","tok","toLowerCase","trim","filter","length","findOccurrences","haystackLower","terms","occ","term","idx","found","indexOf","push","start","end","sort","a","b","bestWindowStart","windowSize","best","bestDistinct","bestCount","i","winEnd","seen","Set","count","j","add","size","computeExcerpt","text","Math","min","excerpt","slice","offset","out","cursor","o","prefix","suffix","lineNumberAt","line","charCodeAt","lineRangeFor","db","path","row","prepare","get","start_line","end_line","cfg","opts","hits","chunkLines","chunkSimilarity","effective","k","allPaths","scopePaths","scopeActive","allowedPaths","fetch","wantsVectors","semanticEnabled","scope","whereJoin","whereCond","scopeCond","matchSql","matchRows","oversized","candidates","edges","linked","seeds","ranked","vec","insert","c","where","similarityCol","rows","bareTerms","resolveSearch","all","r","rawScope","narrowByWhere","err","searchError","max","semantic","anyPresetEmbeds","modelPresent","SenseError","presetName","scopeHasEmbeddings","materializeScope","contentTokenize","undefined","segmentMatch","Map","hit","forEach","set","score","via","featureEnabled","linkEdges","flat","personalizedRank","has","existing","semanticCandidates","lines","similarity","exec","run","readFileSync","join","baseDir"],"mappings":";;;;+BAgIsBA;;;eAAAA;;;sBAhIO;wBACR;uBAG2D;wBACrD;uBACsB;wBACvB;uBACO;6BAEL;yBACC;uBACiD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAE9E,6FAA6F;AAC7F,iEAAiE;AACjE,IAAMC,gBAAgB;AACtB,IAAMC,QAAQ;AAEd,iFAAiF;AACjF,8EAA8E;AAC9E,IAAMC,gBAAgB;AACtB,IAAMC,iBAAiB;AAEvB,8FAA8F;AAC9F,qEAAqE;AACrE,SAASC,iBAAiBC,KAAa;IACrC,IAAMC,UAAUD,MACbE,OAAO,CAAC,MAAM,KACdA,OAAO,CAAC,UAAU,KAClBA,OAAO,CAAC,mCAAmC;IAC9C,OAAOD,QACJE,KAAK,CAAC,OACNC,GAAG,CAAC,SAACC;eAAQA,IAAIH,OAAO,CAAC,kBAAkB;OAAK,iCAAiC;KACjFE,GAAG,CAAC,SAACC;eAAQA,IAAIC,WAAW,GAAGC,IAAI;OACnCC,MAAM,CAAC,SAACH;eAAQA,IAAII,MAAM,GAAG;;AAClC;AAEA,SAASC,gBAAgBC,aAAqB,EAAEC,KAAe;IAC7D,IAAMC,MAA2D,EAAE;QAC9D,kCAAA,2BAAA;;QAAL,QAAK,YAAcD,0BAAd,SAAA,6BAAA,QAAA,yBAAA,iCAAqB;YAArB,IAAME,OAAN;YACH,IAAIC,MAAM;YACV,OAAS;gBACP,IAAMC,QAAQL,cAAcM,OAAO,CAACH,MAAMC;gBAC1C,IAAIC,UAAU,CAAC,GAAG;gBAClBH,IAAIK,IAAI,CAAC;oBAAEC,OAAOH;oBAAOI,KAAKJ,QAAQF,KAAKL,MAAM;oBAAEK,MAAAA;gBAAK;gBACxDC,MAAMC,QAAQF,KAAKL,MAAM;YAC3B;QACF;;QARK;QAAA;;;iBAAA,6BAAA;gBAAA;;;gBAAA;sBAAA;;;;IASL,mFAAmF;IACnF,+EAA+E;IAC/E,OAAOI,IAAIQ,IAAI,CAAC,SAACC,GAAGC;eAAMD,EAAEH,KAAK,GAAGI,EAAEJ,KAAK,IAAII,EAAEH,GAAG,GAAGE,EAAEF,GAAG;;AAC9D;AAEA,wFAAwF;AACxF,6DAA6D;AAC7D,SAASI,gBAAgBX,GAAwD,EAAEY,UAAkB;IACnG,IAAIC,OAAOb,GAAG,CAAC,EAAE,CAACM,KAAK;IACvB,IAAIQ,eAAe;IACnB,IAAIC,YAAY;IAChB,IAAK,IAAIC,IAAI,GAAGA,IAAIhB,IAAIJ,MAAM,EAAEoB,IAAK;QACnC,IAAMC,SAASjB,GAAG,CAACgB,EAAE,CAACV,KAAK,GAAGM;QAC9B,IAAMM,OAAO,IAAIC;QACjB,IAAIC,QAAQ;QACZ,IAAK,IAAIC,IAAIL,GAAGK,IAAIrB,IAAIJ,MAAM,IAAII,GAAG,CAACqB,EAAE,CAACf,KAAK,GAAGW,QAAQI,IAAK;YAC5DH,KAAKI,GAAG,CAACtB,GAAG,CAACqB,EAAE,CAACpB,IAAI;YACpBmB;QACF;QACA,IAAIF,KAAKK,IAAI,GAAGT,gBAAiBI,KAAKK,IAAI,KAAKT,gBAAgBM,QAAQL,WAAY;YACjFF,OAAOb,GAAG,CAACgB,EAAE,CAACV,KAAK;YACnBQ,eAAeI,KAAKK,IAAI;YACxBR,YAAYK;QACd;IACF;IACA,OAAOP;AACT;AAEA,uFAAuF;AACvF,uFAAuF;AACvF,SAASW,eAAeC,IAAY,EAAE1B,KAAe;IACnD,IAAMC,MAAMH,gBAAgB4B,KAAKhC,WAAW,IAAIM;IAChD,IAAIC,IAAIJ,MAAM,KAAK,GAAG;QACpB,iFAAiF;QACjF,qFAAqF;QACrF,sFAAsF;QACtF,IAAMW,MAAMmB,KAAKC,GAAG,CAACF,KAAK7B,MAAM,EAAEX;QAClC,OAAO;YAAE2C,SAAS,AAAC,GAAmDrB,OAAjDkB,KAAKI,KAAK,CAAC,GAAGtB,KAAKlB,OAAO,CAAC,QAAQ,KAAKK,IAAI,IAAkC,OAA7Ba,MAAMkB,KAAK7B,MAAM,GAAG,MAAM;YAAMkC,QAAQ;QAAE;IAClH;IACA,IAAMxB,QAAQK,gBAAgBX,KAAKf;IACnC,IAAMsB,OAAMmB,KAAKC,GAAG,CAACF,KAAK7B,MAAM,EAAEU,QAAQrB;IAC1C,IAAI8C,MAAM;IACV,IAAIC,SAAS1B;QACR,kCAAA,2BAAA;;QAAL,QAAK,YAAWN,wBAAX,SAAA,6BAAA,QAAA,yBAAA,iCAAgB;YAAhB,IAAMiC,IAAN;YACH,IAAIA,EAAE3B,KAAK,GAAGA,SAAS2B,EAAE1B,GAAG,GAAGA,MAAK;YACpC,sFAAsF;YACtF,gFAAgF;YAChF,IAAI0B,EAAE3B,KAAK,GAAG0B,QAAQ;YACtBD,OAAO,AAAC,GAAiCN,OAA/BA,KAAKI,KAAK,CAACG,QAAQC,EAAE3B,KAAK,GAAE,QAA8B,OAA3BmB,KAAKI,KAAK,CAACI,EAAE3B,KAAK,EAAE2B,EAAE1B,GAAG,GAAE;YACpEyB,SAASC,EAAE1B,GAAG;QAChB;;QAPK;QAAA;;;iBAAA,6BAAA;gBAAA;;;gBAAA;sBAAA;;;;IAQLwB,OAAON,KAAKI,KAAK,CAACG,QAAQzB;IAC1B,IAAM2B,SAAS5B,QAAQ,IAAI,MAAM;IACjC,IAAM6B,SAAS5B,OAAMkB,KAAK7B,MAAM,GAAG,MAAM;IACzC,OAAO;QAAEgC,SAAS,AAAC,GAAWG,OAATG,QAAoCC,OAA3BJ,IAAI1C,OAAO,CAAC,QAAQ,MAAc,OAAP8C;QAAUL,QAAQxB;IAAM;AACnF;AAEA,SAAS8B,aAAaX,IAAY,EAAEK,MAAc;IAChD,IAAIO,OAAO;IACX,IAAK,IAAIrB,IAAI,GAAGA,IAAIc,QAAQd,IAAK,IAAIS,KAAKa,UAAU,CAACtB,OAAO,IAAIqB;IAChE,OAAOA;AACT;AAEA,SAASE,aAAaC,EAAgB,EAAEC,IAAY,EAAEJ,IAAY;IAChE,IAAMK,MAAMF,GAAGG,OAAO,CAAC,qIAAqIC,GAAG,CAACH,MAAMJ,MAAMA;IAC5K,OAAOK,MAAM,AAAC,IAAqBA,OAAlBA,IAAIG,UAAU,EAAC,KAAgB,OAAbH,IAAII,QAAQ,IAAK;AACtD;AAaO,SAAejE;wCAAO2D,EAAgB,EAAEO,GAAmB,EAAEhD,KAAa;YAAEiD,MA+GZC,WAAwBC,iBAA8BC,sBA9GrHC,WACEC,GAEFC,UACAC,YACAC,aAIFC,cAMEC,OAIAC,cAIAC,iBAYAC,OACAC,WACAC,WAMAC,WAMAC,UACA9E,OACF+E,WAOEjB,MAGAkB,WACAC,YAKAC,OAIEC,QACAC,OACAC,QAiBFtB,YACAC,iBAEEsB,KAgBFC,QACD,2BAAA,mBAAA,gBAAA,WAAA,oBAAOjC,MAAMkC,GAKZC,OAGAC,eACAC,MASEC,WACD,4BAAA,oBAAA,iBAAA,YAAA,QAAMrC,KAELjB,MAMwBD,iBAApBI,SAASE;;;;;oBA1I4DkB,OAAAA,oEAAsB,CAAC;oBAClGI,YAAY4B,IAAAA,sBAAa,EAACjC,KAAKC;oBAC7BK,IAAMD,UAANC;oBAEFC,WAAW,AAACd,GAAGG,OAAO,CAAC,kCAAkCsC,GAAG,GAA+B1F,GAAG,CAAC,SAAC2F;+BAAMA,EAAEzC,IAAI;;oBAC5Gc,aAAa4B,IAAAA,iBAAQ,EAAC3C,IAAIO,KAAKC,MAAMM;oBACrCE,cAAcD,WAAWhC,IAAI,GAAG+B,SAAS1D,MAAM;oBAKrD,IAAI;wBACF6D,eAAe2B,IAAAA,sBAAa,EAAC5C,IAAIe,YAAYH,UAAUwB,KAAK;oBAC9D,EAAE,OAAOS,KAAK;wBACZ,MAAMC,IAAAA,0BAAW,EAACD,KAActF,OAAOqD,UAAUwB,KAAK;oBACxD;oBACMlB,QAAQhC,KAAK6D,GAAG,CAAClC,IAAI,GAAG;oBAE9B,6FAA6F;oBAC7F,6FAA6F;oBACvFM,eAAeP,UAAUoC,QAAQ,IAAIC,IAAAA,wBAAe,EAAC1C;oBAC3D,IAAIY,gBAAgB,CAAC+B,IAAAA,qBAAY,EAAC3C,MAAM;wBACtC,MAAM,IAAI4C,oBAAU,CAAC,uBAAuB,AAAC,WAA+B,OAArBvC,UAAUwC,UAAU,EAAC;oBAC9E;oBACMhC,kBAAkBD,gBAAgBkC,IAAAA,2BAAkB,EAACrD,IAAIO,KAAKU;oBAEpE,4FAA4F;oBAC5F,0FAA0F;oBAC1F,yFAAyF;oBACzF,wFAAwF;oBACxF,sFAAsF;oBACtF,6FAA6F;oBAC7F,mDAAmD;oBACnD,UAAU;oBACV,6FAA6F;oBAC7F,wDAAwD;oBAClDI,QAAQT,UAAUwB,KAAK;oBACvBd,YAAYD,QAAQ,kDAAkD;oBACtEE,YAAYF,QAAQ,AAAC,QAAa,OAANA,OAAM,OAAK;oBAC7C,yFAAyF;oBACzF,uFAAuF;oBACvF,sFAAsF;oBACtF,uEAAuE;oBACvE,IAAIL,aAAasC,IAAAA,yBAAgB,EAACtD,IAAI,iBAAiBe;oBACjDS,YAAYR,cAAc,2DAA2D;oBAC3F,4DAA4D;oBAC5D,yEAAyE;oBACzE,yFAAyF;oBACzF,sFAAsF;oBACtF,6EAA6E;oBACvES,WAAW,AAAC,kEAAgKH,OAA/F9E,eAAc,yFAAoH+E,OAAnCD,WAAU,2BAAsCE,OAAbD,WAAU,KAAyBjF,OAAtBkF,WAAU,cAAmCN,OAAvB5E,eAAc,WAAe,OAAN4E;oBACzQvE,QAAQ4G,IAAAA,wBAAe,EAAChD,SAASiD,YAAYC,IAAAA,uBAAY,EAAClG,SAASA;oBAEzE,IAAI;wBACFmE,YAAY1B,GAAGG,OAAO,CAACsB,UAAUgB,GAAG,CAAC9F;oBACvC,EAAE,OAAOkG,KAAK;wBACZ,MAAMC,IAAAA,0BAAW,EAACD,KAActF,OAAO8D;oBACzC;oBAEMZ,OAAO,IAAIiD,IAAIhC,UAAU3E,GAAG,CAAC,SAAC2F;+BAAM;4BAACA,EAAEzC,IAAI;4BAAEyC,EAAEiB,GAAG;yBAAC;;oBACzD,wFAAwF;oBACxF,4EAA4E;oBACtEhC,YAAY,IAAIhD,IAAI+C,UAAUvE,MAAM,CAAC,SAACuF;+BAAMA,EAAEiB,GAAG,KAAK;uBAAM5G,GAAG,CAAC,SAAC2F;+BAAMA,EAAEzC,IAAI;;oBAC7E2B,aAAa,IAAI8B;oBACvBhC,UAAUkC,OAAO,CAAC,SAAClB,GAAGlE;wBACpBoD,WAAWiC,GAAG,CAACnB,EAAEzC,IAAI,EAAE;4BAAE6D,OAAO,IAAKvH,CAAAA,QAAQiC,CAAAA;4BAAIuF,KAAK;wBAAQ;oBAChE;oBAEMlC,QAAQmC,IAAAA,uBAAc,EAACzD,KAAK,YAAYmB,UAAUtE,MAAM,GAAG,IAAI6G,IAAAA,mBAAS,EAACjE;oBAC/E,IAAI6B,MAAMzE,MAAM,GAAG,GAAG;wBACpB,yFAAyF;wBACzF,uFAAuF;wBACjF0E,SAAS,IAAInD,IAAIkD,MAAMqC,IAAI;wBAC3BnC,QAAQ,IAAI2B,IAAIhC,UAAU3E,GAAG,CAAC,SAAC2F,GAAGlE;mCAAM;gCAACkE,EAAEzC,IAAI;gCAAE,IAAKzB,CAAAA,IAAI,CAAA;6BAAG;;wBAC7DwD,SAAS,AAAC,qBAAGmC,IAAAA,yBAAgB,EAACrD,UAAUe,OAAOE,QAClD5E,MAAM,CAAC;qEAAE8C,kBAAM6D;mCAAWA,QAAQ,QAAQ7C,aAAamD,GAAG,CAACnE;2BAC3DjC,IAAI,CAAC,SAACC,GAAGC;mCAAMA,CAAC,CAAC,EAAE,GAAGD,CAAC,CAAC,EAAE;2BAC1BoB,KAAK,CAAC,GAAG6B;wBACZc,OAAO4B,OAAO,CAAC,gBAASpF;qEAAPyB;4BACf,IAAMoE,WAAWzC,WAAWxB,GAAG,CAACH;4BAChC,IAAIoE,UAAU;gCACZA,SAASP,KAAK,IAAI,IAAKvH,CAAAA,QAAQiC,CAAAA;gCAC/B,IAAIsD,OAAOsC,GAAG,CAACnE,OAAOoE,SAASN,GAAG,GAAG;4BACvC,OAAO;gCACLnC,WAAWiC,GAAG,CAAC5D,MAAM;oCAAE6D,OAAO,IAAKvH,CAAAA,QAAQiC,CAAAA;oCAAIuF,KAAK;gCAAO;4BAC7D;wBACF;oBACF;oBAEA,sFAAsF;oBACtF,0EAA0E;oBACpErD,aAAa,IAAIgD;oBACjB/C,kBAAkB,IAAI+C;yBACxBtC,iBAAAA;;;;oBACU;;wBAAMkD,IAAAA,2BAAkB,EAACtE,IAAIO,KAAKhD,OAAO2D,OAAOD;;;oBAAtDgB,MAAM;oBACZA,IAAI2B,OAAO,CAAC,gBAA8BpF;4BAA3ByB,aAAAA,MAAMsE,cAAAA,OAAOC,mBAAAA;wBAC1B9D,WAAWmD,GAAG,CAAC5D,MAAMsE;wBACrB5D,gBAAgBkD,GAAG,CAAC5D,MAAMuE;wBAC1B,IAAMH,WAAWzC,WAAWxB,GAAG,CAACH;wBAChC,IAAIoE,UAAU;4BACZA,SAASP,KAAK,IAAI,IAAKvH,CAAAA,QAAQiC,CAAAA;4BAC/B6F,SAASN,GAAG,GAAG,AAAC,GAAe,OAAbM,SAASN,GAAG,EAAC;wBACjC,OAAO;4BACLnC,WAAWiC,GAAG,CAAC5D,MAAM;gCAAE6D,OAAO,IAAKvH,CAAAA,QAAQiC,CAAAA;gCAAIuF,KAAK;4BAAS;wBAC/D;oBACF;;;oBAGF/D,GAAGyE,IAAI,CAAC;oBACRzE,GAAGyE,IAAI,CAAC;oBACFvC,SAASlC,GAAGG,OAAO,CAAC;oBACrB,kCAAA,2BAAA;;wBAAL,IAAK,YAAmByB,iCAAnB,6BAAA,QAAA,yBAAA;2DAAA,iBAAO3B,uBAAMkC;4BAAkBD,OAAOwC,GAAG,CAACzE,MAAMkC,EAAE2B,KAAK,EAAE3B,EAAE4B,GAAG,GAAEtD,YAAAA,KAAKL,GAAG,CAACH,mBAATQ,uBAAAA,YAAkB,OAAMC,kBAAAA,WAAWN,GAAG,CAACH,mBAAfS,6BAAAA,kBAAwB,OAAMC,uBAAAA,gBAAgBP,GAAG,CAACH,mBAApBU,kCAAAA,uBAA6B;;;wBAAnJ;wBAAA;;;iCAAA,6BAAA;gCAAA;;;gCAAA;sCAAA;;;;oBAEL,yFAAyF;oBACzF,2FAA2F;oBAC3F,8BAA8B;oBACxByB,QAAQf,QAAQ,AAAC,UAAe,OAANA,OAAM,OAAK;oBAC3C,sFAAsF;oBACtF,2FAA2F;oBACrFgB,gBAAgBjB,kBAAkB,yBAAyB;oBAC3DkB,OAAOtC,GACVG,OAAO,CACN,AAAC,qIAEEiC,OAFkIC,eAAc,gIAE1I,OAAND,OAAM,yCAEVK,GAAG,CAAC5B;oBAEP,IAAIc,UAAU5C,IAAI,GAAG,GAAG;wBAChBwD,YAAY7F,iBAAiBa;wBAC9B,mCAAA,4BAAA;;4BAAL,IAAK,aAAa+E,2BAAb,8BAAA,SAAA,0BAAA,kCAAmB;gCAAbpC,MAAN;gCACH,IAAIA,IAAIyD,GAAG,KAAK,QAAQ,CAAChC,UAAUyC,GAAG,CAAClE,IAAID,IAAI,GAAa;gCACxDhB,OAAAA,KAAAA;gCACJ,IAAI;oCACFA,OAAO0F,IAAAA,oBAAY,EAACC,IAAAA,cAAI,EAACrE,IAAIsE,OAAO,EAAE3E,IAAID,IAAI,GAAa;gCAC7D,EAAE,eAAM;oCACN,UAAU,mEAAmE;gCAC/E;gCAC4BjB,kBAAAA,eAAeC,MAAMsD,YAAzCnD,UAAoBJ,gBAApBI,SAASE,SAAWN,gBAAXM;gCACjBY,IAAIyD,GAAG,GAAGvE;gCACV,IAAIc,IAAIqE,KAAK,IAAI,MAAMrE,IAAIqE,KAAK,GAAGP,IAAAA,uBAAc,EAACzD,KAAK,cAAcR,aAAaC,IAAIE,IAAID,IAAI,EAAYL,aAAaX,MAAMK,WAAW;4BAC1I;;4BAXK;4BAAA;;;qCAAA,8BAAA;oCAAA;;;oCAAA;0CAAA;;;;oBAYP;oBAEA;;wBAAOgD;;;;IACT"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/commands/search.ts"],"sourcesContent":["import { readFileSync } from 'node:fs';\nimport { join } from 'node:path';\nimport type { DatabaseSync } from 'node:sqlite';\nimport type { ResolvedConfig } from '../config/index.ts';\nimport { contentTokenize, embedConfig, featureEnabled, resolveSearch } from '../config/index.ts';\nimport { localModelMissing, MODEL_FILENAMES } from '../embed/store.ts';\nimport { SenseError } from '../errors.ts';\nimport type { Row } from '../output.ts';\nimport { searchError } from '../search-error.ts';\nimport { segmentMatch } from '../segment.ts';\nimport { materializeScope, narrowByWhere, rawScope, scopeHasEmbeddings } from './scope.ts';\nimport { linksCandidates, vectorsCandidates, wordsCandidates } from './signals.ts';\n\n// snippet() re-tokenizes each candidate doc, superlinearly: ~10s for one 1MB doc\n// (BENCHMARKING.md). Past this bound, rows get the linear JS excerpt instead.\nconst EXCERPT_WINDOW = 160;\n\n// Bare terms from an FTS5 query string: strips operators/quoting so the oversized-doc excerpt\n// scan matches the same words the query matched on, not FTS5 syntax.\nfunction extractBareTerms(query: string): string[] {\n  const cleaned = query\n    .replace(/\"/g, ' ')\n    .replace(/[()*]/g, ' ')\n    .replace(/\\b(AND|OR|NOT|NEAR)\\b(\\/\\d+)?/gi, ' ');\n  return cleaned\n    .split(/\\s+/)\n    .map((tok) => tok.replace(/^[A-Za-z_]\\w*:/, '')) // column filter, e.g. title:term\n    .map((tok) => tok.toLowerCase().trim())\n    .filter((tok) => tok.length > 0);\n}\n\nfunction findOccurrences(haystackLower: string, terms: string[]): Array<{ start: number; end: number; term: string }> {\n  const occ: Array<{ start: number; end: number; term: string }> = [];\n  for (const term of terms) {\n    let idx = 0;\n    for (;;) {\n      const found = haystackLower.indexOf(term, idx);\n      if (found === -1) break;\n      occ.push({ start: found, end: found + term.length, term });\n      idx = found + term.length;\n    }\n  }\n  // Longest span first at equal start, so \"tests\" beats its substring \"test\" and the\n  // whole word gets highlighted; the emit loop then absorbs the shorter overlap.\n  return occ.sort((a, b) => a.start - b.start || b.end - a.end);\n}\n\n// Slides a window over the sorted occurrences, same semantics as FTS5's own best-window\n// pick: most distinct terms, ties broken by most total hits.\nfunction bestWindowStart(occ: Array<{ start: number; end: number; term: string }>, windowSize: number): number {\n  let best = occ[0].start;\n  let bestDistinct = 0;\n  let bestCount = 0;\n  for (let i = 0; i < occ.length; i++) {\n    const winEnd = occ[i].start + windowSize;\n    const seen = new Set<string>();\n    let count = 0;\n    for (let j = i; j < occ.length && occ[j].start < winEnd; j++) {\n      seen.add(occ[j].term);\n      count++;\n    }\n    if (seen.size > bestDistinct || (seen.size === bestDistinct && count > bestCount)) {\n      best = occ[i].start;\n      bestDistinct = seen.size;\n      bestCount = count;\n    }\n  }\n  return best;\n}\n\n// snippet()-shaped excerpt: matched terms wrapped in «», … at cut edges. Linear in doc\n// length, only run for rows actually returned (<= k), unlike snippet()'s per-row cost.\nfunction computeExcerpt(text: string, terms: string[]): { excerpt: string; offset: number } {\n  const occ = findOccurrences(text.toLowerCase(), terms);\n  if (occ.length === 0) {\n    // This is a raw substring scan, not porter-stemmed: a doc matched only through a\n    // stemmed variant (query \"negotiate\" vs doc \"negotiating\") finds no occurrence here.\n    // Fall back to the doc's start, unmarked, rather than claim a match that isn't there.\n    const end = Math.min(text.length, EXCERPT_WINDOW);\n    return { excerpt: `${text.slice(0, end).replace(/\\s+/g, ' ').trim()}${end < text.length ? '…' : ''}`, offset: 0 };\n  }\n  const start = bestWindowStart(occ, EXCERPT_WINDOW);\n  const end = Math.min(text.length, start + EXCERPT_WINDOW);\n  let out = '';\n  let cursor = start;\n  for (const o of occ) {\n    if (o.start < start || o.end > end) continue;\n    // Occurrences of duplicate or substring-overlapping terms (\"test tests\") can overlap;\n    // emitting each would duplicate document text. Keep the first, absorb the rest.\n    if (o.start < cursor) continue;\n    out += `${text.slice(cursor, o.start)}«${text.slice(o.start, o.end)}»`;\n    cursor = o.end;\n  }\n  out += text.slice(cursor, end);\n  const prefix = start > 0 ? '…' : '';\n  const suffix = end < text.length ? '…' : '';\n  return { excerpt: `${prefix}${out.replace(/\\s+/g, ' ')}${suffix}`, offset: start };\n}\n\nfunction lineNumberAt(text: string, offset: number): number {\n  let line = 1;\n  for (let i = 0; i < offset; i++) if (text.charCodeAt(i) === 10) line++;\n  return line;\n}\n\nfunction lineRangeFor(db: DatabaseSync, path: string, line: number): string | null {\n  const row = db.prepare('SELECT start_line, end_line FROM sections WHERE \"path\" = ? AND start_line <= ? AND end_line >= ? ORDER BY start_line DESC LIMIT 1').get(path, line, line) as { start_line: number; end_line: number } | undefined;\n  return row ? `L${row.start_line}-${row.end_line}` : null;\n}\n\nexport interface SearchOptions {\n  k?: number;\n  where?: string; // SQL fragment against frontmatter alias `f`, e.g. \"f.status = 'active'\"\n  preset?: string; // named preset; unknown name throws listing declared presets, undefined -> \"default\"\n  include?: string[]; // ad hoc scope override (repeatable --include); independent of exclude\n  exclude?: string[]; // ad hoc scope override (repeatable --exclude); independent of include\n  noExclude?: boolean; // --no-exclude: drop the preset's exclude for this command\n}\n\n// The declared or defaulted signals compose via RRF; `via` names which ones produced each row.\n// `opts` arrives already resolved (config.ts:resolveSearch).\nexport async function search(db: DatabaseSync, cfg: ResolvedConfig, terms: string, opts: SearchOptions = {}): Promise<Row[]> {\n  const effective = resolveSearch(cfg, opts);\n  const { k, signals } = effective;\n\n  const allPaths = (db.prepare('SELECT \"path\" FROM frontmatter').all() as Array<{ path: string }>).map((r) => r.path);\n  const scopePaths = rawScope(db, cfg, opts, allPaths);\n  const scopeActive = scopePaths.size < allPaths.length;\n  // The set every candidate pool must be filtered to before truncation: scope narrowed by\n  // --where, the same composition scopedPaths() gives the other commands. Runs the same where\n  // fragment matchSql does, so a bad column gets the same attributed error either way.\n  let allowedPaths: Set<string>;\n  try {\n    allowedPaths = narrowByWhere(db, scopePaths, effective.where);\n  } catch (err) {\n    throw searchError(err as Error, terms, effective.where);\n  }\n  const fetch = Math.max(k * 3, 30);\n\n  // A downloadable HF id proceeds -- getProvider fetches it lazily on consent. Only a\n  // local path with missing files errors here, since nothing will ever fetch it for itself.\n  const wantsVectors = signals.vectors !== undefined;\n  if (wantsVectors) {\n    const e = embedConfig(cfg); // validate.ts guarantees this is set whenever \"vectors\" is declared\n    if (localModelMissing(e)) {\n      throw new SenseError('EMBED_MODEL_MISSING', `preset \"${effective.presetName}\" searches with vectors, but the local model path \"${e.model}\" is missing ${MODEL_FILENAMES}; point embed.model at a directory containing them, or drop \"vectors\" from that preset's signals to search without them`);\n    }\n  }\n  const semanticEnabled = wantsVectors && scopeHasEmbeddings(db, cfg, allowedPaths);\n\n  // Terms pass to MATCH as written, with one transform: a run of text whose language marks no\n  // word boundaries becomes a quoted grapheme phrase (and a title:/summary:/text: qualifier\n  // ahead of it retargets its _seg column), matching how it was indexed. Text that already\n  // carries boundaries comes through untouched, so this is a no-op for the languages that\n  // never needed it. Gated on the same predicate that decided whether the sidecars were\n  // populated (contentTokenize(cfg) === undefined), so index and query can't disagree. Invalid\n  // syntax still errors rather than being rewritten.\n  // --where\n  // applies inside the candidate query (a post-filter would drop matches ranked past the pool)\n  // and again on the final select, for link-derived rows.\n  const scope = effective.where;\n  const whereJoin = scope ? `JOIN frontmatter f ON f.\"path\" = content.path` : '';\n  const whereCond = scope ? `AND (${scope})` : '';\n  // A scope narrower than the whole index must filter the candidate pool before LIMIT, not\n  // after -- otherwise scoped notes ranking below the global top-`fetch` never reach the\n  // filter. A join against a temp table, not a bound parameter list: real scopes run to\n  // thousands of paths, past SQLITE_MAX_VARIABLE_NUMBER on older builds.\n  if (scopeActive) materializeScope(db, '_search_scope', scopePaths);\n  const scopeCond = scopeActive ? `AND content.path IN (SELECT \"path\" FROM _search_scope)` : '';\n  const query = contentTokenize(cfg) === undefined ? segmentMatch(terms) : terms;\n\n  const candidates = new Map<string, { score: number; via: string }>();\n  let matchRows: Array<{ path: string; hit: string | null }> = [];\n  // A query that rewrites to nothing (e.g. only unspaced-script punctuation) has no\n  // searchable words: zero lexical rows, and the other signals still compose.\n  if (signals.words !== undefined && query.trim() !== '') {\n    try {\n      matchRows = wordsCandidates(db, candidates, query, whereJoin, whereCond, scopeCond, fetch, signals.words);\n    } catch (err) {\n      throw searchError(err as Error, terms, scope);\n    }\n  }\n  const hits = new Map(matchRows.map((r) => [r.path, r.hit]));\n  // hit === null here means the bound suppressed snippet(), not \"no match\" -- distinguish\n  // from via='link' rows (never in matchRows, so absent from this set) below.\n  const oversized = new Set(matchRows.filter((r) => r.hit === null).map((r) => r.path));\n\n  if (signals.links !== undefined) linksCandidates(db, candidates, matchRows, allPaths, allowedPaths, fetch, signals.links);\n\n  let chunkLines = new Map<string, string>();\n  let chunkSimilarity = new Map<string, number>();\n  if (semanticEnabled) {\n    ({ chunkLines, chunkSimilarity } = await vectorsCandidates(db, cfg, candidates, terms, fetch, allowedPaths, signals.vectors as number));\n  }\n\n  db.exec('DROP TABLE IF EXISTS _search');\n  db.exec('CREATE TEMP TABLE _search (\"path\" TEXT PRIMARY KEY, score REAL, via TEXT, hit TEXT, lines TEXT, similarity REAL)');\n  const insert = db.prepare('INSERT INTO _search (\"path\", score, via, hit, lines, similarity) VALUES (?, ?, ?, ?, ?, ?)');\n  for (const [path, c] of candidates) insert.run(path, c.score, c.via, hits.get(path) ?? null, chunkLines.get(path) ?? null, chunkSimilarity.get(path) ?? null);\n\n  // All three candidate paths (match, link, vector) already filtered to scope+where before\n  // reaching _search; this reapplies --where anyway since the join to frontmatter is already\n  // needed for the path column.\n  const where = scope ? `WHERE (${scope})` : '';\n  // lines is always present now: semantic rows carry their chunk's range, oversized-doc\n  // lexical rows gain one below, everything else stays null. similarity stays semantic-only.\n  const similarityCol = semanticEnabled ? ', _search.similarity' : '';\n  const rows = db\n    .prepare(\n      `SELECT f.\"path\" AS path, content.title, content.summary, _search.hit, _search.via, round(_search.score, 4) AS score, _search.lines${similarityCol}\n       FROM _search JOIN frontmatter f ON f.\"path\" = _search.\"path\" JOIN content ON content.path = _search.\"path\"\n       ${where} ORDER BY _search.score DESC LIMIT ?`\n    )\n    .all(k) as Row[];\n\n  if (oversized.size > 0) {\n    const bareTerms = extractBareTerms(terms);\n    for (const row of rows) {\n      if (row.hit !== null || !oversized.has(row.path as string)) continue;\n      let text: string;\n      try {\n        text = readFileSync(join(cfg.baseDir, row.path as string), 'utf8');\n      } catch {\n        continue; // vanished since the match; leave hit/lines null rather than throw\n      }\n      const { excerpt, offset } = computeExcerpt(text, bareTerms);\n      row.hit = excerpt;\n      if (row.lines == null) row.lines = featureEnabled(cfg, 'sections') ? lineRangeFor(db, row.path as string, lineNumberAt(text, offset)) : null;\n    }\n  }\n\n  return rows;\n}\n"],"names":["search","EXCERPT_WINDOW","extractBareTerms","query","cleaned","replace","split","map","tok","toLowerCase","trim","filter","length","findOccurrences","haystackLower","terms","occ","term","idx","found","indexOf","push","start","end","sort","a","b","bestWindowStart","windowSize","best","bestDistinct","bestCount","i","winEnd","seen","Set","count","j","add","size","computeExcerpt","text","Math","min","excerpt","slice","offset","out","cursor","o","prefix","suffix","lineNumberAt","line","charCodeAt","lineRangeFor","db","path","row","prepare","get","start_line","end_line","cfg","opts","hits","chunkLines","chunkSimilarity","effective","k","signals","allPaths","scopePaths","scopeActive","allowedPaths","fetch","wantsVectors","e","semanticEnabled","scope","whereJoin","whereCond","scopeCond","candidates","matchRows","oversized","insert","c","where","similarityCol","rows","bareTerms","resolveSearch","all","r","rawScope","narrowByWhere","err","searchError","max","vectors","undefined","embedConfig","localModelMissing","SenseError","presetName","MODEL_FILENAMES","model","scopeHasEmbeddings","materializeScope","contentTokenize","segmentMatch","Map","words","wordsCandidates","hit","links","linksCandidates","vectorsCandidates","exec","run","score","via","has","readFileSync","join","baseDir","lines","featureEnabled"],"mappings":";;;;+BAyHsBA;;;eAAAA;;;sBAzHO;wBACR;uBAGuD;uBACzB;wBACxB;6BAEC;yBACC;uBACiD;yBACV;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAEpE,iFAAiF;AACjF,8EAA8E;AAC9E,IAAMC,iBAAiB;AAEvB,8FAA8F;AAC9F,qEAAqE;AACrE,SAASC,iBAAiBC,KAAa;IACrC,IAAMC,UAAUD,MACbE,OAAO,CAAC,MAAM,KACdA,OAAO,CAAC,UAAU,KAClBA,OAAO,CAAC,mCAAmC;IAC9C,OAAOD,QACJE,KAAK,CAAC,OACNC,GAAG,CAAC,SAACC;eAAQA,IAAIH,OAAO,CAAC,kBAAkB;OAAK,iCAAiC;KACjFE,GAAG,CAAC,SAACC;eAAQA,IAAIC,WAAW,GAAGC,IAAI;OACnCC,MAAM,CAAC,SAACH;eAAQA,IAAII,MAAM,GAAG;;AAClC;AAEA,SAASC,gBAAgBC,aAAqB,EAAEC,KAAe;IAC7D,IAAMC,MAA2D,EAAE;QAC9D,kCAAA,2BAAA;;QAAL,QAAK,YAAcD,0BAAd,SAAA,6BAAA,QAAA,yBAAA,iCAAqB;YAArB,IAAME,OAAN;YACH,IAAIC,MAAM;YACV,OAAS;gBACP,IAAMC,QAAQL,cAAcM,OAAO,CAACH,MAAMC;gBAC1C,IAAIC,UAAU,CAAC,GAAG;gBAClBH,IAAIK,IAAI,CAAC;oBAAEC,OAAOH;oBAAOI,KAAKJ,QAAQF,KAAKL,MAAM;oBAAEK,MAAAA;gBAAK;gBACxDC,MAAMC,QAAQF,KAAKL,MAAM;YAC3B;QACF;;QARK;QAAA;;;iBAAA,6BAAA;gBAAA;;;gBAAA;sBAAA;;;;IASL,mFAAmF;IACnF,+EAA+E;IAC/E,OAAOI,IAAIQ,IAAI,CAAC,SAACC,GAAGC;eAAMD,EAAEH,KAAK,GAAGI,EAAEJ,KAAK,IAAII,EAAEH,GAAG,GAAGE,EAAEF,GAAG;;AAC9D;AAEA,wFAAwF;AACxF,6DAA6D;AAC7D,SAASI,gBAAgBX,GAAwD,EAAEY,UAAkB;IACnG,IAAIC,OAAOb,GAAG,CAAC,EAAE,CAACM,KAAK;IACvB,IAAIQ,eAAe;IACnB,IAAIC,YAAY;IAChB,IAAK,IAAIC,IAAI,GAAGA,IAAIhB,IAAIJ,MAAM,EAAEoB,IAAK;QACnC,IAAMC,SAASjB,GAAG,CAACgB,EAAE,CAACV,KAAK,GAAGM;QAC9B,IAAMM,OAAO,IAAIC;QACjB,IAAIC,QAAQ;QACZ,IAAK,IAAIC,IAAIL,GAAGK,IAAIrB,IAAIJ,MAAM,IAAII,GAAG,CAACqB,EAAE,CAACf,KAAK,GAAGW,QAAQI,IAAK;YAC5DH,KAAKI,GAAG,CAACtB,GAAG,CAACqB,EAAE,CAACpB,IAAI;YACpBmB;QACF;QACA,IAAIF,KAAKK,IAAI,GAAGT,gBAAiBI,KAAKK,IAAI,KAAKT,gBAAgBM,QAAQL,WAAY;YACjFF,OAAOb,GAAG,CAACgB,EAAE,CAACV,KAAK;YACnBQ,eAAeI,KAAKK,IAAI;YACxBR,YAAYK;QACd;IACF;IACA,OAAOP;AACT;AAEA,uFAAuF;AACvF,uFAAuF;AACvF,SAASW,eAAeC,IAAY,EAAE1B,KAAe;IACnD,IAAMC,MAAMH,gBAAgB4B,KAAKhC,WAAW,IAAIM;IAChD,IAAIC,IAAIJ,MAAM,KAAK,GAAG;QACpB,iFAAiF;QACjF,qFAAqF;QACrF,sFAAsF;QACtF,IAAMW,MAAMmB,KAAKC,GAAG,CAACF,KAAK7B,MAAM,EAAEX;QAClC,OAAO;YAAE2C,SAAS,AAAC,GAAmDrB,OAAjDkB,KAAKI,KAAK,CAAC,GAAGtB,KAAKlB,OAAO,CAAC,QAAQ,KAAKK,IAAI,IAAkC,OAA7Ba,MAAMkB,KAAK7B,MAAM,GAAG,MAAM;YAAMkC,QAAQ;QAAE;IAClH;IACA,IAAMxB,QAAQK,gBAAgBX,KAAKf;IACnC,IAAMsB,OAAMmB,KAAKC,GAAG,CAACF,KAAK7B,MAAM,EAAEU,QAAQrB;IAC1C,IAAI8C,MAAM;IACV,IAAIC,SAAS1B;QACR,kCAAA,2BAAA;;QAAL,QAAK,YAAWN,wBAAX,SAAA,6BAAA,QAAA,yBAAA,iCAAgB;YAAhB,IAAMiC,IAAN;YACH,IAAIA,EAAE3B,KAAK,GAAGA,SAAS2B,EAAE1B,GAAG,GAAGA,MAAK;YACpC,sFAAsF;YACtF,gFAAgF;YAChF,IAAI0B,EAAE3B,KAAK,GAAG0B,QAAQ;YACtBD,OAAO,AAAC,GAAiCN,OAA/BA,KAAKI,KAAK,CAACG,QAAQC,EAAE3B,KAAK,GAAE,QAA8B,OAA3BmB,KAAKI,KAAK,CAACI,EAAE3B,KAAK,EAAE2B,EAAE1B,GAAG,GAAE;YACpEyB,SAASC,EAAE1B,GAAG;QAChB;;QAPK;QAAA;;;iBAAA,6BAAA;gBAAA;;;gBAAA;sBAAA;;;;IAQLwB,OAAON,KAAKI,KAAK,CAACG,QAAQzB;IAC1B,IAAM2B,SAAS5B,QAAQ,IAAI,MAAM;IACjC,IAAM6B,SAAS5B,OAAMkB,KAAK7B,MAAM,GAAG,MAAM;IACzC,OAAO;QAAEgC,SAAS,AAAC,GAAWG,OAATG,QAAoCC,OAA3BJ,IAAI1C,OAAO,CAAC,QAAQ,MAAc,OAAP8C;QAAUL,QAAQxB;IAAM;AACnF;AAEA,SAAS8B,aAAaX,IAAY,EAAEK,MAAc;IAChD,IAAIO,OAAO;IACX,IAAK,IAAIrB,IAAI,GAAGA,IAAIc,QAAQd,IAAK,IAAIS,KAAKa,UAAU,CAACtB,OAAO,IAAIqB;IAChE,OAAOA;AACT;AAEA,SAASE,aAAaC,EAAgB,EAAEC,IAAY,EAAEJ,IAAY;IAChE,IAAMK,MAAMF,GAAGG,OAAO,CAAC,qIAAqIC,GAAG,CAACH,MAAMJ,MAAMA;IAC5K,OAAOK,MAAM,AAAC,IAAqBA,OAAlBA,IAAIG,UAAU,EAAC,KAAgB,OAAbH,IAAII,QAAQ,IAAK;AACtD;AAaO,SAAe9D;wCAAOwD,EAAgB,EAAEO,GAAmB,EAAEhD,KAAa;YAAEiD,MA6EZC,WAAwBC,iBAA8BC,sBA5ErHC,WACEC,GAAGC,SAELC,UACAC,YACAC,aAIFC,cAMEC,OAIAC,cAEEC,GAKFC,iBAYAC,OACAC,WACAC,WAMAC,WACA/E,OAEAgF,YACFC,WAUEnB,MAGAoB,WAIFnB,YACAC,sBAOEmB,QACD,2BAAA,mBAAA,gBAAA,WAAA,oBAAO7B,MAAM8B,GAKZC,OAGAC,eACAC,MASEC,WACD,4BAAA,oBAAA,iBAAA,YAAA,QAAMjC,KAELjB,MAMwBD,iBAApBI,SAASE;;;;;oBAxG4DkB,OAAAA,oEAAsB,CAAC;oBAClGI,YAAYwB,IAAAA,sBAAa,EAAC7B,KAAKC;oBAC7BK,IAAeD,UAAfC,GAAGC,UAAYF,UAAZE;oBAELC,WAAW,AAACf,GAAGG,OAAO,CAAC,kCAAkCkC,GAAG,GAA+BtF,GAAG,CAAC,SAACuF;+BAAMA,EAAErC,IAAI;;oBAC5Ge,aAAauB,IAAAA,iBAAQ,EAACvC,IAAIO,KAAKC,MAAMO;oBACrCE,cAAcD,WAAWjC,IAAI,GAAGgC,SAAS3D,MAAM;oBAKrD,IAAI;wBACF8D,eAAesB,IAAAA,sBAAa,EAACxC,IAAIgB,YAAYJ,UAAUoB,KAAK;oBAC9D,EAAE,OAAOS,KAAK;wBACZ,MAAMC,IAAAA,0BAAW,EAACD,KAAclF,OAAOqD,UAAUoB,KAAK;oBACxD;oBACMb,QAAQjC,KAAKyD,GAAG,CAAC9B,IAAI,GAAG;oBAE9B,oFAAoF;oBACpF,0FAA0F;oBACpFO,eAAeN,QAAQ8B,OAAO,KAAKC;oBACzC,IAAIzB,cAAc;wBACVC,IAAIyB,IAAAA,oBAAW,EAACvC,MAAM,oEAAoE;wBAChG,IAAIwC,IAAAA,0BAAiB,EAAC1B,IAAI;4BACxB,MAAM,IAAI2B,oBAAU,CAAC,uBAAuB,AAAC,WAAoF3B,OAA1ET,UAAUqC,UAAU,EAAC,uDAA4EC,OAAvB7B,EAAE8B,KAAK,EAAC,iBAA+B,OAAhBD,wBAAe,EAAC;wBAC1K;oBACF;oBACM5B,kBAAkBF,gBAAgBgC,IAAAA,2BAAkB,EAACpD,IAAIO,KAAKW;oBAEpE,4FAA4F;oBAC5F,0FAA0F;oBAC1F,yFAAyF;oBACzF,wFAAwF;oBACxF,sFAAsF;oBACtF,6FAA6F;oBAC7F,mDAAmD;oBACnD,UAAU;oBACV,6FAA6F;oBAC7F,wDAAwD;oBAClDK,QAAQX,UAAUoB,KAAK;oBACvBR,YAAYD,QAAQ,kDAAkD;oBACtEE,YAAYF,QAAQ,AAAC,QAAa,OAANA,OAAM,OAAK;oBAC7C,yFAAyF;oBACzF,uFAAuF;oBACvF,sFAAsF;oBACtF,uEAAuE;oBACvE,IAAIN,aAAaoC,IAAAA,yBAAgB,EAACrD,IAAI,iBAAiBgB;oBACjDU,YAAYT,cAAc,2DAA2D;oBACrFtE,QAAQ2G,IAAAA,wBAAe,EAAC/C,SAASsC,YAAYU,IAAAA,uBAAY,EAAChG,SAASA;oBAEnEoE,aAAa,IAAI6B;oBACnB5B;oBACJ,kFAAkF;oBAClF,4EAA4E;oBAC5E,IAAId,QAAQ2C,KAAK,KAAKZ,aAAalG,MAAMO,IAAI,OAAO,IAAI;wBACtD,IAAI;4BACF0E,YAAY8B,IAAAA,0BAAe,EAAC1D,IAAI2B,YAAYhF,OAAO6E,WAAWC,WAAWC,WAAWP,OAAOL,QAAQ2C,KAAK;wBAC1G,EAAE,OAAOhB,KAAK;4BACZ,MAAMC,IAAAA,0BAAW,EAACD,KAAclF,OAAOgE;wBACzC;oBACF;oBACMd,OAAO,IAAI+C,IAAI5B,UAAU7E,GAAG,CAAC,SAACuF;+BAAM;4BAACA,EAAErC,IAAI;4BAAEqC,EAAEqB,GAAG;yBAAC;;oBACzD,wFAAwF;oBACxF,4EAA4E;oBACtE9B,YAAY,IAAIlD,IAAIiD,UAAUzE,MAAM,CAAC,SAACmF;+BAAMA,EAAEqB,GAAG,KAAK;uBAAM5G,GAAG,CAAC,SAACuF;+BAAMA,EAAErC,IAAI;;oBAEnF,IAAIa,QAAQ8C,KAAK,KAAKf,WAAWgB,IAAAA,0BAAe,EAAC7D,IAAI2B,YAAYC,WAAWb,UAAUG,cAAcC,OAAOL,QAAQ8C,KAAK;oBAEpHlD,aAAa,IAAI8C;oBACjB7C,kBAAkB,IAAI6C;yBACtBlC,iBAAAA;;;;oBACiC;;wBAAMwC,IAAAA,4BAAiB,EAAC9D,IAAIO,KAAKoB,YAAYpE,OAAO4D,OAAOD,cAAcJ,QAAQ8B,OAAO;;;0BAAxF,eAAhClC,iBAAAA,YAAYC,sBAAAA;;;oBAGjBX,GAAG+D,IAAI,CAAC;oBACR/D,GAAG+D,IAAI,CAAC;oBACFjC,SAAS9B,GAAGG,OAAO,CAAC;oBACrB,kCAAA,2BAAA;;wBAAL,IAAK,YAAmBwB,iCAAnB,6BAAA,QAAA,yBAAA;2DAAA,iBAAO1B,uBAAM8B;4BAAkBD,OAAOkC,GAAG,CAAC/D,MAAM8B,EAAEkC,KAAK,EAAElC,EAAEmC,GAAG,GAAEzD,YAAAA,KAAKL,GAAG,CAACH,mBAATQ,uBAAAA,YAAkB,OAAMC,kBAAAA,WAAWN,GAAG,CAACH,mBAAfS,6BAAAA,kBAAwB,OAAMC,uBAAAA,gBAAgBP,GAAG,CAACH,mBAApBU,kCAAAA,uBAA6B;;;wBAAnJ;wBAAA;;;iCAAA,6BAAA;gCAAA;;;gCAAA;sCAAA;;;;oBAEL,yFAAyF;oBACzF,2FAA2F;oBAC3F,8BAA8B;oBACxBqB,QAAQT,QAAQ,AAAC,UAAe,OAANA,OAAM,OAAK;oBAC3C,sFAAsF;oBACtF,2FAA2F;oBACrFU,gBAAgBX,kBAAkB,yBAAyB;oBAC3DY,OAAOlC,GACVG,OAAO,CACN,AAAC,qIAEE6B,OAFkIC,eAAc,gIAE1I,OAAND,OAAM,yCAEVK,GAAG,CAACxB;oBAEP,IAAIgB,UAAU9C,IAAI,GAAG,GAAG;wBAChBoD,YAAYzF,iBAAiBa;wBAC9B,mCAAA,4BAAA;;4BAAL,IAAK,aAAa2E,2BAAb,8BAAA,SAAA,0BAAA,kCAAmB;gCAAbhC,MAAN;gCACH,IAAIA,IAAIyD,GAAG,KAAK,QAAQ,CAAC9B,UAAUsC,GAAG,CAACjE,IAAID,IAAI,GAAa;gCACxDhB,OAAAA,KAAAA;gCACJ,IAAI;oCACFA,OAAOmF,IAAAA,oBAAY,EAACC,IAAAA,cAAI,EAAC9D,IAAI+D,OAAO,EAAEpE,IAAID,IAAI,GAAa;gCAC7D,EAAE,eAAM;oCACN,UAAU,mEAAmE;gCAC/E;gCAC4BjB,kBAAAA,eAAeC,MAAMkD,YAAzC/C,UAAoBJ,gBAApBI,SAASE,SAAWN,gBAAXM;gCACjBY,IAAIyD,GAAG,GAAGvE;gCACV,IAAIc,IAAIqE,KAAK,IAAI,MAAMrE,IAAIqE,KAAK,GAAGC,IAAAA,uBAAc,EAACjE,KAAK,cAAcR,aAAaC,IAAIE,IAAID,IAAI,EAAYL,aAAaX,MAAMK,WAAW;4BAC1I;;4BAXK;4BAAA;;;qCAAA,8BAAA;oCAAA;;;oCAAA;0CAAA;;;;oBAYP;oBAEA;;wBAAO4C;;;;IACT"}

package/dist/cjs/commands/signals.d.cts

@@ -0,0 +1,18 @@
+import type { DatabaseSync } from 'node:sqlite';
+import type { ResolvedConfig } from '../config/index.js';
+export type Candidates = Map<string, {
+    score: number;
+    via: string;
+}>;
+export declare function wordsCandidates(db: DatabaseSync, candidates: Candidates, query: string, whereJoin: string, whereCond: string, scopeCond: string, fetch: number, weight: number): Array<{
+    path: string;
+    hit: string | null;
+}>;
+export declare function linksCandidates(db: DatabaseSync, candidates: Candidates, matchRows: Array<{
+    path: string;
+    hit: string | null;
+}>, allPaths: string[], allowedPaths: Set<string>, fetch: number, weight: number): void;
+export declare function vectorsCandidates(db: DatabaseSync, cfg: ResolvedConfig, candidates: Candidates, terms: string, fetch: number, allowedPaths: Set<string>, weight: number): Promise<{
+    chunkLines: Map<string, string>;
+    chunkSimilarity: Map<string, number>;
+}>;

package/dist/cjs/commands/signals.d.ts

@@ -0,0 +1,18 @@
+import type { DatabaseSync } from 'node:sqlite';
+import type { ResolvedConfig } from '../config/index.js';
+export type Candidates = Map<string, {
+    score: number;
+    via: string;
+}>;
+export declare function wordsCandidates(db: DatabaseSync, candidates: Candidates, query: string, whereJoin: string, whereCond: string, scopeCond: string, fetch: number, weight: number): Array<{
+    path: string;
+    hit: string | null;
+}>;
+export declare function linksCandidates(db: DatabaseSync, candidates: Candidates, matchRows: Array<{
+    path: string;
+    hit: string | null;
+}>, allPaths: string[], allowedPaths: Set<string>, fetch: number, weight: number): void;
+export declare function vectorsCandidates(db: DatabaseSync, cfg: ResolvedConfig, candidates: Candidates, terms: string, fetch: number, allowedPaths: Set<string>, weight: number): Promise<{
+    chunkLines: Map<string, string>;
+    chunkSimilarity: Map<string, number>;
+}>;