@kolisachint/hoocode-agent 0.4.167 → 0.4.168

package/dist/core/search/eval.js

@@ -50,6 +50,13 @@ export const EVAL_CONFIGS = [
     // Three-way fusion: dense + BM25 + grep, all fused here so `k` is ours and
     // the per-leg ranks survive into the trace. `bm25+dense` isolates what the
     // grep leg still contributes once BM25 is present.
+    //
+    // `3-way` used to lose badly to `bm25+dense` (MRR 0.608 -> 0.552, 18 of 62
+    // queries worse) because two lexical views of the same corpus dilute each
+    // other in fusion. Its grep leg is now scoped to files the index has not
+    // read, so the two rows are identical on this corpus by construction —
+    // the difference only appears under `--live-edits`, where `3-way` scores
+    // 1.000 and `bm25+dense` scores 0.000.
     { label: "bm25+dense +rr", mode: "semantic", rerank: true, bm25Leg: true },
     { label: "3-way +rr", mode: "hybrid", rerank: true, bm25Leg: true },
     // Cross-encoder reranking, against the same retrieval each deterministic

package/dist/core/search/eval.js.map

@@ -1 +1 @@
-{"version":3,"file":"eval.js","sourceRoot":"","sources":["../../../src/core/search/eval.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAGH,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AA2DxD;;;;;;;;;;GAUG;AACH,MAAM,CAAC,MAAM,YAAY,GAA0B;IAClD,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE;IACrC,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,UAAU,EAAE;IACvC,EAAE,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE;IAChD,EAAE,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE;IAChD,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE;IAClD,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE;IAClD,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE;IAC/B,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,EAAE,IAAI,EAAE;IACvD,EAAE,KAAK,EAAE,cAAc,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE;IACzD,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE;IAClE,EAAE,KAAK,EAAE,iBAAiB,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE;IACpE,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE;IACjD,4EAA4E;IAC5E,sEAAsE;IACtE,4EAA4E;IAC5E,wCAAwC;IACxC,EAAE,KAAK,EAAE,eAAe,EAAE,IAAI,EAAE,UAAU,EAAE,YAAY,EAAE,IAAI,EAAE;IAChE,EAAE,KAAK,EAAE,mBAAmB,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE;IAClF,2EAA2E;IAC3E,2EAA2E;IAC3E,mDAAmD;IACnD,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE;IAC1E,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE;IACnE,yEAAyE;IACzE,oEAAkE;IAClE,EAAE,KAAK,EAAE,cAAc,EAAE,IAAI,EAAE,UAAU,EAAE,YAAY,EAAE,IAAI,EAAE;IAC/D,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,YAAY,EAAE,IAAI,EAAE;IACvD,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE;CAChF,CAAC;AAEF,+EAA6E;AAC7E,MAAM,gBAAgB,GAAG,EAAE,CAAC;AAE5B,MAAM,UAAU,eAAe,CAAC,IAAmB,EAAE,IAAkB,EAAW;IACjF,IAAI,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI;QAAE,OAAO,KAAK,CAAC;IAC1C,IAAI,IAAI,CAAC,KAAK,KAAK,MAAM;QAAE,OAAO,IAAI,CAAC;IACvC,IAAI,IAAI,CAAC,SAAS,KAAK,SAAS,IAAI,IAAI,CAAC,OAAO,KAAK,SAAS;QAAE,OAAO,IAAI,CAAC;IAC5E,OAAO,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,SAAS,CAAC;AAAA,CACxE;AAED,gFAAgF;AAChF,MAAM,UAAU,SAAS,CAAC,UAAoC,EAAE,IAA6B,EAAE,CAAS,EAAU;IACjH,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IAChC,MAAM,GAAG,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;IACnC,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,KAAK,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC;QACtB,IAAI,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,eAAe,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YAAE,OAAO,EAAE,CAAC;IACvD,CAAC;IACD,OAAO,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC;AAAA,CAC7B;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,GAAG,CAAC,UAAoC,EAAE,IAA6B,EAAU;IAChG,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IAChC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC;QACtB,MAAM,KAAK,GAAG,UAAU,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,eAAe,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QACjE,IAAI,KAAK,IAAI,CAAC;YAAE,KAAK,IAAI,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;IAC1C,CAAC;IACD,OAAO,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC;AAAA,CAC3B;AAaD,MAAM,CAAC,KAAK,UAAU,aAAa,CAClC,GAAW,EACX,SAAoB,EACpB,OAAO,GAA0B,YAAY,EAC7C,OAA0B,EAC1B,aAAgC,EACH;IAC7B,MAAM,OAAO,GAAsB,EAAE,CAAC;IACtC,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC9B,oEAAoE;QACpE,yEAAyE;QACzE,IAAI,CAAC,MAAM,CAAC,YAAY,IAAI,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,aAAa;YAAE,SAAS;QACxE,mEAAmE;QACnE,kEAAkE;QAClE,IAAI,MAAM,CAAC,YAAY,IAAI,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,OAAO,CAAC,EAAE,oBAAoB,EAAE;YAAE,SAAS;QACzG,MAAM,SAAS,GAAG,MAAM,kBAAkB,CAAC;YAC1C,GAAG;YACH,KAAK,EAAE,SAAS,CAAC,KAAK;YACtB,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,MAAM,EAAE,MAAM,CAAC,MAAM,IAAI,KAAK;YAC9B,KAAK,EAAE,gBAAgB;YACvB,OAAO,EAAE,MAAM,CAAC,YAAY,IAAI,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,OAAO;YACxE,YAAY,EAAE,MAAM,CAAC,YAAY;YACjC,OAAO,EAAE,MAAM,CAAC,OAAO;YACvB,YAAY,EAAE,MAAM,CAAC,YAAY;SACjC,CAAC,CAAC;QACH,OAAO,CAAC,IAAI,CAAC;YACZ,KAAK,EAAE,MAAM,CAAC,KAAK;YACnB,YAAY,EAAE,SAAS,CAAC,YAAY;YACpC,QAAQ,EAAE,SAAS,CAAC,cAAc,KAAK,SAAS;YAChD,SAAS,EAAE,SAAS,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC;YAC7D,SAAS,EAAE,SAAS,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC;YAC7D,UAAU,EAAE,SAAS,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,EAAE,EAAE,CAAC;YAC/D,UAAU,EAAE,SAAS,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,EAAE,EAAE,CAAC;YAC/D,GAAG,EAAE,GAAG,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,CAAC;SAC9C,CAAC,CAAC;IACJ,CAAC;IACD,OAAO,OAAO,CAAC;AAAA,CACf","sourcesContent":["/**\n * Retrieval evaluation gate (docs/hybrid-retrieval-design.md, step 6 of the\n * shipping order).\n *\n * Measures Recall@K and MRR for lexical, semantic, and hybrid retrieval across\n * an RRF `k` sweep, against a gold set keyed by **path + line range matched by\n * span overlap** — never by chunkId, which is only stable per index build.\n * Recall@50 doubles as the reranker gate: a gold span that never reaches the\n * fused top-50 cannot be rescued by any reranker.\n *\n * Rank-sensitive metrics are the point of Recall@1 and MRR: an agent reads the\n * top result or two, so \"the gold span is somewhere in the top 5\" understates\n * how much ordering matters. Recall@5/10/50 are kept for continuity with the\n * numbers already published in the design note.\n *\n * Driven by `scripts/search-eval.ts` via `eval-harness.ts`; the scoring logic\n * lives here so it is unit-testable without an embedding index.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { retrieveCandidates } from \"./hybrid-search.js\";\nimport type { CandidateSpan, ResolvedSearchMode, SearchMode } from \"./types.js\";\n\nexport interface EvalGoldSpan {\n\t/** Repo-relative POSIX path. */\n\tpath: string;\n\t/** 1-based inclusive; omit both to accept any span in the file. */\n\tstartLine?: number;\n\tendLine?: number;\n\t/**\n\t * Literal source text that must occur inside `[startLine, endLine]`. Not\n\t * used for scoring — it is how the gold set survives the corpus moving\n\t * underneath it: `scripts/search-eval-gold.ts` re-resolves the line range\n\t * from this anchor, and the fixture test fails when an anchor no longer\n\t * sits inside its recorded range. Omitted for `scope: \"file\"` entries.\n\t */\n\tanchor?: string;\n\t/**\n\t * `\"span\"` (default) scores by overlap with the recorded line range;\n\t * `\"file\"` accepts any span in the file. File scope is deliberate for\n\t * path-class queries, where the whole file *is* the answer — it is\n\t * recorded explicitly so a file-level score is never mistaken for a\n\t * span-level one.\n\t */\n\tscope?: \"span\" | \"file\";\n}\n\nexport interface EvalQuery {\n\tid: string;\n\t/** Query class from the design doc (exact-symbol, path, error-fragment,\n\t *  conceptual, cross-file, boundary). Reporting only. */\n\tclass: string;\n\tquery: string;\n\tgold: EvalGoldSpan[];\n}\n\nexport interface EvalConfig {\n\tlabel: string;\n\tmode: SearchMode;\n\trrfK?: number;\n\trerank?: boolean;\n\t/**\n\t * Score this config against a daemon-side hybrid store (BM25 fused with\n\t * vectors inside the Rust daemon) instead of the dense-only index. Skipped\n\t * when the harness has no hybrid service, so records without one simply\n\t * omit the row rather than silently scoring it as plain semantic.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as a separate leg and fuse it here with\n\t * dense and grep. Like `daemonHybrid` it needs a hybrid store, so it is\n\t * skipped when the harness has no hybrid service.\n\t */\n\tbm25Leg?: boolean;\n\t/** Rerank with the daemon's cross-encoder rather than the deterministic\n\t *  scorer. Needs a service, so it is skipped without one. */\n\tcrossEncoder?: boolean;\n}\n\n/**\n * The sweep from the design doc — single retrievers, hybrid across k, the\n * routed auto mode — plus reranked (`+rr`) variants for the step 7 gate.\n *\n * Exported because `scripts/search-eval.ts` and `test/search-eval.test.ts`\n * both consume it. It previously lost its `export` to a knip-driven\n * dead-export sweep (02efaab): the only importer was a `.mjs` script reading\n * from `dist/`, which static analysis cannot see, and the eval gate silently\n * stopped running. The TypeScript importers are the fix — do not \"clean up\"\n * this export without checking them.\n */\nexport const EVAL_CONFIGS: readonly EvalConfig[] = [\n\t{ label: \"lexical\", mode: \"lexical\" },\n\t{ label: \"semantic\", mode: \"semantic\" },\n\t{ label: \"hybrid k=0\", mode: \"hybrid\", rrfK: 0 },\n\t{ label: \"hybrid k=2\", mode: \"hybrid\", rrfK: 2 },\n\t{ label: \"hybrid k=10\", mode: \"hybrid\", rrfK: 10 },\n\t{ label: \"hybrid k=60\", mode: \"hybrid\", rrfK: 60 },\n\t{ label: \"auto\", mode: \"auto\" },\n\t{ label: \"lexical +rr\", mode: \"lexical\", rerank: true },\n\t{ label: \"semantic +rr\", mode: \"semantic\", rerank: true },\n\t{ label: \"hybrid k=2 +rr\", mode: \"hybrid\", rrfK: 2, rerank: true },\n\t{ label: \"hybrid k=60 +rr\", mode: \"hybrid\", rrfK: 60, rerank: true },\n\t{ label: \"auto +rr\", mode: \"auto\", rerank: true },\n\t// Is BM25 a better lexical leg than ripgrep? These two run the daemon's own\n\t// vector+BM25 fusion and no ripgrep at all, so comparing them against\n\t// \"semantic\" isolates what BM25 adds, and against \"hybrid k=2\" compares the\n\t// two lexical legs at the system level.\n\t{ label: \"daemon-hybrid\", mode: \"semantic\", daemonHybrid: true },\n\t{ label: \"daemon-hybrid +rr\", mode: \"semantic\", rerank: true, daemonHybrid: true },\n\t// Three-way fusion: dense + BM25 + grep, all fused here so `k` is ours and\n\t// the per-leg ranks survive into the trace. `bm25+dense` isolates what the\n\t// grep leg still contributes once BM25 is present.\n\t{ label: \"bm25+dense +rr\", mode: \"semantic\", rerank: true, bm25Leg: true },\n\t{ label: \"3-way +rr\", mode: \"hybrid\", rerank: true, bm25Leg: true },\n\t// Cross-encoder reranking, against the same retrieval each deterministic\n\t// `+rr` row uses — so the delta is the reranker and nothing else.\n\t{ label: \"semantic +ce\", mode: \"semantic\", crossEncoder: true },\n\t{ label: \"auto +ce\", mode: \"auto\", crossEncoder: true },\n\t{ label: \"bm25+dense +ce\", mode: \"semantic\", bm25Leg: true, crossEncoder: true },\n];\n\n/** Candidates fetched per eval query — deep enough for the reranker gate. */\nconst EVAL_FETCH_LIMIT = 50;\n\nexport function spanMatchesGold(span: CandidateSpan, gold: EvalGoldSpan): boolean {\n\tif (span.path !== gold.path) return false;\n\tif (gold.scope === \"file\") return true;\n\tif (gold.startLine === undefined || gold.endLine === undefined) return true;\n\treturn span.startLine <= gold.endLine && span.endLine >= gold.startLine;\n}\n\n/** Fraction of gold spans matched by at least one of the top-`k` candidates. */\nexport function recallAtK(candidates: readonly CandidateSpan[], gold: readonly EvalGoldSpan[], k: number): number {\n\tif (gold.length === 0) return 0;\n\tconst top = candidates.slice(0, k);\n\tlet matched = 0;\n\tfor (const g of gold) {\n\t\tif (top.some((c) => spanMatchesGold(c, g))) matched++;\n\t}\n\treturn matched / gold.length;\n}\n\n/**\n * Mean reciprocal rank, averaged over gold spans: for each gold span, `1/rank`\n * of the first candidate matching it (0 when it never appears).\n *\n * Averaging per gold span rather than taking the single first hit keeps\n * multi-span queries (the cross-file class) honest — finding one of two\n * required sites should not score like finding both. For single-span queries\n * this reduces to textbook MRR.\n */\nexport function mrr(candidates: readonly CandidateSpan[], gold: readonly EvalGoldSpan[]): number {\n\tif (gold.length === 0) return 0;\n\tlet total = 0;\n\tfor (const g of gold) {\n\t\tconst index = candidates.findIndex((c) => spanMatchesGold(c, g));\n\t\tif (index >= 0) total += 1 / (index + 1);\n\t}\n\treturn total / gold.length;\n}\n\nexport interface EvalQueryResult {\n\tlabel: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegraded: boolean;\n\trecallAt1: number;\n\trecallAt5: number;\n\trecallAt10: number;\n\trecallAt50: number;\n\tmrr: number;\n}\n\nexport async function evaluateQuery(\n\tcwd: string,\n\tevalQuery: EvalQuery,\n\tconfigs: readonly EvalConfig[] = EVAL_CONFIGS,\n\tservice?: EmbsearchService,\n\thybridService?: EmbsearchService,\n): Promise<EvalQueryResult[]> {\n\tconst results: EvalQueryResult[] = [];\n\tfor (const config of configs) {\n\t\t// A daemon-hybrid config against the plain store would error in the\n\t\t// daemon (`query_hybrid requires a hybrid store`); omit the row instead.\n\t\tif ((config.daemonHybrid || config.bm25Leg) && !hybridService) continue;\n\t\t// Cross-encoder rows need a live daemon; omit rather than silently\n\t\t// falling back to the deterministic reranker under a \"+ce\" label.\n\t\tif (config.crossEncoder && !(config.bm25Leg ? hybridService : service)?.supportsCrossEncoder()) continue;\n\t\tconst retrieved = await retrieveCandidates({\n\t\t\tcwd,\n\t\t\tquery: evalQuery.query,\n\t\t\tmode: config.mode,\n\t\t\trrfK: config.rrfK,\n\t\t\trerank: config.rerank ?? false,\n\t\t\tlimit: EVAL_FETCH_LIMIT,\n\t\t\tservice: config.daemonHybrid || config.bm25Leg ? hybridService : service,\n\t\t\tdaemonHybrid: config.daemonHybrid,\n\t\t\tbm25Leg: config.bm25Leg,\n\t\t\tcrossEncoder: config.crossEncoder,\n\t\t});\n\t\tresults.push({\n\t\t\tlabel: config.label,\n\t\t\tresolvedMode: retrieved.resolvedMode,\n\t\t\tdegraded: retrieved.degradedReason !== undefined,\n\t\t\trecallAt1: recallAtK(retrieved.candidates, evalQuery.gold, 1),\n\t\t\trecallAt5: recallAtK(retrieved.candidates, evalQuery.gold, 5),\n\t\t\trecallAt10: recallAtK(retrieved.candidates, evalQuery.gold, 10),\n\t\t\trecallAt50: recallAtK(retrieved.candidates, evalQuery.gold, 50),\n\t\t\tmrr: mrr(retrieved.candidates, evalQuery.gold),\n\t\t});\n\t}\n\treturn results;\n}\n"]}
+{"version":3,"file":"eval.js","sourceRoot":"","sources":["../../../src/core/search/eval.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAGH,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AA2DxD;;;;;;;;;;GAUG;AACH,MAAM,CAAC,MAAM,YAAY,GAA0B;IAClD,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE;IACrC,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,UAAU,EAAE;IACvC,EAAE,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE;IAChD,EAAE,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE;IAChD,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE;IAClD,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE;IAClD,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE;IAC/B,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,EAAE,IAAI,EAAE;IACvD,EAAE,KAAK,EAAE,cAAc,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE;IACzD,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE;IAClE,EAAE,KAAK,EAAE,iBAAiB,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE;IACpE,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE;IACjD,4EAA4E;IAC5E,sEAAsE;IACtE,4EAA4E;IAC5E,wCAAwC;IACxC,EAAE,KAAK,EAAE,eAAe,EAAE,IAAI,EAAE,UAAU,EAAE,YAAY,EAAE,IAAI,EAAE;IAChE,EAAE,KAAK,EAAE,mBAAmB,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE;IAClF,2EAA2E;IAC3E,2EAA2E;IAC3E,mDAAmD;IACnD,EAAE;IACF,2EAA2E;IAC3E,0EAA0E;IAC1E,yEAAyE;IACzE,yEAAuE;IACvE,yEAAyE;IACzE,uCAAuC;IACvC,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE;IAC1E,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE;IACnE,yEAAyE;IACzE,oEAAkE;IAClE,EAAE,KAAK,EAAE,cAAc,EAAE,IAAI,EAAE,UAAU,EAAE,YAAY,EAAE,IAAI,EAAE;IAC/D,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,EAAE,YAAY,EAAE,IAAI,EAAE;IACvD,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE;CAChF,CAAC;AAEF,+EAA6E;AAC7E,MAAM,gBAAgB,GAAG,EAAE,CAAC;AAE5B,MAAM,UAAU,eAAe,CAAC,IAAmB,EAAE,IAAkB,EAAW;IACjF,IAAI,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI;QAAE,OAAO,KAAK,CAAC;IAC1C,IAAI,IAAI,CAAC,KAAK,KAAK,MAAM;QAAE,OAAO,IAAI,CAAC;IACvC,IAAI,IAAI,CAAC,SAAS,KAAK,SAAS,IAAI,IAAI,CAAC,OAAO,KAAK,SAAS;QAAE,OAAO,IAAI,CAAC;IAC5E,OAAO,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,SAAS,CAAC;AAAA,CACxE;AAED,gFAAgF;AAChF,MAAM,UAAU,SAAS,CAAC,UAAoC,EAAE,IAA6B,EAAE,CAAS,EAAU;IACjH,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IAChC,MAAM,GAAG,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;IACnC,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,KAAK,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC;QACtB,IAAI,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,eAAe,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YAAE,OAAO,EAAE,CAAC;IACvD,CAAC;IACD,OAAO,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC;AAAA,CAC7B;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,GAAG,CAAC,UAAoC,EAAE,IAA6B,EAAU;IAChG,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IAChC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC;QACtB,MAAM,KAAK,GAAG,UAAU,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,eAAe,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QACjE,IAAI,KAAK,IAAI,CAAC;YAAE,KAAK,IAAI,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;IAC1C,CAAC;IACD,OAAO,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC;AAAA,CAC3B;AAaD,MAAM,CAAC,KAAK,UAAU,aAAa,CAClC,GAAW,EACX,SAAoB,EACpB,OAAO,GAA0B,YAAY,EAC7C,OAA0B,EAC1B,aAAgC,EACH;IAC7B,MAAM,OAAO,GAAsB,EAAE,CAAC;IACtC,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC9B,oEAAoE;QACpE,yEAAyE;QACzE,IAAI,CAAC,MAAM,CAAC,YAAY,IAAI,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,aAAa;YAAE,SAAS;QACxE,mEAAmE;QACnE,kEAAkE;QAClE,IAAI,MAAM,CAAC,YAAY,IAAI,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,OAAO,CAAC,EAAE,oBAAoB,EAAE;YAAE,SAAS;QACzG,MAAM,SAAS,GAAG,MAAM,kBAAkB,CAAC;YAC1C,GAAG;YACH,KAAK,EAAE,SAAS,CAAC,KAAK;YACtB,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,MAAM,EAAE,MAAM,CAAC,MAAM,IAAI,KAAK;YAC9B,KAAK,EAAE,gBAAgB;YACvB,OAAO,EAAE,MAAM,CAAC,YAAY,IAAI,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,OAAO;YACxE,YAAY,EAAE,MAAM,CAAC,YAAY;YACjC,OAAO,EAAE,MAAM,CAAC,OAAO;YACvB,YAAY,EAAE,MAAM,CAAC,YAAY;SACjC,CAAC,CAAC;QACH,OAAO,CAAC,IAAI,CAAC;YACZ,KAAK,EAAE,MAAM,CAAC,KAAK;YACnB,YAAY,EAAE,SAAS,CAAC,YAAY;YACpC,QAAQ,EAAE,SAAS,CAAC,cAAc,KAAK,SAAS;YAChD,SAAS,EAAE,SAAS,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC;YAC7D,SAAS,EAAE,SAAS,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC;YAC7D,UAAU,EAAE,SAAS,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,EAAE,EAAE,CAAC;YAC/D,UAAU,EAAE,SAAS,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,EAAE,EAAE,CAAC;YAC/D,GAAG,EAAE,GAAG,CAAC,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,IAAI,CAAC;SAC9C,CAAC,CAAC;IACJ,CAAC;IACD,OAAO,OAAO,CAAC;AAAA,CACf","sourcesContent":["/**\n * Retrieval evaluation gate (docs/hybrid-retrieval-design.md, step 6 of the\n * shipping order).\n *\n * Measures Recall@K and MRR for lexical, semantic, and hybrid retrieval across\n * an RRF `k` sweep, against a gold set keyed by **path + line range matched by\n * span overlap** — never by chunkId, which is only stable per index build.\n * Recall@50 doubles as the reranker gate: a gold span that never reaches the\n * fused top-50 cannot be rescued by any reranker.\n *\n * Rank-sensitive metrics are the point of Recall@1 and MRR: an agent reads the\n * top result or two, so \"the gold span is somewhere in the top 5\" understates\n * how much ordering matters. Recall@5/10/50 are kept for continuity with the\n * numbers already published in the design note.\n *\n * Driven by `scripts/search-eval.ts` via `eval-harness.ts`; the scoring logic\n * lives here so it is unit-testable without an embedding index.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { retrieveCandidates } from \"./hybrid-search.js\";\nimport type { CandidateSpan, ResolvedSearchMode, SearchMode } from \"./types.js\";\n\nexport interface EvalGoldSpan {\n\t/** Repo-relative POSIX path. */\n\tpath: string;\n\t/** 1-based inclusive; omit both to accept any span in the file. */\n\tstartLine?: number;\n\tendLine?: number;\n\t/**\n\t * Literal source text that must occur inside `[startLine, endLine]`. Not\n\t * used for scoring — it is how the gold set survives the corpus moving\n\t * underneath it: `scripts/search-eval-gold.ts` re-resolves the line range\n\t * from this anchor, and the fixture test fails when an anchor no longer\n\t * sits inside its recorded range. Omitted for `scope: \"file\"` entries.\n\t */\n\tanchor?: string;\n\t/**\n\t * `\"span\"` (default) scores by overlap with the recorded line range;\n\t * `\"file\"` accepts any span in the file. File scope is deliberate for\n\t * path-class queries, where the whole file *is* the answer — it is\n\t * recorded explicitly so a file-level score is never mistaken for a\n\t * span-level one.\n\t */\n\tscope?: \"span\" | \"file\";\n}\n\nexport interface EvalQuery {\n\tid: string;\n\t/** Query class from the design doc (exact-symbol, path, error-fragment,\n\t *  conceptual, cross-file, boundary). Reporting only. */\n\tclass: string;\n\tquery: string;\n\tgold: EvalGoldSpan[];\n}\n\nexport interface EvalConfig {\n\tlabel: string;\n\tmode: SearchMode;\n\trrfK?: number;\n\trerank?: boolean;\n\t/**\n\t * Score this config against a daemon-side hybrid store (BM25 fused with\n\t * vectors inside the Rust daemon) instead of the dense-only index. Skipped\n\t * when the harness has no hybrid service, so records without one simply\n\t * omit the row rather than silently scoring it as plain semantic.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as a separate leg and fuse it here with\n\t * dense and grep. Like `daemonHybrid` it needs a hybrid store, so it is\n\t * skipped when the harness has no hybrid service.\n\t */\n\tbm25Leg?: boolean;\n\t/** Rerank with the daemon's cross-encoder rather than the deterministic\n\t *  scorer. Needs a service, so it is skipped without one. */\n\tcrossEncoder?: boolean;\n}\n\n/**\n * The sweep from the design doc — single retrievers, hybrid across k, the\n * routed auto mode — plus reranked (`+rr`) variants for the step 7 gate.\n *\n * Exported because `scripts/search-eval.ts` and `test/search-eval.test.ts`\n * both consume it. It previously lost its `export` to a knip-driven\n * dead-export sweep (02efaab): the only importer was a `.mjs` script reading\n * from `dist/`, which static analysis cannot see, and the eval gate silently\n * stopped running. The TypeScript importers are the fix — do not \"clean up\"\n * this export without checking them.\n */\nexport const EVAL_CONFIGS: readonly EvalConfig[] = [\n\t{ label: \"lexical\", mode: \"lexical\" },\n\t{ label: \"semantic\", mode: \"semantic\" },\n\t{ label: \"hybrid k=0\", mode: \"hybrid\", rrfK: 0 },\n\t{ label: \"hybrid k=2\", mode: \"hybrid\", rrfK: 2 },\n\t{ label: \"hybrid k=10\", mode: \"hybrid\", rrfK: 10 },\n\t{ label: \"hybrid k=60\", mode: \"hybrid\", rrfK: 60 },\n\t{ label: \"auto\", mode: \"auto\" },\n\t{ label: \"lexical +rr\", mode: \"lexical\", rerank: true },\n\t{ label: \"semantic +rr\", mode: \"semantic\", rerank: true },\n\t{ label: \"hybrid k=2 +rr\", mode: \"hybrid\", rrfK: 2, rerank: true },\n\t{ label: \"hybrid k=60 +rr\", mode: \"hybrid\", rrfK: 60, rerank: true },\n\t{ label: \"auto +rr\", mode: \"auto\", rerank: true },\n\t// Is BM25 a better lexical leg than ripgrep? These two run the daemon's own\n\t// vector+BM25 fusion and no ripgrep at all, so comparing them against\n\t// \"semantic\" isolates what BM25 adds, and against \"hybrid k=2\" compares the\n\t// two lexical legs at the system level.\n\t{ label: \"daemon-hybrid\", mode: \"semantic\", daemonHybrid: true },\n\t{ label: \"daemon-hybrid +rr\", mode: \"semantic\", rerank: true, daemonHybrid: true },\n\t// Three-way fusion: dense + BM25 + grep, all fused here so `k` is ours and\n\t// the per-leg ranks survive into the trace. `bm25+dense` isolates what the\n\t// grep leg still contributes once BM25 is present.\n\t//\n\t// `3-way` used to lose badly to `bm25+dense` (MRR 0.608 -> 0.552, 18 of 62\n\t// queries worse) because two lexical views of the same corpus dilute each\n\t// other in fusion. Its grep leg is now scoped to files the index has not\n\t// read, so the two rows are identical on this corpus by construction —\n\t// the difference only appears under `--live-edits`, where `3-way` scores\n\t// 1.000 and `bm25+dense` scores 0.000.\n\t{ label: \"bm25+dense +rr\", mode: \"semantic\", rerank: true, bm25Leg: true },\n\t{ label: \"3-way +rr\", mode: \"hybrid\", rerank: true, bm25Leg: true },\n\t// Cross-encoder reranking, against the same retrieval each deterministic\n\t// `+rr` row uses — so the delta is the reranker and nothing else.\n\t{ label: \"semantic +ce\", mode: \"semantic\", crossEncoder: true },\n\t{ label: \"auto +ce\", mode: \"auto\", crossEncoder: true },\n\t{ label: \"bm25+dense +ce\", mode: \"semantic\", bm25Leg: true, crossEncoder: true },\n];\n\n/** Candidates fetched per eval query — deep enough for the reranker gate. */\nconst EVAL_FETCH_LIMIT = 50;\n\nexport function spanMatchesGold(span: CandidateSpan, gold: EvalGoldSpan): boolean {\n\tif (span.path !== gold.path) return false;\n\tif (gold.scope === \"file\") return true;\n\tif (gold.startLine === undefined || gold.endLine === undefined) return true;\n\treturn span.startLine <= gold.endLine && span.endLine >= gold.startLine;\n}\n\n/** Fraction of gold spans matched by at least one of the top-`k` candidates. */\nexport function recallAtK(candidates: readonly CandidateSpan[], gold: readonly EvalGoldSpan[], k: number): number {\n\tif (gold.length === 0) return 0;\n\tconst top = candidates.slice(0, k);\n\tlet matched = 0;\n\tfor (const g of gold) {\n\t\tif (top.some((c) => spanMatchesGold(c, g))) matched++;\n\t}\n\treturn matched / gold.length;\n}\n\n/**\n * Mean reciprocal rank, averaged over gold spans: for each gold span, `1/rank`\n * of the first candidate matching it (0 when it never appears).\n *\n * Averaging per gold span rather than taking the single first hit keeps\n * multi-span queries (the cross-file class) honest — finding one of two\n * required sites should not score like finding both. For single-span queries\n * this reduces to textbook MRR.\n */\nexport function mrr(candidates: readonly CandidateSpan[], gold: readonly EvalGoldSpan[]): number {\n\tif (gold.length === 0) return 0;\n\tlet total = 0;\n\tfor (const g of gold) {\n\t\tconst index = candidates.findIndex((c) => spanMatchesGold(c, g));\n\t\tif (index >= 0) total += 1 / (index + 1);\n\t}\n\treturn total / gold.length;\n}\n\nexport interface EvalQueryResult {\n\tlabel: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegraded: boolean;\n\trecallAt1: number;\n\trecallAt5: number;\n\trecallAt10: number;\n\trecallAt50: number;\n\tmrr: number;\n}\n\nexport async function evaluateQuery(\n\tcwd: string,\n\tevalQuery: EvalQuery,\n\tconfigs: readonly EvalConfig[] = EVAL_CONFIGS,\n\tservice?: EmbsearchService,\n\thybridService?: EmbsearchService,\n): Promise<EvalQueryResult[]> {\n\tconst results: EvalQueryResult[] = [];\n\tfor (const config of configs) {\n\t\t// A daemon-hybrid config against the plain store would error in the\n\t\t// daemon (`query_hybrid requires a hybrid store`); omit the row instead.\n\t\tif ((config.daemonHybrid || config.bm25Leg) && !hybridService) continue;\n\t\t// Cross-encoder rows need a live daemon; omit rather than silently\n\t\t// falling back to the deterministic reranker under a \"+ce\" label.\n\t\tif (config.crossEncoder && !(config.bm25Leg ? hybridService : service)?.supportsCrossEncoder()) continue;\n\t\tconst retrieved = await retrieveCandidates({\n\t\t\tcwd,\n\t\t\tquery: evalQuery.query,\n\t\t\tmode: config.mode,\n\t\t\trrfK: config.rrfK,\n\t\t\trerank: config.rerank ?? false,\n\t\t\tlimit: EVAL_FETCH_LIMIT,\n\t\t\tservice: config.daemonHybrid || config.bm25Leg ? hybridService : service,\n\t\t\tdaemonHybrid: config.daemonHybrid,\n\t\t\tbm25Leg: config.bm25Leg,\n\t\t\tcrossEncoder: config.crossEncoder,\n\t\t});\n\t\tresults.push({\n\t\t\tlabel: config.label,\n\t\t\tresolvedMode: retrieved.resolvedMode,\n\t\t\tdegraded: retrieved.degradedReason !== undefined,\n\t\t\trecallAt1: recallAtK(retrieved.candidates, evalQuery.gold, 1),\n\t\t\trecallAt5: recallAtK(retrieved.candidates, evalQuery.gold, 5),\n\t\t\trecallAt10: recallAtK(retrieved.candidates, evalQuery.gold, 10),\n\t\t\trecallAt50: recallAtK(retrieved.candidates, evalQuery.gold, 50),\n\t\t\tmrr: mrr(retrieved.candidates, evalQuery.gold),\n\t\t});\n\t}\n\treturn results;\n}\n"]}

package/dist/core/search/hybrid-search.d.ts

@@ -80,6 +80,23 @@ export interface RunSearchResult {
         total: number;
     };
 }
+/**
+ * Move candidates from files the index has not read to the front.
+ *
+ * They are there because grep found them and nothing else could: the index is
+ * ranking a stale copy of the file, or has never seen it. Left to fuse, they
+ * lose — RRF rewards agreement, and one leg reporting a single document is
+ * outvoted by two legs agreeing on hundreds. Measured: scoping grep to stale
+ * files without this hoist scored 25% on the live-edit set where unscoped grep
+ * scored 100%, because the fused window filled with consensus hits about the
+ * indexed copy.
+ *
+ * Capped, because "stale" scales with how far behind the index is. A few
+ * edited files is the case this exists for; a fresh checkout makes everything
+ * stale, and hoisting all of it would quietly turn hybrid search back into
+ * grep. Past the cap the remainder keeps its fused position.
+ */
+export declare function hoistStaleCandidates(candidates: readonly FusedCandidate[], staleFiles: ReadonlySet<string>): FusedCandidate[];
 export declare function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult>;
 export declare function runSearch(options: RunSearchOptions): Promise<RunSearchResult>;
 //# sourceMappingURL=hybrid-search.d.ts.map

package/dist/core/search/hybrid-search.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"hybrid-search.d.ts","sourceRoot":"","sources":["../../../src/core/search/hybrid-search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAS1E,OAAO,KAAK,EAAiB,cAAc,EAAa,kBAAkB,EAAE,UAAU,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAmCxH,MAAM,WAAW,eAAe;IAC/B,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,CAAC,EAAE,UAAU,CAAC;IAClB,kDAAkD;IAClD,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,yCAAyC;IACzC,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,wFAAwF;IACxF,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,wEAAwE;IACxE,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;;;;OAQG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;;;OAIG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;;;;OAIG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB,OAAO,CAAC,EAAE,gBAAgB,CAAC;IAC3B,MAAM,CAAC,EAAE,WAAW,CAAC;CACrB;AAED,MAAM,WAAW,cAAc;IAC9B,UAAU,EAAE,cAAc,EAAE,CAAC;IAC7B,YAAY,EAAE,kBAAkB,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,UAAU,EAAE,WAAW,CAAC,YAAY,CAAC,CAAC;IACtC,UAAU,EAAE,WAAW,CAAC,YAAY,CAAC,CAAC;IACtC,uDAAuD;IACvD,QAAQ,CAAC,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;IAC3C,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,WAAW,CAAC,QAAQ,CAAC,CAAC;CAC/B;AAED,MAAM,WAAW,gBAAiB,SAAQ,eAAe;IACxD,oDAAoD;IACpD,WAAW,CAAC,EAAE,MAAM,CAAC;CACrB;AAED,MAAM,WAAW,eAAe;IAC/B,IAAI,EAAE,MAAM,CAAC;IACb,YAAY,EAAE,kBAAkB,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,WAAW,EAAE,MAAM,CAAC;IACpB,uDAAuD;IACvD,QAAQ,CAAC,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;CAC3C;AAYD,wBAAsB,kBAAkB,CAAC,OAAO,EAAE,eAAe,GAAG,OAAO,CAAC,cAAc,CAAC,CA+I1F;AAED,wBAAsB,SAAS,CAAC,OAAO,EAAE,gBAAgB,GAAG,OAAO,CAAC,eAAe,CAAC,CAyBnF","sourcesContent":["/**\n * Hybrid search orchestrator: resolve mode, run retrievers in parallel, fuse\n * by rank, expand within budget, trace everything\n * (docs/hybrid-retrieval-design.md).\n *\n * Single-retriever modes flow through the same pipeline — rrfFuse over one\n * list preserves its order — so lexical, semantic, and hybrid all produce the\n * same result shape and the same trace record.\n *\n * `retrieveCandidates` is the candidate-level core (also used by the eval\n * harness, which needs forced modes, a configurable `k`, and no trace\n * pollution); `runSearch` wraps it with span expansion and tracing for the\n * tool.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { adaptGrepHits, type ChunkLookup } from \"./adapter.js\";\nimport { assembleContext } from \"./context-assembler.js\";\nimport { crossEncoderRerank } from \"./cross-rerank.js\";\nimport { runLexicalRetriever } from \"./lexical-retriever.js\";\nimport { resolveSearchMode } from \"./mode.js\";\nimport { rerankCandidates } from \"./rerank.js\";\nimport { DEFAULT_RRF_K, rrfFuse } from \"./rrf.js\";\nimport { writeSearchTrace } from \"./trace.js\";\nimport type { CandidateSpan, FusedCandidate, RankedHit, ResolvedSearchMode, SearchMode, SearchTrace } from \"./types.js\";\n\n/** Raw grep line-hits fetched per query (pre-collapse). */\nconst LEXICAL_MATCH_LIMIT = 200;\n/** Adapted lexical candidates entering fusion. The eval gate showed the\n *  uncapped lexical tail diluting hybrid below plain semantic: lexical\n *  precision is front-loaded by the adapter's term-evidence ranking, while\n *  RRF weighs a rank-30 lexical candidate like a rank-30 embedding hit. */\nconst LEXICAL_FUSION_CAP = 20;\n/**\n * Embedding hits fetched per query when fusing with the grep leg.\n *\n * Deliberately shallow. The grep list is capped at {@link LEXICAL_FUSION_CAP}\n * because its tail is unranked noise, so a deep dense pool here simply\n * outnumbers it: raising this to 200 alongside a 20-candidate grep list cost\n * `auto +rr` 0.472 -> 0.450 MRR.\n */\nconst EMBED_TOP_K = 50;\n/**\n * Per-leg depth when both retrievers are ranked ones (dense + BM25).\n *\n * Four times {@link FUSED_WINDOW}, matching what the daemon's `query_hybrid`\n * does internally (`pool = 4·k`). Fusing at the same depth as the window loses\n * any candidate ranked well by one retriever but just outside the other's\n * top-50 — worth 12pp R@10 and 8pp R@50 for `bm25+dense`, which is exactly\n * what closed the gap to daemon-side fusion.\n */\nconst FUSION_POOL_TOP_K = 200;\n/** BM25 depth, matching {@link FUSION_POOL_TOP_K} so neither ranked leg is\n *  structurally advantaged by pool size. The daemon returns only documents\n *  sharing a query term, so this is an upper bound, not a fill. */\nconst BM25_TOP_K = FUSION_POOL_TOP_K;\n/** Fused candidates kept for reranking / final slicing. */\nconst FUSED_WINDOW = 50;\n\nexport interface RetrieveOptions {\n\tcwd: string;\n\tquery: string;\n\tmode?: SearchMode;\n\t/** Optional glob filter applied to file paths. */\n\tglob?: string;\n\t/** Maximum fused candidates returned. */\n\tlimit?: number;\n\t/** RRF constant override (eval harness sweeps this). Default: {@link DEFAULT_RRF_K}. */\n\trrfK?: number;\n\t/** Rerank the fused top-50 before slicing to `limit`. Default: true. */\n\trerank?: boolean;\n\t/**\n\t * Ask the daemon to fuse its own BM25 index with the vectors and return one\n\t * already-fused ranking, instead of taking a dense-only list. Needs a store\n\t * built with `--hybrid`. The fused list arrives as a single \"embed\" leg,\n\t * because a pre-fused ranking has no per-retriever structure left to record.\n\t *\n\t * Prefer {@link bm25Leg}: fusing here keeps the legs separable in the trace\n\t * and lets the grep leg participate.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as its own ranked list and fuse it here,\n\t * alongside dense and grep. Needs a store built with `--hybrid` and an\n\t * embsearch new enough to serve `retriever: \"lexical\"`.\n\t */\n\tbm25Leg?: boolean;\n\t/**\n\t * Reorder the fused shortlist with the daemon's cross-encoder instead of\n\t * the deterministic reranker. Needs embsearch >= 0.3.0; costs one model\n\t * pass per scored candidate.\n\t */\n\tcrossEncoder?: boolean;\n\tservice?: EmbsearchService;\n\tsignal?: AbortSignal;\n}\n\nexport interface RetrieveResult {\n\tcandidates: FusedCandidate[];\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tindexPhase: SearchTrace[\"indexPhase\"];\n\tretrievers: SearchTrace[\"retrievers\"];\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n\trrfK: number;\n\trerank?: SearchTrace[\"rerank\"];\n}\n\nexport interface RunSearchOptions extends RetrieveOptions {\n\t/** Approximate token budget for the result text. */\n\ttokenBudget?: number;\n}\n\nexport interface RunSearchResult {\n\ttext: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tresultCount: number;\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n}\n\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\nexport async function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult> {\n\tconst { cwd, query, service, signal } = options;\n\tconst glob = normalizeSearchGlob(options.glob);\n\tconst requestedMode = options.mode ?? \"auto\";\n\tconst limit = Math.max(1, options.limit ?? 10);\n\tconst rrfK = options.rrfK ?? DEFAULT_RRF_K;\n\n\tconst state = service?.getState();\n\tconst embedAvailable = service?.isAvailable() ?? false;\n\tconst embedUnavailableReason =\n\t\tstate === undefined\n\t\t\t? \"semantic index is not enabled\"\n\t\t\t: state.phase === \"unavailable\" || state.phase === \"skipped\"\n\t\t\t\t? state.reason\n\t\t\t\t: state.phase === \"idle\"\n\t\t\t\t\t? \"semantic index has not started\"\n\t\t\t\t\t: undefined;\n\n\tconst resolution = resolveSearchMode(query, requestedMode, embedAvailable, embedUnavailableReason);\n\tconst mode = resolution.mode;\n\n\t// Map lexical hits onto indexed chunk ids whenever the sidecar is usable,\n\t// even in lexical-only mode, so identities line up across modes.\n\tconst lookupChunk: ChunkLookup | undefined = embedAvailable\n\t\t? (rel, line) => service!.findEnclosingChunk(rel, line)\n\t\t: undefined;\n\n\tconst spans = new Map<string, CandidateSpan>();\n\tconst lists: RankedHit[][] = [];\n\tconst retrieverStats: SearchTrace[\"retrievers\"] = {};\n\tconst errors: Error[] = [];\n\n\tconst runLexical = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\tconst lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal });\n\t\t\tconst adapted = adaptGrepHits(lineHits, lookupChunk);\n\t\t\t// In single-retriever lexical mode the full list is the result; in\n\t\t\t// hybrid, only the front-loaded head is trustworthy enough to vote.\n\t\t\tconst hits = mode === \"hybrid\" ? adapted.hits.slice(0, LEXICAL_FUSION_CAP) : adapted.hits;\n\t\t\tfor (const [id, span] of adapted.spans) if (!spans.has(id)) spans.set(id, span);\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runEmbed = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// The flat index pads top-k with whatever exists; with the cosine\n\t\t\t// metric the store uses, score <= 0 means \"no relation at all\", so\n\t\t\t// those padding hits would cast RRF votes on pure noise.\n\t\t\t// A BM25 leg is itself ranked, so the pair can afford — and needs — the\n\t\t\t// deeper pool; the grep leg cannot (see EMBED_TOP_K).\n\t\t\tconst topK = options.bm25Leg ? FUSION_POOL_TOP_K : EMBED_TOP_K;\n\t\t\tconst chunkHits = (\n\t\t\t\tawait service!.searchChunks(query, topK, glob, options.daemonHybrid ? \"hybrid\" : \"dense\")\n\t\t\t).filter((hit) => hit.score > 0);\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"embed\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runBm25 = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Raw BM25 sums; only the ordering enters fusion, the score is a\n\t\t\t// diagnostic. Unlike the dense leg there is no zero-score padding to\n\t\t\t// filter — the daemon omits documents sharing no query term.\n\t\t\tconst chunkHits = await service!.searchChunks(query, BM25_TOP_K, glob, \"lexical\");\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"bm25\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tif (!spans.has(hit.id)) {\n\t\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t\t}\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runs: Promise<void>[] = [];\n\tif (mode === \"lexical\" || mode === \"hybrid\") runs.push(runLexical());\n\tif (mode === \"semantic\" || mode === \"hybrid\") runs.push(runEmbed());\n\tif (options.bm25Leg && embedAvailable && mode !== \"lexical\") runs.push(runBm25());\n\tawait Promise.all(runs);\n\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t// A partial failure in hybrid degrades to whichever retriever survived;\n\t// only a total loss is an error.\n\tif (lists.length === 0) throw errors[0] ?? new Error(\"search produced no retriever results\");\n\n\tconst fused = rrfFuse(lists, rrfK).slice(0, FUSED_WINDOW);\n\tlet candidates: FusedCandidate[] = [];\n\tfor (const hit of fused) {\n\t\tconst span = spans.get(hit.id);\n\t\tif (span) candidates.push({ ...hit, ...span });\n\t}\n\n\tlet rerankInfo: SearchTrace[\"rerank\"];\n\tif (options.crossEncoder && service?.supportsCrossEncoder()) {\n\t\tconst reranked = await crossEncoderRerank(query, candidates, cwd, service);\n\t\trerankInfo = { applied: true, candidateCount: reranked.scored, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t} else if (options.rerank !== false) {\n\t\tconst reranked = rerankCandidates(query, candidates, cwd);\n\t\trerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t}\n\tcandidates = candidates.slice(0, limit);\n\n\treturn {\n\t\tcandidates,\n\t\tresolvedMode: mode,\n\t\tdegradedReason: resolution.degradedReason,\n\t\tindexPhase: state?.phase === \"ready\" ? \"ready\" : state?.phase === \"indexing\" ? \"indexing\" : \"unavailable\",\n\t\tretrievers: retrieverStats,\n\t\tindexing: state?.phase === \"indexing\" ? { done: state.done, total: state.total } : undefined,\n\t\trrfK,\n\t\trerank: rerankInfo,\n\t};\n}\n\nexport async function runSearch(options: RunSearchOptions): Promise<RunSearchResult> {\n\tconst retrieved = await retrieveCandidates(options);\n\n\tconst assembled = assembleContext(retrieved.candidates, { cwd: options.cwd, tokenBudget: options.tokenBudget });\n\n\twriteSearchTrace(options.cwd, {\n\t\ttimestampMs: Date.now(),\n\t\tquery: options.query,\n\t\trequestedMode: options.mode ?? \"auto\",\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tindexPhase: retrieved.indexPhase,\n\t\trrfK: retrieved.resolvedMode === \"hybrid\" ? retrieved.rrfK : undefined,\n\t\tretrievers: retrieved.retrievers,\n\t\tfused: retrieved.candidates.map(({ id, rrfScore, ranks, rawScores }) => ({ id, rrfScore, ranks, rawScores })),\n\t\trerank: retrieved.rerank,\n\t});\n\n\treturn {\n\t\ttext: assembled.text,\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tresultCount: retrieved.candidates.length,\n\t\tindexing: retrieved.indexing,\n\t};\n}\n"]}
+{"version":3,"file":"hybrid-search.d.ts","sourceRoot":"","sources":["../../../src/core/search/hybrid-search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAS1E,OAAO,KAAK,EAAiB,cAAc,EAAa,kBAAkB,EAAE,UAAU,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAuCxH,MAAM,WAAW,eAAe;IAC/B,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,CAAC,EAAE,UAAU,CAAC;IAClB,kDAAkD;IAClD,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,yCAAyC;IACzC,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,wFAAwF;IACxF,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,wEAAwE;IACxE,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;;;;OAQG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;;;OAIG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;;;;OAIG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB,OAAO,CAAC,EAAE,gBAAgB,CAAC;IAC3B,MAAM,CAAC,EAAE,WAAW,CAAC;CACrB;AAED,MAAM,WAAW,cAAc;IAC9B,UAAU,EAAE,cAAc,EAAE,CAAC;IAC7B,YAAY,EAAE,kBAAkB,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,UAAU,EAAE,WAAW,CAAC,YAAY,CAAC,CAAC;IACtC,UAAU,EAAE,WAAW,CAAC,YAAY,CAAC,CAAC;IACtC,uDAAuD;IACvD,QAAQ,CAAC,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;IAC3C,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,WAAW,CAAC,QAAQ,CAAC,CAAC;CAC/B;AAED,MAAM,WAAW,gBAAiB,SAAQ,eAAe;IACxD,oDAAoD;IACpD,WAAW,CAAC,EAAE,MAAM,CAAC;CACrB;AAED,MAAM,WAAW,eAAe;IAC/B,IAAI,EAAE,MAAM,CAAC;IACb,YAAY,EAAE,kBAAkB,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,WAAW,EAAE,MAAM,CAAC;IACpB,uDAAuD;IACvD,QAAQ,CAAC,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;CAC3C;AAYD;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,oBAAoB,CACnC,UAAU,EAAE,SAAS,cAAc,EAAE,EACrC,UAAU,EAAE,WAAW,CAAC,MAAM,CAAC,GAC7B,cAAc,EAAE,CAQlB;AAED,wBAAsB,kBAAkB,CAAC,OAAO,EAAE,eAAe,GAAG,OAAO,CAAC,cAAc,CAAC,CAsK1F;AAED,wBAAsB,SAAS,CAAC,OAAO,EAAE,gBAAgB,GAAG,OAAO,CAAC,eAAe,CAAC,CAyBnF","sourcesContent":["/**\n * Hybrid search orchestrator: resolve mode, run retrievers in parallel, fuse\n * by rank, expand within budget, trace everything\n * (docs/hybrid-retrieval-design.md).\n *\n * Single-retriever modes flow through the same pipeline — rrfFuse over one\n * list preserves its order — so lexical, semantic, and hybrid all produce the\n * same result shape and the same trace record.\n *\n * `retrieveCandidates` is the candidate-level core (also used by the eval\n * harness, which needs forced modes, a configurable `k`, and no trace\n * pollution); `runSearch` wraps it with span expansion and tracing for the\n * tool.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { adaptGrepHits, type ChunkLookup } from \"./adapter.js\";\nimport { assembleContext } from \"./context-assembler.js\";\nimport { crossEncoderRerank } from \"./cross-rerank.js\";\nimport { runLexicalRetriever } from \"./lexical-retriever.js\";\nimport { resolveSearchMode } from \"./mode.js\";\nimport { rerankCandidates } from \"./rerank.js\";\nimport { DEFAULT_RRF_K, rrfFuse } from \"./rrf.js\";\nimport { writeSearchTrace } from \"./trace.js\";\nimport type { CandidateSpan, FusedCandidate, RankedHit, ResolvedSearchMode, SearchMode, SearchTrace } from \"./types.js\";\n\n/** Raw grep line-hits fetched per query (pre-collapse). */\nconst LEXICAL_MATCH_LIMIT = 200;\n/** Adapted lexical candidates entering fusion. The eval gate showed the\n *  uncapped lexical tail diluting hybrid below plain semantic: lexical\n *  precision is front-loaded by the adapter's term-evidence ranking, while\n *  RRF weighs a rank-30 lexical candidate like a rank-30 embedding hit. */\nconst LEXICAL_FUSION_CAP = 20;\n/**\n * Embedding hits fetched per query when fusing with the grep leg.\n *\n * Deliberately shallow. The grep list is capped at {@link LEXICAL_FUSION_CAP}\n * because its tail is unranked noise, so a deep dense pool here simply\n * outnumbers it: raising this to 200 alongside a 20-candidate grep list cost\n * `auto +rr` 0.472 -> 0.450 MRR.\n */\nconst EMBED_TOP_K = 50;\n/**\n * Per-leg depth when both retrievers are ranked ones (dense + BM25).\n *\n * Four times {@link FUSED_WINDOW}, matching what the daemon's `query_hybrid`\n * does internally (`pool = 4·k`). Fusing at the same depth as the window loses\n * any candidate ranked well by one retriever but just outside the other's\n * top-50 — worth 12pp R@10 and 8pp R@50 for `bm25+dense`, which is exactly\n * what closed the gap to daemon-side fusion.\n */\nconst FUSION_POOL_TOP_K = 200;\n/** BM25 depth, matching {@link FUSION_POOL_TOP_K} so neither ranked leg is\n *  structurally advantaged by pool size. The daemon returns only documents\n *  sharing a query term, so this is an upper bound, not a fill. */\nconst BM25_TOP_K = FUSION_POOL_TOP_K;\n/** Candidates from unindexed/stale files moved ahead of the fused window.\n *  Small on purpose: this answers \"what did I just write\", not \"search the\n *  working tree\". */\nconst STALE_HOIST_CAP = 5;\n/** Fused candidates kept for reranking / final slicing. */\nconst FUSED_WINDOW = 50;\n\nexport interface RetrieveOptions {\n\tcwd: string;\n\tquery: string;\n\tmode?: SearchMode;\n\t/** Optional glob filter applied to file paths. */\n\tglob?: string;\n\t/** Maximum fused candidates returned. */\n\tlimit?: number;\n\t/** RRF constant override (eval harness sweeps this). Default: {@link DEFAULT_RRF_K}. */\n\trrfK?: number;\n\t/** Rerank the fused top-50 before slicing to `limit`. Default: true. */\n\trerank?: boolean;\n\t/**\n\t * Ask the daemon to fuse its own BM25 index with the vectors and return one\n\t * already-fused ranking, instead of taking a dense-only list. Needs a store\n\t * built with `--hybrid`. The fused list arrives as a single \"embed\" leg,\n\t * because a pre-fused ranking has no per-retriever structure left to record.\n\t *\n\t * Prefer {@link bm25Leg}: fusing here keeps the legs separable in the trace\n\t * and lets the grep leg participate.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as its own ranked list and fuse it here,\n\t * alongside dense and grep. Needs a store built with `--hybrid` and an\n\t * embsearch new enough to serve `retriever: \"lexical\"`.\n\t */\n\tbm25Leg?: boolean;\n\t/**\n\t * Reorder the fused shortlist with the daemon's cross-encoder instead of\n\t * the deterministic reranker. Needs embsearch >= 0.3.0; costs one model\n\t * pass per scored candidate.\n\t */\n\tcrossEncoder?: boolean;\n\tservice?: EmbsearchService;\n\tsignal?: AbortSignal;\n}\n\nexport interface RetrieveResult {\n\tcandidates: FusedCandidate[];\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tindexPhase: SearchTrace[\"indexPhase\"];\n\tretrievers: SearchTrace[\"retrievers\"];\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n\trrfK: number;\n\trerank?: SearchTrace[\"rerank\"];\n}\n\nexport interface RunSearchOptions extends RetrieveOptions {\n\t/** Approximate token budget for the result text. */\n\ttokenBudget?: number;\n}\n\nexport interface RunSearchResult {\n\ttext: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tresultCount: number;\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n}\n\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\n/**\n * Move candidates from files the index has not read to the front.\n *\n * They are there because grep found them and nothing else could: the index is\n * ranking a stale copy of the file, or has never seen it. Left to fuse, they\n * lose — RRF rewards agreement, and one leg reporting a single document is\n * outvoted by two legs agreeing on hundreds. Measured: scoping grep to stale\n * files without this hoist scored 25% on the live-edit set where unscoped grep\n * scored 100%, because the fused window filled with consensus hits about the\n * indexed copy.\n *\n * Capped, because \"stale\" scales with how far behind the index is. A few\n * edited files is the case this exists for; a fresh checkout makes everything\n * stale, and hoisting all of it would quietly turn hybrid search back into\n * grep. Past the cap the remainder keeps its fused position.\n */\nexport function hoistStaleCandidates(\n\tcandidates: readonly FusedCandidate[],\n\tstaleFiles: ReadonlySet<string>,\n): FusedCandidate[] {\n\tif (staleFiles.size === 0) return [...candidates];\n\tconst hoisted: FusedCandidate[] = [];\n\tconst rest: FusedCandidate[] = [];\n\tfor (const candidate of candidates) {\n\t\t(staleFiles.has(candidate.path) && hoisted.length < STALE_HOIST_CAP ? hoisted : rest).push(candidate);\n\t}\n\treturn [...hoisted, ...rest];\n}\n\nexport async function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult> {\n\tconst { cwd, query, service, signal } = options;\n\tconst glob = normalizeSearchGlob(options.glob);\n\tconst requestedMode = options.mode ?? \"auto\";\n\tconst limit = Math.max(1, options.limit ?? 10);\n\tconst rrfK = options.rrfK ?? DEFAULT_RRF_K;\n\n\tconst state = service?.getState();\n\tconst embedAvailable = service?.isAvailable() ?? false;\n\tconst embedUnavailableReason =\n\t\tstate === undefined\n\t\t\t? \"semantic index is not enabled\"\n\t\t\t: state.phase === \"unavailable\" || state.phase === \"skipped\"\n\t\t\t\t? state.reason\n\t\t\t\t: state.phase === \"idle\"\n\t\t\t\t\t? \"semantic index has not started\"\n\t\t\t\t\t: undefined;\n\n\tconst resolution = resolveSearchMode(query, requestedMode, embedAvailable, embedUnavailableReason);\n\tconst mode = resolution.mode;\n\n\t// Map lexical hits onto indexed chunk ids whenever the sidecar is usable,\n\t// even in lexical-only mode, so identities line up across modes.\n\tconst lookupChunk: ChunkLookup | undefined = embedAvailable\n\t\t? (rel, line) => service!.findEnclosingChunk(rel, line)\n\t\t: undefined;\n\n\tconst spans = new Map<string, CandidateSpan>();\n\t/** Files the index does not have current content for, when grep is scoped\n\t *  to them. Read again after fusion — see the hoist below. */\n\tconst staleSet = new Set<string>();\n\tconst lists: RankedHit[][] = [];\n\tconst retrieverStats: SearchTrace[\"retrievers\"] = {};\n\tconst errors: Error[] = [];\n\n\tconst runLexical = async (scopeToStale = false): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Scoped run: grep covers only what the index has not read yet, so it\n\t\t\t// adds the one thing BM25 cannot see without re-voting on documents\n\t\t\t// BM25 already ranked. Fusing two lexical views of the same corpus is\n\t\t\t// what made the three-leg configuration lose (18 of 62 queries worse).\n\t\t\tconst paths = scopeToStale ? service?.staleFiles(signal) : undefined;\n\t\t\tif (paths) for (const rel of paths) staleSet.add(rel);\n\t\t\tconst lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal, paths });\n\t\t\tconst adapted = adaptGrepHits(lineHits, lookupChunk);\n\t\t\t// In single-retriever lexical mode the full list is the result; in\n\t\t\t// hybrid, only the front-loaded head is trustworthy enough to vote.\n\t\t\tconst hits = mode === \"hybrid\" ? adapted.hits.slice(0, LEXICAL_FUSION_CAP) : adapted.hits;\n\t\t\tfor (const [id, span] of adapted.spans) if (!spans.has(id)) spans.set(id, span);\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runEmbed = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// The flat index pads top-k with whatever exists; with the cosine\n\t\t\t// metric the store uses, score <= 0 means \"no relation at all\", so\n\t\t\t// those padding hits would cast RRF votes on pure noise.\n\t\t\t// A BM25 leg is itself ranked, so the pair can afford — and needs — the\n\t\t\t// deeper pool; the grep leg cannot (see EMBED_TOP_K).\n\t\t\tconst topK = options.bm25Leg ? FUSION_POOL_TOP_K : EMBED_TOP_K;\n\t\t\tconst chunkHits = (\n\t\t\t\tawait service!.searchChunks(query, topK, glob, options.daemonHybrid ? \"hybrid\" : \"dense\")\n\t\t\t).filter((hit) => hit.score > 0);\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"embed\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runBm25 = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Raw BM25 sums; only the ordering enters fusion, the score is a\n\t\t\t// diagnostic. Unlike the dense leg there is no zero-score padding to\n\t\t\t// filter — the daemon omits documents sharing no query term.\n\t\t\tconst chunkHits = await service!.searchChunks(query, BM25_TOP_K, glob, \"lexical\");\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"bm25\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tif (!spans.has(hit.id)) {\n\t\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t\t}\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\t// BM25 is the better lexical leg where the index is current: it beats grep\n\t// on Recall@50 (0.790 -> 0.879, 6 queries better and 0 worse, p <= 0.05)\n\t// because it ranks the whole corpus rather than truncating a match stream.\n\t// It is also blind to anything indexed later than it was written, which is\n\t// precisely where grep still wins — so grep runs scoped to that set instead\n\t// of being dropped or left to duplicate BM25 over the whole tree.\n\tconst bm25AsLexicalLeg = options.bm25Leg && embedAvailable && mode !== \"lexical\";\n\n\tconst runs: Promise<void>[] = [];\n\tif (mode === \"lexical\" || mode === \"hybrid\") runs.push(runLexical(bm25AsLexicalLeg));\n\tif (mode === \"semantic\" || mode === \"hybrid\") runs.push(runEmbed());\n\t// `semantic` gets the BM25 leg too: the caller asked for the index, and\n\t// BM25 is part of it. Only an explicit `lexical` request excludes it.\n\tif (bm25AsLexicalLeg) runs.push(runBm25());\n\tawait Promise.all(runs);\n\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t// A partial failure in hybrid degrades to whichever retriever survived;\n\t// only a total loss is an error.\n\tif (lists.length === 0) throw errors[0] ?? new Error(\"search produced no retriever results\");\n\n\tconst fused = rrfFuse(lists, rrfK).slice(0, FUSED_WINDOW);\n\tlet candidates: FusedCandidate[] = [];\n\tfor (const hit of fused) {\n\t\tconst span = spans.get(hit.id);\n\t\tif (span) candidates.push({ ...hit, ...span });\n\t}\n\n\tlet rerankInfo: SearchTrace[\"rerank\"];\n\tif (options.crossEncoder && service?.supportsCrossEncoder()) {\n\t\tconst reranked = await crossEncoderRerank(query, candidates, cwd, service);\n\t\trerankInfo = { applied: true, candidateCount: reranked.scored, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t} else if (options.rerank !== false) {\n\t\tconst reranked = rerankCandidates(query, candidates, cwd);\n\t\trerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t}\n\t// After reranking, not before: both rerankers weight the fused prior, and a\n\t// lone grep hit from an unindexed file has the lowest prior there is, so\n\t// hoisting first would simply be undone. The reranker still orders the\n\t// hoisted set against itself.\n\tcandidates = hoistStaleCandidates(candidates, staleSet).slice(0, limit);\n\n\treturn {\n\t\tcandidates,\n\t\tresolvedMode: mode,\n\t\tdegradedReason: resolution.degradedReason,\n\t\tindexPhase: state?.phase === \"ready\" ? \"ready\" : state?.phase === \"indexing\" ? \"indexing\" : \"unavailable\",\n\t\tretrievers: retrieverStats,\n\t\tindexing: state?.phase === \"indexing\" ? { done: state.done, total: state.total } : undefined,\n\t\trrfK,\n\t\trerank: rerankInfo,\n\t};\n}\n\nexport async function runSearch(options: RunSearchOptions): Promise<RunSearchResult> {\n\tconst retrieved = await retrieveCandidates(options);\n\n\tconst assembled = assembleContext(retrieved.candidates, { cwd: options.cwd, tokenBudget: options.tokenBudget });\n\n\twriteSearchTrace(options.cwd, {\n\t\ttimestampMs: Date.now(),\n\t\tquery: options.query,\n\t\trequestedMode: options.mode ?? \"auto\",\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tindexPhase: retrieved.indexPhase,\n\t\trrfK: retrieved.resolvedMode === \"hybrid\" ? retrieved.rrfK : undefined,\n\t\tretrievers: retrieved.retrievers,\n\t\tfused: retrieved.candidates.map(({ id, rrfScore, ranks, rawScores }) => ({ id, rrfScore, ranks, rawScores })),\n\t\trerank: retrieved.rerank,\n\t});\n\n\treturn {\n\t\ttext: assembled.text,\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tresultCount: retrieved.candidates.length,\n\t\tindexing: retrieved.indexing,\n\t};\n}\n"]}

package/dist/core/search/hybrid-search.js

@@ -50,6 +50,10 @@ const FUSION_POOL_TOP_K = 200;
  *  structurally advantaged by pool size. The daemon returns only documents
  *  sharing a query term, so this is an upper bound, not a fill. */
 const BM25_TOP_K = FUSION_POOL_TOP_K;
+/** Candidates from unindexed/stale files moved ahead of the fused window.
+ *  Small on purpose: this answers "what did I just write", not "search the
+ *  working tree". */
+const STALE_HOIST_CAP = 5;
 /** Fused candidates kept for reranking / final slicing. */
 const FUSED_WINDOW = 50;
 function normalizeSearchGlob(glob) {
@@ -62,6 +66,32 @@ function normalizeSearchGlob(glob) {
     }
     return glob;
 }
+/**
+ * Move candidates from files the index has not read to the front.
+ *
+ * They are there because grep found them and nothing else could: the index is
+ * ranking a stale copy of the file, or has never seen it. Left to fuse, they
+ * lose — RRF rewards agreement, and one leg reporting a single document is
+ * outvoted by two legs agreeing on hundreds. Measured: scoping grep to stale
+ * files without this hoist scored 25% on the live-edit set where unscoped grep
+ * scored 100%, because the fused window filled with consensus hits about the
+ * indexed copy.
+ *
+ * Capped, because "stale" scales with how far behind the index is. A few
+ * edited files is the case this exists for; a fresh checkout makes everything
+ * stale, and hoisting all of it would quietly turn hybrid search back into
+ * grep. Past the cap the remainder keeps its fused position.
+ */
+export function hoistStaleCandidates(candidates, staleFiles) {
+    if (staleFiles.size === 0)
+        return [...candidates];
+    const hoisted = [];
+    const rest = [];
+    for (const candidate of candidates) {
+        (staleFiles.has(candidate.path) && hoisted.length < STALE_HOIST_CAP ? hoisted : rest).push(candidate);
+    }
+    return [...hoisted, ...rest];
+}
 export async function retrieveCandidates(options) {
     const { cwd, query, service, signal } = options;
     const glob = normalizeSearchGlob(options.glob);
@@ -85,13 +115,24 @@ export async function retrieveCandidates(options) {
         ? (rel, line) => service.findEnclosingChunk(rel, line)
         : undefined;
     const spans = new Map();
+    /** Files the index does not have current content for, when grep is scoped
+     *  to them. Read again after fusion — see the hoist below. */
+    const staleSet = new Set();
     const lists = [];
     const retrieverStats = {};
     const errors = [];
-    const runLexical = async () => {
+    const runLexical = async (scopeToStale = false) => {
         const startedMs = Date.now();
         try {
-            const lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal });
+            // Scoped run: grep covers only what the index has not read yet, so it
+            // adds the one thing BM25 cannot see without re-voting on documents
+            // BM25 already ranked. Fusing two lexical views of the same corpus is
+            // what made the three-leg configuration lose (18 of 62 queries worse).
+            const paths = scopeToStale ? service?.staleFiles(signal) : undefined;
+            if (paths)
+                for (const rel of paths)
+                    staleSet.add(rel);
+            const lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal, paths });
             const adapted = adaptGrepHits(lineHits, lookupChunk);
             // In single-retriever lexical mode the full list is the result; in
             // hybrid, only the front-loaded head is trustworthy enough to vote.
@@ -160,12 +201,21 @@ export async function retrieveCandidates(options) {
             retrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };
         }
     };
+    // BM25 is the better lexical leg where the index is current: it beats grep
+    // on Recall@50 (0.790 -> 0.879, 6 queries better and 0 worse, p <= 0.05)
+    // because it ranks the whole corpus rather than truncating a match stream.
+    // It is also blind to anything indexed later than it was written, which is
+    // precisely where grep still wins — so grep runs scoped to that set instead
+    // of being dropped or left to duplicate BM25 over the whole tree.
+    const bm25AsLexicalLeg = options.bm25Leg && embedAvailable && mode !== "lexical";
     const runs = [];
     if (mode === "lexical" || mode === "hybrid")
-        runs.push(runLexical());
+        runs.push(runLexical(bm25AsLexicalLeg));
     if (mode === "semantic" || mode === "hybrid")
         runs.push(runEmbed());
-    if (options.bm25Leg && embedAvailable && mode !== "lexical")
+    // `semantic` gets the BM25 leg too: the caller asked for the index, and
+    // BM25 is part of it. Only an explicit `lexical` request excludes it.
+    if (bm25AsLexicalLeg)
         runs.push(runBm25());
     await Promise.all(runs);
     if (signal?.aborted)
@@ -192,7 +242,11 @@ export async function retrieveCandidates(options) {
         rerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };
         candidates = reranked.candidates;
     }
-    candidates = candidates.slice(0, limit);
+    // After reranking, not before: both rerankers weight the fused prior, and a
+    // lone grep hit from an unindexed file has the lowest prior there is, so
+    // hoisting first would simply be undone. The reranker still orders the
+    // hoisted set against itself.
+    candidates = hoistStaleCandidates(candidates, staleSet).slice(0, limit);
     return {
         candidates,
         resolvedMode: mode,

package/dist/core/search/hybrid-search.js.map

@@ -1 +1 @@
-{"version":3,"file":"hybrid-search.js","sourceRoot":"","sources":["../../../src/core/search/hybrid-search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAGH,OAAO,EAAE,aAAa,EAAoB,MAAM,cAAc,CAAC;AAC/D,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AACvD,OAAO,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAC;AAC7D,OAAO,EAAE,iBAAiB,EAAE,MAAM,WAAW,CAAC;AAC9C,OAAO,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAC/C,OAAO,EAAE,aAAa,EAAE,OAAO,EAAE,MAAM,UAAU,CAAC;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAG9C,2DAA2D;AAC3D,MAAM,mBAAmB,GAAG,GAAG,CAAC;AAChC;;;2EAG2E;AAC3E,MAAM,kBAAkB,GAAG,EAAE,CAAC;AAC9B;;;;;;;GAOG;AACH,MAAM,WAAW,GAAG,EAAE,CAAC;AACvB;;;;;;;;GAQG;AACH,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAC9B;;mEAEmE;AACnE,MAAM,UAAU,GAAG,iBAAiB,CAAC;AACrC,2DAA2D;AAC3D,MAAM,YAAY,GAAG,EAAE,CAAC;AAkExB,SAAS,mBAAmB,CAAC,IAAwB,EAAsB;IAC1E,IAAI,CAAC,IAAI;QAAE,OAAO,SAAS,CAAC;IAC5B,yEAAyE;IACzE,6EAA6E;IAC7E,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;QAC5E,OAAO,MAAM,IAAI,EAAE,CAAC;IACrB,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,OAAwB,EAA2B;IAC3F,MAAM,EAAE,GAAG,EAAE,KAAK,EAAE,OAAO,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IAChD,MAAM,IAAI,GAAG,mBAAmB,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IAC/C,MAAM,aAAa,GAAG,OAAO,CAAC,IAAI,IAAI,MAAM,CAAC;IAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;IAC/C,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,IAAI,aAAa,CAAC;IAE3C,MAAM,KAAK,GAAG,OAAO,EAAE,QAAQ,EAAE,CAAC;IAClC,MAAM,cAAc,GAAG,OAAO,EAAE,WAAW,EAAE,IAAI,KAAK,CAAC;IACvD,MAAM,sBAAsB,GAC3B,KAAK,KAAK,SAAS;QAClB,CAAC,CAAC,+BAA+B;QACjC,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,aAAa,IAAI,KAAK,CAAC,KAAK,KAAK,SAAS;YAC3D,CAAC,CAAC,KAAK,CAAC,MAAM;YACd,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,MAAM;gBACvB,CAAC,CAAC,gCAAgC;gBAClC,CAAC,CAAC,SAAS,CAAC;IAEhB,MAAM,UAAU,GAAG,iBAAiB,CAAC,KAAK,EAAE,aAAa,EAAE,cAAc,EAAE,sBAAsB,CAAC,CAAC;IACnG,MAAM,IAAI,GAAG,UAAU,CAAC,IAAI,CAAC;IAE7B,0EAA0E;IAC1E,iEAAiE;IACjE,MAAM,WAAW,GAA4B,cAAc;QAC1D,CAAC,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,CAAC,OAAQ,CAAC,kBAAkB,CAAC,GAAG,EAAE,IAAI,CAAC;QACvD,CAAC,CAAC,SAAS,CAAC;IAEb,MAAM,KAAK,GAAG,IAAI,GAAG,EAAyB,CAAC;IAC/C,MAAM,KAAK,GAAkB,EAAE,CAAC;IAChC,MAAM,cAAc,GAA8B,EAAE,CAAC;IACrD,MAAM,MAAM,GAAY,EAAE,CAAC;IAE3B,MAAM,UAAU,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC7C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,MAAM,QAAQ,GAAG,MAAM,mBAAmB,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,KAAK,EAAE,mBAAmB,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,CAAC;YACrG,MAAM,OAAO,GAAG,aAAa,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;YACrD,mEAAmE;YACnE,oEAAoE;YACpE,MAAM,IAAI,GAAG,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC;YAC1F,KAAK,MAAM,CAAC,EAAE,EAAE,IAAI,CAAC,IAAI,OAAO,CAAC,KAAK;gBAAE,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;oBAAE,KAAK,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;YAChF,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC1E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,QAAQ,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC3C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,kEAAkE;YAClE,mEAAmE;YACnE,yDAAyD;YACzD,4EAAwE;YACxE,sDAAsD;YACtD,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,WAAW,CAAC;YAC/D,MAAM,SAAS,GAAG,CACjB,MAAM,OAAQ,CAAC,YAAY,CAAC,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,CACzF,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;YACjC,MAAM,IAAI,GAAgB,SAAS,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;gBACpD,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC;gBACX,KAAK,EAAE,GAAG,CAAC,KAAK;gBAChB,MAAM,EAAE,OAAO;aACf,CAAC,CAAC,CAAC;YACJ,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;gBAC7B,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;YACvF,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,KAAK,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACrF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,KAAK,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC3E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,OAAO,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC1C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,iEAAiE;YACjE,qEAAqE;YACrE,+DAA6D;YAC7D,MAAM,SAAS,GAAG,MAAM,OAAQ,CAAC,YAAY,CAAC,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,CAAC,CAAC;YAClF,MAAM,IAAI,GAAgB,SAAS,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;gBACpD,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC;gBACX,KAAK,EAAE,GAAG,CAAC,KAAK;gBAChB,MAAM,EAAE,MAAM;aACd,CAAC,CAAC,CAAC;YACJ,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;gBAC7B,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,CAAC;oBACxB,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;gBACvF,CAAC;YACF,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC1E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,IAAI,GAAoB,EAAE,CAAC;IACjC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,QAAQ;QAAE,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,CAAC,CAAC;IACrE,IAAI,IAAI,KAAK,UAAU,IAAI,IAAI,KAAK,QAAQ;QAAE,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC;IACpE,IAAI,OAAO,CAAC,OAAO,IAAI,cAAc,IAAI,IAAI,KAAK,SAAS;QAAE,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC;IAClF,MAAM,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACxB,IAAI,MAAM,EAAE,OAAO;QAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;IAC1D,wEAAwE;IACxE,iCAAiC;IACjC,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,MAAM,MAAM,CAAC,CAAC,CAAC,IAAI,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC;IAE7F,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,YAAY,CAAC,CAAC;IAC1D,IAAI,UAAU,GAAqB,EAAE,CAAC;IACtC,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAC/B,IAAI,IAAI;YAAE,UAAU,CAAC,IAAI,CAAC,EAAE,GAAG,GAAG,EAAE,GAAG,IAAI,EAAE,CAAC,CAAC;IAChD,CAAC;IAED,IAAI,UAAiC,CAAC;IACtC,IAAI,OAAO,CAAC,YAAY,IAAI,OAAO,EAAE,oBAAoB,EAAE,EAAE,CAAC;QAC7D,MAAM,QAAQ,GAAG,MAAM,kBAAkB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,EAAE,OAAO,CAAC,CAAC;QAC3E,UAAU,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,QAAQ,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,SAAS,EAAE,CAAC;QAC/F,UAAU,GAAG,QAAQ,CAAC,UAAU,CAAC;IAClC,CAAC;SAAM,IAAI,OAAO,CAAC,MAAM,KAAK,KAAK,EAAE,CAAC;QACrC,MAAM,QAAQ,GAAG,gBAAgB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,CAAC,CAAC;QAC1D,UAAU,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,UAAU,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,SAAS,EAAE,CAAC;QACjG,UAAU,GAAG,QAAQ,CAAC,UAAU,CAAC;IAClC,CAAC;IACD,UAAU,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAExC,OAAO;QACN,UAAU;QACV,YAAY,EAAE,IAAI;QAClB,cAAc,EAAE,UAAU,CAAC,cAAc;QACzC,UAAU,EAAE,KAAK,EAAE,KAAK,KAAK,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,EAAE,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,aAAa;QACzG,UAAU,EAAE,cAAc;QAC1B,QAAQ,EAAE,KAAK,EAAE,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS;QAC5F,IAAI;QACJ,MAAM,EAAE,UAAU;KAClB,CAAC;AAAA,CACF;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,OAAyB,EAA4B;IACpF,MAAM,SAAS,GAAG,MAAM,kBAAkB,CAAC,OAAO,CAAC,CAAC;IAEpD,MAAM,SAAS,GAAG,eAAe,CAAC,SAAS,CAAC,UAAU,EAAE,EAAE,GAAG,EAAE,OAAO,CAAC,GAAG,EAAE,WAAW,EAAE,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC;IAEhH,gBAAgB,CAAC,OAAO,CAAC,GAAG,EAAE;QAC7B,WAAW,EAAE,IAAI,CAAC,GAAG,EAAE;QACvB,KAAK,EAAE,OAAO,CAAC,KAAK;QACpB,aAAa,EAAE,OAAO,CAAC,IAAI,IAAI,MAAM;QACrC,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;QACxC,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,IAAI,EAAE,SAAS,CAAC,YAAY,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS;QACtE,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,KAAK,EAAE,SAAS,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC,CAAC;QAC7G,MAAM,EAAE,SAAS,CAAC,MAAM;KACxB,CAAC,CAAC;IAEH,OAAO;QACN,IAAI,EAAE,SAAS,CAAC,IAAI;QACpB,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;QACxC,WAAW,EAAE,SAAS,CAAC,UAAU,CAAC,MAAM;QACxC,QAAQ,EAAE,SAAS,CAAC,QAAQ;KAC5B,CAAC;AAAA,CACF","sourcesContent":["/**\n * Hybrid search orchestrator: resolve mode, run retrievers in parallel, fuse\n * by rank, expand within budget, trace everything\n * (docs/hybrid-retrieval-design.md).\n *\n * Single-retriever modes flow through the same pipeline — rrfFuse over one\n * list preserves its order — so lexical, semantic, and hybrid all produce the\n * same result shape and the same trace record.\n *\n * `retrieveCandidates` is the candidate-level core (also used by the eval\n * harness, which needs forced modes, a configurable `k`, and no trace\n * pollution); `runSearch` wraps it with span expansion and tracing for the\n * tool.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { adaptGrepHits, type ChunkLookup } from \"./adapter.js\";\nimport { assembleContext } from \"./context-assembler.js\";\nimport { crossEncoderRerank } from \"./cross-rerank.js\";\nimport { runLexicalRetriever } from \"./lexical-retriever.js\";\nimport { resolveSearchMode } from \"./mode.js\";\nimport { rerankCandidates } from \"./rerank.js\";\nimport { DEFAULT_RRF_K, rrfFuse } from \"./rrf.js\";\nimport { writeSearchTrace } from \"./trace.js\";\nimport type { CandidateSpan, FusedCandidate, RankedHit, ResolvedSearchMode, SearchMode, SearchTrace } from \"./types.js\";\n\n/** Raw grep line-hits fetched per query (pre-collapse). */\nconst LEXICAL_MATCH_LIMIT = 200;\n/** Adapted lexical candidates entering fusion. The eval gate showed the\n *  uncapped lexical tail diluting hybrid below plain semantic: lexical\n *  precision is front-loaded by the adapter's term-evidence ranking, while\n *  RRF weighs a rank-30 lexical candidate like a rank-30 embedding hit. */\nconst LEXICAL_FUSION_CAP = 20;\n/**\n * Embedding hits fetched per query when fusing with the grep leg.\n *\n * Deliberately shallow. The grep list is capped at {@link LEXICAL_FUSION_CAP}\n * because its tail is unranked noise, so a deep dense pool here simply\n * outnumbers it: raising this to 200 alongside a 20-candidate grep list cost\n * `auto +rr` 0.472 -> 0.450 MRR.\n */\nconst EMBED_TOP_K = 50;\n/**\n * Per-leg depth when both retrievers are ranked ones (dense + BM25).\n *\n * Four times {@link FUSED_WINDOW}, matching what the daemon's `query_hybrid`\n * does internally (`pool = 4·k`). Fusing at the same depth as the window loses\n * any candidate ranked well by one retriever but just outside the other's\n * top-50 — worth 12pp R@10 and 8pp R@50 for `bm25+dense`, which is exactly\n * what closed the gap to daemon-side fusion.\n */\nconst FUSION_POOL_TOP_K = 200;\n/** BM25 depth, matching {@link FUSION_POOL_TOP_K} so neither ranked leg is\n *  structurally advantaged by pool size. The daemon returns only documents\n *  sharing a query term, so this is an upper bound, not a fill. */\nconst BM25_TOP_K = FUSION_POOL_TOP_K;\n/** Fused candidates kept for reranking / final slicing. */\nconst FUSED_WINDOW = 50;\n\nexport interface RetrieveOptions {\n\tcwd: string;\n\tquery: string;\n\tmode?: SearchMode;\n\t/** Optional glob filter applied to file paths. */\n\tglob?: string;\n\t/** Maximum fused candidates returned. */\n\tlimit?: number;\n\t/** RRF constant override (eval harness sweeps this). Default: {@link DEFAULT_RRF_K}. */\n\trrfK?: number;\n\t/** Rerank the fused top-50 before slicing to `limit`. Default: true. */\n\trerank?: boolean;\n\t/**\n\t * Ask the daemon to fuse its own BM25 index with the vectors and return one\n\t * already-fused ranking, instead of taking a dense-only list. Needs a store\n\t * built with `--hybrid`. The fused list arrives as a single \"embed\" leg,\n\t * because a pre-fused ranking has no per-retriever structure left to record.\n\t *\n\t * Prefer {@link bm25Leg}: fusing here keeps the legs separable in the trace\n\t * and lets the grep leg participate.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as its own ranked list and fuse it here,\n\t * alongside dense and grep. Needs a store built with `--hybrid` and an\n\t * embsearch new enough to serve `retriever: \"lexical\"`.\n\t */\n\tbm25Leg?: boolean;\n\t/**\n\t * Reorder the fused shortlist with the daemon's cross-encoder instead of\n\t * the deterministic reranker. Needs embsearch >= 0.3.0; costs one model\n\t * pass per scored candidate.\n\t */\n\tcrossEncoder?: boolean;\n\tservice?: EmbsearchService;\n\tsignal?: AbortSignal;\n}\n\nexport interface RetrieveResult {\n\tcandidates: FusedCandidate[];\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tindexPhase: SearchTrace[\"indexPhase\"];\n\tretrievers: SearchTrace[\"retrievers\"];\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n\trrfK: number;\n\trerank?: SearchTrace[\"rerank\"];\n}\n\nexport interface RunSearchOptions extends RetrieveOptions {\n\t/** Approximate token budget for the result text. */\n\ttokenBudget?: number;\n}\n\nexport interface RunSearchResult {\n\ttext: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tresultCount: number;\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n}\n\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\nexport async function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult> {\n\tconst { cwd, query, service, signal } = options;\n\tconst glob = normalizeSearchGlob(options.glob);\n\tconst requestedMode = options.mode ?? \"auto\";\n\tconst limit = Math.max(1, options.limit ?? 10);\n\tconst rrfK = options.rrfK ?? DEFAULT_RRF_K;\n\n\tconst state = service?.getState();\n\tconst embedAvailable = service?.isAvailable() ?? false;\n\tconst embedUnavailableReason =\n\t\tstate === undefined\n\t\t\t? \"semantic index is not enabled\"\n\t\t\t: state.phase === \"unavailable\" || state.phase === \"skipped\"\n\t\t\t\t? state.reason\n\t\t\t\t: state.phase === \"idle\"\n\t\t\t\t\t? \"semantic index has not started\"\n\t\t\t\t\t: undefined;\n\n\tconst resolution = resolveSearchMode(query, requestedMode, embedAvailable, embedUnavailableReason);\n\tconst mode = resolution.mode;\n\n\t// Map lexical hits onto indexed chunk ids whenever the sidecar is usable,\n\t// even in lexical-only mode, so identities line up across modes.\n\tconst lookupChunk: ChunkLookup | undefined = embedAvailable\n\t\t? (rel, line) => service!.findEnclosingChunk(rel, line)\n\t\t: undefined;\n\n\tconst spans = new Map<string, CandidateSpan>();\n\tconst lists: RankedHit[][] = [];\n\tconst retrieverStats: SearchTrace[\"retrievers\"] = {};\n\tconst errors: Error[] = [];\n\n\tconst runLexical = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\tconst lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal });\n\t\t\tconst adapted = adaptGrepHits(lineHits, lookupChunk);\n\t\t\t// In single-retriever lexical mode the full list is the result; in\n\t\t\t// hybrid, only the front-loaded head is trustworthy enough to vote.\n\t\t\tconst hits = mode === \"hybrid\" ? adapted.hits.slice(0, LEXICAL_FUSION_CAP) : adapted.hits;\n\t\t\tfor (const [id, span] of adapted.spans) if (!spans.has(id)) spans.set(id, span);\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runEmbed = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// The flat index pads top-k with whatever exists; with the cosine\n\t\t\t// metric the store uses, score <= 0 means \"no relation at all\", so\n\t\t\t// those padding hits would cast RRF votes on pure noise.\n\t\t\t// A BM25 leg is itself ranked, so the pair can afford — and needs — the\n\t\t\t// deeper pool; the grep leg cannot (see EMBED_TOP_K).\n\t\t\tconst topK = options.bm25Leg ? FUSION_POOL_TOP_K : EMBED_TOP_K;\n\t\t\tconst chunkHits = (\n\t\t\t\tawait service!.searchChunks(query, topK, glob, options.daemonHybrid ? \"hybrid\" : \"dense\")\n\t\t\t).filter((hit) => hit.score > 0);\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"embed\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runBm25 = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Raw BM25 sums; only the ordering enters fusion, the score is a\n\t\t\t// diagnostic. Unlike the dense leg there is no zero-score padding to\n\t\t\t// filter — the daemon omits documents sharing no query term.\n\t\t\tconst chunkHits = await service!.searchChunks(query, BM25_TOP_K, glob, \"lexical\");\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"bm25\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tif (!spans.has(hit.id)) {\n\t\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t\t}\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runs: Promise<void>[] = [];\n\tif (mode === \"lexical\" || mode === \"hybrid\") runs.push(runLexical());\n\tif (mode === \"semantic\" || mode === \"hybrid\") runs.push(runEmbed());\n\tif (options.bm25Leg && embedAvailable && mode !== \"lexical\") runs.push(runBm25());\n\tawait Promise.all(runs);\n\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t// A partial failure in hybrid degrades to whichever retriever survived;\n\t// only a total loss is an error.\n\tif (lists.length === 0) throw errors[0] ?? new Error(\"search produced no retriever results\");\n\n\tconst fused = rrfFuse(lists, rrfK).slice(0, FUSED_WINDOW);\n\tlet candidates: FusedCandidate[] = [];\n\tfor (const hit of fused) {\n\t\tconst span = spans.get(hit.id);\n\t\tif (span) candidates.push({ ...hit, ...span });\n\t}\n\n\tlet rerankInfo: SearchTrace[\"rerank\"];\n\tif (options.crossEncoder && service?.supportsCrossEncoder()) {\n\t\tconst reranked = await crossEncoderRerank(query, candidates, cwd, service);\n\t\trerankInfo = { applied: true, candidateCount: reranked.scored, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t} else if (options.rerank !== false) {\n\t\tconst reranked = rerankCandidates(query, candidates, cwd);\n\t\trerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t}\n\tcandidates = candidates.slice(0, limit);\n\n\treturn {\n\t\tcandidates,\n\t\tresolvedMode: mode,\n\t\tdegradedReason: resolution.degradedReason,\n\t\tindexPhase: state?.phase === \"ready\" ? \"ready\" : state?.phase === \"indexing\" ? \"indexing\" : \"unavailable\",\n\t\tretrievers: retrieverStats,\n\t\tindexing: state?.phase === \"indexing\" ? { done: state.done, total: state.total } : undefined,\n\t\trrfK,\n\t\trerank: rerankInfo,\n\t};\n}\n\nexport async function runSearch(options: RunSearchOptions): Promise<RunSearchResult> {\n\tconst retrieved = await retrieveCandidates(options);\n\n\tconst assembled = assembleContext(retrieved.candidates, { cwd: options.cwd, tokenBudget: options.tokenBudget });\n\n\twriteSearchTrace(options.cwd, {\n\t\ttimestampMs: Date.now(),\n\t\tquery: options.query,\n\t\trequestedMode: options.mode ?? \"auto\",\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tindexPhase: retrieved.indexPhase,\n\t\trrfK: retrieved.resolvedMode === \"hybrid\" ? retrieved.rrfK : undefined,\n\t\tretrievers: retrieved.retrievers,\n\t\tfused: retrieved.candidates.map(({ id, rrfScore, ranks, rawScores }) => ({ id, rrfScore, ranks, rawScores })),\n\t\trerank: retrieved.rerank,\n\t});\n\n\treturn {\n\t\ttext: assembled.text,\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tresultCount: retrieved.candidates.length,\n\t\tindexing: retrieved.indexing,\n\t};\n}\n"]}
+{"version":3,"file":"hybrid-search.js","sourceRoot":"","sources":["../../../src/core/search/hybrid-search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAGH,OAAO,EAAE,aAAa,EAAoB,MAAM,cAAc,CAAC;AAC/D,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AACvD,OAAO,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAC;AAC7D,OAAO,EAAE,iBAAiB,EAAE,MAAM,WAAW,CAAC;AAC9C,OAAO,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAC/C,OAAO,EAAE,aAAa,EAAE,OAAO,EAAE,MAAM,UAAU,CAAC;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAG9C,2DAA2D;AAC3D,MAAM,mBAAmB,GAAG,GAAG,CAAC;AAChC;;;2EAG2E;AAC3E,MAAM,kBAAkB,GAAG,EAAE,CAAC;AAC9B;;;;;;;GAOG;AACH,MAAM,WAAW,GAAG,EAAE,CAAC;AACvB;;;;;;;;GAQG;AACH,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAC9B;;mEAEmE;AACnE,MAAM,UAAU,GAAG,iBAAiB,CAAC;AACrC;;qBAEqB;AACrB,MAAM,eAAe,GAAG,CAAC,CAAC;AAC1B,2DAA2D;AAC3D,MAAM,YAAY,GAAG,EAAE,CAAC;AAkExB,SAAS,mBAAmB,CAAC,IAAwB,EAAsB;IAC1E,IAAI,CAAC,IAAI;QAAE,OAAO,SAAS,CAAC;IAC5B,yEAAyE;IACzE,6EAA6E;IAC7E,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;QAC5E,OAAO,MAAM,IAAI,EAAE,CAAC;IACrB,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,oBAAoB,CACnC,UAAqC,EACrC,UAA+B,EACZ;IACnB,IAAI,UAAU,CAAC,IAAI,KAAK,CAAC;QAAE,OAAO,CAAC,GAAG,UAAU,CAAC,CAAC;IAClD,MAAM,OAAO,GAAqB,EAAE,CAAC;IACrC,MAAM,IAAI,GAAqB,EAAE,CAAC;IAClC,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACpC,CAAC,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,OAAO,CAAC,MAAM,GAAG,eAAe,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IACvG,CAAC;IACD,OAAO,CAAC,GAAG,OAAO,EAAE,GAAG,IAAI,CAAC,CAAC;AAAA,CAC7B;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,OAAwB,EAA2B;IAC3F,MAAM,EAAE,GAAG,EAAE,KAAK,EAAE,OAAO,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IAChD,MAAM,IAAI,GAAG,mBAAmB,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IAC/C,MAAM,aAAa,GAAG,OAAO,CAAC,IAAI,IAAI,MAAM,CAAC;IAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;IAC/C,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,IAAI,aAAa,CAAC;IAE3C,MAAM,KAAK,GAAG,OAAO,EAAE,QAAQ,EAAE,CAAC;IAClC,MAAM,cAAc,GAAG,OAAO,EAAE,WAAW,EAAE,IAAI,KAAK,CAAC;IACvD,MAAM,sBAAsB,GAC3B,KAAK,KAAK,SAAS;QAClB,CAAC,CAAC,+BAA+B;QACjC,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,aAAa,IAAI,KAAK,CAAC,KAAK,KAAK,SAAS;YAC3D,CAAC,CAAC,KAAK,CAAC,MAAM;YACd,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,MAAM;gBACvB,CAAC,CAAC,gCAAgC;gBAClC,CAAC,CAAC,SAAS,CAAC;IAEhB,MAAM,UAAU,GAAG,iBAAiB,CAAC,KAAK,EAAE,aAAa,EAAE,cAAc,EAAE,sBAAsB,CAAC,CAAC;IACnG,MAAM,IAAI,GAAG,UAAU,CAAC,IAAI,CAAC;IAE7B,0EAA0E;IAC1E,iEAAiE;IACjE,MAAM,WAAW,GAA4B,cAAc;QAC1D,CAAC,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,CAAC,OAAQ,CAAC,kBAAkB,CAAC,GAAG,EAAE,IAAI,CAAC;QACvD,CAAC,CAAC,SAAS,CAAC;IAEb,MAAM,KAAK,GAAG,IAAI,GAAG,EAAyB,CAAC;IAC/C;oEAC8D;IAC9D,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAC;IACnC,MAAM,KAAK,GAAkB,EAAE,CAAC;IAChC,MAAM,cAAc,GAA8B,EAAE,CAAC;IACrD,MAAM,MAAM,GAAY,EAAE,CAAC;IAE3B,MAAM,UAAU,GAAG,KAAK,EAAE,YAAY,GAAG,KAAK,EAAiB,EAAE,CAAC;QACjE,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,sEAAsE;YACtE,oEAAoE;YACpE,sEAAsE;YACtE,uEAAuE;YACvE,MAAM,KAAK,GAAG,YAAY,CAAC,CAAC,CAAC,OAAO,EAAE,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;YACrE,IAAI,KAAK;gBAAE,KAAK,MAAM,GAAG,IAAI,KAAK;oBAAE,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACtD,MAAM,QAAQ,GAAG,MAAM,mBAAmB,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,KAAK,EAAE,mBAAmB,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;YAC5G,MAAM,OAAO,GAAG,aAAa,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;YACrD,mEAAmE;YACnE,oEAAoE;YACpE,MAAM,IAAI,GAAG,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC;YAC1F,KAAK,MAAM,CAAC,EAAE,EAAE,IAAI,CAAC,IAAI,OAAO,CAAC,KAAK;gBAAE,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;oBAAE,KAAK,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;YAChF,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC1E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,QAAQ,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC3C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,kEAAkE;YAClE,mEAAmE;YACnE,yDAAyD;YACzD,4EAAwE;YACxE,sDAAsD;YACtD,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,WAAW,CAAC;YAC/D,MAAM,SAAS,GAAG,CACjB,MAAM,OAAQ,CAAC,YAAY,CAAC,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,CACzF,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;YACjC,MAAM,IAAI,GAAgB,SAAS,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;gBACpD,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC;gBACX,KAAK,EAAE,GAAG,CAAC,KAAK;gBAChB,MAAM,EAAE,OAAO;aACf,CAAC,CAAC,CAAC;YACJ,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;gBAC7B,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;YACvF,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,KAAK,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACrF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,KAAK,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC3E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,OAAO,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC1C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,iEAAiE;YACjE,qEAAqE;YACrE,+DAA6D;YAC7D,MAAM,SAAS,GAAG,MAAM,OAAQ,CAAC,YAAY,CAAC,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,CAAC,CAAC;YAClF,MAAM,IAAI,GAAgB,SAAS,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;gBACpD,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC;gBACX,KAAK,EAAE,GAAG,CAAC,KAAK;gBAChB,MAAM,EAAE,MAAM;aACd,CAAC,CAAC,CAAC;YACJ,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;gBAC7B,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,CAAC;oBACxB,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;gBACvF,CAAC;YACF,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC1E,CAAC;IAAA,CACD,CAAC;IAEF,2EAA2E;IAC3E,yEAAyE;IACzE,2EAA2E;IAC3E,2EAA2E;IAC3E,8EAA4E;IAC5E,kEAAkE;IAClE,MAAM,gBAAgB,GAAG,OAAO,CAAC,OAAO,IAAI,cAAc,IAAI,IAAI,KAAK,SAAS,CAAC;IAEjF,MAAM,IAAI,GAAoB,EAAE,CAAC;IACjC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,QAAQ;QAAE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,gBAAgB,CAAC,CAAC,CAAC;IACrF,IAAI,IAAI,KAAK,UAAU,IAAI,IAAI,KAAK,QAAQ;QAAE,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC;IACpE,wEAAwE;IACxE,sEAAsE;IACtE,IAAI,gBAAgB;QAAE,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC;IAC3C,MAAM,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACxB,IAAI,MAAM,EAAE,OAAO;QAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;IAC1D,wEAAwE;IACxE,iCAAiC;IACjC,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,MAAM,MAAM,CAAC,CAAC,CAAC,IAAI,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC;IAE7F,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,YAAY,CAAC,CAAC;IAC1D,IAAI,UAAU,GAAqB,EAAE,CAAC;IACtC,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAC/B,IAAI,IAAI;YAAE,UAAU,CAAC,IAAI,CAAC,EAAE,GAAG,GAAG,EAAE,GAAG,IAAI,EAAE,CAAC,CAAC;IAChD,CAAC;IAED,IAAI,UAAiC,CAAC;IACtC,IAAI,OAAO,CAAC,YAAY,IAAI,OAAO,EAAE,oBAAoB,EAAE,EAAE,CAAC;QAC7D,MAAM,QAAQ,GAAG,MAAM,kBAAkB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,EAAE,OAAO,CAAC,CAAC;QAC3E,UAAU,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,QAAQ,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,SAAS,EAAE,CAAC;QAC/F,UAAU,GAAG,QAAQ,CAAC,UAAU,CAAC;IAClC,CAAC;SAAM,IAAI,OAAO,CAAC,MAAM,KAAK,KAAK,EAAE,CAAC;QACrC,MAAM,QAAQ,GAAG,gBAAgB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,CAAC,CAAC;QAC1D,UAAU,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,UAAU,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,SAAS,EAAE,CAAC;QACjG,UAAU,GAAG,QAAQ,CAAC,UAAU,CAAC;IAClC,CAAC;IACD,4EAA4E;IAC5E,yEAAyE;IACzE,uEAAuE;IACvE,8BAA8B;IAC9B,UAAU,GAAG,oBAAoB,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAExE,OAAO;QACN,UAAU;QACV,YAAY,EAAE,IAAI;QAClB,cAAc,EAAE,UAAU,CAAC,cAAc;QACzC,UAAU,EAAE,KAAK,EAAE,KAAK,KAAK,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,EAAE,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,aAAa;QACzG,UAAU,EAAE,cAAc;QAC1B,QAAQ,EAAE,KAAK,EAAE,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS;QAC5F,IAAI;QACJ,MAAM,EAAE,UAAU;KAClB,CAAC;AAAA,CACF;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,OAAyB,EAA4B;IACpF,MAAM,SAAS,GAAG,MAAM,kBAAkB,CAAC,OAAO,CAAC,CAAC;IAEpD,MAAM,SAAS,GAAG,eAAe,CAAC,SAAS,CAAC,UAAU,EAAE,EAAE,GAAG,EAAE,OAAO,CAAC,GAAG,EAAE,WAAW,EAAE,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC;IAEhH,gBAAgB,CAAC,OAAO,CAAC,GAAG,EAAE;QAC7B,WAAW,EAAE,IAAI,CAAC,GAAG,EAAE;QACvB,KAAK,EAAE,OAAO,CAAC,KAAK;QACpB,aAAa,EAAE,OAAO,CAAC,IAAI,IAAI,MAAM;QACrC,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;QACxC,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,IAAI,EAAE,SAAS,CAAC,YAAY,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS;QACtE,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,KAAK,EAAE,SAAS,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC,CAAC;QAC7G,MAAM,EAAE,SAAS,CAAC,MAAM;KACxB,CAAC,CAAC;IAEH,OAAO;QACN,IAAI,EAAE,SAAS,CAAC,IAAI;QACpB,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;QACxC,WAAW,EAAE,SAAS,CAAC,UAAU,CAAC,MAAM;QACxC,QAAQ,EAAE,SAAS,CAAC,QAAQ;KAC5B,CAAC;AAAA,CACF","sourcesContent":["/**\n * Hybrid search orchestrator: resolve mode, run retrievers in parallel, fuse\n * by rank, expand within budget, trace everything\n * (docs/hybrid-retrieval-design.md).\n *\n * Single-retriever modes flow through the same pipeline — rrfFuse over one\n * list preserves its order — so lexical, semantic, and hybrid all produce the\n * same result shape and the same trace record.\n *\n * `retrieveCandidates` is the candidate-level core (also used by the eval\n * harness, which needs forced modes, a configurable `k`, and no trace\n * pollution); `runSearch` wraps it with span expansion and tracing for the\n * tool.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { adaptGrepHits, type ChunkLookup } from \"./adapter.js\";\nimport { assembleContext } from \"./context-assembler.js\";\nimport { crossEncoderRerank } from \"./cross-rerank.js\";\nimport { runLexicalRetriever } from \"./lexical-retriever.js\";\nimport { resolveSearchMode } from \"./mode.js\";\nimport { rerankCandidates } from \"./rerank.js\";\nimport { DEFAULT_RRF_K, rrfFuse } from \"./rrf.js\";\nimport { writeSearchTrace } from \"./trace.js\";\nimport type { CandidateSpan, FusedCandidate, RankedHit, ResolvedSearchMode, SearchMode, SearchTrace } from \"./types.js\";\n\n/** Raw grep line-hits fetched per query (pre-collapse). */\nconst LEXICAL_MATCH_LIMIT = 200;\n/** Adapted lexical candidates entering fusion. The eval gate showed the\n *  uncapped lexical tail diluting hybrid below plain semantic: lexical\n *  precision is front-loaded by the adapter's term-evidence ranking, while\n *  RRF weighs a rank-30 lexical candidate like a rank-30 embedding hit. */\nconst LEXICAL_FUSION_CAP = 20;\n/**\n * Embedding hits fetched per query when fusing with the grep leg.\n *\n * Deliberately shallow. The grep list is capped at {@link LEXICAL_FUSION_CAP}\n * because its tail is unranked noise, so a deep dense pool here simply\n * outnumbers it: raising this to 200 alongside a 20-candidate grep list cost\n * `auto +rr` 0.472 -> 0.450 MRR.\n */\nconst EMBED_TOP_K = 50;\n/**\n * Per-leg depth when both retrievers are ranked ones (dense + BM25).\n *\n * Four times {@link FUSED_WINDOW}, matching what the daemon's `query_hybrid`\n * does internally (`pool = 4·k`). Fusing at the same depth as the window loses\n * any candidate ranked well by one retriever but just outside the other's\n * top-50 — worth 12pp R@10 and 8pp R@50 for `bm25+dense`, which is exactly\n * what closed the gap to daemon-side fusion.\n */\nconst FUSION_POOL_TOP_K = 200;\n/** BM25 depth, matching {@link FUSION_POOL_TOP_K} so neither ranked leg is\n *  structurally advantaged by pool size. The daemon returns only documents\n *  sharing a query term, so this is an upper bound, not a fill. */\nconst BM25_TOP_K = FUSION_POOL_TOP_K;\n/** Candidates from unindexed/stale files moved ahead of the fused window.\n *  Small on purpose: this answers \"what did I just write\", not \"search the\n *  working tree\". */\nconst STALE_HOIST_CAP = 5;\n/** Fused candidates kept for reranking / final slicing. */\nconst FUSED_WINDOW = 50;\n\nexport interface RetrieveOptions {\n\tcwd: string;\n\tquery: string;\n\tmode?: SearchMode;\n\t/** Optional glob filter applied to file paths. */\n\tglob?: string;\n\t/** Maximum fused candidates returned. */\n\tlimit?: number;\n\t/** RRF constant override (eval harness sweeps this). Default: {@link DEFAULT_RRF_K}. */\n\trrfK?: number;\n\t/** Rerank the fused top-50 before slicing to `limit`. Default: true. */\n\trerank?: boolean;\n\t/**\n\t * Ask the daemon to fuse its own BM25 index with the vectors and return one\n\t * already-fused ranking, instead of taking a dense-only list. Needs a store\n\t * built with `--hybrid`. The fused list arrives as a single \"embed\" leg,\n\t * because a pre-fused ranking has no per-retriever structure left to record.\n\t *\n\t * Prefer {@link bm25Leg}: fusing here keeps the legs separable in the trace\n\t * and lets the grep leg participate.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as its own ranked list and fuse it here,\n\t * alongside dense and grep. Needs a store built with `--hybrid` and an\n\t * embsearch new enough to serve `retriever: \"lexical\"`.\n\t */\n\tbm25Leg?: boolean;\n\t/**\n\t * Reorder the fused shortlist with the daemon's cross-encoder instead of\n\t * the deterministic reranker. Needs embsearch >= 0.3.0; costs one model\n\t * pass per scored candidate.\n\t */\n\tcrossEncoder?: boolean;\n\tservice?: EmbsearchService;\n\tsignal?: AbortSignal;\n}\n\nexport interface RetrieveResult {\n\tcandidates: FusedCandidate[];\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tindexPhase: SearchTrace[\"indexPhase\"];\n\tretrievers: SearchTrace[\"retrievers\"];\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n\trrfK: number;\n\trerank?: SearchTrace[\"rerank\"];\n}\n\nexport interface RunSearchOptions extends RetrieveOptions {\n\t/** Approximate token budget for the result text. */\n\ttokenBudget?: number;\n}\n\nexport interface RunSearchResult {\n\ttext: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tresultCount: number;\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n}\n\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\n/**\n * Move candidates from files the index has not read to the front.\n *\n * They are there because grep found them and nothing else could: the index is\n * ranking a stale copy of the file, or has never seen it. Left to fuse, they\n * lose — RRF rewards agreement, and one leg reporting a single document is\n * outvoted by two legs agreeing on hundreds. Measured: scoping grep to stale\n * files without this hoist scored 25% on the live-edit set where unscoped grep\n * scored 100%, because the fused window filled with consensus hits about the\n * indexed copy.\n *\n * Capped, because \"stale\" scales with how far behind the index is. A few\n * edited files is the case this exists for; a fresh checkout makes everything\n * stale, and hoisting all of it would quietly turn hybrid search back into\n * grep. Past the cap the remainder keeps its fused position.\n */\nexport function hoistStaleCandidates(\n\tcandidates: readonly FusedCandidate[],\n\tstaleFiles: ReadonlySet<string>,\n): FusedCandidate[] {\n\tif (staleFiles.size === 0) return [...candidates];\n\tconst hoisted: FusedCandidate[] = [];\n\tconst rest: FusedCandidate[] = [];\n\tfor (const candidate of candidates) {\n\t\t(staleFiles.has(candidate.path) && hoisted.length < STALE_HOIST_CAP ? hoisted : rest).push(candidate);\n\t}\n\treturn [...hoisted, ...rest];\n}\n\nexport async function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult> {\n\tconst { cwd, query, service, signal } = options;\n\tconst glob = normalizeSearchGlob(options.glob);\n\tconst requestedMode = options.mode ?? \"auto\";\n\tconst limit = Math.max(1, options.limit ?? 10);\n\tconst rrfK = options.rrfK ?? DEFAULT_RRF_K;\n\n\tconst state = service?.getState();\n\tconst embedAvailable = service?.isAvailable() ?? false;\n\tconst embedUnavailableReason =\n\t\tstate === undefined\n\t\t\t? \"semantic index is not enabled\"\n\t\t\t: state.phase === \"unavailable\" || state.phase === \"skipped\"\n\t\t\t\t? state.reason\n\t\t\t\t: state.phase === \"idle\"\n\t\t\t\t\t? \"semantic index has not started\"\n\t\t\t\t\t: undefined;\n\n\tconst resolution = resolveSearchMode(query, requestedMode, embedAvailable, embedUnavailableReason);\n\tconst mode = resolution.mode;\n\n\t// Map lexical hits onto indexed chunk ids whenever the sidecar is usable,\n\t// even in lexical-only mode, so identities line up across modes.\n\tconst lookupChunk: ChunkLookup | undefined = embedAvailable\n\t\t? (rel, line) => service!.findEnclosingChunk(rel, line)\n\t\t: undefined;\n\n\tconst spans = new Map<string, CandidateSpan>();\n\t/** Files the index does not have current content for, when grep is scoped\n\t *  to them. Read again after fusion — see the hoist below. */\n\tconst staleSet = new Set<string>();\n\tconst lists: RankedHit[][] = [];\n\tconst retrieverStats: SearchTrace[\"retrievers\"] = {};\n\tconst errors: Error[] = [];\n\n\tconst runLexical = async (scopeToStale = false): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Scoped run: grep covers only what the index has not read yet, so it\n\t\t\t// adds the one thing BM25 cannot see without re-voting on documents\n\t\t\t// BM25 already ranked. Fusing two lexical views of the same corpus is\n\t\t\t// what made the three-leg configuration lose (18 of 62 queries worse).\n\t\t\tconst paths = scopeToStale ? service?.staleFiles(signal) : undefined;\n\t\t\tif (paths) for (const rel of paths) staleSet.add(rel);\n\t\t\tconst lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal, paths });\n\t\t\tconst adapted = adaptGrepHits(lineHits, lookupChunk);\n\t\t\t// In single-retriever lexical mode the full list is the result; in\n\t\t\t// hybrid, only the front-loaded head is trustworthy enough to vote.\n\t\t\tconst hits = mode === \"hybrid\" ? adapted.hits.slice(0, LEXICAL_FUSION_CAP) : adapted.hits;\n\t\t\tfor (const [id, span] of adapted.spans) if (!spans.has(id)) spans.set(id, span);\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runEmbed = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// The flat index pads top-k with whatever exists; with the cosine\n\t\t\t// metric the store uses, score <= 0 means \"no relation at all\", so\n\t\t\t// those padding hits would cast RRF votes on pure noise.\n\t\t\t// A BM25 leg is itself ranked, so the pair can afford — and needs — the\n\t\t\t// deeper pool; the grep leg cannot (see EMBED_TOP_K).\n\t\t\tconst topK = options.bm25Leg ? FUSION_POOL_TOP_K : EMBED_TOP_K;\n\t\t\tconst chunkHits = (\n\t\t\t\tawait service!.searchChunks(query, topK, glob, options.daemonHybrid ? \"hybrid\" : \"dense\")\n\t\t\t).filter((hit) => hit.score > 0);\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"embed\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runBm25 = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Raw BM25 sums; only the ordering enters fusion, the score is a\n\t\t\t// diagnostic. Unlike the dense leg there is no zero-score padding to\n\t\t\t// filter — the daemon omits documents sharing no query term.\n\t\t\tconst chunkHits = await service!.searchChunks(query, BM25_TOP_K, glob, \"lexical\");\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"bm25\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tif (!spans.has(hit.id)) {\n\t\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t\t}\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\t// BM25 is the better lexical leg where the index is current: it beats grep\n\t// on Recall@50 (0.790 -> 0.879, 6 queries better and 0 worse, p <= 0.05)\n\t// because it ranks the whole corpus rather than truncating a match stream.\n\t// It is also blind to anything indexed later than it was written, which is\n\t// precisely where grep still wins — so grep runs scoped to that set instead\n\t// of being dropped or left to duplicate BM25 over the whole tree.\n\tconst bm25AsLexicalLeg = options.bm25Leg && embedAvailable && mode !== \"lexical\";\n\n\tconst runs: Promise<void>[] = [];\n\tif (mode === \"lexical\" || mode === \"hybrid\") runs.push(runLexical(bm25AsLexicalLeg));\n\tif (mode === \"semantic\" || mode === \"hybrid\") runs.push(runEmbed());\n\t// `semantic` gets the BM25 leg too: the caller asked for the index, and\n\t// BM25 is part of it. Only an explicit `lexical` request excludes it.\n\tif (bm25AsLexicalLeg) runs.push(runBm25());\n\tawait Promise.all(runs);\n\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t// A partial failure in hybrid degrades to whichever retriever survived;\n\t// only a total loss is an error.\n\tif (lists.length === 0) throw errors[0] ?? new Error(\"search produced no retriever results\");\n\n\tconst fused = rrfFuse(lists, rrfK).slice(0, FUSED_WINDOW);\n\tlet candidates: FusedCandidate[] = [];\n\tfor (const hit of fused) {\n\t\tconst span = spans.get(hit.id);\n\t\tif (span) candidates.push({ ...hit, ...span });\n\t}\n\n\tlet rerankInfo: SearchTrace[\"rerank\"];\n\tif (options.crossEncoder && service?.supportsCrossEncoder()) {\n\t\tconst reranked = await crossEncoderRerank(query, candidates, cwd, service);\n\t\trerankInfo = { applied: true, candidateCount: reranked.scored, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t} else if (options.rerank !== false) {\n\t\tconst reranked = rerankCandidates(query, candidates, cwd);\n\t\trerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t}\n\t// After reranking, not before: both rerankers weight the fused prior, and a\n\t// lone grep hit from an unindexed file has the lowest prior there is, so\n\t// hoisting first would simply be undone. The reranker still orders the\n\t// hoisted set against itself.\n\tcandidates = hoistStaleCandidates(candidates, staleSet).slice(0, limit);\n\n\treturn {\n\t\tcandidates,\n\t\tresolvedMode: mode,\n\t\tdegradedReason: resolution.degradedReason,\n\t\tindexPhase: state?.phase === \"ready\" ? \"ready\" : state?.phase === \"indexing\" ? \"indexing\" : \"unavailable\",\n\t\tretrievers: retrieverStats,\n\t\tindexing: state?.phase === \"indexing\" ? { done: state.done, total: state.total } : undefined,\n\t\trrfK,\n\t\trerank: rerankInfo,\n\t};\n}\n\nexport async function runSearch(options: RunSearchOptions): Promise<RunSearchResult> {\n\tconst retrieved = await retrieveCandidates(options);\n\n\tconst assembled = assembleContext(retrieved.candidates, { cwd: options.cwd, tokenBudget: options.tokenBudget });\n\n\twriteSearchTrace(options.cwd, {\n\t\ttimestampMs: Date.now(),\n\t\tquery: options.query,\n\t\trequestedMode: options.mode ?? \"auto\",\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tindexPhase: retrieved.indexPhase,\n\t\trrfK: retrieved.resolvedMode === \"hybrid\" ? retrieved.rrfK : undefined,\n\t\tretrievers: retrieved.retrievers,\n\t\tfused: retrieved.candidates.map(({ id, rrfScore, ranks, rawScores }) => ({ id, rrfScore, ranks, rawScores })),\n\t\trerank: retrieved.rerank,\n\t});\n\n\treturn {\n\t\ttext: assembled.text,\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tresultCount: retrieved.candidates.length,\n\t\tindexing: retrieved.indexing,\n\t};\n}\n"]}

package/dist/core/search/lexical-retriever.d.ts

@@ -32,6 +32,14 @@ export interface RunLexicalOptions {
     limit: number;
     glob?: string;
     signal?: AbortSignal;
+    /**
+     * Restrict the search to these repo-relative files instead of walking `cwd`.
+     *
+     * Used to scope the grep leg to files the embedding index does not have
+     * current content for. An empty array means "no files to search", which is
+     * a real answer — not "search everything".
+     */
+    paths?: readonly string[];
 }
 export declare function runLexicalRetriever(options: RunLexicalOptions): Promise<GrepLineHit[]>;
 //# sourceMappingURL=lexical-retriever.d.ts.map

package/dist/core/search/lexical-retriever.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"lexical-retriever.d.ts","sourceRoot":"","sources":["../../../src/core/search/lexical-retriever.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAQH,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,cAAc,CAAC;AAWhD,MAAM,WAAW,gBAAgB;IAChC,8BAA8B;IAC9B,OAAO,EAAE,MAAM,CAAC;IAChB,wEAAwE;IACxE,KAAK,EAAE,MAAM,EAAE,CAAC;CAChB;AAED;;;;GAIG;AACH,wBAAgB,qBAAqB,CAAC,KAAK,EAAE,MAAM,GAAG,gBAAgB,GAAG,SAAS,CAgBjF;AAED,0DAA0D;AAC1D,wBAAgB,mBAAmB,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAErE;AAUD;;;GAGG;AACH,MAAM,WAAW,iBAAiB;IACjC,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,WAAW,CAAC;CACrB;AAiBD,wBAAsB,mBAAmB,CAAC,OAAO,EAAE,iBAAiB,GAAG,OAAO,CAAC,WAAW,EAAE,CAAC,CAyG5F","sourcesContent":["/**\n * Internal lexical retriever for hybrid search.\n *\n * This is not the grep *tool* — it is the lexical recall backend: it turns a\n * natural query into a ripgrep pattern, streams matches, and returns bare\n * `rel:line` hits for the grep→chunk adapter. rg drives the fast path; the\n * pure-JS nativeGrep fallback keeps restricted environments working, same as\n * the grep tool.\n */\n\nimport { createInterface } from \"node:readline\";\nimport { spawn } from \"child_process\";\nimport { readFileSync } from \"fs\";\nimport path from \"path\";\nimport { ensureTool } from \"../../utils/tools-manager.js\";\nimport { isNativeSearchForced, nativeGrep } from \"../tools/native-search.js\";\nimport type { GrepLineHit } from \"./adapter.js\";\n\n/** Terms considered per query (longest first) when building the pattern. */\nconst MAX_TERMS = 4;\n/** Minimum token length worth matching on. */\nconst MIN_TERM_LENGTH = 3;\n\nfunction escapeRegExp(value: string): string {\n\treturn value.replace(/[.*+?^${}()|[\\]\\\\]/g, \"\\\\$&\");\n}\n\nexport interface LexicalQueryPlan {\n\t/** rg-ready regex pattern. */\n\tpattern: string;\n\t/** Raw (unescaped, lowercased) terms, for per-line term attribution. */\n\tterms: string[];\n}\n\n/**\n * Build the retrieval plan for a query: a quoted segment is searched\n * verbatim; otherwise the longest few identifier-ish tokens are OR-ed\n * together. Returns undefined when the query yields nothing searchable.\n */\nexport function buildLexicalQueryPlan(query: string): LexicalQueryPlan | undefined {\n\tconst quoted = [...query.matchAll(/[\"'`]([^\"'`]+)[\"'`]/g)]\n\t\t.map((m) => m[1].trim())\n\t\t.filter((s) => s.length > 0)\n\t\t.sort((a, b) => b.length - a.length)[0];\n\tif (quoted) return { pattern: escapeRegExp(quoted), terms: [quoted.toLowerCase()] };\n\n\tconst tokens = [...new Set(query.match(/[A-Za-z0-9_$][\\w$.-]*/g) ?? [])]\n\t\t.filter((t) => t.length >= MIN_TERM_LENGTH)\n\t\t.sort((a, b) => b.length - a.length || a.localeCompare(b))\n\t\t.slice(0, MAX_TERMS);\n\tif (tokens.length === 0) {\n\t\tconst trimmed = query.trim();\n\t\treturn trimmed ? { pattern: escapeRegExp(trimmed), terms: [trimmed.toLowerCase()] } : undefined;\n\t}\n\treturn { pattern: tokens.map(escapeRegExp).join(\"|\"), terms: tokens.map((t) => t.toLowerCase()) };\n}\n\n/** Pattern-only view of {@link buildLexicalQueryPlan}. */\nexport function buildLexicalPattern(query: string): string | undefined {\n\treturn buildLexicalQueryPlan(query)?.pattern;\n}\n\n/** Which plan terms appear on a matched line (retrieval is case-insensitive,\n *  so attribution is too). */\nfunction termsOnLine(plan: LexicalQueryPlan, lineText: string | undefined): string[] {\n\tif (!lineText) return [];\n\tconst lower = lineText.toLowerCase();\n\treturn plan.terms.filter((t) => lower.includes(t));\n}\n\n/**\n * Run lexical retrieval over `cwd`, returning up to `limit` line-hits in\n * output order with POSIX repo-relative paths.\n */\nexport interface RunLexicalOptions {\n\tcwd: string;\n\tquery: string;\n\tlimit: number;\n\tglob?: string;\n\tsignal?: AbortSignal;\n}\n\n/**\n * Run the lexical retriever for the search tool. Optional glob filter scopes\n * file paths (slashless matches basename anywhere, slash patterns match the\n * full repo-relative path).\n */\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\nexport async function runLexicalRetriever(options: RunLexicalOptions): Promise<GrepLineHit[]> {\n\tconst { cwd, query, limit, glob: rawGlob, signal } = options;\n\tconst glob = normalizeSearchGlob(rawGlob);\n\tconst plan = buildLexicalQueryPlan(query);\n\tif (!plan) return [];\n\tconst { pattern } = plan;\n\n\tconst toRel = (filePath: string): string => {\n\t\tconst rel = path.relative(cwd, filePath);\n\t\treturn (rel && !rel.startsWith(\"..\") ? rel : filePath).replace(/\\\\/g, \"/\");\n\t};\n\n\tconst rgPath = isNativeSearchForced() ? undefined : await ensureTool(\"rg\", true);\n\tif (!rgPath) {\n\t\tconst result = await nativeGrep(cwd, {\n\t\t\tpattern,\n\t\t\tisDirectory: true,\n\t\t\tignoreCase: true,\n\t\t\tlimit,\n\t\t\tglob,\n\t\t\tsignal,\n\t\t\treadFile: (p) => readFileSync(p, \"utf-8\"),\n\t\t});\n\t\treturn result.matches.map((m) => ({\n\t\t\trel: toRel(m.filePath),\n\t\t\tline: m.lineNumber,\n\t\t\tterms: termsOnLine(plan, m.lineText),\n\t\t}));\n\t}\n\n\treturn new Promise<GrepLineHit[]>((resolve, reject) => {\n\t\t// --sort path forces a deterministic (single-threaded) walk: with the\n\t\t// match cap truncating the stream, a parallel walk would return a\n\t\t// different hit subset per run — \"same query, different context\".\n\t\t// `!.git` because --hidden would otherwise search .git contents;\n\t\t// --no-require-git so .gitignore is honored outside git repos too.\n\t\tconst args = [\n\t\t\t\"--json\",\n\t\t\t\"--line-number\",\n\t\t\t\"--color=never\",\n\t\t\t\"--hidden\",\n\t\t\t\"--no-require-git\",\n\t\t\t\"--ignore-case\",\n\t\t\t\"--sort\",\n\t\t\t\"path\",\n\t\t\t\"--glob\",\n\t\t\t\"!**/.git/**\",\n\t\t];\n\t\tif (glob) args.push(\"--glob\", glob);\n\t\targs.push(\"--\", pattern, cwd);\n\t\tconst child = spawn(rgPath, args, { stdio: [\"ignore\", \"pipe\", \"pipe\"] });\n\t\tconst rl = createInterface({ input: child.stdout });\n\t\tconst hits: GrepLineHit[] = [];\n\t\tlet stderr = \"\";\n\t\tlet killedDueToLimit = false;\n\t\tlet aborted = false;\n\n\t\tconst onAbort = () => {\n\t\t\taborted = true;\n\t\t\tif (!child.killed) child.kill();\n\t\t};\n\t\tsignal?.addEventListener(\"abort\", onAbort, { once: true });\n\t\tchild.stderr?.on(\"data\", (chunk) => {\n\t\t\tstderr += chunk.toString();\n\t\t});\n\n\t\trl.on(\"line\", (line) => {\n\t\t\tif (!line.trim() || hits.length >= limit) return;\n\t\t\tlet event: any;\n\t\t\ttry {\n\t\t\t\tevent = JSON.parse(line);\n\t\t\t} catch {\n\t\t\t\treturn;\n\t\t\t}\n\t\t\tif (event.type !== \"match\") return;\n\t\t\tconst filePath = event.data?.path?.text;\n\t\t\tconst lineNumber = event.data?.line_number;\n\t\t\tif (filePath && typeof lineNumber === \"number\") {\n\t\t\t\thits.push({ rel: toRel(filePath), line: lineNumber, terms: termsOnLine(plan, event.data?.lines?.text) });\n\t\t\t}\n\t\t\tif (hits.length >= limit && !child.killed) {\n\t\t\t\tkilledDueToLimit = true;\n\t\t\t\tchild.kill();\n\t\t\t}\n\t\t});\n\n\t\tchild.on(\"error\", (error) => {\n\t\t\tsignal?.removeEventListener(\"abort\", onAbort);\n\t\t\treject(new Error(`Failed to run ripgrep: ${error.message}`));\n\t\t});\n\t\tchild.on(\"close\", (code) => {\n\t\t\trl.close();\n\t\t\tsignal?.removeEventListener(\"abort\", onAbort);\n\t\t\tif (aborted) {\n\t\t\t\treject(new Error(\"Operation aborted\"));\n\t\t\t\treturn;\n\t\t\t}\n\t\t\t// rg exits 1 on \"no matches\" — that is a valid empty result.\n\t\t\tif (!killedDueToLimit && code !== 0 && code !== 1) {\n\t\t\t\treject(new Error(stderr.trim() || `ripgrep exited with code ${code}`));\n\t\t\t\treturn;\n\t\t\t}\n\t\t\tresolve(hits);\n\t\t});\n\t});\n}\n"]}
+{"version":3,"file":"lexical-retriever.d.ts","sourceRoot":"","sources":["../../../src/core/search/lexical-retriever.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAQH,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,cAAc,CAAC;AAWhD,MAAM,WAAW,gBAAgB;IAChC,8BAA8B;IAC9B,OAAO,EAAE,MAAM,CAAC;IAChB,wEAAwE;IACxE,KAAK,EAAE,MAAM,EAAE,CAAC;CAChB;AAED;;;;GAIG;AACH,wBAAgB,qBAAqB,CAAC,KAAK,EAAE,MAAM,GAAG,gBAAgB,GAAG,SAAS,CAgBjF;AAED,0DAA0D;AAC1D,wBAAgB,mBAAmB,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAErE;AAUD;;;GAGG;AACH,MAAM,WAAW,iBAAiB;IACjC,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB;;;;;;OAMG;IACH,KAAK,CAAC,EAAE,SAAS,MAAM,EAAE,CAAC;CAC1B;AAiBD,wBAAsB,mBAAmB,CAAC,OAAO,EAAE,iBAAiB,GAAG,OAAO,CAAC,WAAW,EAAE,CAAC,CAmH5F","sourcesContent":["/**\n * Internal lexical retriever for hybrid search.\n *\n * This is not the grep *tool* — it is the lexical recall backend: it turns a\n * natural query into a ripgrep pattern, streams matches, and returns bare\n * `rel:line` hits for the grep→chunk adapter. rg drives the fast path; the\n * pure-JS nativeGrep fallback keeps restricted environments working, same as\n * the grep tool.\n */\n\nimport { createInterface } from \"node:readline\";\nimport { spawn } from \"child_process\";\nimport { readFileSync } from \"fs\";\nimport path from \"path\";\nimport { ensureTool } from \"../../utils/tools-manager.js\";\nimport { isNativeSearchForced, nativeGrep } from \"../tools/native-search.js\";\nimport type { GrepLineHit } from \"./adapter.js\";\n\n/** Terms considered per query (longest first) when building the pattern. */\nconst MAX_TERMS = 4;\n/** Minimum token length worth matching on. */\nconst MIN_TERM_LENGTH = 3;\n\nfunction escapeRegExp(value: string): string {\n\treturn value.replace(/[.*+?^${}()|[\\]\\\\]/g, \"\\\\$&\");\n}\n\nexport interface LexicalQueryPlan {\n\t/** rg-ready regex pattern. */\n\tpattern: string;\n\t/** Raw (unescaped, lowercased) terms, for per-line term attribution. */\n\tterms: string[];\n}\n\n/**\n * Build the retrieval plan for a query: a quoted segment is searched\n * verbatim; otherwise the longest few identifier-ish tokens are OR-ed\n * together. Returns undefined when the query yields nothing searchable.\n */\nexport function buildLexicalQueryPlan(query: string): LexicalQueryPlan | undefined {\n\tconst quoted = [...query.matchAll(/[\"'`]([^\"'`]+)[\"'`]/g)]\n\t\t.map((m) => m[1].trim())\n\t\t.filter((s) => s.length > 0)\n\t\t.sort((a, b) => b.length - a.length)[0];\n\tif (quoted) return { pattern: escapeRegExp(quoted), terms: [quoted.toLowerCase()] };\n\n\tconst tokens = [...new Set(query.match(/[A-Za-z0-9_$][\\w$.-]*/g) ?? [])]\n\t\t.filter((t) => t.length >= MIN_TERM_LENGTH)\n\t\t.sort((a, b) => b.length - a.length || a.localeCompare(b))\n\t\t.slice(0, MAX_TERMS);\n\tif (tokens.length === 0) {\n\t\tconst trimmed = query.trim();\n\t\treturn trimmed ? { pattern: escapeRegExp(trimmed), terms: [trimmed.toLowerCase()] } : undefined;\n\t}\n\treturn { pattern: tokens.map(escapeRegExp).join(\"|\"), terms: tokens.map((t) => t.toLowerCase()) };\n}\n\n/** Pattern-only view of {@link buildLexicalQueryPlan}. */\nexport function buildLexicalPattern(query: string): string | undefined {\n\treturn buildLexicalQueryPlan(query)?.pattern;\n}\n\n/** Which plan terms appear on a matched line (retrieval is case-insensitive,\n *  so attribution is too). */\nfunction termsOnLine(plan: LexicalQueryPlan, lineText: string | undefined): string[] {\n\tif (!lineText) return [];\n\tconst lower = lineText.toLowerCase();\n\treturn plan.terms.filter((t) => lower.includes(t));\n}\n\n/**\n * Run lexical retrieval over `cwd`, returning up to `limit` line-hits in\n * output order with POSIX repo-relative paths.\n */\nexport interface RunLexicalOptions {\n\tcwd: string;\n\tquery: string;\n\tlimit: number;\n\tglob?: string;\n\tsignal?: AbortSignal;\n\t/**\n\t * Restrict the search to these repo-relative files instead of walking `cwd`.\n\t *\n\t * Used to scope the grep leg to files the embedding index does not have\n\t * current content for. An empty array means \"no files to search\", which is\n\t * a real answer — not \"search everything\".\n\t */\n\tpaths?: readonly string[];\n}\n\n/**\n * Run the lexical retriever for the search tool. Optional glob filter scopes\n * file paths (slashless matches basename anywhere, slash patterns match the\n * full repo-relative path).\n */\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\nexport async function runLexicalRetriever(options: RunLexicalOptions): Promise<GrepLineHit[]> {\n\tconst { cwd, query, limit, glob: rawGlob, signal, paths } = options;\n\tconst glob = normalizeSearchGlob(rawGlob);\n\tconst plan = buildLexicalQueryPlan(query);\n\tif (!plan) return [];\n\tconst { pattern } = plan;\n\t// Scoped to an empty set: nothing to search, and handing rg no paths would\n\t// make it read stdin and hang.\n\tif (paths?.length === 0) return [];\n\n\tconst toRel = (filePath: string): string => {\n\t\tconst rel = path.relative(cwd, filePath);\n\t\treturn (rel && !rel.startsWith(\"..\") ? rel : filePath).replace(/\\\\/g, \"/\");\n\t};\n\n\tconst rgPath = isNativeSearchForced() ? undefined : await ensureTool(\"rg\", true);\n\tif (!rgPath) {\n\t\t// nativeGrep takes one root, so a scoped run visits each file directly.\n\t\t// The scoped set is small by construction — it is what changed since the\n\t\t// last index pass — so this stays bounded.\n\t\tconst roots = paths ? paths.map((rel) => path.resolve(cwd, rel)) : [cwd];\n\t\tconst hits: GrepLineHit[] = [];\n\t\tfor (const root of roots) {\n\t\t\tif (hits.length >= limit) break;\n\t\t\tconst result = await nativeGrep(root, {\n\t\t\t\tpattern,\n\t\t\t\tisDirectory: paths === undefined,\n\t\t\t\tignoreCase: true,\n\t\t\t\tlimit: limit - hits.length,\n\t\t\t\tglob,\n\t\t\t\tsignal,\n\t\t\t\treadFile: (p) => readFileSync(p, \"utf-8\"),\n\t\t\t});\n\t\t\tfor (const m of result.matches) {\n\t\t\t\thits.push({ rel: toRel(m.filePath), line: m.lineNumber, terms: termsOnLine(plan, m.lineText) });\n\t\t\t}\n\t\t}\n\t\treturn hits;\n\t}\n\n\treturn new Promise<GrepLineHit[]>((resolve, reject) => {\n\t\t// --sort path forces a deterministic (single-threaded) walk: with the\n\t\t// match cap truncating the stream, a parallel walk would return a\n\t\t// different hit subset per run — \"same query, different context\".\n\t\t// `!.git` because --hidden would otherwise search .git contents;\n\t\t// --no-require-git so .gitignore is honored outside git repos too.\n\t\tconst args = [\n\t\t\t\"--json\",\n\t\t\t\"--line-number\",\n\t\t\t\"--color=never\",\n\t\t\t\"--hidden\",\n\t\t\t\"--no-require-git\",\n\t\t\t\"--ignore-case\",\n\t\t\t\"--sort\",\n\t\t\t\"path\",\n\t\t\t\"--glob\",\n\t\t\t\"!**/.git/**\",\n\t\t];\n\t\tif (glob) args.push(\"--glob\", glob);\n\t\targs.push(\"--\", pattern, ...(paths ? paths.map((rel) => path.resolve(cwd, rel)) : [cwd]));\n\t\tconst child = spawn(rgPath, args, { stdio: [\"ignore\", \"pipe\", \"pipe\"] });\n\t\tconst rl = createInterface({ input: child.stdout });\n\t\tconst hits: GrepLineHit[] = [];\n\t\tlet stderr = \"\";\n\t\tlet killedDueToLimit = false;\n\t\tlet aborted = false;\n\n\t\tconst onAbort = () => {\n\t\t\taborted = true;\n\t\t\tif (!child.killed) child.kill();\n\t\t};\n\t\tsignal?.addEventListener(\"abort\", onAbort, { once: true });\n\t\tchild.stderr?.on(\"data\", (chunk) => {\n\t\t\tstderr += chunk.toString();\n\t\t});\n\n\t\trl.on(\"line\", (line) => {\n\t\t\tif (!line.trim() || hits.length >= limit) return;\n\t\t\tlet event: any;\n\t\t\ttry {\n\t\t\t\tevent = JSON.parse(line);\n\t\t\t} catch {\n\t\t\t\treturn;\n\t\t\t}\n\t\t\tif (event.type !== \"match\") return;\n\t\t\tconst filePath = event.data?.path?.text;\n\t\t\tconst lineNumber = event.data?.line_number;\n\t\t\tif (filePath && typeof lineNumber === \"number\") {\n\t\t\t\thits.push({ rel: toRel(filePath), line: lineNumber, terms: termsOnLine(plan, event.data?.lines?.text) });\n\t\t\t}\n\t\t\tif (hits.length >= limit && !child.killed) {\n\t\t\t\tkilledDueToLimit = true;\n\t\t\t\tchild.kill();\n\t\t\t}\n\t\t});\n\n\t\tchild.on(\"error\", (error) => {\n\t\t\tsignal?.removeEventListener(\"abort\", onAbort);\n\t\t\treject(new Error(`Failed to run ripgrep: ${error.message}`));\n\t\t});\n\t\tchild.on(\"close\", (code) => {\n\t\t\trl.close();\n\t\t\tsignal?.removeEventListener(\"abort\", onAbort);\n\t\t\tif (aborted) {\n\t\t\t\treject(new Error(\"Operation aborted\"));\n\t\t\t\treturn;\n\t\t\t}\n\t\t\t// rg exits 1 on \"no matches\" — that is a valid empty result.\n\t\t\tif (!killedDueToLimit && code !== 0 && code !== 1) {\n\t\t\t\treject(new Error(stderr.trim() || `ripgrep exited with code ${code}`));\n\t\t\t\treturn;\n\t\t\t}\n\t\t\tresolve(hits);\n\t\t});\n\t});\n}\n"]}

package/dist/core/search/lexical-retriever.js

@@ -70,32 +70,44 @@ function normalizeSearchGlob(glob) {
     return glob;
 }
 export async function runLexicalRetriever(options) {
-    const { cwd, query, limit, glob: rawGlob, signal } = options;
+    const { cwd, query, limit, glob: rawGlob, signal, paths } = options;
     const glob = normalizeSearchGlob(rawGlob);
     const plan = buildLexicalQueryPlan(query);
     if (!plan)
         return [];
     const { pattern } = plan;
+    // Scoped to an empty set: nothing to search, and handing rg no paths would
+    // make it read stdin and hang.
+    if (paths?.length === 0)
+        return [];
     const toRel = (filePath) => {
         const rel = path.relative(cwd, filePath);
         return (rel && !rel.startsWith("..") ? rel : filePath).replace(/\\/g, "/");
     };
     const rgPath = isNativeSearchForced() ? undefined : await ensureTool("rg", true);
     if (!rgPath) {
-        const result = await nativeGrep(cwd, {
-            pattern,
-            isDirectory: true,
-            ignoreCase: true,
-            limit,
-            glob,
-            signal,
-            readFile: (p) => readFileSync(p, "utf-8"),
-        });
-        return result.matches.map((m) => ({
-            rel: toRel(m.filePath),
-            line: m.lineNumber,
-            terms: termsOnLine(plan, m.lineText),
-        }));
+        // nativeGrep takes one root, so a scoped run visits each file directly.
+        // The scoped set is small by construction — it is what changed since the
+        // last index pass — so this stays bounded.
+        const roots = paths ? paths.map((rel) => path.resolve(cwd, rel)) : [cwd];
+        const hits = [];
+        for (const root of roots) {
+            if (hits.length >= limit)
+                break;
+            const result = await nativeGrep(root, {
+                pattern,
+                isDirectory: paths === undefined,
+                ignoreCase: true,
+                limit: limit - hits.length,
+                glob,
+                signal,
+                readFile: (p) => readFileSync(p, "utf-8"),
+            });
+            for (const m of result.matches) {
+                hits.push({ rel: toRel(m.filePath), line: m.lineNumber, terms: termsOnLine(plan, m.lineText) });
+            }
+        }
+        return hits;
     }
     return new Promise((resolve, reject) => {
         // --sort path forces a deterministic (single-threaded) walk: with the
@@ -117,7 +129,7 @@ export async function runLexicalRetriever(options) {
         ];
         if (glob)
             args.push("--glob", glob);
-        args.push("--", pattern, cwd);
+        args.push("--", pattern, ...(paths ? paths.map((rel) => path.resolve(cwd, rel)) : [cwd]));
         const child = spawn(rgPath, args, { stdio: ["ignore", "pipe", "pipe"] });
         const rl = createInterface({ input: child.stdout });
         const hits = [];