@kolisachint/hoocode-agent 0.4.167 → 0.4.168

package/CHANGELOG.md

@@ -1,5 +1,7 @@
 # Changelog
 
+## [0.4.168] - 2026-08-08
+
 ## [0.4.167] - 2026-08-08
 
 ## [0.4.166] - 2026-08-08

package/dist/core/embsearch/client.d.ts

@@ -14,6 +14,16 @@ export interface EmbSearchDaemonInfo {
     modelId: string;
     dim: number;
     count: number;
+    /**
+     * Whether this daemon can actually serve `rerank`, or `undefined` from a
+     * daemon too old to say.
+     *
+     * Reported separately from the version because the two came apart:
+     * released binaries carry the `rerank` op but no longer bundle the ~23 MB
+     * cross-encoder weights, so a version check alone would advertise a
+     * reranker that fails on first call.
+     */
+    rerank?: boolean;
 }
 /** One candidate sent for cross-encoder scoring. */
 export interface EmbSearchRerankPassage {

package/dist/core/embsearch/client.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"client.d.ts","sourceRoot":"","sources":["../../../src/core/embsearch/client.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAIH,MAAM,WAAW,eAAe;IAC/B,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,mBAAmB;IACnC,OAAO,EAAE,MAAM,CAAC;IAChB,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;CACd;AAED,oDAAoD;AACpD,MAAM,WAAW,sBAAsB;IACtC,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;CACb;AAED;wDACwD;AACxD,MAAM,WAAW,qBAAqB;IACrC,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,mBAAmB;IACnC,QAAQ,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,MAAM,CAAC;CAChB;AAED,wEAAwE;AACxE,MAAM,MAAM,eAAe,GAAG,OAAO,GAAG,SAAS,GAAG,QAAQ,CAAC;AAE7D,MAAM,WAAW,sBAAsB;IACtC,6EAA6E;IAC7E,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,sCAAsC;IACtC,UAAU,EAAE,MAAM,CAAC;IACnB,0CAA0C;IAC1C,SAAS,EAAE,MAAM,CAAC;IAClB,6DAA6D;IAC7D,MAAM,CAAC,EAAE,QAAQ,GAAG,KAAK,GAAG,WAAW,CAAC;CACxC;AAuBD,qBAAa,eAAe;IAC3B,OAAO,CAAC,IAAI,CAAiC;IAC7C,OAAO,CAAC,KAAK,CAAiB;IAC9B,OAAO,CAAC,MAAM,CAAM;IACpB,OAAO,CAAC,MAAM,CAAS;IACvB,OAAO,CAAC,YAAY,CAAgB;IAEpC,YAAY,IAAI,EAAE,sBAAsB,EA+BvC;IAED,6DAA6D;IAC7D,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAErB;IAED,IAAI,QAAQ,IAAI,OAAO,CAEtB;IAED,OAAO,CAAC,QAAQ;IAqBhB,OAAO,CAAC,IAAI;IAQZ;;;;;;;;;;;;OAYG;IACG,KAAK,CAAC,IAAI,EAAE,MAAM,EAAE,CAAC,SAAK,EAAE,SAAS,GAAE,eAAyB,GAAG,OAAO,CAAC,eAAe,EAAE,CAAC,CAKlG;IAED;;;;;;;;OAQG;IACG,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,sBAAsB,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,qBAAqB,EAAE,CAAC,CAG3G;IAED;;;;OAIG;IACG,IAAI,CAAC,KAAK,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAGnF;IAED,qEAAqE;IAC/D,IAAI,IAAI,OAAO,CAAC,mBAAmB,CAAC,CAGzC;IAED,yDAAyD;IACnD,MAAM,CAAC,EAAE,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,CAGzC;IAED,uDAAuD;IACjD,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAE7B;IAED,gDAAgD;IAC1C,IAAI,IAAI,OAAO,CAAC,IAAI,CAAC,CAE1B;IAED,+DAA+D;IACzD,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAI3B;CACD","sourcesContent":["/**\n * Client for the `embsearch` stdio daemon (vendored from\n * github.com/kolisachint/embeddingsearchtools ts/client.ts).\n *\n * Spawns `embsearch serve` once and talks newline-delimited JSON over its\n * stdin/stdout. The process stays alive so the model and index load a single\n * time; every query after startup is hot.\n */\n\nimport { type ChildProcessWithoutNullStreams, spawn } from \"child_process\";\n\nexport interface EmbSearchResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchDaemonInfo {\n\tmodelId: string;\n\tdim: number;\n\tcount: number;\n}\n\n/** One candidate sent for cross-encoder scoring. */\nexport interface EmbSearchRerankPassage {\n\tid: string;\n\ttext: string;\n}\n\n/** A cross-encoder relevance logit. Higher is more relevant, but the scale is\n *  unnormalized and comparable only within one call. */\nexport interface EmbSearchRerankResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchBulkResult {\n\tinserted: number;\n\tupdated: number;\n}\n\n/** Which daemon-side retriever answers a query (embsearch >= 0.2.0). */\nexport type DaemonRetriever = \"dense\" | \"lexical\" | \"hybrid\";\n\nexport interface EmbSearchClientOptions {\n\t/** Open/create the store with a BM25 lexical index alongside the vectors. */\n\thybrid?: boolean;\n\t/** Path to the `embsearch` binary. */\n\tbinaryPath: string;\n\t/** Store directory passed as `--path`. */\n\tstorePath: string;\n\t/** Metric for a freshly created store. Default: \"cosine\". */\n\tmetric?: \"cosine\" | \"dot\" | \"euclidean\";\n}\n\ninterface Pending {\n\tresolve: (value: EmbSearchRawResponse) => void;\n\treject: (err: Error) => void;\n}\n\ninterface EmbSearchRawResponse {\n\tok: boolean;\n\terror?: string;\n\tresults?: EmbSearchResult[];\n\tinserted?: boolean;\n\tremoved?: boolean;\n\tcount?: number;\n\tinserted_count?: number;\n\tupdated_count?: number;\n\tmodel_id?: string;\n\tdim?: number;\n\t/** Cross-encoder scores. Separate from `results` because the daemon's\n\t *  logits are not comparable to retrieval scores. */\n\treranked?: EmbSearchRerankResult[];\n}\n\nexport class EmbSearchClient {\n\tprivate proc: ChildProcessWithoutNullStreams;\n\tprivate queue: Pending[] = [];\n\tprivate buffer = \"\";\n\tprivate closed = false;\n\tprivate readyPromise: Promise<void>;\n\n\tconstructor(opts: EmbSearchClientOptions) {\n\t\tconst args = [\"serve\", \"--path\", opts.storePath];\n\t\tif (opts.metric) args.push(\"--metric\", opts.metric);\n\t\t// Hybrid-ness is fixed when a store is created: passing --hybrid against\n\t\t// an existing non-hybrid store warns and is ignored daemon-side, so a\n\t\t// hybrid store needs its own directory.\n\t\tif (opts.hybrid) args.push(\"--hybrid\");\n\n\t\tthis.proc = spawn(opts.binaryPath, args, { stdio: [\"pipe\", \"pipe\", \"pipe\"] });\n\n\t\tthis.proc.stdout.setEncoding(\"utf8\");\n\t\tthis.proc.stdout.on(\"data\", (chunk: string) => this.onStdout(chunk));\n\t\t// Swallow the informational stderr banner; errors surface via responses/exit.\n\t\tthis.proc.stderr.resume();\n\n\t\t// Readiness = the daemon answering a ping (probes the actual request loop).\n\t\tthis.readyPromise = this.send({ op: \"ping\" }).then(() => undefined);\n\t\t// Spawn failures reject the pending ping via the exit handler; mark the\n\t\t// promise handled so a failed spawn doesn't raise an unhandled rejection\n\t\t// before ready() is awaited.\n\t\tthis.readyPromise.catch(() => {});\n\n\t\tthis.proc.on(\"error\", (err) => {\n\t\t\tthis.closed = true;\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t\tthis.proc.on(\"exit\", (code) => {\n\t\t\tthis.closed = true;\n\t\t\tconst err = new Error(`embsearch daemon exited (code ${code})`);\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t}\n\n\t/** Resolves once the daemon has loaded the model + index. */\n\tready(): Promise<void> {\n\t\treturn this.readyPromise;\n\t}\n\n\tget isClosed(): boolean {\n\t\treturn this.closed;\n\t}\n\n\tprivate onStdout(chunk: string): void {\n\t\tthis.buffer += chunk;\n\t\t// Each response is one line; dispatch FIFO against the pending queue.\n\t\tfor (let nl = this.buffer.indexOf(\"\\n\"); nl !== -1; nl = this.buffer.indexOf(\"\\n\")) {\n\t\t\tconst line = this.buffer.slice(0, nl).trim();\n\t\t\tthis.buffer = this.buffer.slice(nl + 1);\n\t\t\tif (!line) continue;\n\t\t\tconst pending = this.queue.shift();\n\t\t\tif (!pending) continue;\n\t\t\tlet msg: EmbSearchRawResponse;\n\t\t\ttry {\n\t\t\t\tmsg = JSON.parse(line) as EmbSearchRawResponse;\n\t\t\t} catch {\n\t\t\t\tpending.reject(new Error(`bad response: ${line}`));\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tif (msg.ok) pending.resolve(msg);\n\t\t\telse pending.reject(new Error(msg.error ?? \"unknown error\"));\n\t\t}\n\t}\n\n\tprivate send(req: Record<string, unknown>): Promise<EmbSearchRawResponse> {\n\t\tif (this.closed) return Promise.reject(new Error(\"embsearch client is closed\"));\n\t\treturn new Promise<EmbSearchRawResponse>((resolve, reject) => {\n\t\t\tthis.queue.push({ resolve, reject });\n\t\t\tthis.proc.stdin.write(`${JSON.stringify(req)}\\n`);\n\t\t});\n\t}\n\n\t/**\n\t * Search for the top-`k` matches for `text`.\n\t *\n\t * `retriever` selects which leg answers:\n\t *  - `dense` (default) — vector search, the historical behaviour;\n\t *  - `lexical` — BM25 only, raw scores, no embedding computed;\n\t *  - `hybrid` — both, pre-fused by the daemon's own RRF constant.\n\t *\n\t * `lexical` and `hybrid` need a store created with `--hybrid`. Prefer\n\t * `lexical` over `hybrid` when fusing here: `hybrid` collapses both legs\n\t * into one RRF score, discarding the per-retriever ranks the trace records\n\t * and preventing n-way fusion with the grep leg.\n\t */\n\tasync query(text: string, k = 10, retriever: DaemonRetriever = \"dense\"): Promise<EmbSearchResult[]> {\n\t\tconst res = await this.send(\n\t\t\tretriever === \"dense\" ? { op: \"query\", text, k } : { op: \"query\", text, k, retriever },\n\t\t);\n\t\treturn res.results ?? [];\n\t}\n\n\t/**\n\t * Score `passages` against `query` with the daemon's cross-encoder and\n\t * return the best `k`, best first.\n\t *\n\t * Passages are sent inline rather than referenced by id: the caller has the\n\t * exact spans it intends to show the model, and a cross-encoder scores the\n\t * text it is given, so sending anything else would score the wrong thing.\n\t * Requires an onnx build with reranker weights (embsearch >= 0.3.0).\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tconst res = await this.send({ op: \"rerank\", query, passages, k });\n\t\treturn res.reranked ?? [];\n\t}\n\n\t/**\n\t * Batched insert-or-replace. One embedding inference for the whole batch —\n\t * the fast path for bulk indexing. Keep batches modest (e.g. 32–64) so a\n\t * concurrent query is not stuck behind a huge inference.\n\t */\n\tasync bulk(items: Array<{ id: string; text: string }>): Promise<EmbSearchBulkResult> {\n\t\tconst res = await this.send({ op: \"bulk\", items });\n\t\treturn { inserted: res.inserted_count ?? 0, updated: res.updated_count ?? 0 };\n\t}\n\n\t/** Model id, dimensionality, and live vector count of the daemon. */\n\tasync info(): Promise<EmbSearchDaemonInfo> {\n\t\tconst res = await this.send({ op: \"info\" });\n\t\treturn { modelId: res.model_id ?? \"\", dim: res.dim ?? 0, count: res.count ?? 0 };\n\t}\n\n\t/** Remove a record. Resolves to `true` if it existed. */\n\tasync remove(id: string): Promise<boolean> {\n\t\tconst res = await this.send({ op: \"remove\", id });\n\t\treturn res.removed === true;\n\t}\n\n\t/** Reclaim tombstoned rows left behind by `remove`. */\n\tasync compact(): Promise<void> {\n\t\tawait this.send({ op: \"compact\" });\n\t}\n\n\t/** Persist the index to the store directory. */\n\tasync save(): Promise<void> {\n\t\tawait this.send({ op: \"save\" });\n\t}\n\n\t/** Shut the daemon down, closing stdin so it exits cleanly. */\n\tasync close(): Promise<void> {\n\t\tif (this.closed) return;\n\t\tthis.proc.stdin.end();\n\t\tawait new Promise<void>((resolve) => this.proc.on(\"exit\", () => resolve()));\n\t}\n}\n"]}
+{"version":3,"file":"client.d.ts","sourceRoot":"","sources":["../../../src/core/embsearch/client.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAIH,MAAM,WAAW,eAAe;IAC/B,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,mBAAmB;IACnC,OAAO,EAAE,MAAM,CAAC;IAChB,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd;;;;;;;;OAQG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;CACjB;AAED,oDAAoD;AACpD,MAAM,WAAW,sBAAsB;IACtC,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;CACb;AAED;wDACwD;AACxD,MAAM,WAAW,qBAAqB;IACrC,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,mBAAmB;IACnC,QAAQ,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,MAAM,CAAC;CAChB;AAED,wEAAwE;AACxE,MAAM,MAAM,eAAe,GAAG,OAAO,GAAG,SAAS,GAAG,QAAQ,CAAC;AAE7D,MAAM,WAAW,sBAAsB;IACtC,6EAA6E;IAC7E,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,sCAAsC;IACtC,UAAU,EAAE,MAAM,CAAC;IACnB,0CAA0C;IAC1C,SAAS,EAAE,MAAM,CAAC;IAClB,6DAA6D;IAC7D,MAAM,CAAC,EAAE,QAAQ,GAAG,KAAK,GAAG,WAAW,CAAC;CACxC;AA0BD,qBAAa,eAAe;IAC3B,OAAO,CAAC,IAAI,CAAiC;IAC7C,OAAO,CAAC,KAAK,CAAiB;IAC9B,OAAO,CAAC,MAAM,CAAM;IACpB,OAAO,CAAC,MAAM,CAAS;IACvB,OAAO,CAAC,YAAY,CAAgB;IAEpC,YAAY,IAAI,EAAE,sBAAsB,EA+BvC;IAED,6DAA6D;IAC7D,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAErB;IAED,IAAI,QAAQ,IAAI,OAAO,CAEtB;IAED,OAAO,CAAC,QAAQ;IAqBhB,OAAO,CAAC,IAAI;IAQZ;;;;;;;;;;;;OAYG;IACG,KAAK,CAAC,IAAI,EAAE,MAAM,EAAE,CAAC,SAAK,EAAE,SAAS,GAAE,eAAyB,GAAG,OAAO,CAAC,eAAe,EAAE,CAAC,CAKlG;IAED;;;;;;;;OAQG;IACG,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,sBAAsB,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,qBAAqB,EAAE,CAAC,CAG3G;IAED;;;;OAIG;IACG,IAAI,CAAC,KAAK,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAGnF;IAED,qEAAqE;IAC/D,IAAI,IAAI,OAAO,CAAC,mBAAmB,CAAC,CAGzC;IAED,yDAAyD;IACnD,MAAM,CAAC,EAAE,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,CAGzC;IAED,uDAAuD;IACjD,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAE7B;IAED,gDAAgD;IAC1C,IAAI,IAAI,OAAO,CAAC,IAAI,CAAC,CAE1B;IAED,+DAA+D;IACzD,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAI3B;CACD","sourcesContent":["/**\n * Client for the `embsearch` stdio daemon (vendored from\n * github.com/kolisachint/embeddingsearchtools ts/client.ts).\n *\n * Spawns `embsearch serve` once and talks newline-delimited JSON over its\n * stdin/stdout. The process stays alive so the model and index load a single\n * time; every query after startup is hot.\n */\n\nimport { type ChildProcessWithoutNullStreams, spawn } from \"child_process\";\n\nexport interface EmbSearchResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchDaemonInfo {\n\tmodelId: string;\n\tdim: number;\n\tcount: number;\n\t/**\n\t * Whether this daemon can actually serve `rerank`, or `undefined` from a\n\t * daemon too old to say.\n\t *\n\t * Reported separately from the version because the two came apart:\n\t * released binaries carry the `rerank` op but no longer bundle the ~23 MB\n\t * cross-encoder weights, so a version check alone would advertise a\n\t * reranker that fails on first call.\n\t */\n\trerank?: boolean;\n}\n\n/** One candidate sent for cross-encoder scoring. */\nexport interface EmbSearchRerankPassage {\n\tid: string;\n\ttext: string;\n}\n\n/** A cross-encoder relevance logit. Higher is more relevant, but the scale is\n *  unnormalized and comparable only within one call. */\nexport interface EmbSearchRerankResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchBulkResult {\n\tinserted: number;\n\tupdated: number;\n}\n\n/** Which daemon-side retriever answers a query (embsearch >= 0.2.0). */\nexport type DaemonRetriever = \"dense\" | \"lexical\" | \"hybrid\";\n\nexport interface EmbSearchClientOptions {\n\t/** Open/create the store with a BM25 lexical index alongside the vectors. */\n\thybrid?: boolean;\n\t/** Path to the `embsearch` binary. */\n\tbinaryPath: string;\n\t/** Store directory passed as `--path`. */\n\tstorePath: string;\n\t/** Metric for a freshly created store. Default: \"cosine\". */\n\tmetric?: \"cosine\" | \"dot\" | \"euclidean\";\n}\n\ninterface Pending {\n\tresolve: (value: EmbSearchRawResponse) => void;\n\treject: (err: Error) => void;\n}\n\ninterface EmbSearchRawResponse {\n\tok: boolean;\n\terror?: string;\n\tresults?: EmbSearchResult[];\n\tinserted?: boolean;\n\tremoved?: boolean;\n\tcount?: number;\n\tinserted_count?: number;\n\tupdated_count?: number;\n\tmodel_id?: string;\n\tdim?: number;\n\t/** Cross-encoder scores. Separate from `results` because the daemon's\n\t *  logits are not comparable to retrieval scores. */\n\treranked?: EmbSearchRerankResult[];\n\t/** `info` only: whether `rerank` will work. Absent on daemons too old to\n\t *  report it. */\n\trerank?: boolean;\n}\n\nexport class EmbSearchClient {\n\tprivate proc: ChildProcessWithoutNullStreams;\n\tprivate queue: Pending[] = [];\n\tprivate buffer = \"\";\n\tprivate closed = false;\n\tprivate readyPromise: Promise<void>;\n\n\tconstructor(opts: EmbSearchClientOptions) {\n\t\tconst args = [\"serve\", \"--path\", opts.storePath];\n\t\tif (opts.metric) args.push(\"--metric\", opts.metric);\n\t\t// Hybrid-ness is fixed when a store is created: passing --hybrid against\n\t\t// an existing non-hybrid store warns and is ignored daemon-side, so a\n\t\t// hybrid store needs its own directory.\n\t\tif (opts.hybrid) args.push(\"--hybrid\");\n\n\t\tthis.proc = spawn(opts.binaryPath, args, { stdio: [\"pipe\", \"pipe\", \"pipe\"] });\n\n\t\tthis.proc.stdout.setEncoding(\"utf8\");\n\t\tthis.proc.stdout.on(\"data\", (chunk: string) => this.onStdout(chunk));\n\t\t// Swallow the informational stderr banner; errors surface via responses/exit.\n\t\tthis.proc.stderr.resume();\n\n\t\t// Readiness = the daemon answering a ping (probes the actual request loop).\n\t\tthis.readyPromise = this.send({ op: \"ping\" }).then(() => undefined);\n\t\t// Spawn failures reject the pending ping via the exit handler; mark the\n\t\t// promise handled so a failed spawn doesn't raise an unhandled rejection\n\t\t// before ready() is awaited.\n\t\tthis.readyPromise.catch(() => {});\n\n\t\tthis.proc.on(\"error\", (err) => {\n\t\t\tthis.closed = true;\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t\tthis.proc.on(\"exit\", (code) => {\n\t\t\tthis.closed = true;\n\t\t\tconst err = new Error(`embsearch daemon exited (code ${code})`);\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t}\n\n\t/** Resolves once the daemon has loaded the model + index. */\n\tready(): Promise<void> {\n\t\treturn this.readyPromise;\n\t}\n\n\tget isClosed(): boolean {\n\t\treturn this.closed;\n\t}\n\n\tprivate onStdout(chunk: string): void {\n\t\tthis.buffer += chunk;\n\t\t// Each response is one line; dispatch FIFO against the pending queue.\n\t\tfor (let nl = this.buffer.indexOf(\"\\n\"); nl !== -1; nl = this.buffer.indexOf(\"\\n\")) {\n\t\t\tconst line = this.buffer.slice(0, nl).trim();\n\t\t\tthis.buffer = this.buffer.slice(nl + 1);\n\t\t\tif (!line) continue;\n\t\t\tconst pending = this.queue.shift();\n\t\t\tif (!pending) continue;\n\t\t\tlet msg: EmbSearchRawResponse;\n\t\t\ttry {\n\t\t\t\tmsg = JSON.parse(line) as EmbSearchRawResponse;\n\t\t\t} catch {\n\t\t\t\tpending.reject(new Error(`bad response: ${line}`));\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tif (msg.ok) pending.resolve(msg);\n\t\t\telse pending.reject(new Error(msg.error ?? \"unknown error\"));\n\t\t}\n\t}\n\n\tprivate send(req: Record<string, unknown>): Promise<EmbSearchRawResponse> {\n\t\tif (this.closed) return Promise.reject(new Error(\"embsearch client is closed\"));\n\t\treturn new Promise<EmbSearchRawResponse>((resolve, reject) => {\n\t\t\tthis.queue.push({ resolve, reject });\n\t\t\tthis.proc.stdin.write(`${JSON.stringify(req)}\\n`);\n\t\t});\n\t}\n\n\t/**\n\t * Search for the top-`k` matches for `text`.\n\t *\n\t * `retriever` selects which leg answers:\n\t *  - `dense` (default) — vector search, the historical behaviour;\n\t *  - `lexical` — BM25 only, raw scores, no embedding computed;\n\t *  - `hybrid` — both, pre-fused by the daemon's own RRF constant.\n\t *\n\t * `lexical` and `hybrid` need a store created with `--hybrid`. Prefer\n\t * `lexical` over `hybrid` when fusing here: `hybrid` collapses both legs\n\t * into one RRF score, discarding the per-retriever ranks the trace records\n\t * and preventing n-way fusion with the grep leg.\n\t */\n\tasync query(text: string, k = 10, retriever: DaemonRetriever = \"dense\"): Promise<EmbSearchResult[]> {\n\t\tconst res = await this.send(\n\t\t\tretriever === \"dense\" ? { op: \"query\", text, k } : { op: \"query\", text, k, retriever },\n\t\t);\n\t\treturn res.results ?? [];\n\t}\n\n\t/**\n\t * Score `passages` against `query` with the daemon's cross-encoder and\n\t * return the best `k`, best first.\n\t *\n\t * Passages are sent inline rather than referenced by id: the caller has the\n\t * exact spans it intends to show the model, and a cross-encoder scores the\n\t * text it is given, so sending anything else would score the wrong thing.\n\t * Requires an onnx build with reranker weights (embsearch >= 0.3.0).\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tconst res = await this.send({ op: \"rerank\", query, passages, k });\n\t\treturn res.reranked ?? [];\n\t}\n\n\t/**\n\t * Batched insert-or-replace. One embedding inference for the whole batch —\n\t * the fast path for bulk indexing. Keep batches modest (e.g. 32–64) so a\n\t * concurrent query is not stuck behind a huge inference.\n\t */\n\tasync bulk(items: Array<{ id: string; text: string }>): Promise<EmbSearchBulkResult> {\n\t\tconst res = await this.send({ op: \"bulk\", items });\n\t\treturn { inserted: res.inserted_count ?? 0, updated: res.updated_count ?? 0 };\n\t}\n\n\t/** Model id, dimensionality, and live vector count of the daemon. */\n\tasync info(): Promise<EmbSearchDaemonInfo> {\n\t\tconst res = await this.send({ op: \"info\" });\n\t\treturn { modelId: res.model_id ?? \"\", dim: res.dim ?? 0, count: res.count ?? 0, rerank: res.rerank };\n\t}\n\n\t/** Remove a record. Resolves to `true` if it existed. */\n\tasync remove(id: string): Promise<boolean> {\n\t\tconst res = await this.send({ op: \"remove\", id });\n\t\treturn res.removed === true;\n\t}\n\n\t/** Reclaim tombstoned rows left behind by `remove`. */\n\tasync compact(): Promise<void> {\n\t\tawait this.send({ op: \"compact\" });\n\t}\n\n\t/** Persist the index to the store directory. */\n\tasync save(): Promise<void> {\n\t\tawait this.send({ op: \"save\" });\n\t}\n\n\t/** Shut the daemon down, closing stdin so it exits cleanly. */\n\tasync close(): Promise<void> {\n\t\tif (this.closed) return;\n\t\tthis.proc.stdin.end();\n\t\tawait new Promise<void>((resolve) => this.proc.on(\"exit\", () => resolve()));\n\t}\n}\n"]}

package/dist/core/embsearch/client.js

@@ -127,7 +127,7 @@ export class EmbSearchClient {
     /** Model id, dimensionality, and live vector count of the daemon. */
     async info() {
         const res = await this.send({ op: "info" });
-        return { modelId: res.model_id ?? "", dim: res.dim ?? 0, count: res.count ?? 0 };
+        return { modelId: res.model_id ?? "", dim: res.dim ?? 0, count: res.count ?? 0, rerank: res.rerank };
     }
     /** Remove a record. Resolves to `true` if it existed. */
     async remove(id) {

package/dist/core/embsearch/client.js.map

@@ -1 +1 @@
-{"version":3,"file":"client.js","sourceRoot":"","sources":["../../../src/core/embsearch/client.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAuC,KAAK,EAAE,MAAM,eAAe,CAAC;AAkE3E,MAAM,OAAO,eAAe;IACnB,IAAI,CAAiC;IACrC,KAAK,GAAc,EAAE,CAAC;IACtB,MAAM,GAAG,EAAE,CAAC;IACZ,MAAM,GAAG,KAAK,CAAC;IACf,YAAY,CAAgB;IAEpC,YAAY,IAA4B,EAAE;QACzC,MAAM,IAAI,GAAG,CAAC,OAAO,EAAE,QAAQ,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;QACjD,IAAI,IAAI,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;QACpD,yEAAyE;QACzE,sEAAsE;QACtE,wCAAwC;QACxC,IAAI,IAAI,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAEvC,IAAI,CAAC,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,KAAK,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;QAE9E,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;QACrC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,KAAa,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;QACrE,8EAA8E;QAC9E,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;QAE1B,4EAA4E;QAC5E,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;QACpE,wEAAwE;QACxE,yEAAyE;QACzE,6BAA6B;QAC7B,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAC,CAAC,CAAC,CAAC;QAElC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;YAC9B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;YACnB,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAA,CACpD,CAAC,CAAC;QACH,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC;YAC9B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;YACnB,MAAM,GAAG,GAAG,IAAI,KAAK,CAAC,iCAAiC,IAAI,GAAG,CAAC,CAAC;YAChE,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAA,CACpD,CAAC,CAAC;IAAA,CACH;IAED,6DAA6D;IAC7D,KAAK,GAAkB;QACtB,OAAO,IAAI,CAAC,YAAY,CAAC;IAAA,CACzB;IAED,IAAI,QAAQ,GAAY;QACvB,OAAO,IAAI,CAAC,MAAM,CAAC;IAAA,CACnB;IAEO,QAAQ,CAAC,KAAa,EAAQ;QACrC,IAAI,CAAC,MAAM,IAAI,KAAK,CAAC;QACrB,sEAAsE;QACtE,KAAK,IAAI,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,EAAE,KAAK,CAAC,CAAC,EAAE,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;YACpF,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;YAC7C,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;YACxC,IAAI,CAAC,IAAI;gBAAE,SAAS;YACpB,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;YACnC,IAAI,CAAC,OAAO;gBAAE,SAAS;YACvB,IAAI,GAAyB,CAAC;YAC9B,IAAI,CAAC;gBACJ,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAyB,CAAC;YAChD,CAAC;YAAC,MAAM,CAAC;gBACR,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,iBAAiB,IAAI,EAAE,CAAC,CAAC,CAAC;gBACnD,SAAS;YACV,CAAC;YACD,IAAI,GAAG,CAAC,EAAE;gBAAE,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;;gBAC5B,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,GAAG,CAAC,KAAK,IAAI,eAAe,CAAC,CAAC,CAAC;QAC9D,CAAC;IAAA,CACD;IAEO,IAAI,CAAC,GAA4B,EAAiC;QACzE,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC,CAAC;QAChF,OAAO,IAAI,OAAO,CAAuB,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,CAAC;YAC7D,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;YACrC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAAA,CAClD,CAAC,CAAC;IAAA,CACH;IAED;;;;;;;;;;;;OAYG;IACH,KAAK,CAAC,KAAK,CAAC,IAAY,EAAE,CAAC,GAAG,EAAE,EAAE,SAAS,GAAoB,OAAO,EAA8B;QACnG,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAC1B,SAAS,KAAK,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,SAAS,EAAE,CACtF,CAAC;QACF,OAAO,GAAG,CAAC,OAAO,IAAI,EAAE,CAAC;IAAA,CACzB;IAED;;;;;;;;OAQG;IACH,KAAK,CAAC,MAAM,CAAC,KAAa,EAAE,QAAkC,EAAE,CAAS,EAAoC;QAC5G,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QAClE,OAAO,GAAG,CAAC,QAAQ,IAAI,EAAE,CAAC;IAAA,CAC1B;IAED;;;;OAIG;IACH,KAAK,CAAC,IAAI,CAAC,KAA0C,EAAgC;QACpF,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;QACnD,OAAO,EAAE,QAAQ,EAAE,GAAG,CAAC,cAAc,IAAI,CAAC,EAAE,OAAO,EAAE,GAAG,CAAC,aAAa,IAAI,CAAC,EAAE,CAAC;IAAA,CAC9E;IAED,qEAAqE;IACrE,KAAK,CAAC,IAAI,GAAiC;QAC1C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;QAC5C,OAAO,EAAE,OAAO,EAAE,GAAG,CAAC,QAAQ,IAAI,EAAE,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,CAAC,KAAK,IAAI,CAAC,EAAE,CAAC;IAAA,CACjF;IAED,yDAAyD;IACzD,KAAK,CAAC,MAAM,CAAC,EAAU,EAAoB;QAC1C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;QAClD,OAAO,GAAG,CAAC,OAAO,KAAK,IAAI,CAAC;IAAA,CAC5B;IAED,uDAAuD;IACvD,KAAK,CAAC,OAAO,GAAkB;QAC9B,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,CAAC;IAAA,CACnC;IAED,gDAAgD;IAChD,KAAK,CAAC,IAAI,GAAkB;QAC3B,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;IAAA,CAChC;IAED,+DAA+D;IAC/D,KAAK,CAAC,KAAK,GAAkB;QAC5B,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO;QACxB,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC;QACtB,MAAM,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;IAAA,CAC5E;CACD","sourcesContent":["/**\n * Client for the `embsearch` stdio daemon (vendored from\n * github.com/kolisachint/embeddingsearchtools ts/client.ts).\n *\n * Spawns `embsearch serve` once and talks newline-delimited JSON over its\n * stdin/stdout. The process stays alive so the model and index load a single\n * time; every query after startup is hot.\n */\n\nimport { type ChildProcessWithoutNullStreams, spawn } from \"child_process\";\n\nexport interface EmbSearchResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchDaemonInfo {\n\tmodelId: string;\n\tdim: number;\n\tcount: number;\n}\n\n/** One candidate sent for cross-encoder scoring. */\nexport interface EmbSearchRerankPassage {\n\tid: string;\n\ttext: string;\n}\n\n/** A cross-encoder relevance logit. Higher is more relevant, but the scale is\n *  unnormalized and comparable only within one call. */\nexport interface EmbSearchRerankResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchBulkResult {\n\tinserted: number;\n\tupdated: number;\n}\n\n/** Which daemon-side retriever answers a query (embsearch >= 0.2.0). */\nexport type DaemonRetriever = \"dense\" | \"lexical\" | \"hybrid\";\n\nexport interface EmbSearchClientOptions {\n\t/** Open/create the store with a BM25 lexical index alongside the vectors. */\n\thybrid?: boolean;\n\t/** Path to the `embsearch` binary. */\n\tbinaryPath: string;\n\t/** Store directory passed as `--path`. */\n\tstorePath: string;\n\t/** Metric for a freshly created store. Default: \"cosine\". */\n\tmetric?: \"cosine\" | \"dot\" | \"euclidean\";\n}\n\ninterface Pending {\n\tresolve: (value: EmbSearchRawResponse) => void;\n\treject: (err: Error) => void;\n}\n\ninterface EmbSearchRawResponse {\n\tok: boolean;\n\terror?: string;\n\tresults?: EmbSearchResult[];\n\tinserted?: boolean;\n\tremoved?: boolean;\n\tcount?: number;\n\tinserted_count?: number;\n\tupdated_count?: number;\n\tmodel_id?: string;\n\tdim?: number;\n\t/** Cross-encoder scores. Separate from `results` because the daemon's\n\t *  logits are not comparable to retrieval scores. */\n\treranked?: EmbSearchRerankResult[];\n}\n\nexport class EmbSearchClient {\n\tprivate proc: ChildProcessWithoutNullStreams;\n\tprivate queue: Pending[] = [];\n\tprivate buffer = \"\";\n\tprivate closed = false;\n\tprivate readyPromise: Promise<void>;\n\n\tconstructor(opts: EmbSearchClientOptions) {\n\t\tconst args = [\"serve\", \"--path\", opts.storePath];\n\t\tif (opts.metric) args.push(\"--metric\", opts.metric);\n\t\t// Hybrid-ness is fixed when a store is created: passing --hybrid against\n\t\t// an existing non-hybrid store warns and is ignored daemon-side, so a\n\t\t// hybrid store needs its own directory.\n\t\tif (opts.hybrid) args.push(\"--hybrid\");\n\n\t\tthis.proc = spawn(opts.binaryPath, args, { stdio: [\"pipe\", \"pipe\", \"pipe\"] });\n\n\t\tthis.proc.stdout.setEncoding(\"utf8\");\n\t\tthis.proc.stdout.on(\"data\", (chunk: string) => this.onStdout(chunk));\n\t\t// Swallow the informational stderr banner; errors surface via responses/exit.\n\t\tthis.proc.stderr.resume();\n\n\t\t// Readiness = the daemon answering a ping (probes the actual request loop).\n\t\tthis.readyPromise = this.send({ op: \"ping\" }).then(() => undefined);\n\t\t// Spawn failures reject the pending ping via the exit handler; mark the\n\t\t// promise handled so a failed spawn doesn't raise an unhandled rejection\n\t\t// before ready() is awaited.\n\t\tthis.readyPromise.catch(() => {});\n\n\t\tthis.proc.on(\"error\", (err) => {\n\t\t\tthis.closed = true;\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t\tthis.proc.on(\"exit\", (code) => {\n\t\t\tthis.closed = true;\n\t\t\tconst err = new Error(`embsearch daemon exited (code ${code})`);\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t}\n\n\t/** Resolves once the daemon has loaded the model + index. */\n\tready(): Promise<void> {\n\t\treturn this.readyPromise;\n\t}\n\n\tget isClosed(): boolean {\n\t\treturn this.closed;\n\t}\n\n\tprivate onStdout(chunk: string): void {\n\t\tthis.buffer += chunk;\n\t\t// Each response is one line; dispatch FIFO against the pending queue.\n\t\tfor (let nl = this.buffer.indexOf(\"\\n\"); nl !== -1; nl = this.buffer.indexOf(\"\\n\")) {\n\t\t\tconst line = this.buffer.slice(0, nl).trim();\n\t\t\tthis.buffer = this.buffer.slice(nl + 1);\n\t\t\tif (!line) continue;\n\t\t\tconst pending = this.queue.shift();\n\t\t\tif (!pending) continue;\n\t\t\tlet msg: EmbSearchRawResponse;\n\t\t\ttry {\n\t\t\t\tmsg = JSON.parse(line) as EmbSearchRawResponse;\n\t\t\t} catch {\n\t\t\t\tpending.reject(new Error(`bad response: ${line}`));\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tif (msg.ok) pending.resolve(msg);\n\t\t\telse pending.reject(new Error(msg.error ?? \"unknown error\"));\n\t\t}\n\t}\n\n\tprivate send(req: Record<string, unknown>): Promise<EmbSearchRawResponse> {\n\t\tif (this.closed) return Promise.reject(new Error(\"embsearch client is closed\"));\n\t\treturn new Promise<EmbSearchRawResponse>((resolve, reject) => {\n\t\t\tthis.queue.push({ resolve, reject });\n\t\t\tthis.proc.stdin.write(`${JSON.stringify(req)}\\n`);\n\t\t});\n\t}\n\n\t/**\n\t * Search for the top-`k` matches for `text`.\n\t *\n\t * `retriever` selects which leg answers:\n\t *  - `dense` (default) — vector search, the historical behaviour;\n\t *  - `lexical` — BM25 only, raw scores, no embedding computed;\n\t *  - `hybrid` — both, pre-fused by the daemon's own RRF constant.\n\t *\n\t * `lexical` and `hybrid` need a store created with `--hybrid`. Prefer\n\t * `lexical` over `hybrid` when fusing here: `hybrid` collapses both legs\n\t * into one RRF score, discarding the per-retriever ranks the trace records\n\t * and preventing n-way fusion with the grep leg.\n\t */\n\tasync query(text: string, k = 10, retriever: DaemonRetriever = \"dense\"): Promise<EmbSearchResult[]> {\n\t\tconst res = await this.send(\n\t\t\tretriever === \"dense\" ? { op: \"query\", text, k } : { op: \"query\", text, k, retriever },\n\t\t);\n\t\treturn res.results ?? [];\n\t}\n\n\t/**\n\t * Score `passages` against `query` with the daemon's cross-encoder and\n\t * return the best `k`, best first.\n\t *\n\t * Passages are sent inline rather than referenced by id: the caller has the\n\t * exact spans it intends to show the model, and a cross-encoder scores the\n\t * text it is given, so sending anything else would score the wrong thing.\n\t * Requires an onnx build with reranker weights (embsearch >= 0.3.0).\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tconst res = await this.send({ op: \"rerank\", query, passages, k });\n\t\treturn res.reranked ?? [];\n\t}\n\n\t/**\n\t * Batched insert-or-replace. One embedding inference for the whole batch —\n\t * the fast path for bulk indexing. Keep batches modest (e.g. 32–64) so a\n\t * concurrent query is not stuck behind a huge inference.\n\t */\n\tasync bulk(items: Array<{ id: string; text: string }>): Promise<EmbSearchBulkResult> {\n\t\tconst res = await this.send({ op: \"bulk\", items });\n\t\treturn { inserted: res.inserted_count ?? 0, updated: res.updated_count ?? 0 };\n\t}\n\n\t/** Model id, dimensionality, and live vector count of the daemon. */\n\tasync info(): Promise<EmbSearchDaemonInfo> {\n\t\tconst res = await this.send({ op: \"info\" });\n\t\treturn { modelId: res.model_id ?? \"\", dim: res.dim ?? 0, count: res.count ?? 0 };\n\t}\n\n\t/** Remove a record. Resolves to `true` if it existed. */\n\tasync remove(id: string): Promise<boolean> {\n\t\tconst res = await this.send({ op: \"remove\", id });\n\t\treturn res.removed === true;\n\t}\n\n\t/** Reclaim tombstoned rows left behind by `remove`. */\n\tasync compact(): Promise<void> {\n\t\tawait this.send({ op: \"compact\" });\n\t}\n\n\t/** Persist the index to the store directory. */\n\tasync save(): Promise<void> {\n\t\tawait this.send({ op: \"save\" });\n\t}\n\n\t/** Shut the daemon down, closing stdin so it exits cleanly. */\n\tasync close(): Promise<void> {\n\t\tif (this.closed) return;\n\t\tthis.proc.stdin.end();\n\t\tawait new Promise<void>((resolve) => this.proc.on(\"exit\", () => resolve()));\n\t}\n}\n"]}
+{"version":3,"file":"client.js","sourceRoot":"","sources":["../../../src/core/embsearch/client.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAuC,KAAK,EAAE,MAAM,eAAe,CAAC;AA+E3E,MAAM,OAAO,eAAe;IACnB,IAAI,CAAiC;IACrC,KAAK,GAAc,EAAE,CAAC;IACtB,MAAM,GAAG,EAAE,CAAC;IACZ,MAAM,GAAG,KAAK,CAAC;IACf,YAAY,CAAgB;IAEpC,YAAY,IAA4B,EAAE;QACzC,MAAM,IAAI,GAAG,CAAC,OAAO,EAAE,QAAQ,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;QACjD,IAAI,IAAI,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;QACpD,yEAAyE;QACzE,sEAAsE;QACtE,wCAAwC;QACxC,IAAI,IAAI,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAEvC,IAAI,CAAC,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,KAAK,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;QAE9E,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;QACrC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,KAAa,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;QACrE,8EAA8E;QAC9E,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;QAE1B,4EAA4E;QAC5E,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;QACpE,wEAAwE;QACxE,yEAAyE;QACzE,6BAA6B;QAC7B,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAC,CAAC,CAAC,CAAC;QAElC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;YAC9B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;YACnB,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAA,CACpD,CAAC,CAAC;QACH,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC;YAC9B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;YACnB,MAAM,GAAG,GAAG,IAAI,KAAK,CAAC,iCAAiC,IAAI,GAAG,CAAC,CAAC;YAChE,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAA,CACpD,CAAC,CAAC;IAAA,CACH;IAED,6DAA6D;IAC7D,KAAK,GAAkB;QACtB,OAAO,IAAI,CAAC,YAAY,CAAC;IAAA,CACzB;IAED,IAAI,QAAQ,GAAY;QACvB,OAAO,IAAI,CAAC,MAAM,CAAC;IAAA,CACnB;IAEO,QAAQ,CAAC,KAAa,EAAQ;QACrC,IAAI,CAAC,MAAM,IAAI,KAAK,CAAC;QACrB,sEAAsE;QACtE,KAAK,IAAI,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,EAAE,KAAK,CAAC,CAAC,EAAE,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;YACpF,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;YAC7C,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;YACxC,IAAI,CAAC,IAAI;gBAAE,SAAS;YACpB,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;YACnC,IAAI,CAAC,OAAO;gBAAE,SAAS;YACvB,IAAI,GAAyB,CAAC;YAC9B,IAAI,CAAC;gBACJ,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAyB,CAAC;YAChD,CAAC;YAAC,MAAM,CAAC;gBACR,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,iBAAiB,IAAI,EAAE,CAAC,CAAC,CAAC;gBACnD,SAAS;YACV,CAAC;YACD,IAAI,GAAG,CAAC,EAAE;gBAAE,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;;gBAC5B,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,GAAG,CAAC,KAAK,IAAI,eAAe,CAAC,CAAC,CAAC;QAC9D,CAAC;IAAA,CACD;IAEO,IAAI,CAAC,GAA4B,EAAiC;QACzE,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC,CAAC;QAChF,OAAO,IAAI,OAAO,CAAuB,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,CAAC;YAC7D,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;YACrC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAAA,CAClD,CAAC,CAAC;IAAA,CACH;IAED;;;;;;;;;;;;OAYG;IACH,KAAK,CAAC,KAAK,CAAC,IAAY,EAAE,CAAC,GAAG,EAAE,EAAE,SAAS,GAAoB,OAAO,EAA8B;QACnG,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAC1B,SAAS,KAAK,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,SAAS,EAAE,CACtF,CAAC;QACF,OAAO,GAAG,CAAC,OAAO,IAAI,EAAE,CAAC;IAAA,CACzB;IAED;;;;;;;;OAQG;IACH,KAAK,CAAC,MAAM,CAAC,KAAa,EAAE,QAAkC,EAAE,CAAS,EAAoC;QAC5G,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QAClE,OAAO,GAAG,CAAC,QAAQ,IAAI,EAAE,CAAC;IAAA,CAC1B;IAED;;;;OAIG;IACH,KAAK,CAAC,IAAI,CAAC,KAA0C,EAAgC;QACpF,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;QACnD,OAAO,EAAE,QAAQ,EAAE,GAAG,CAAC,cAAc,IAAI,CAAC,EAAE,OAAO,EAAE,GAAG,CAAC,aAAa,IAAI,CAAC,EAAE,CAAC;IAAA,CAC9E;IAED,qEAAqE;IACrE,KAAK,CAAC,IAAI,GAAiC;QAC1C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;QAC5C,OAAO,EAAE,OAAO,EAAE,GAAG,CAAC,QAAQ,IAAI,EAAE,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,CAAC,KAAK,IAAI,CAAC,EAAE,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,CAAC;IAAA,CACrG;IAED,yDAAyD;IACzD,KAAK,CAAC,MAAM,CAAC,EAAU,EAAoB;QAC1C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;QAClD,OAAO,GAAG,CAAC,OAAO,KAAK,IAAI,CAAC;IAAA,CAC5B;IAED,uDAAuD;IACvD,KAAK,CAAC,OAAO,GAAkB;QAC9B,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,CAAC;IAAA,CACnC;IAED,gDAAgD;IAChD,KAAK,CAAC,IAAI,GAAkB;QAC3B,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;IAAA,CAChC;IAED,+DAA+D;IAC/D,KAAK,CAAC,KAAK,GAAkB;QAC5B,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO;QACxB,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC;QACtB,MAAM,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;IAAA,CAC5E;CACD","sourcesContent":["/**\n * Client for the `embsearch` stdio daemon (vendored from\n * github.com/kolisachint/embeddingsearchtools ts/client.ts).\n *\n * Spawns `embsearch serve` once and talks newline-delimited JSON over its\n * stdin/stdout. The process stays alive so the model and index load a single\n * time; every query after startup is hot.\n */\n\nimport { type ChildProcessWithoutNullStreams, spawn } from \"child_process\";\n\nexport interface EmbSearchResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchDaemonInfo {\n\tmodelId: string;\n\tdim: number;\n\tcount: number;\n\t/**\n\t * Whether this daemon can actually serve `rerank`, or `undefined` from a\n\t * daemon too old to say.\n\t *\n\t * Reported separately from the version because the two came apart:\n\t * released binaries carry the `rerank` op but no longer bundle the ~23 MB\n\t * cross-encoder weights, so a version check alone would advertise a\n\t * reranker that fails on first call.\n\t */\n\trerank?: boolean;\n}\n\n/** One candidate sent for cross-encoder scoring. */\nexport interface EmbSearchRerankPassage {\n\tid: string;\n\ttext: string;\n}\n\n/** A cross-encoder relevance logit. Higher is more relevant, but the scale is\n *  unnormalized and comparable only within one call. */\nexport interface EmbSearchRerankResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchBulkResult {\n\tinserted: number;\n\tupdated: number;\n}\n\n/** Which daemon-side retriever answers a query (embsearch >= 0.2.0). */\nexport type DaemonRetriever = \"dense\" | \"lexical\" | \"hybrid\";\n\nexport interface EmbSearchClientOptions {\n\t/** Open/create the store with a BM25 lexical index alongside the vectors. */\n\thybrid?: boolean;\n\t/** Path to the `embsearch` binary. */\n\tbinaryPath: string;\n\t/** Store directory passed as `--path`. */\n\tstorePath: string;\n\t/** Metric for a freshly created store. Default: \"cosine\". */\n\tmetric?: \"cosine\" | \"dot\" | \"euclidean\";\n}\n\ninterface Pending {\n\tresolve: (value: EmbSearchRawResponse) => void;\n\treject: (err: Error) => void;\n}\n\ninterface EmbSearchRawResponse {\n\tok: boolean;\n\terror?: string;\n\tresults?: EmbSearchResult[];\n\tinserted?: boolean;\n\tremoved?: boolean;\n\tcount?: number;\n\tinserted_count?: number;\n\tupdated_count?: number;\n\tmodel_id?: string;\n\tdim?: number;\n\t/** Cross-encoder scores. Separate from `results` because the daemon's\n\t *  logits are not comparable to retrieval scores. */\n\treranked?: EmbSearchRerankResult[];\n\t/** `info` only: whether `rerank` will work. Absent on daemons too old to\n\t *  report it. */\n\trerank?: boolean;\n}\n\nexport class EmbSearchClient {\n\tprivate proc: ChildProcessWithoutNullStreams;\n\tprivate queue: Pending[] = [];\n\tprivate buffer = \"\";\n\tprivate closed = false;\n\tprivate readyPromise: Promise<void>;\n\n\tconstructor(opts: EmbSearchClientOptions) {\n\t\tconst args = [\"serve\", \"--path\", opts.storePath];\n\t\tif (opts.metric) args.push(\"--metric\", opts.metric);\n\t\t// Hybrid-ness is fixed when a store is created: passing --hybrid against\n\t\t// an existing non-hybrid store warns and is ignored daemon-side, so a\n\t\t// hybrid store needs its own directory.\n\t\tif (opts.hybrid) args.push(\"--hybrid\");\n\n\t\tthis.proc = spawn(opts.binaryPath, args, { stdio: [\"pipe\", \"pipe\", \"pipe\"] });\n\n\t\tthis.proc.stdout.setEncoding(\"utf8\");\n\t\tthis.proc.stdout.on(\"data\", (chunk: string) => this.onStdout(chunk));\n\t\t// Swallow the informational stderr banner; errors surface via responses/exit.\n\t\tthis.proc.stderr.resume();\n\n\t\t// Readiness = the daemon answering a ping (probes the actual request loop).\n\t\tthis.readyPromise = this.send({ op: \"ping\" }).then(() => undefined);\n\t\t// Spawn failures reject the pending ping via the exit handler; mark the\n\t\t// promise handled so a failed spawn doesn't raise an unhandled rejection\n\t\t// before ready() is awaited.\n\t\tthis.readyPromise.catch(() => {});\n\n\t\tthis.proc.on(\"error\", (err) => {\n\t\t\tthis.closed = true;\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t\tthis.proc.on(\"exit\", (code) => {\n\t\t\tthis.closed = true;\n\t\t\tconst err = new Error(`embsearch daemon exited (code ${code})`);\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t}\n\n\t/** Resolves once the daemon has loaded the model + index. */\n\tready(): Promise<void> {\n\t\treturn this.readyPromise;\n\t}\n\n\tget isClosed(): boolean {\n\t\treturn this.closed;\n\t}\n\n\tprivate onStdout(chunk: string): void {\n\t\tthis.buffer += chunk;\n\t\t// Each response is one line; dispatch FIFO against the pending queue.\n\t\tfor (let nl = this.buffer.indexOf(\"\\n\"); nl !== -1; nl = this.buffer.indexOf(\"\\n\")) {\n\t\t\tconst line = this.buffer.slice(0, nl).trim();\n\t\t\tthis.buffer = this.buffer.slice(nl + 1);\n\t\t\tif (!line) continue;\n\t\t\tconst pending = this.queue.shift();\n\t\t\tif (!pending) continue;\n\t\t\tlet msg: EmbSearchRawResponse;\n\t\t\ttry {\n\t\t\t\tmsg = JSON.parse(line) as EmbSearchRawResponse;\n\t\t\t} catch {\n\t\t\t\tpending.reject(new Error(`bad response: ${line}`));\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tif (msg.ok) pending.resolve(msg);\n\t\t\telse pending.reject(new Error(msg.error ?? \"unknown error\"));\n\t\t}\n\t}\n\n\tprivate send(req: Record<string, unknown>): Promise<EmbSearchRawResponse> {\n\t\tif (this.closed) return Promise.reject(new Error(\"embsearch client is closed\"));\n\t\treturn new Promise<EmbSearchRawResponse>((resolve, reject) => {\n\t\t\tthis.queue.push({ resolve, reject });\n\t\t\tthis.proc.stdin.write(`${JSON.stringify(req)}\\n`);\n\t\t});\n\t}\n\n\t/**\n\t * Search for the top-`k` matches for `text`.\n\t *\n\t * `retriever` selects which leg answers:\n\t *  - `dense` (default) — vector search, the historical behaviour;\n\t *  - `lexical` — BM25 only, raw scores, no embedding computed;\n\t *  - `hybrid` — both, pre-fused by the daemon's own RRF constant.\n\t *\n\t * `lexical` and `hybrid` need a store created with `--hybrid`. Prefer\n\t * `lexical` over `hybrid` when fusing here: `hybrid` collapses both legs\n\t * into one RRF score, discarding the per-retriever ranks the trace records\n\t * and preventing n-way fusion with the grep leg.\n\t */\n\tasync query(text: string, k = 10, retriever: DaemonRetriever = \"dense\"): Promise<EmbSearchResult[]> {\n\t\tconst res = await this.send(\n\t\t\tretriever === \"dense\" ? { op: \"query\", text, k } : { op: \"query\", text, k, retriever },\n\t\t);\n\t\treturn res.results ?? [];\n\t}\n\n\t/**\n\t * Score `passages` against `query` with the daemon's cross-encoder and\n\t * return the best `k`, best first.\n\t *\n\t * Passages are sent inline rather than referenced by id: the caller has the\n\t * exact spans it intends to show the model, and a cross-encoder scores the\n\t * text it is given, so sending anything else would score the wrong thing.\n\t * Requires an onnx build with reranker weights (embsearch >= 0.3.0).\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tconst res = await this.send({ op: \"rerank\", query, passages, k });\n\t\treturn res.reranked ?? [];\n\t}\n\n\t/**\n\t * Batched insert-or-replace. One embedding inference for the whole batch —\n\t * the fast path for bulk indexing. Keep batches modest (e.g. 32–64) so a\n\t * concurrent query is not stuck behind a huge inference.\n\t */\n\tasync bulk(items: Array<{ id: string; text: string }>): Promise<EmbSearchBulkResult> {\n\t\tconst res = await this.send({ op: \"bulk\", items });\n\t\treturn { inserted: res.inserted_count ?? 0, updated: res.updated_count ?? 0 };\n\t}\n\n\t/** Model id, dimensionality, and live vector count of the daemon. */\n\tasync info(): Promise<EmbSearchDaemonInfo> {\n\t\tconst res = await this.send({ op: \"info\" });\n\t\treturn { modelId: res.model_id ?? \"\", dim: res.dim ?? 0, count: res.count ?? 0, rerank: res.rerank };\n\t}\n\n\t/** Remove a record. Resolves to `true` if it existed. */\n\tasync remove(id: string): Promise<boolean> {\n\t\tconst res = await this.send({ op: \"remove\", id });\n\t\treturn res.removed === true;\n\t}\n\n\t/** Reclaim tombstoned rows left behind by `remove`. */\n\tasync compact(): Promise<void> {\n\t\tawait this.send({ op: \"compact\" });\n\t}\n\n\t/** Persist the index to the store directory. */\n\tasync save(): Promise<void> {\n\t\tawait this.send({ op: \"save\" });\n\t}\n\n\t/** Shut the daemon down, closing stdin so it exits cleanly. */\n\tasync close(): Promise<void> {\n\t\tif (this.closed) return;\n\t\tthis.proc.stdin.end();\n\t\tawait new Promise<void>((resolve) => this.proc.on(\"exit\", () => resolve()));\n\t}\n}\n"]}

package/dist/core/embsearch/embsearch-service.d.ts

@@ -81,6 +81,27 @@ export declare class EmbsearchService {
     private setState;
     /** Whether the running daemon can serve a BM25-only query. */
     supportsLexicalRetriever(): boolean;
+    /**
+     * Repo files whose on-disk content the index does not have — unknown to it,
+     * or changed since it last read them.
+     *
+     * This is the set BM25 is structurally blind to, and the only place the
+     * grep leg still earns its keep once BM25 is available. An agent that edits
+     * a file and immediately searches for what it wrote is asking about exactly
+     * these files; the index cannot answer until the next pass.
+     *
+     * Compares mtime and size only, never hashing: the check runs per query, and
+     * a false positive merely lets grep cover a file BM25 already covers, while
+     * a false negative would lose the edit.
+     *
+     * Deliberately uncached. A cache here caches the *absence* of an edit, which
+     * is the one thing this must never do — an agent writes a file and searches
+     * for it in the same breath. A 1s TTL was tried and cost the live-edit set
+     * 75% to 100% of its score depending on how the timing fell, which is worse
+     * than wrong: it was non-deterministic. One scan is ~25ms over ~1k files and
+     * happens once per search, against retrieval that already costs more.
+     */
+    staleFiles(signal?: AbortSignal): string[];
     /** Whether the running daemon can score with a cross-encoder. */
     supportsCrossEncoder(): boolean;
     /**

package/dist/core/embsearch/embsearch-service.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"embsearch-service.d.ts","sourceRoot":"","sources":["../../../src/core/embsearch/embsearch-service.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAOH,OAAO,EACN,KAAK,eAAe,EAEpB,KAAK,sBAAsB,EAC3B,KAAK,qBAAqB,EAC1B,MAAM,aAAa,CAAC;AA8CrB,MAAM,MAAM,cAAc,GACvB;IAAE,KAAK,EAAE,MAAM,CAAA;CAAE,GACjB;IAAE,KAAK,EAAE,SAAS,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,GACpC;IAAE,KAAK,EAAE,aAAa,CAAC;IAAC,aAAa,EAAE,MAAM,CAAC;IAAC,UAAU,EAAE,MAAM,GAAG,IAAI,CAAA;CAAE,GAC1E;IAAE,KAAK,EAAE,UAAU,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,GAClD;IAAE,KAAK,EAAE,OAAO,CAAC;IAAC,UAAU,EAAE,MAAM,CAAA;CAAE,GACtC;IAAE,KAAK,EAAE,aAAa,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,CAAC;AAE5C,MAAM,WAAW,uBAAuB;IACvC,GAAG,EAAE,MAAM,CAAC;IACZ,gFAAgF;IAChF,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,wDAAwD;IACxD,cAAc,EAAE,MAAM,CAAC;IACvB;;;;OAIG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB;;;OAGG;IACH,WAAW,CAAC,EAAE,OAAO,CAAC;IACtB,wDAAwD;IACxD,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,cAAc,KAAK,IAAI,CAAC;CAC7C;AAED,MAAM,WAAW,WAAW;IAC3B,IAAI,EAAE,MAAM,CAAC;IACb,SAAS,EAAE,MAAM,CAAC;IAClB,OAAO,EAAE,MAAM,CAAC;IAChB,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,gBAAiB,SAAQ,WAAW;IACpD,sFAAoF;IACpF,EAAE,EAAE,MAAM,CAAC;CACX;AAED,qBAAa,gBAAgB;IAC5B,OAAO,CAAC,QAAQ,CAAC,OAAO,CAA0B;IAClD,OAAO,CAAC,MAAM,CAA8B;IAC5C,OAAO,CAAC,IAAI,CAAwB;IACpC,OAAO,CAAC,KAAK,CAAqC;IAClD,OAAO,CAAC,QAAQ,CAAS;IACzB,iEAAiE;IACjE,OAAO,CAAC,gBAAgB,CAAS;IACjC,wEAAwE;IACxE,OAAO,CAAC,YAAY,CAAS;IAE7B,YAAY,OAAO,EAAE,uBAAuB,EAE3C;IAED,QAAQ,IAAI,cAAc,CAEzB;IAED,oFAAoF;IACpF,WAAW,IAAI,OAAO,CAErB;IAED,OAAO,CAAC,QAAQ;IAKhB,8DAA8D;IAC9D,wBAAwB,IAAI,OAAO,CAElC;IAED,iEAAiE;IACjE,oBAAoB,IAAI,OAAO,CAE9B;IAED;;;;;;OAMG;IACG,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,sBAAsB,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,qBAAqB,EAAE,CAAC,CAW3G;IAED,OAAO,CAAC,kBAAkB;YAQZ,aAAa;IAY3B;;;OAGG;IACG,KAAK,CAAC,MAAM,CAAC,EAAE,WAAW,GAAG,OAAO,CAAC,IAAI,CAAC,CAQ/C;YAEa,GAAG;YA2CH,iBAAiB;IAqF/B,OAAO,CAAC,WAAW;IAMnB,4DAA4D;IACtD,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,SAAK,GAAG,OAAO,CAAC,WAAW,EAAE,CAAC,CAE1D;IAED,wEAAwE;IAClE,YAAY,CACjB,KAAK,EAAE,MAAM,EACb,CAAC,SAAK,EACN,IAAI,CAAC,EAAE,MAAM,EACb,SAAS,GAAE,eAAyB,GAClC,OAAO,CAAC,gBAAgB,EAAE,CAAC,CA2B7B;IAED;;;;;OAKG;IACH,kBAAkB,CACjB,GAAG,EAAE,MAAM,EACX,IAAI,EAAE,MAAM,GACV;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAC;QAAC,SAAS,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,MAAM,CAAA;KAAE,GAAG,SAAS,CAU9E;YAEa,WAAW;IAYzB,kEAAkE;IAC5D,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAW7B;CACD;AAWD,wBAAgB,wBAAwB,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,EAAE,gBAAgB,GAAG,IAAI,CAMrF;AAED,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,GAAG,gBAAgB,GAAG,SAAS,CAE7E;AAED,wBAAgB,0BAA0B,CAAC,GAAG,EAAE,MAAM,GAAG,IAAI,CAE5D","sourcesContent":["/**\n * Orchestrates semantic indexing and search for a repository.\n *\n * Lifecycle (all behind --enable-embsearchtools):\n *  1. `start()` — resolve the embsearch binary, scan the repo (ignore-aware),\n *     apply the byte threshold. Under threshold → dormant. Over → spawn the\n *     daemon, verify the backend is not the mock embedder, then index changed\n *     files in the background in small batches, reporting progress.\n *  2. `search()` — top-k semantic query, mapping chunk ids back to\n *     `path:start-end` via the sidecar metadata.\n *  3. `dispose()` — save + close the daemon.\n *\n * Every failure degrades to `unavailable` with a reason; nothing here ever\n * blocks session startup or affects grep/find.\n */\n\nimport { execFileSync } from \"child_process\";\nimport { readFileSync } from \"fs\";\nimport { minimatch } from \"minimatch\";\nimport { ensureTool } from \"../../utils/tools-manager.js\";\nimport { chunkFile } from \"./chunker.js\";\nimport {\n\ttype DaemonRetriever,\n\tEmbSearchClient,\n\ttype EmbSearchRerankPassage,\n\ttype EmbSearchRerankResult,\n} from \"./client.js\";\nimport {\n\temptyIndexMeta,\n\ttype FileMeta,\n\tgetEmbsearchStoreDir,\n\tgetVectorStoreDir,\n\thashContent,\n\ttype IndexMeta,\n\tloadIndexMeta,\n\tsaveIndexMeta,\n} from \"./index-meta.js\";\nimport { type RepoScanFile, scanRepo } from \"./repo-scan.js\";\n\n/** Chunks per bulk request. Small enough that a concurrent query is never\n *  stuck long behind one padded batch inference. */\nconst BULK_BATCH_SIZE = 48;\n/** Yield between batches so background indexing doesn't starve the session. */\nconst BATCH_YIELD_MS = 15;\n/** The Rust mock backend's model id — semantically meaningless, never index with it. */\nconst MOCK_MODEL_ID = \"mock-hash-v1\";\n/**\n * First embsearch release serving `retriever: \"lexical\"`.\n *\n * The guard matters because the daemon does not reject unknown request fields:\n * an older binary silently ignores `retriever` and answers with dense results.\n * Fusing that list a second time as a \"bm25\" leg would double-count it and\n * corrupt the ranking with no error anywhere — so refuse instead of degrading.\n */\nconst MIN_LEXICAL_RETRIEVER_VERSION = [0, 2, 0] as const;\n/** First embsearch release bundling cross-encoder reranker weights. */\nconst MIN_RERANK_VERSION = [0, 3, 0] as const;\n\n/** `embsearch 0.2.0` -> [0, 2, 0]; undefined when it cannot be parsed. */\nfunction parseBinaryVersion(output: string): number[] | undefined {\n\tconst match = output.match(/(\\d+)\\.(\\d+)\\.(\\d+)/);\n\treturn match ? [Number(match[1]), Number(match[2]), Number(match[3])] : undefined;\n}\n\nfunction atLeast(version: readonly number[], minimum: readonly number[]): boolean {\n\tfor (let i = 0; i < minimum.length; i++) {\n\t\tconst part = version[i] ?? 0;\n\t\tif (part !== minimum[i]) return part > minimum[i];\n\t}\n\treturn true;\n}\n\nexport type EmbsearchState =\n\t| { phase: \"idle\" }\n\t| { phase: \"skipped\"; reason: string }\n\t| { phase: \"downloading\"; receivedBytes: number; totalBytes: number | null }\n\t| { phase: \"indexing\"; done: number; total: number }\n\t| { phase: \"ready\"; chunkCount: number }\n\t| { phase: \"unavailable\"; reason: string };\n\nexport interface EmbsearchServiceOptions {\n\tcwd: string;\n\t/** Explicit binary path (settings override). Default: \"embsearch\" from PATH. */\n\tbinaryPath?: string;\n\t/** Minimum indexable bytes before indexing kicks in. */\n\tthresholdBytes: number;\n\t/**\n\t * Override the store location. Only the eval harness sets this, so a\n\t * second index (e.g. a BM25-hybrid store) can exist for the same repo\n\t * without colliding with the primary one.\n\t */\n\tstoreDir?: string;\n\t/**\n\t * Create the store with the daemon's BM25 lexical index. Fixed at store\n\t * creation, so this must pair with a distinct `storeDir`.\n\t */\n\thybridStore?: boolean;\n\t/** Progress callback for UI (footer / stderr lines). */\n\tonProgress?: (state: EmbsearchState) => void;\n}\n\nexport interface SemanticHit {\n\tpath: string;\n\tstartLine: number;\n\tendLine: number;\n\tscore: number;\n}\n\nexport interface SemanticChunkHit extends SemanticHit {\n\t/** Per-build chunk id (`relpath#index`) — the fusion identity for hybrid search. */\n\tid: string;\n}\n\nexport class EmbsearchService {\n\tprivate readonly options: EmbsearchServiceOptions;\n\tprivate client: EmbSearchClient | undefined;\n\tprivate meta: IndexMeta | undefined;\n\tprivate state: EmbsearchState = { phase: \"idle\" };\n\tprivate disposed = false;\n\t/** Whether the resolved binary serves `retriever: \"lexical\"`. */\n\tprivate lexicalRetriever = false;\n\t/** Whether the resolved binary serves the cross-encoder `rerank` op. */\n\tprivate crossEncoder = false;\n\n\tconstructor(options: EmbsearchServiceOptions) {\n\t\tthis.options = options;\n\t}\n\n\tgetState(): EmbsearchState {\n\t\treturn this.state;\n\t}\n\n\t/** Semantic search is usable (index ready, or still building with partial data). */\n\tisAvailable(): boolean {\n\t\treturn this.state.phase === \"ready\" || this.state.phase === \"indexing\";\n\t}\n\n\tprivate setState(state: EmbsearchState): void {\n\t\tthis.state = state;\n\t\tthis.options.onProgress?.(state);\n\t}\n\n\t/** Whether the running daemon can serve a BM25-only query. */\n\tsupportsLexicalRetriever(): boolean {\n\t\treturn this.lexicalRetriever;\n\t}\n\n\t/** Whether the running daemon can score with a cross-encoder. */\n\tsupportsCrossEncoder(): boolean {\n\t\treturn this.crossEncoder;\n\t}\n\n\t/**\n\t * Cross-encoder rerank of caller-supplied passages.\n\t *\n\t * Unlike the retrievers this does not consult the index at all — it scores\n\t * exactly the text passed in, which is why the caller sends its expanded\n\t * windows rather than chunk ids.\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tif (!this.client || this.client.isClosed) {\n\t\t\tthrow new Error(\"semantic index is not available\");\n\t\t}\n\t\tif (!this.crossEncoder) {\n\t\t\tthrow new Error(\n\t\t\t\t`embsearch is too old to rerank (needs >= ${MIN_RERANK_VERSION.join(\".\")}); ` +\n\t\t\t\t\t\"an older daemon has no cross-encoder weights\",\n\t\t\t);\n\t\t}\n\t\treturn await this.client.rerank(query, passages, k);\n\t}\n\n\tprivate probeBinaryVersion(binary: string): number[] | undefined {\n\t\ttry {\n\t\t\treturn parseBinaryVersion(execFileSync(binary, [\"--version\"], { encoding: \"utf-8\", timeout: 10_000 }));\n\t\t} catch {\n\t\t\treturn undefined;\n\t\t}\n\t}\n\n\tprivate async resolveBinary(): Promise<string | undefined> {\n\t\tif (this.options.binaryPath) {\n\t\t\treturn this.options.binaryPath;\n\t\t}\n\t\t// Surface the on-demand binary download through the same progress channel as\n\t\t// indexing, so the first-run fetch renders a progress bar instead of a stall.\n\t\t// A cached binary resolves without ever invoking this callback.\n\t\treturn await ensureTool(\"embsearch\", true, (receivedBytes, totalBytes) => {\n\t\t\tthis.setState({ phase: \"downloading\", receivedBytes, totalBytes });\n\t\t});\n\t}\n\n\t/**\n\t * Scan, threshold-check, and (when needed) index in the background.\n\t * Resolves when indexing completes or the feature settles dormant.\n\t */\n\tasync start(signal?: AbortSignal): Promise<void> {\n\t\ttry {\n\t\t\tawait this.run(signal);\n\t\t} catch (e) {\n\t\t\tconst reason = e instanceof Error ? e.message : String(e);\n\t\t\tthis.setState({ phase: \"unavailable\", reason });\n\t\t\tawait this.closeClient();\n\t\t}\n\t}\n\n\tprivate async run(signal?: AbortSignal): Promise<void> {\n\t\tconst binary = await this.resolveBinary();\n\t\tif (!binary) {\n\t\t\tthis.setState({\n\t\t\t\tphase: \"unavailable\",\n\t\t\t\treason: \"embsearch binary not found (PATH or embsearchBinaryPath setting)\",\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst scan = scanRepo(this.options.cwd, signal);\n\t\tif (scan.totalBytes < this.options.thresholdBytes) {\n\t\t\tthis.setState({\n\t\t\t\tphase: \"skipped\",\n\t\t\t\treason: `repo under threshold (${scan.totalBytes} < ${this.options.thresholdBytes} bytes)`,\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst binaryVersion = this.probeBinaryVersion(binary);\n\t\tthis.lexicalRetriever = binaryVersion !== undefined && atLeast(binaryVersion, MIN_LEXICAL_RETRIEVER_VERSION);\n\t\tthis.crossEncoder = binaryVersion !== undefined && atLeast(binaryVersion, MIN_RERANK_VERSION);\n\n\t\tconst storeDir = this.options.storeDir ?? getEmbsearchStoreDir(this.options.cwd);\n\t\tthis.client = new EmbSearchClient({\n\t\t\tbinaryPath: binary,\n\t\t\tstorePath: getVectorStoreDir(storeDir),\n\t\t\thybrid: this.options.hybridStore,\n\t\t});\n\t\tawait this.client.ready();\n\n\t\tconst info = await this.client.info();\n\t\tif (info.modelId === MOCK_MODEL_ID) {\n\t\t\tthrow new Error(\"embsearch binary uses the mock embedder (not semantic); install an onnx build\");\n\t\t}\n\n\t\t// Missing/stale sidecar (format, chunker, or model changed) → clean rebuild.\n\t\tthis.meta = loadIndexMeta(storeDir, info.modelId) ?? emptyIndexMeta(this.options.cwd, info.modelId);\n\t\tthis.meta.lastUsedMs = Date.now();\n\n\t\tawait this.indexChangedFiles(scan.files, storeDir, signal);\n\t}\n\n\tprivate async indexChangedFiles(files: RepoScanFile[], storeDir: string, signal?: AbortSignal): Promise<void> {\n\t\tconst meta = this.meta!;\n\t\tconst client = this.client!;\n\n\t\t// Diff scan vs sidecar: cheap mtime+size check first, hash only on delta.\n\t\tconst toIndex: Array<{ file: RepoScanFile; content: string; hash: string }> = [];\n\t\tconst seen = new Set<string>();\n\t\tfor (const file of files) {\n\t\t\tseen.add(file.rel);\n\t\t\tconst known = meta.files[file.rel];\n\t\t\tif (known && known.mtimeMs === file.mtimeMs && known.size === file.size) continue;\n\t\t\tlet content: string;\n\t\t\ttry {\n\t\t\t\tcontent = readFileSync(file.abs, \"utf-8\");\n\t\t\t} catch {\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tconst hash = hashContent(content);\n\t\t\tif (known && known.hash === hash) {\n\t\t\t\t// Touched but unchanged — refresh stat info only.\n\t\t\t\tknown.mtimeMs = file.mtimeMs;\n\t\t\t\tknown.size = file.size;\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\ttoIndex.push({ file, content, hash });\n\t\t}\n\t\tconst toRemove = Object.keys(meta.files).filter((rel) => !seen.has(rel));\n\n\t\tif (toIndex.length === 0 && toRemove.length === 0) {\n\t\t\tsaveIndexMeta(storeDir, meta);\n\t\t\tthis.setState({ phase: \"ready\", chunkCount: this.countChunks(meta) });\n\t\t\treturn;\n\t\t}\n\n\t\t// Chunk changed files; count total upserts for exact progress.\n\t\tconst work: Array<{ rel: string; fileMeta: FileMeta; chunks: Array<{ id: string; text: string }> }> = [];\n\t\tlet totalChunks = 0;\n\t\tfor (const { file, content, hash } of toIndex) {\n\t\t\tconst chunks = chunkFile(file.rel, content);\n\t\t\twork.push({\n\t\t\t\trel: file.rel,\n\t\t\t\tfileMeta: {\n\t\t\t\t\tmtimeMs: file.mtimeMs,\n\t\t\t\t\tsize: file.size,\n\t\t\t\t\thash,\n\t\t\t\t\tchunks: chunks.map((c) => [c.startLine, c.endLine]),\n\t\t\t\t},\n\t\t\t\tchunks: chunks.map((c) => ({ id: c.id, text: c.text })),\n\t\t\t});\n\t\t\ttotalChunks += chunks.length;\n\t\t}\n\n\t\tthis.setState({ phase: \"indexing\", done: 0, total: totalChunks });\n\n\t\t// Drop vectors of deleted files and superseded chunk tails.\n\t\tfor (const rel of toRemove) {\n\t\t\tfor (let i = 0; i < meta.files[rel].chunks.length; i++) await client.remove(`${rel}#${i}`);\n\t\t\tdelete meta.files[rel];\n\t\t}\n\n\t\tlet done = 0;\n\t\tfor (const item of work) {\n\t\t\tif (signal?.aborted || this.disposed) return;\n\t\t\tconst oldChunkCount = meta.files[item.rel]?.chunks.length ?? 0;\n\t\t\t// Remove old chunks beyond the new count (upsert covers the rest).\n\t\t\tfor (let i = item.chunks.length; i < oldChunkCount; i++) await client.remove(`${item.rel}#${i}`);\n\n\t\t\tfor (let offset = 0; offset < item.chunks.length; offset += BULK_BATCH_SIZE) {\n\t\t\t\tif (signal?.aborted || this.disposed) return;\n\t\t\t\tconst batch = item.chunks.slice(offset, offset + BULK_BATCH_SIZE);\n\t\t\t\tawait client.bulk(batch);\n\t\t\t\tdone += batch.length;\n\t\t\t\tthis.setState({ phase: \"indexing\", done, total: totalChunks });\n\t\t\t\t// Yield so queries and the event loop stay responsive.\n\t\t\t\tawait new Promise((resolve) => setTimeout(resolve, BATCH_YIELD_MS));\n\t\t\t}\n\t\t\tmeta.files[item.rel] = item.fileMeta;\n\t\t}\n\n\t\tawait client.compact();\n\t\tawait client.save();\n\t\tsaveIndexMeta(storeDir, meta);\n\t\tthis.setState({ phase: \"ready\", chunkCount: this.countChunks(meta) });\n\t}\n\n\tprivate countChunks(meta: IndexMeta): number {\n\t\tlet n = 0;\n\t\tfor (const rel of Object.keys(meta.files)) n += meta.files[rel].chunks.length;\n\t\treturn n;\n\t}\n\n\t/** Top-`k` semantic hits as `path` + line range + score. */\n\tasync search(query: string, k = 10): Promise<SemanticHit[]> {\n\t\treturn await this.searchChunks(query, k);\n\t}\n\n\t/** Top-`k` semantic hits including their chunk ids, for rank fusion. */\n\tasync searchChunks(\n\t\tquery: string,\n\t\tk = 10,\n\t\tglob?: string,\n\t\tretriever: DaemonRetriever = \"dense\",\n\t): Promise<SemanticChunkHit[]> {\n\t\tif (!this.client || this.client.isClosed || !this.meta) {\n\t\t\tthrow new Error(\"semantic index is not available\");\n\t\t}\n\t\tif (retriever === \"lexical\" && !this.lexicalRetriever) {\n\t\t\tthrow new Error(\n\t\t\t\t`embsearch is too old for retriever \"lexical\" (needs >= ${MIN_LEXICAL_RETRIEVER_VERSION.join(\".\")}); ` +\n\t\t\t\t\t\"an older daemon ignores the field and answers with dense results\",\n\t\t\t);\n\t\t}\n\t\tconst results = await this.client.query(query, k, retriever);\n\t\tconst hits: SemanticChunkHit[] = [];\n\t\tconst matchGlob = (rel: string): boolean => {\n\t\t\tif (!glob) return true;\n\t\t\treturn minimatch(rel, glob, { dot: true, matchBase: !glob.includes(\"/\") });\n\t\t};\n\t\tfor (const result of results) {\n\t\t\tconst sep = result.id.lastIndexOf(\"#\");\n\t\t\tif (sep === -1) continue;\n\t\t\tconst rel = result.id.slice(0, sep);\n\t\t\tif (!matchGlob(rel)) continue;\n\t\t\tconst chunkIndex = Number.parseInt(result.id.slice(sep + 1), 10);\n\t\t\tconst range = this.meta.files[rel]?.chunks[chunkIndex];\n\t\t\tif (!range) continue;\n\t\t\thits.push({ id: result.id, path: rel, startLine: range[0], endLine: range[1], score: result.score });\n\t\t}\n\t\treturn hits;\n\t}\n\n\t/**\n\t * Resolve a repo-relative path + line to its enclosing indexed chunk, or\n\t * undefined when the file/line is not covered by the index. Chunks overlap\n\t * by a few lines; the first (lowest-index) containing chunk wins so the\n\t * mapping is deterministic.\n\t */\n\tfindEnclosingChunk(\n\t\trel: string,\n\t\tline: number,\n\t): { id: string; path: string; startLine: number; endLine: number } | undefined {\n\t\tconst file = this.meta?.files[rel];\n\t\tif (!file) return undefined;\n\t\tfor (let i = 0; i < file.chunks.length; i++) {\n\t\t\tconst [startLine, endLine] = file.chunks[i];\n\t\t\tif (line >= startLine && line <= endLine) {\n\t\t\t\treturn { id: `${rel}#${i}`, path: rel, startLine, endLine };\n\t\t\t}\n\t\t}\n\t\treturn undefined;\n\t}\n\n\tprivate async closeClient(): Promise<void> {\n\t\tconst client = this.client;\n\t\tthis.client = undefined;\n\t\tif (client && !client.isClosed) {\n\t\t\ttry {\n\t\t\t\tawait client.close();\n\t\t\t} catch {\n\t\t\t\t// already dead\n\t\t\t}\n\t\t}\n\t}\n\n\t/** Persist state and shut the daemon down. Safe to call twice. */\n\tasync dispose(): Promise<void> {\n\t\tif (this.disposed) return;\n\t\tthis.disposed = true;\n\t\tif (this.client && !this.client.isClosed) {\n\t\t\ttry {\n\t\t\t\tawait this.client.save();\n\t\t\t} catch {\n\t\t\t\t// daemon may have exited; nothing to save\n\t\t\t}\n\t\t}\n\t\tawait this.closeClient();\n\t}\n}\n\n// --- Per-cwd service registry ---\n//\n// The search tool is constructed by the generic tool factory table and\n// only receives `cwd`; the service is created later during session init (flag\n// gated). This registry connects the two without threading a service instance\n// through every layer between main.ts and the tool factories.\n\nconst services = new Map<string, EmbsearchService>();\n\nexport function registerEmbsearchService(cwd: string, service: EmbsearchService): void {\n\tconst old = services.get(cwd);\n\tif (old && old !== service) {\n\t\told.dispose().catch(() => {});\n\t}\n\tservices.set(cwd, service);\n}\n\nexport function getEmbsearchService(cwd: string): EmbsearchService | undefined {\n\treturn services.get(cwd);\n}\n\nexport function unregisterEmbsearchService(cwd: string): void {\n\tservices.delete(cwd);\n}\n"]}
+{"version":3,"file":"embsearch-service.d.ts","sourceRoot":"","sources":["../../../src/core/embsearch/embsearch-service.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAOH,OAAO,EACN,KAAK,eAAe,EAEpB,KAAK,sBAAsB,EAC3B,KAAK,qBAAqB,EAC1B,MAAM,aAAa,CAAC;AAuDrB,MAAM,MAAM,cAAc,GACvB;IAAE,KAAK,EAAE,MAAM,CAAA;CAAE,GACjB;IAAE,KAAK,EAAE,SAAS,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,GACpC;IAAE,KAAK,EAAE,aAAa,CAAC;IAAC,aAAa,EAAE,MAAM,CAAC;IAAC,UAAU,EAAE,MAAM,GAAG,IAAI,CAAA;CAAE,GAC1E;IAAE,KAAK,EAAE,UAAU,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,GAClD;IAAE,KAAK,EAAE,OAAO,CAAC;IAAC,UAAU,EAAE,MAAM,CAAA;CAAE,GACtC;IAAE,KAAK,EAAE,aAAa,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,CAAC;AAE5C,MAAM,WAAW,uBAAuB;IACvC,GAAG,EAAE,MAAM,CAAC;IACZ,gFAAgF;IAChF,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,wDAAwD;IACxD,cAAc,EAAE,MAAM,CAAC;IACvB;;;;OAIG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB;;;OAGG;IACH,WAAW,CAAC,EAAE,OAAO,CAAC;IACtB,wDAAwD;IACxD,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,cAAc,KAAK,IAAI,CAAC;CAC7C;AAED,MAAM,WAAW,WAAW;IAC3B,IAAI,EAAE,MAAM,CAAC;IACb,SAAS,EAAE,MAAM,CAAC;IAClB,OAAO,EAAE,MAAM,CAAC;IAChB,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,gBAAiB,SAAQ,WAAW;IACpD,sFAAoF;IACpF,EAAE,EAAE,MAAM,CAAC;CACX;AAED,qBAAa,gBAAgB;IAC5B,OAAO,CAAC,QAAQ,CAAC,OAAO,CAA0B;IAClD,OAAO,CAAC,MAAM,CAA8B;IAC5C,OAAO,CAAC,IAAI,CAAwB;IACpC,OAAO,CAAC,KAAK,CAAqC;IAClD,OAAO,CAAC,QAAQ,CAAS;IACzB,iEAAiE;IACjE,OAAO,CAAC,gBAAgB,CAAS;IACjC,wEAAwE;IACxE,OAAO,CAAC,YAAY,CAAS;IAE7B,YAAY,OAAO,EAAE,uBAAuB,EAE3C;IAED,QAAQ,IAAI,cAAc,CAEzB;IAED,oFAAoF;IACpF,WAAW,IAAI,OAAO,CAErB;IAED,OAAO,CAAC,QAAQ;IAKhB,8DAA8D;IAC9D,wBAAwB,IAAI,OAAO,CAElC;IAED;;;;;;;;;;;;;;;;;;;OAmBG;IACH,UAAU,CAAC,MAAM,CAAC,EAAE,WAAW,GAAG,MAAM,EAAE,CAezC;IAED,iEAAiE;IACjE,oBAAoB,IAAI,OAAO,CAE9B;IAED;;;;;;OAMG;IACG,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,sBAAsB,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,qBAAqB,EAAE,CAAC,CAY3G;IAED,OAAO,CAAC,kBAAkB;YAQZ,aAAa;IAY3B;;;OAGG;IACG,KAAK,CAAC,MAAM,CAAC,EAAE,WAAW,GAAG,OAAO,CAAC,IAAI,CAAC,CAQ/C;YAEa,GAAG;YAgDH,iBAAiB;IAqF/B,OAAO,CAAC,WAAW;IAMnB,4DAA4D;IACtD,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,SAAK,GAAG,OAAO,CAAC,WAAW,EAAE,CAAC,CAE1D;IAED,wEAAwE;IAClE,YAAY,CACjB,KAAK,EAAE,MAAM,EACb,CAAC,SAAK,EACN,IAAI,CAAC,EAAE,MAAM,EACb,SAAS,GAAE,eAAyB,GAClC,OAAO,CAAC,gBAAgB,EAAE,CAAC,CA2B7B;IAED;;;;;OAKG;IACH,kBAAkB,CACjB,GAAG,EAAE,MAAM,EACX,IAAI,EAAE,MAAM,GACV;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAC;QAAC,SAAS,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,MAAM,CAAA;KAAE,GAAG,SAAS,CAU9E;YAEa,WAAW;IAYzB,kEAAkE;IAC5D,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAW7B;CACD;AAWD,wBAAgB,wBAAwB,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,EAAE,gBAAgB,GAAG,IAAI,CAMrF;AAED,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,GAAG,gBAAgB,GAAG,SAAS,CAE7E;AAED,wBAAgB,0BAA0B,CAAC,GAAG,EAAE,MAAM,GAAG,IAAI,CAE5D","sourcesContent":["/**\n * Orchestrates semantic indexing and search for a repository.\n *\n * Lifecycle (all behind --enable-embsearchtools):\n *  1. `start()` — resolve the embsearch binary, scan the repo (ignore-aware),\n *     apply the byte threshold. Under threshold → dormant. Over → spawn the\n *     daemon, verify the backend is not the mock embedder, then index changed\n *     files in the background in small batches, reporting progress.\n *  2. `search()` — top-k semantic query, mapping chunk ids back to\n *     `path:start-end` via the sidecar metadata.\n *  3. `dispose()` — save + close the daemon.\n *\n * Every failure degrades to `unavailable` with a reason; nothing here ever\n * blocks session startup or affects grep/find.\n */\n\nimport { execFileSync } from \"child_process\";\nimport { readFileSync } from \"fs\";\nimport { minimatch } from \"minimatch\";\nimport { ensureTool } from \"../../utils/tools-manager.js\";\nimport { chunkFile } from \"./chunker.js\";\nimport {\n\ttype DaemonRetriever,\n\tEmbSearchClient,\n\ttype EmbSearchRerankPassage,\n\ttype EmbSearchRerankResult,\n} from \"./client.js\";\nimport {\n\temptyIndexMeta,\n\ttype FileMeta,\n\tgetEmbsearchStoreDir,\n\tgetVectorStoreDir,\n\thashContent,\n\ttype IndexMeta,\n\tloadIndexMeta,\n\tsaveIndexMeta,\n} from \"./index-meta.js\";\nimport { type RepoScanFile, scanRepo } from \"./repo-scan.js\";\n\n/** Chunks per bulk request. Small enough that a concurrent query is never\n *  stuck long behind one padded batch inference. */\nconst BULK_BATCH_SIZE = 48;\n/** Yield between batches so background indexing doesn't starve the session. */\nconst BATCH_YIELD_MS = 15;\n/** The Rust mock backend's model id — semantically meaningless, never index with it. */\nconst MOCK_MODEL_ID = \"mock-hash-v1\";\n/**\n * First embsearch release serving `retriever: \"lexical\"`.\n *\n * The guard matters because the daemon does not reject unknown request fields:\n * an older binary silently ignores `retriever` and answers with dense results.\n * Fusing that list a second time as a \"bm25\" leg would double-count it and\n * corrupt the ranking with no error anywhere — so refuse instead of degrading.\n */\nconst MIN_LEXICAL_RETRIEVER_VERSION = [0, 2, 0] as const;\n/**\n * First embsearch release serving the `rerank` op.\n *\n * Necessary but no longer sufficient. Releases from 0.3.1 carry the op without\n * the ~23 MB cross-encoder weights — they measured worse than the\n * deterministic reranker on five of six query classes, so they are no longer\n * bundled — and such a daemon answers `rerank` with an error. The version is\n * therefore only the floor for *asking*; `info.rerank` is the answer, and\n * {@link EmbsearchService.supportsCrossEncoder} needs both.\n */\nconst MIN_RERANK_VERSION = [0, 3, 0] as const;\n\n/** `embsearch 0.2.0` -> [0, 2, 0]; undefined when it cannot be parsed. */\nfunction parseBinaryVersion(output: string): number[] | undefined {\n\tconst match = output.match(/(\\d+)\\.(\\d+)\\.(\\d+)/);\n\treturn match ? [Number(match[1]), Number(match[2]), Number(match[3])] : undefined;\n}\n\nfunction atLeast(version: readonly number[], minimum: readonly number[]): boolean {\n\tfor (let i = 0; i < minimum.length; i++) {\n\t\tconst part = version[i] ?? 0;\n\t\tif (part !== minimum[i]) return part > minimum[i];\n\t}\n\treturn true;\n}\n\nexport type EmbsearchState =\n\t| { phase: \"idle\" }\n\t| { phase: \"skipped\"; reason: string }\n\t| { phase: \"downloading\"; receivedBytes: number; totalBytes: number | null }\n\t| { phase: \"indexing\"; done: number; total: number }\n\t| { phase: \"ready\"; chunkCount: number }\n\t| { phase: \"unavailable\"; reason: string };\n\nexport interface EmbsearchServiceOptions {\n\tcwd: string;\n\t/** Explicit binary path (settings override). Default: \"embsearch\" from PATH. */\n\tbinaryPath?: string;\n\t/** Minimum indexable bytes before indexing kicks in. */\n\tthresholdBytes: number;\n\t/**\n\t * Override the store location. Only the eval harness sets this, so a\n\t * second index (e.g. a BM25-hybrid store) can exist for the same repo\n\t * without colliding with the primary one.\n\t */\n\tstoreDir?: string;\n\t/**\n\t * Create the store with the daemon's BM25 lexical index. Fixed at store\n\t * creation, so this must pair with a distinct `storeDir`.\n\t */\n\thybridStore?: boolean;\n\t/** Progress callback for UI (footer / stderr lines). */\n\tonProgress?: (state: EmbsearchState) => void;\n}\n\nexport interface SemanticHit {\n\tpath: string;\n\tstartLine: number;\n\tendLine: number;\n\tscore: number;\n}\n\nexport interface SemanticChunkHit extends SemanticHit {\n\t/** Per-build chunk id (`relpath#index`) — the fusion identity for hybrid search. */\n\tid: string;\n}\n\nexport class EmbsearchService {\n\tprivate readonly options: EmbsearchServiceOptions;\n\tprivate client: EmbSearchClient | undefined;\n\tprivate meta: IndexMeta | undefined;\n\tprivate state: EmbsearchState = { phase: \"idle\" };\n\tprivate disposed = false;\n\t/** Whether the resolved binary serves `retriever: \"lexical\"`. */\n\tprivate lexicalRetriever = false;\n\t/** Whether the resolved binary serves the cross-encoder `rerank` op. */\n\tprivate crossEncoder = false;\n\n\tconstructor(options: EmbsearchServiceOptions) {\n\t\tthis.options = options;\n\t}\n\n\tgetState(): EmbsearchState {\n\t\treturn this.state;\n\t}\n\n\t/** Semantic search is usable (index ready, or still building with partial data). */\n\tisAvailable(): boolean {\n\t\treturn this.state.phase === \"ready\" || this.state.phase === \"indexing\";\n\t}\n\n\tprivate setState(state: EmbsearchState): void {\n\t\tthis.state = state;\n\t\tthis.options.onProgress?.(state);\n\t}\n\n\t/** Whether the running daemon can serve a BM25-only query. */\n\tsupportsLexicalRetriever(): boolean {\n\t\treturn this.lexicalRetriever;\n\t}\n\n\t/**\n\t * Repo files whose on-disk content the index does not have — unknown to it,\n\t * or changed since it last read them.\n\t *\n\t * This is the set BM25 is structurally blind to, and the only place the\n\t * grep leg still earns its keep once BM25 is available. An agent that edits\n\t * a file and immediately searches for what it wrote is asking about exactly\n\t * these files; the index cannot answer until the next pass.\n\t *\n\t * Compares mtime and size only, never hashing: the check runs per query, and\n\t * a false positive merely lets grep cover a file BM25 already covers, while\n\t * a false negative would lose the edit.\n\t *\n\t * Deliberately uncached. A cache here caches the *absence* of an edit, which\n\t * is the one thing this must never do — an agent writes a file and searches\n\t * for it in the same breath. A 1s TTL was tried and cost the live-edit set\n\t * 75% to 100% of its score depending on how the timing fell, which is worse\n\t * than wrong: it was non-deterministic. One scan is ~25ms over ~1k files and\n\t * happens once per search, against retrieval that already costs more.\n\t */\n\tstaleFiles(signal?: AbortSignal): string[] {\n\t\tif (!this.meta) return [];\n\t\tconst meta = this.meta;\n\t\tconst files: string[] = [];\n\t\ttry {\n\t\t\tfor (const file of scanRepo(this.options.cwd, signal).files) {\n\t\t\t\tconst known = meta.files[file.rel];\n\t\t\t\tif (!known || known.mtimeMs !== file.mtimeMs || known.size !== file.size) files.push(file.rel);\n\t\t\t}\n\t\t} catch {\n\t\t\t// A failed scan must not silently narrow the grep leg to nothing;\n\t\t\t// report no staleness and let the indexed legs answer.\n\t\t\treturn [];\n\t\t}\n\t\treturn files;\n\t}\n\n\t/** Whether the running daemon can score with a cross-encoder. */\n\tsupportsCrossEncoder(): boolean {\n\t\treturn this.crossEncoder;\n\t}\n\n\t/**\n\t * Cross-encoder rerank of caller-supplied passages.\n\t *\n\t * Unlike the retrievers this does not consult the index at all — it scores\n\t * exactly the text passed in, which is why the caller sends its expanded\n\t * windows rather than chunk ids.\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tif (!this.client || this.client.isClosed) {\n\t\t\tthrow new Error(\"semantic index is not available\");\n\t\t}\n\t\tif (!this.crossEncoder) {\n\t\t\tthrow new Error(\n\t\t\t\t`this embsearch daemon cannot rerank (needs >= ${MIN_RERANK_VERSION.join(\".\")} reporting ` +\n\t\t\t\t\t\"`rerank: true`); released binaries ship without cross-encoder weights — start the daemon \" +\n\t\t\t\t\t\"with --reranker-model <dir>\",\n\t\t\t);\n\t\t}\n\t\treturn await this.client.rerank(query, passages, k);\n\t}\n\n\tprivate probeBinaryVersion(binary: string): number[] | undefined {\n\t\ttry {\n\t\t\treturn parseBinaryVersion(execFileSync(binary, [\"--version\"], { encoding: \"utf-8\", timeout: 10_000 }));\n\t\t} catch {\n\t\t\treturn undefined;\n\t\t}\n\t}\n\n\tprivate async resolveBinary(): Promise<string | undefined> {\n\t\tif (this.options.binaryPath) {\n\t\t\treturn this.options.binaryPath;\n\t\t}\n\t\t// Surface the on-demand binary download through the same progress channel as\n\t\t// indexing, so the first-run fetch renders a progress bar instead of a stall.\n\t\t// A cached binary resolves without ever invoking this callback.\n\t\treturn await ensureTool(\"embsearch\", true, (receivedBytes, totalBytes) => {\n\t\t\tthis.setState({ phase: \"downloading\", receivedBytes, totalBytes });\n\t\t});\n\t}\n\n\t/**\n\t * Scan, threshold-check, and (when needed) index in the background.\n\t * Resolves when indexing completes or the feature settles dormant.\n\t */\n\tasync start(signal?: AbortSignal): Promise<void> {\n\t\ttry {\n\t\t\tawait this.run(signal);\n\t\t} catch (e) {\n\t\t\tconst reason = e instanceof Error ? e.message : String(e);\n\t\t\tthis.setState({ phase: \"unavailable\", reason });\n\t\t\tawait this.closeClient();\n\t\t}\n\t}\n\n\tprivate async run(signal?: AbortSignal): Promise<void> {\n\t\tconst binary = await this.resolveBinary();\n\t\tif (!binary) {\n\t\t\tthis.setState({\n\t\t\t\tphase: \"unavailable\",\n\t\t\t\treason: \"embsearch binary not found (PATH or embsearchBinaryPath setting)\",\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst scan = scanRepo(this.options.cwd, signal);\n\t\tif (scan.totalBytes < this.options.thresholdBytes) {\n\t\t\tthis.setState({\n\t\t\t\tphase: \"skipped\",\n\t\t\t\treason: `repo under threshold (${scan.totalBytes} < ${this.options.thresholdBytes} bytes)`,\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst binaryVersion = this.probeBinaryVersion(binary);\n\t\tthis.lexicalRetriever = binaryVersion !== undefined && atLeast(binaryVersion, MIN_LEXICAL_RETRIEVER_VERSION);\n\t\t// Provisional: the version says the daemon understands `rerank`. Whether\n\t\t// it can serve one is answered by `info` below, once the client is up.\n\t\tthis.crossEncoder = binaryVersion !== undefined && atLeast(binaryVersion, MIN_RERANK_VERSION);\n\n\t\tconst storeDir = this.options.storeDir ?? getEmbsearchStoreDir(this.options.cwd);\n\t\tthis.client = new EmbSearchClient({\n\t\t\tbinaryPath: binary,\n\t\t\tstorePath: getVectorStoreDir(storeDir),\n\t\t\thybrid: this.options.hybridStore,\n\t\t});\n\t\tawait this.client.ready();\n\n\t\tconst info = await this.client.info();\n\t\t// A daemon old enough to omit the field is left on the version verdict —\n\t\t// back then the weights were bundled, so version did imply capability.\n\t\tif (info.rerank === false) this.crossEncoder = false;\n\t\tif (info.modelId === MOCK_MODEL_ID) {\n\t\t\tthrow new Error(\"embsearch binary uses the mock embedder (not semantic); install an onnx build\");\n\t\t}\n\n\t\t// Missing/stale sidecar (format, chunker, or model changed) → clean rebuild.\n\t\tthis.meta = loadIndexMeta(storeDir, info.modelId) ?? emptyIndexMeta(this.options.cwd, info.modelId);\n\t\tthis.meta.lastUsedMs = Date.now();\n\n\t\tawait this.indexChangedFiles(scan.files, storeDir, signal);\n\t}\n\n\tprivate async indexChangedFiles(files: RepoScanFile[], storeDir: string, signal?: AbortSignal): Promise<void> {\n\t\tconst meta = this.meta!;\n\t\tconst client = this.client!;\n\n\t\t// Diff scan vs sidecar: cheap mtime+size check first, hash only on delta.\n\t\tconst toIndex: Array<{ file: RepoScanFile; content: string; hash: string }> = [];\n\t\tconst seen = new Set<string>();\n\t\tfor (const file of files) {\n\t\t\tseen.add(file.rel);\n\t\t\tconst known = meta.files[file.rel];\n\t\t\tif (known && known.mtimeMs === file.mtimeMs && known.size === file.size) continue;\n\t\t\tlet content: string;\n\t\t\ttry {\n\t\t\t\tcontent = readFileSync(file.abs, \"utf-8\");\n\t\t\t} catch {\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tconst hash = hashContent(content);\n\t\t\tif (known && known.hash === hash) {\n\t\t\t\t// Touched but unchanged — refresh stat info only.\n\t\t\t\tknown.mtimeMs = file.mtimeMs;\n\t\t\t\tknown.size = file.size;\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\ttoIndex.push({ file, content, hash });\n\t\t}\n\t\tconst toRemove = Object.keys(meta.files).filter((rel) => !seen.has(rel));\n\n\t\tif (toIndex.length === 0 && toRemove.length === 0) {\n\t\t\tsaveIndexMeta(storeDir, meta);\n\t\t\tthis.setState({ phase: \"ready\", chunkCount: this.countChunks(meta) });\n\t\t\treturn;\n\t\t}\n\n\t\t// Chunk changed files; count total upserts for exact progress.\n\t\tconst work: Array<{ rel: string; fileMeta: FileMeta; chunks: Array<{ id: string; text: string }> }> = [];\n\t\tlet totalChunks = 0;\n\t\tfor (const { file, content, hash } of toIndex) {\n\t\t\tconst chunks = chunkFile(file.rel, content);\n\t\t\twork.push({\n\t\t\t\trel: file.rel,\n\t\t\t\tfileMeta: {\n\t\t\t\t\tmtimeMs: file.mtimeMs,\n\t\t\t\t\tsize: file.size,\n\t\t\t\t\thash,\n\t\t\t\t\tchunks: chunks.map((c) => [c.startLine, c.endLine]),\n\t\t\t\t},\n\t\t\t\tchunks: chunks.map((c) => ({ id: c.id, text: c.text })),\n\t\t\t});\n\t\t\ttotalChunks += chunks.length;\n\t\t}\n\n\t\tthis.setState({ phase: \"indexing\", done: 0, total: totalChunks });\n\n\t\t// Drop vectors of deleted files and superseded chunk tails.\n\t\tfor (const rel of toRemove) {\n\t\t\tfor (let i = 0; i < meta.files[rel].chunks.length; i++) await client.remove(`${rel}#${i}`);\n\t\t\tdelete meta.files[rel];\n\t\t}\n\n\t\tlet done = 0;\n\t\tfor (const item of work) {\n\t\t\tif (signal?.aborted || this.disposed) return;\n\t\t\tconst oldChunkCount = meta.files[item.rel]?.chunks.length ?? 0;\n\t\t\t// Remove old chunks beyond the new count (upsert covers the rest).\n\t\t\tfor (let i = item.chunks.length; i < oldChunkCount; i++) await client.remove(`${item.rel}#${i}`);\n\n\t\t\tfor (let offset = 0; offset < item.chunks.length; offset += BULK_BATCH_SIZE) {\n\t\t\t\tif (signal?.aborted || this.disposed) return;\n\t\t\t\tconst batch = item.chunks.slice(offset, offset + BULK_BATCH_SIZE);\n\t\t\t\tawait client.bulk(batch);\n\t\t\t\tdone += batch.length;\n\t\t\t\tthis.setState({ phase: \"indexing\", done, total: totalChunks });\n\t\t\t\t// Yield so queries and the event loop stay responsive.\n\t\t\t\tawait new Promise((resolve) => setTimeout(resolve, BATCH_YIELD_MS));\n\t\t\t}\n\t\t\tmeta.files[item.rel] = item.fileMeta;\n\t\t}\n\n\t\tawait client.compact();\n\t\tawait client.save();\n\t\tsaveIndexMeta(storeDir, meta);\n\t\tthis.setState({ phase: \"ready\", chunkCount: this.countChunks(meta) });\n\t}\n\n\tprivate countChunks(meta: IndexMeta): number {\n\t\tlet n = 0;\n\t\tfor (const rel of Object.keys(meta.files)) n += meta.files[rel].chunks.length;\n\t\treturn n;\n\t}\n\n\t/** Top-`k` semantic hits as `path` + line range + score. */\n\tasync search(query: string, k = 10): Promise<SemanticHit[]> {\n\t\treturn await this.searchChunks(query, k);\n\t}\n\n\t/** Top-`k` semantic hits including their chunk ids, for rank fusion. */\n\tasync searchChunks(\n\t\tquery: string,\n\t\tk = 10,\n\t\tglob?: string,\n\t\tretriever: DaemonRetriever = \"dense\",\n\t): Promise<SemanticChunkHit[]> {\n\t\tif (!this.client || this.client.isClosed || !this.meta) {\n\t\t\tthrow new Error(\"semantic index is not available\");\n\t\t}\n\t\tif (retriever === \"lexical\" && !this.lexicalRetriever) {\n\t\t\tthrow new Error(\n\t\t\t\t`embsearch is too old for retriever \"lexical\" (needs >= ${MIN_LEXICAL_RETRIEVER_VERSION.join(\".\")}); ` +\n\t\t\t\t\t\"an older daemon ignores the field and answers with dense results\",\n\t\t\t);\n\t\t}\n\t\tconst results = await this.client.query(query, k, retriever);\n\t\tconst hits: SemanticChunkHit[] = [];\n\t\tconst matchGlob = (rel: string): boolean => {\n\t\t\tif (!glob) return true;\n\t\t\treturn minimatch(rel, glob, { dot: true, matchBase: !glob.includes(\"/\") });\n\t\t};\n\t\tfor (const result of results) {\n\t\t\tconst sep = result.id.lastIndexOf(\"#\");\n\t\t\tif (sep === -1) continue;\n\t\t\tconst rel = result.id.slice(0, sep);\n\t\t\tif (!matchGlob(rel)) continue;\n\t\t\tconst chunkIndex = Number.parseInt(result.id.slice(sep + 1), 10);\n\t\t\tconst range = this.meta.files[rel]?.chunks[chunkIndex];\n\t\t\tif (!range) continue;\n\t\t\thits.push({ id: result.id, path: rel, startLine: range[0], endLine: range[1], score: result.score });\n\t\t}\n\t\treturn hits;\n\t}\n\n\t/**\n\t * Resolve a repo-relative path + line to its enclosing indexed chunk, or\n\t * undefined when the file/line is not covered by the index. Chunks overlap\n\t * by a few lines; the first (lowest-index) containing chunk wins so the\n\t * mapping is deterministic.\n\t */\n\tfindEnclosingChunk(\n\t\trel: string,\n\t\tline: number,\n\t): { id: string; path: string; startLine: number; endLine: number } | undefined {\n\t\tconst file = this.meta?.files[rel];\n\t\tif (!file) return undefined;\n\t\tfor (let i = 0; i < file.chunks.length; i++) {\n\t\t\tconst [startLine, endLine] = file.chunks[i];\n\t\t\tif (line >= startLine && line <= endLine) {\n\t\t\t\treturn { id: `${rel}#${i}`, path: rel, startLine, endLine };\n\t\t\t}\n\t\t}\n\t\treturn undefined;\n\t}\n\n\tprivate async closeClient(): Promise<void> {\n\t\tconst client = this.client;\n\t\tthis.client = undefined;\n\t\tif (client && !client.isClosed) {\n\t\t\ttry {\n\t\t\t\tawait client.close();\n\t\t\t} catch {\n\t\t\t\t// already dead\n\t\t\t}\n\t\t}\n\t}\n\n\t/** Persist state and shut the daemon down. Safe to call twice. */\n\tasync dispose(): Promise<void> {\n\t\tif (this.disposed) return;\n\t\tthis.disposed = true;\n\t\tif (this.client && !this.client.isClosed) {\n\t\t\ttry {\n\t\t\t\tawait this.client.save();\n\t\t\t} catch {\n\t\t\t\t// daemon may have exited; nothing to save\n\t\t\t}\n\t\t}\n\t\tawait this.closeClient();\n\t}\n}\n\n// --- Per-cwd service registry ---\n//\n// The search tool is constructed by the generic tool factory table and\n// only receives `cwd`; the service is created later during session init (flag\n// gated). This registry connects the two without threading a service instance\n// through every layer between main.ts and the tool factories.\n\nconst services = new Map<string, EmbsearchService>();\n\nexport function registerEmbsearchService(cwd: string, service: EmbsearchService): void {\n\tconst old = services.get(cwd);\n\tif (old && old !== service) {\n\t\told.dispose().catch(() => {});\n\t}\n\tservices.set(cwd, service);\n}\n\nexport function getEmbsearchService(cwd: string): EmbsearchService | undefined {\n\treturn services.get(cwd);\n}\n\nexport function unregisterEmbsearchService(cwd: string): void {\n\tservices.delete(cwd);\n}\n"]}

package/dist/core/embsearch/embsearch-service.js

@@ -37,7 +37,16 @@ const MOCK_MODEL_ID = "mock-hash-v1";
  * corrupt the ranking with no error anywhere — so refuse instead of degrading.
  */
 const MIN_LEXICAL_RETRIEVER_VERSION = [0, 2, 0];
-/** First embsearch release bundling cross-encoder reranker weights. */
+/**
+ * First embsearch release serving the `rerank` op.
+ *
+ * Necessary but no longer sufficient. Releases from 0.3.1 carry the op without
+ * the ~23 MB cross-encoder weights — they measured worse than the
+ * deterministic reranker on five of six query classes, so they are no longer
+ * bundled — and such a daemon answers `rerank` with an error. The version is
+ * therefore only the floor for *asking*; `info.rerank` is the answer, and
+ * {@link EmbsearchService.supportsCrossEncoder} needs both.
+ */
 const MIN_RERANK_VERSION = [0, 3, 0];
 /** `embsearch 0.2.0` -> [0, 2, 0]; undefined when it cannot be parsed. */
 function parseBinaryVersion(output) {
@@ -80,6 +89,45 @@ export class EmbsearchService {
     supportsLexicalRetriever() {
         return this.lexicalRetriever;
     }
+    /**
+     * Repo files whose on-disk content the index does not have — unknown to it,
+     * or changed since it last read them.
+     *
+     * This is the set BM25 is structurally blind to, and the only place the
+     * grep leg still earns its keep once BM25 is available. An agent that edits
+     * a file and immediately searches for what it wrote is asking about exactly
+     * these files; the index cannot answer until the next pass.
+     *
+     * Compares mtime and size only, never hashing: the check runs per query, and
+     * a false positive merely lets grep cover a file BM25 already covers, while
+     * a false negative would lose the edit.
+     *
+     * Deliberately uncached. A cache here caches the *absence* of an edit, which
+     * is the one thing this must never do — an agent writes a file and searches
+     * for it in the same breath. A 1s TTL was tried and cost the live-edit set
+     * 75% to 100% of its score depending on how the timing fell, which is worse
+     * than wrong: it was non-deterministic. One scan is ~25ms over ~1k files and
+     * happens once per search, against retrieval that already costs more.
+     */
+    staleFiles(signal) {
+        if (!this.meta)
+            return [];
+        const meta = this.meta;
+        const files = [];
+        try {
+            for (const file of scanRepo(this.options.cwd, signal).files) {
+                const known = meta.files[file.rel];
+                if (!known || known.mtimeMs !== file.mtimeMs || known.size !== file.size)
+                    files.push(file.rel);
+            }
+        }
+        catch {
+            // A failed scan must not silently narrow the grep leg to nothing;
+            // report no staleness and let the indexed legs answer.
+            return [];
+        }
+        return files;
+    }
     /** Whether the running daemon can score with a cross-encoder. */
     supportsCrossEncoder() {
         return this.crossEncoder;
@@ -96,8 +144,9 @@ export class EmbsearchService {
             throw new Error("semantic index is not available");
         }
         if (!this.crossEncoder) {
-            throw new Error(`embsearch is too old to rerank (needs >= ${MIN_RERANK_VERSION.join(".")}); ` +
-                "an older daemon has no cross-encoder weights");
+            throw new Error(`this embsearch daemon cannot rerank (needs >= ${MIN_RERANK_VERSION.join(".")} reporting ` +
+                "`rerank: true`); released binaries ship without cross-encoder weights — start the daemon " +
+                "with --reranker-model <dir>");
         }
         return await this.client.rerank(query, passages, k);
     }
@@ -153,6 +202,8 @@ export class EmbsearchService {
         }
         const binaryVersion = this.probeBinaryVersion(binary);
         this.lexicalRetriever = binaryVersion !== undefined && atLeast(binaryVersion, MIN_LEXICAL_RETRIEVER_VERSION);
+        // Provisional: the version says the daemon understands `rerank`. Whether
+        // it can serve one is answered by `info` below, once the client is up.
         this.crossEncoder = binaryVersion !== undefined && atLeast(binaryVersion, MIN_RERANK_VERSION);
         const storeDir = this.options.storeDir ?? getEmbsearchStoreDir(this.options.cwd);
         this.client = new EmbSearchClient({
@@ -162,6 +213,10 @@ export class EmbsearchService {
         });
         await this.client.ready();
         const info = await this.client.info();
+        // A daemon old enough to omit the field is left on the version verdict —
+        // back then the weights were bundled, so version did imply capability.
+        if (info.rerank === false)
+            this.crossEncoder = false;
         if (info.modelId === MOCK_MODEL_ID) {
             throw new Error("embsearch binary uses the mock embedder (not semantic); install an onnx build");
         }

package/dist/core/embsearch/embsearch-service.js.map

@@ -1 +1 @@
-{"version":3,"file":"embsearch-service.js","sourceRoot":"","sources":["../../../src/core/embsearch/embsearch-service.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,eAAe,CAAC;AAC7C,OAAO,EAAE,YAAY,EAAE,MAAM,IAAI,CAAC;AAClC,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AACtC,OAAO,EAAE,UAAU,EAAE,MAAM,8BAA8B,CAAC;AAC1D,OAAO,EAAE,SAAS,EAAE,MAAM,cAAc,CAAC;AACzC,OAAO,EAEN,eAAe,GAGf,MAAM,aAAa,CAAC;AACrB,OAAO,EACN,cAAc,EAEd,oBAAoB,EACpB,iBAAiB,EACjB,WAAW,EAEX,aAAa,EACb,aAAa,GACb,MAAM,iBAAiB,CAAC;AACzB,OAAO,EAAqB,QAAQ,EAAE,MAAM,gBAAgB,CAAC;AAE7D;oDACoD;AACpD,MAAM,eAAe,GAAG,EAAE,CAAC;AAC3B,+EAA+E;AAC/E,MAAM,cAAc,GAAG,EAAE,CAAC;AAC1B,0FAAwF;AACxF,MAAM,aAAa,GAAG,cAAc,CAAC;AACrC;;;;;;;GAOG;AACH,MAAM,6BAA6B,GAAG,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAU,CAAC;AACzD,uEAAuE;AACvE,MAAM,kBAAkB,GAAG,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAU,CAAC;AAE9C,0EAA0E;AAC1E,SAAS,kBAAkB,CAAC,MAAc,EAAwB;IACjE,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,qBAAqB,CAAC,CAAC;IAClD,OAAO,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;AAAA,CAClF;AAED,SAAS,OAAO,CAAC,OAA0B,EAAE,OAA0B,EAAW;IACjF,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACzC,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QAC7B,IAAI,IAAI,KAAK,OAAO,CAAC,CAAC,CAAC;YAAE,OAAO,IAAI,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;IACnD,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AA2CD,MAAM,OAAO,gBAAgB;IACX,OAAO,CAA0B;IAC1C,MAAM,CAA8B;IACpC,IAAI,CAAwB;IAC5B,KAAK,GAAmB,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;IAC1C,QAAQ,GAAG,KAAK,CAAC;IACzB,iEAAiE;IACzD,gBAAgB,GAAG,KAAK,CAAC;IACjC,wEAAwE;IAChE,YAAY,GAAG,KAAK,CAAC;IAE7B,YAAY,OAAgC,EAAE;QAC7C,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;IAAA,CACvB;IAED,QAAQ,GAAmB;QAC1B,OAAO,IAAI,CAAC,KAAK,CAAC;IAAA,CAClB;IAED,oFAAoF;IACpF,WAAW,GAAY;QACtB,OAAO,IAAI,CAAC,KAAK,CAAC,KAAK,KAAK,OAAO,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,KAAK,UAAU,CAAC;IAAA,CACvE;IAEO,QAAQ,CAAC,KAAqB,EAAQ;QAC7C,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,OAAO,CAAC,UAAU,EAAE,CAAC,KAAK,CAAC,CAAC;IAAA,CACjC;IAED,8DAA8D;IAC9D,wBAAwB,GAAY;QACnC,OAAO,IAAI,CAAC,gBAAgB,CAAC;IAAA,CAC7B;IAED,iEAAiE;IACjE,oBAAoB,GAAY;QAC/B,OAAO,IAAI,CAAC,YAAY,CAAC;IAAA,CACzB;IAED;;;;;;OAMG;IACH,KAAK,CAAC,MAAM,CAAC,KAAa,EAAE,QAAkC,EAAE,CAAS,EAAoC;QAC5G,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,CAAC,QAAQ,EAAE,CAAC;YAC1C,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAC;QACpD,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,CAAC;YACxB,MAAM,IAAI,KAAK,CACd,4CAA4C,kBAAkB,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK;gBAC5E,8CAA8C,CAC/C,CAAC;QACH,CAAC;QACD,OAAO,MAAM,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC;IAAA,CACpD;IAEO,kBAAkB,CAAC,MAAc,EAAwB;QAChE,IAAI,CAAC;YACJ,OAAO,kBAAkB,CAAC,YAAY,CAAC,MAAM,EAAE,CAAC,WAAW,CAAC,EAAE,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC;QACxG,CAAC;QAAC,MAAM,CAAC;YACR,OAAO,SAAS,CAAC;QAClB,CAAC;IAAA,CACD;IAEO,KAAK,CAAC,aAAa,GAAgC;QAC1D,IAAI,IAAI,CAAC,OAAO,CAAC,UAAU,EAAE,CAAC;YAC7B,OAAO,IAAI,CAAC,OAAO,CAAC,UAAU,CAAC;QAChC,CAAC;QACD,6EAA6E;QAC7E,8EAA8E;QAC9E,gEAAgE;QAChE,OAAO,MAAM,UAAU,CAAC,WAAW,EAAE,IAAI,EAAE,CAAC,aAAa,EAAE,UAAU,EAAE,EAAE,CAAC;YACzE,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,aAAa,EAAE,aAAa,EAAE,UAAU,EAAE,CAAC,CAAC;QAAA,CACnE,CAAC,CAAC;IAAA,CACH;IAED;;;OAGG;IACH,KAAK,CAAC,KAAK,CAAC,MAAoB,EAAiB;QAChD,IAAI,CAAC;YACJ,MAAM,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;QACxB,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,MAAM,GAAG,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;YAC1D,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,aAAa,EAAE,MAAM,EAAE,CAAC,CAAC;YAChD,MAAM,IAAI,CAAC,WAAW,EAAE,CAAC;QAC1B,CAAC;IAAA,CACD;IAEO,KAAK,CAAC,GAAG,CAAC,MAAoB,EAAiB;QACtD,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,aAAa,EAAE,CAAC;QAC1C,IAAI,CAAC,MAAM,EAAE,CAAC;YACb,IAAI,CAAC,QAAQ,CAAC;gBACb,KAAK,EAAE,aAAa;gBACpB,MAAM,EAAE,kEAAkE;aAC1E,CAAC,CAAC;YACH,OAAO;QACR,CAAC;QAED,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;QAChD,IAAI,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,cAAc,EAAE,CAAC;YACnD,IAAI,CAAC,QAAQ,CAAC;gBACb,KAAK,EAAE,SAAS;gBAChB,MAAM,EAAE,yBAAyB,IAAI,CAAC,UAAU,MAAM,IAAI,CAAC,OAAO,CAAC,cAAc,SAAS;aAC1F,CAAC,CAAC;YACH,OAAO;QACR,CAAC;QAED,MAAM,aAAa,GAAG,IAAI,CAAC,kBAAkB,CAAC,MAAM,CAAC,CAAC;QACtD,IAAI,CAAC,gBAAgB,GAAG,aAAa,KAAK,SAAS,IAAI,OAAO,CAAC,aAAa,EAAE,6BAA6B,CAAC,CAAC;QAC7G,IAAI,CAAC,YAAY,GAAG,aAAa,KAAK,SAAS,IAAI,OAAO,CAAC,aAAa,EAAE,kBAAkB,CAAC,CAAC;QAE9F,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,QAAQ,IAAI,oBAAoB,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;QACjF,IAAI,CAAC,MAAM,GAAG,IAAI,eAAe,CAAC;YACjC,UAAU,EAAE,MAAM;YAClB,SAAS,EAAE,iBAAiB,CAAC,QAAQ,CAAC;YACtC,MAAM,EAAE,IAAI,CAAC,OAAO,CAAC,WAAW;SAChC,CAAC,CAAC;QACH,MAAM,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,CAAC;QAE1B,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QACtC,IAAI,IAAI,CAAC,OAAO,KAAK,aAAa,EAAE,CAAC;YACpC,MAAM,IAAI,KAAK,CAAC,+EAA+E,CAAC,CAAC;QAClG,CAAC;QAED,+EAA6E;QAC7E,IAAI,CAAC,IAAI,GAAG,aAAa,CAAC,QAAQ,EAAE,IAAI,CAAC,OAAO,CAAC,IAAI,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;QACpG,IAAI,CAAC,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAElC,MAAM,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,KAAK,EAAE,QAAQ,EAAE,MAAM,CAAC,CAAC;IAAA,CAC3D;IAEO,KAAK,CAAC,iBAAiB,CAAC,KAAqB,EAAE,QAAgB,EAAE,MAAoB,EAAiB;QAC7G,MAAM,IAAI,GAAG,IAAI,CAAC,IAAK,CAAC;QACxB,MAAM,MAAM,GAAG,IAAI,CAAC,MAAO,CAAC;QAE5B,0EAA0E;QAC1E,MAAM,OAAO,GAAiE,EAAE,CAAC;QACjF,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;QAC/B,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YAC1B,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACnB,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACnC,IAAI,KAAK,IAAI,KAAK,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO,IAAI,KAAK,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI;gBAAE,SAAS;YAClF,IAAI,OAAe,CAAC;YACpB,IAAI,CAAC;gBACJ,OAAO,GAAG,YAAY,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;YAC3C,CAAC;YAAC,MAAM,CAAC;gBACR,SAAS;YACV,CAAC;YACD,MAAM,IAAI,GAAG,WAAW,CAAC,OAAO,CAAC,CAAC;YAClC,IAAI,KAAK,IAAI,KAAK,CAAC,IAAI,KAAK,IAAI,EAAE,CAAC;gBAClC,oDAAkD;gBAClD,KAAK,CAAC,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC;gBAC7B,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC;gBACvB,SAAS;YACV,CAAC;YACD,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC;QACvC,CAAC;QACD,MAAM,QAAQ,GAAG,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;QAEzE,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACnD,aAAa,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;YAC9B,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,UAAU,EAAE,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YACtE,OAAO;QACR,CAAC;QAED,+DAA+D;QAC/D,MAAM,IAAI,GAA4F,EAAE,CAAC;QACzG,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,KAAK,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,OAAO,EAAE,CAAC;YAC/C,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;YAC5C,IAAI,CAAC,IAAI,CAAC;gBACT,GAAG,EAAE,IAAI,CAAC,GAAG;gBACb,QAAQ,EAAE;oBACT,OAAO,EAAE,IAAI,CAAC,OAAO;oBACrB,IAAI,EAAE,IAAI,CAAC,IAAI;oBACf,IAAI;oBACJ,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC;iBACnD;gBACD,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;aACvD,CAAC,CAAC;YACH,WAAW,IAAI,MAAM,CAAC,MAAM,CAAC;QAC9B,CAAC;QAED,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,CAAC,CAAC;QAElE,4DAA4D;QAC5D,KAAK,MAAM,GAAG,IAAI,QAAQ,EAAE,CAAC;YAC5B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE;gBAAE,MAAM,MAAM,CAAC,MAAM,CAAC,GAAG,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC;YAC3F,OAAO,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QACxB,CAAC;QAED,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;YACzB,IAAI,MAAM,EAAE,OAAO,IAAI,IAAI,CAAC,QAAQ;gBAAE,OAAO;YAC7C,MAAM,aAAa,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,IAAI,CAAC,CAAC;YAC/D,mEAAmE;YACnE,KAAK,IAAI,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,GAAG,aAAa,EAAE,CAAC,EAAE;gBAAE,MAAM,MAAM,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC;YAEjG,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,IAAI,eAAe,EAAE,CAAC;gBAC7E,IAAI,MAAM,EAAE,OAAO,IAAI,IAAI,CAAC,QAAQ;oBAAE,OAAO;gBAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,EAAE,MAAM,GAAG,eAAe,CAAC,CAAC;gBAClE,MAAM,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;gBACzB,IAAI,IAAI,KAAK,CAAC,MAAM,CAAC;gBACrB,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,KAAK,EAAE,WAAW,EAAE,CAAC,CAAC;gBAC/D,uDAAuD;gBACvD,MAAM,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,UAAU,CAAC,OAAO,EAAE,cAAc,CAAC,CAAC,CAAC;YACrE,CAAC;YACD,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC;QACtC,CAAC;QAED,MAAM,MAAM,CAAC,OAAO,EAAE,CAAC;QACvB,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACpB,aAAa,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;QAC9B,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,UAAU,EAAE,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAAA,CACtE;IAEO,WAAW,CAAC,IAAe,EAAU;QAC5C,IAAI,CAAC,GAAG,CAAC,CAAC;QACV,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC;YAAE,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC;QAC9E,OAAO,CAAC,CAAC;IAAA,CACT;IAED,4DAA4D;IAC5D,KAAK,CAAC,MAAM,CAAC,KAAa,EAAE,CAAC,GAAG,EAAE,EAA0B;QAC3D,OAAO,MAAM,IAAI,CAAC,YAAY,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;IAAA,CACzC;IAED,wEAAwE;IACxE,KAAK,CAAC,YAAY,CACjB,KAAa,EACb,CAAC,GAAG,EAAE,EACN,IAAa,EACb,SAAS,GAAoB,OAAO,EACN;QAC9B,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,CAAC,QAAQ,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;YACxD,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAC;QACpD,CAAC;QACD,IAAI,SAAS,KAAK,SAAS,IAAI,CAAC,IAAI,CAAC,gBAAgB,EAAE,CAAC;YACvD,MAAM,IAAI,KAAK,CACd,0DAA0D,6BAA6B,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK;gBACrG,kEAAkE,CACnE,CAAC;QACH,CAAC;QACD,MAAM,OAAO,GAAG,MAAM,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC,EAAE,SAAS,CAAC,CAAC;QAC7D,MAAM,IAAI,GAAuB,EAAE,CAAC;QACpC,MAAM,SAAS,GAAG,CAAC,GAAW,EAAW,EAAE,CAAC;YAC3C,IAAI,CAAC,IAAI;gBAAE,OAAO,IAAI,CAAC;YACvB,OAAO,SAAS,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,GAAG,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAAA,CAC3E,CAAC;QACF,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC9B,MAAM,GAAG,GAAG,MAAM,CAAC,EAAE,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;YACvC,IAAI,GAAG,KAAK,CAAC,CAAC;gBAAE,SAAS;YACzB,MAAM,GAAG,GAAG,MAAM,CAAC,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;YACpC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC;gBAAE,SAAS;YAC9B,MAAM,UAAU,GAAG,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC,KAAK,CAAC,GAAG,GAAG,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YACjE,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,MAAM,CAAC,UAAU,CAAC,CAAC;YACvD,IAAI,CAAC,KAAK;gBAAE,SAAS;YACrB,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,CAAC,EAAE,EAAE,IAAI,EAAE,GAAG,EAAE,SAAS,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,EAAE,CAAC,CAAC;QACtG,CAAC;QACD,OAAO,IAAI,CAAC;IAAA,CACZ;IAED;;;;;OAKG;IACH,kBAAkB,CACjB,GAAW,EACX,IAAY,EACmE;QAC/E,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC;QACnC,IAAI,CAAC,IAAI;YAAE,OAAO,SAAS,CAAC;QAC5B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAC7C,MAAM,CAAC,SAAS,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;YAC5C,IAAI,IAAI,IAAI,SAAS,IAAI,IAAI,IAAI,OAAO,EAAE,CAAC;gBAC1C,OAAO,EAAE,EAAE,EAAE,GAAG,GAAG,IAAI,CAAC,EAAE,EAAE,IAAI,EAAE,GAAG,EAAE,SAAS,EAAE,OAAO,EAAE,CAAC;YAC7D,CAAC;QACF,CAAC;QACD,OAAO,SAAS,CAAC;IAAA,CACjB;IAEO,KAAK,CAAC,WAAW,GAAkB;QAC1C,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC;QAC3B,IAAI,CAAC,MAAM,GAAG,SAAS,CAAC;QACxB,IAAI,MAAM,IAAI,CAAC,MAAM,CAAC,QAAQ,EAAE,CAAC;YAChC,IAAI,CAAC;gBACJ,MAAM,MAAM,CAAC,KAAK,EAAE,CAAC;YACtB,CAAC;YAAC,MAAM,CAAC;gBACR,eAAe;YAChB,CAAC;QACF,CAAC;IAAA,CACD;IAED,kEAAkE;IAClE,KAAK,CAAC,OAAO,GAAkB;QAC9B,IAAI,IAAI,CAAC,QAAQ;YAAE,OAAO;QAC1B,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC;QACrB,IAAI,IAAI,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,EAAE,CAAC;YAC1C,IAAI,CAAC;gBACJ,MAAM,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;YAC1B,CAAC;YAAC,MAAM,CAAC;gBACR,0CAA0C;YAC3C,CAAC;QACF,CAAC;QACD,MAAM,IAAI,CAAC,WAAW,EAAE,CAAC;IAAA,CACzB;CACD;AAED,mCAAmC;AACnC,EAAE;AACF,uEAAuE;AACvE,8EAA8E;AAC9E,8EAA8E;AAC9E,8DAA8D;AAE9D,MAAM,QAAQ,GAAG,IAAI,GAAG,EAA4B,CAAC;AAErD,MAAM,UAAU,wBAAwB,CAAC,GAAW,EAAE,OAAyB,EAAQ;IACtF,MAAM,GAAG,GAAG,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;IAC9B,IAAI,GAAG,IAAI,GAAG,KAAK,OAAO,EAAE,CAAC;QAC5B,GAAG,CAAC,OAAO,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAC,CAAC,CAAC,CAAC;IAC/B,CAAC;IACD,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;AAAA,CAC3B;AAED,MAAM,UAAU,mBAAmB,CAAC,GAAW,EAAgC;IAC9E,OAAO,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;AAAA,CACzB;AAED,MAAM,UAAU,0BAA0B,CAAC,GAAW,EAAQ;IAC7D,QAAQ,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;AAAA,CACrB","sourcesContent":["/**\n * Orchestrates semantic indexing and search for a repository.\n *\n * Lifecycle (all behind --enable-embsearchtools):\n *  1. `start()` — resolve the embsearch binary, scan the repo (ignore-aware),\n *     apply the byte threshold. Under threshold → dormant. Over → spawn the\n *     daemon, verify the backend is not the mock embedder, then index changed\n *     files in the background in small batches, reporting progress.\n *  2. `search()` — top-k semantic query, mapping chunk ids back to\n *     `path:start-end` via the sidecar metadata.\n *  3. `dispose()` — save + close the daemon.\n *\n * Every failure degrades to `unavailable` with a reason; nothing here ever\n * blocks session startup or affects grep/find.\n */\n\nimport { execFileSync } from \"child_process\";\nimport { readFileSync } from \"fs\";\nimport { minimatch } from \"minimatch\";\nimport { ensureTool } from \"../../utils/tools-manager.js\";\nimport { chunkFile } from \"./chunker.js\";\nimport {\n\ttype DaemonRetriever,\n\tEmbSearchClient,\n\ttype EmbSearchRerankPassage,\n\ttype EmbSearchRerankResult,\n} from \"./client.js\";\nimport {\n\temptyIndexMeta,\n\ttype FileMeta,\n\tgetEmbsearchStoreDir,\n\tgetVectorStoreDir,\n\thashContent,\n\ttype IndexMeta,\n\tloadIndexMeta,\n\tsaveIndexMeta,\n} from \"./index-meta.js\";\nimport { type RepoScanFile, scanRepo } from \"./repo-scan.js\";\n\n/** Chunks per bulk request. Small enough that a concurrent query is never\n *  stuck long behind one padded batch inference. */\nconst BULK_BATCH_SIZE = 48;\n/** Yield between batches so background indexing doesn't starve the session. */\nconst BATCH_YIELD_MS = 15;\n/** The Rust mock backend's model id — semantically meaningless, never index with it. */\nconst MOCK_MODEL_ID = \"mock-hash-v1\";\n/**\n * First embsearch release serving `retriever: \"lexical\"`.\n *\n * The guard matters because the daemon does not reject unknown request fields:\n * an older binary silently ignores `retriever` and answers with dense results.\n * Fusing that list a second time as a \"bm25\" leg would double-count it and\n * corrupt the ranking with no error anywhere — so refuse instead of degrading.\n */\nconst MIN_LEXICAL_RETRIEVER_VERSION = [0, 2, 0] as const;\n/** First embsearch release bundling cross-encoder reranker weights. */\nconst MIN_RERANK_VERSION = [0, 3, 0] as const;\n\n/** `embsearch 0.2.0` -> [0, 2, 0]; undefined when it cannot be parsed. */\nfunction parseBinaryVersion(output: string): number[] | undefined {\n\tconst match = output.match(/(\\d+)\\.(\\d+)\\.(\\d+)/);\n\treturn match ? [Number(match[1]), Number(match[2]), Number(match[3])] : undefined;\n}\n\nfunction atLeast(version: readonly number[], minimum: readonly number[]): boolean {\n\tfor (let i = 0; i < minimum.length; i++) {\n\t\tconst part = version[i] ?? 0;\n\t\tif (part !== minimum[i]) return part > minimum[i];\n\t}\n\treturn true;\n}\n\nexport type EmbsearchState =\n\t| { phase: \"idle\" }\n\t| { phase: \"skipped\"; reason: string }\n\t| { phase: \"downloading\"; receivedBytes: number; totalBytes: number | null }\n\t| { phase: \"indexing\"; done: number; total: number }\n\t| { phase: \"ready\"; chunkCount: number }\n\t| { phase: \"unavailable\"; reason: string };\n\nexport interface EmbsearchServiceOptions {\n\tcwd: string;\n\t/** Explicit binary path (settings override). Default: \"embsearch\" from PATH. */\n\tbinaryPath?: string;\n\t/** Minimum indexable bytes before indexing kicks in. */\n\tthresholdBytes: number;\n\t/**\n\t * Override the store location. Only the eval harness sets this, so a\n\t * second index (e.g. a BM25-hybrid store) can exist for the same repo\n\t * without colliding with the primary one.\n\t */\n\tstoreDir?: string;\n\t/**\n\t * Create the store with the daemon's BM25 lexical index. Fixed at store\n\t * creation, so this must pair with a distinct `storeDir`.\n\t */\n\thybridStore?: boolean;\n\t/** Progress callback for UI (footer / stderr lines). */\n\tonProgress?: (state: EmbsearchState) => void;\n}\n\nexport interface SemanticHit {\n\tpath: string;\n\tstartLine: number;\n\tendLine: number;\n\tscore: number;\n}\n\nexport interface SemanticChunkHit extends SemanticHit {\n\t/** Per-build chunk id (`relpath#index`) — the fusion identity for hybrid search. */\n\tid: string;\n}\n\nexport class EmbsearchService {\n\tprivate readonly options: EmbsearchServiceOptions;\n\tprivate client: EmbSearchClient | undefined;\n\tprivate meta: IndexMeta | undefined;\n\tprivate state: EmbsearchState = { phase: \"idle\" };\n\tprivate disposed = false;\n\t/** Whether the resolved binary serves `retriever: \"lexical\"`. */\n\tprivate lexicalRetriever = false;\n\t/** Whether the resolved binary serves the cross-encoder `rerank` op. */\n\tprivate crossEncoder = false;\n\n\tconstructor(options: EmbsearchServiceOptions) {\n\t\tthis.options = options;\n\t}\n\n\tgetState(): EmbsearchState {\n\t\treturn this.state;\n\t}\n\n\t/** Semantic search is usable (index ready, or still building with partial data). */\n\tisAvailable(): boolean {\n\t\treturn this.state.phase === \"ready\" || this.state.phase === \"indexing\";\n\t}\n\n\tprivate setState(state: EmbsearchState): void {\n\t\tthis.state = state;\n\t\tthis.options.onProgress?.(state);\n\t}\n\n\t/** Whether the running daemon can serve a BM25-only query. */\n\tsupportsLexicalRetriever(): boolean {\n\t\treturn this.lexicalRetriever;\n\t}\n\n\t/** Whether the running daemon can score with a cross-encoder. */\n\tsupportsCrossEncoder(): boolean {\n\t\treturn this.crossEncoder;\n\t}\n\n\t/**\n\t * Cross-encoder rerank of caller-supplied passages.\n\t *\n\t * Unlike the retrievers this does not consult the index at all — it scores\n\t * exactly the text passed in, which is why the caller sends its expanded\n\t * windows rather than chunk ids.\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tif (!this.client || this.client.isClosed) {\n\t\t\tthrow new Error(\"semantic index is not available\");\n\t\t}\n\t\tif (!this.crossEncoder) {\n\t\t\tthrow new Error(\n\t\t\t\t`embsearch is too old to rerank (needs >= ${MIN_RERANK_VERSION.join(\".\")}); ` +\n\t\t\t\t\t\"an older daemon has no cross-encoder weights\",\n\t\t\t);\n\t\t}\n\t\treturn await this.client.rerank(query, passages, k);\n\t}\n\n\tprivate probeBinaryVersion(binary: string): number[] | undefined {\n\t\ttry {\n\t\t\treturn parseBinaryVersion(execFileSync(binary, [\"--version\"], { encoding: \"utf-8\", timeout: 10_000 }));\n\t\t} catch {\n\t\t\treturn undefined;\n\t\t}\n\t}\n\n\tprivate async resolveBinary(): Promise<string | undefined> {\n\t\tif (this.options.binaryPath) {\n\t\t\treturn this.options.binaryPath;\n\t\t}\n\t\t// Surface the on-demand binary download through the same progress channel as\n\t\t// indexing, so the first-run fetch renders a progress bar instead of a stall.\n\t\t// A cached binary resolves without ever invoking this callback.\n\t\treturn await ensureTool(\"embsearch\", true, (receivedBytes, totalBytes) => {\n\t\t\tthis.setState({ phase: \"downloading\", receivedBytes, totalBytes });\n\t\t});\n\t}\n\n\t/**\n\t * Scan, threshold-check, and (when needed) index in the background.\n\t * Resolves when indexing completes or the feature settles dormant.\n\t */\n\tasync start(signal?: AbortSignal): Promise<void> {\n\t\ttry {\n\t\t\tawait this.run(signal);\n\t\t} catch (e) {\n\t\t\tconst reason = e instanceof Error ? e.message : String(e);\n\t\t\tthis.setState({ phase: \"unavailable\", reason });\n\t\t\tawait this.closeClient();\n\t\t}\n\t}\n\n\tprivate async run(signal?: AbortSignal): Promise<void> {\n\t\tconst binary = await this.resolveBinary();\n\t\tif (!binary) {\n\t\t\tthis.setState({\n\t\t\t\tphase: \"unavailable\",\n\t\t\t\treason: \"embsearch binary not found (PATH or embsearchBinaryPath setting)\",\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst scan = scanRepo(this.options.cwd, signal);\n\t\tif (scan.totalBytes < this.options.thresholdBytes) {\n\t\t\tthis.setState({\n\t\t\t\tphase: \"skipped\",\n\t\t\t\treason: `repo under threshold (${scan.totalBytes} < ${this.options.thresholdBytes} bytes)`,\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst binaryVersion = this.probeBinaryVersion(binary);\n\t\tthis.lexicalRetriever = binaryVersion !== undefined && atLeast(binaryVersion, MIN_LEXICAL_RETRIEVER_VERSION);\n\t\tthis.crossEncoder = binaryVersion !== undefined && atLeast(binaryVersion, MIN_RERANK_VERSION);\n\n\t\tconst storeDir = this.options.storeDir ?? getEmbsearchStoreDir(this.options.cwd);\n\t\tthis.client = new EmbSearchClient({\n\t\t\tbinaryPath: binary,\n\t\t\tstorePath: getVectorStoreDir(storeDir),\n\t\t\thybrid: this.options.hybridStore,\n\t\t});\n\t\tawait this.client.ready();\n\n\t\tconst info = await this.client.info();\n\t\tif (info.modelId === MOCK_MODEL_ID) {\n\t\t\tthrow new Error(\"embsearch binary uses the mock embedder (not semantic); install an onnx build\");\n\t\t}\n\n\t\t// Missing/stale sidecar (format, chunker, or model changed) → clean rebuild.\n\t\tthis.meta = loadIndexMeta(storeDir, info.modelId) ?? emptyIndexMeta(this.options.cwd, info.modelId);\n\t\tthis.meta.lastUsedMs = Date.now();\n\n\t\tawait this.indexChangedFiles(scan.files, storeDir, signal);\n\t}\n\n\tprivate async indexChangedFiles(files: RepoScanFile[], storeDir: string, signal?: AbortSignal): Promise<void> {\n\t\tconst meta = this.meta!;\n\t\tconst client = this.client!;\n\n\t\t// Diff scan vs sidecar: cheap mtime+size check first, hash only on delta.\n\t\tconst toIndex: Array<{ file: RepoScanFile; content: string; hash: string }> = [];\n\t\tconst seen = new Set<string>();\n\t\tfor (const file of files) {\n\t\t\tseen.add(file.rel);\n\t\t\tconst known = meta.files[file.rel];\n\t\t\tif (known && known.mtimeMs === file.mtimeMs && known.size === file.size) continue;\n\t\t\tlet content: string;\n\t\t\ttry {\n\t\t\t\tcontent = readFileSync(file.abs, \"utf-8\");\n\t\t\t} catch {\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tconst hash = hashContent(content);\n\t\t\tif (known && known.hash === hash) {\n\t\t\t\t// Touched but unchanged — refresh stat info only.\n\t\t\t\tknown.mtimeMs = file.mtimeMs;\n\t\t\t\tknown.size = file.size;\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\ttoIndex.push({ file, content, hash });\n\t\t}\n\t\tconst toRemove = Object.keys(meta.files).filter((rel) => !seen.has(rel));\n\n\t\tif (toIndex.length === 0 && toRemove.length === 0) {\n\t\t\tsaveIndexMeta(storeDir, meta);\n\t\t\tthis.setState({ phase: \"ready\", chunkCount: this.countChunks(meta) });\n\t\t\treturn;\n\t\t}\n\n\t\t// Chunk changed files; count total upserts for exact progress.\n\t\tconst work: Array<{ rel: string; fileMeta: FileMeta; chunks: Array<{ id: string; text: string }> }> = [];\n\t\tlet totalChunks = 0;\n\t\tfor (const { file, content, hash } of toIndex) {\n\t\t\tconst chunks = chunkFile(file.rel, content);\n\t\t\twork.push({\n\t\t\t\trel: file.rel,\n\t\t\t\tfileMeta: {\n\t\t\t\t\tmtimeMs: file.mtimeMs,\n\t\t\t\t\tsize: file.size,\n\t\t\t\t\thash,\n\t\t\t\t\tchunks: chunks.map((c) => [c.startLine, c.endLine]),\n\t\t\t\t},\n\t\t\t\tchunks: chunks.map((c) => ({ id: c.id, text: c.text })),\n\t\t\t});\n\t\t\ttotalChunks += chunks.length;\n\t\t}\n\n\t\tthis.setState({ phase: \"indexing\", done: 0, total: totalChunks });\n\n\t\t// Drop vectors of deleted files and superseded chunk tails.\n\t\tfor (const rel of toRemove) {\n\t\t\tfor (let i = 0; i < meta.files[rel].chunks.length; i++) await client.remove(`${rel}#${i}`);\n\t\t\tdelete meta.files[rel];\n\t\t}\n\n\t\tlet done = 0;\n\t\tfor (const item of work) {\n\t\t\tif (signal?.aborted || this.disposed) return;\n\t\t\tconst oldChunkCount = meta.files[item.rel]?.chunks.length ?? 0;\n\t\t\t// Remove old chunks beyond the new count (upsert covers the rest).\n\t\t\tfor (let i = item.chunks.length; i < oldChunkCount; i++) await client.remove(`${item.rel}#${i}`);\n\n\t\t\tfor (let offset = 0; offset < item.chunks.length; offset += BULK_BATCH_SIZE) {\n\t\t\t\tif (signal?.aborted || this.disposed) return;\n\t\t\t\tconst batch = item.chunks.slice(offset, offset + BULK_BATCH_SIZE);\n\t\t\t\tawait client.bulk(batch);\n\t\t\t\tdone += batch.length;\n\t\t\t\tthis.setState({ phase: \"indexing\", done, total: totalChunks });\n\t\t\t\t// Yield so queries and the event loop stay responsive.\n\t\t\t\tawait new Promise((resolve) => setTimeout(resolve, BATCH_YIELD_MS));\n\t\t\t}\n\t\t\tmeta.files[item.rel] = item.fileMeta;\n\t\t}\n\n\t\tawait client.compact();\n\t\tawait client.save();\n\t\tsaveIndexMeta(storeDir, meta);\n\t\tthis.setState({ phase: \"ready\", chunkCount: this.countChunks(meta) });\n\t}\n\n\tprivate countChunks(meta: IndexMeta): number {\n\t\tlet n = 0;\n\t\tfor (const rel of Object.keys(meta.files)) n += meta.files[rel].chunks.length;\n\t\treturn n;\n\t}\n\n\t/** Top-`k` semantic hits as `path` + line range + score. */\n\tasync search(query: string, k = 10): Promise<SemanticHit[]> {\n\t\treturn await this.searchChunks(query, k);\n\t}\n\n\t/** Top-`k` semantic hits including their chunk ids, for rank fusion. */\n\tasync searchChunks(\n\t\tquery: string,\n\t\tk = 10,\n\t\tglob?: string,\n\t\tretriever: DaemonRetriever = \"dense\",\n\t): Promise<SemanticChunkHit[]> {\n\t\tif (!this.client || this.client.isClosed || !this.meta) {\n\t\t\tthrow new Error(\"semantic index is not available\");\n\t\t}\n\t\tif (retriever === \"lexical\" && !this.lexicalRetriever) {\n\t\t\tthrow new Error(\n\t\t\t\t`embsearch is too old for retriever \"lexical\" (needs >= ${MIN_LEXICAL_RETRIEVER_VERSION.join(\".\")}); ` +\n\t\t\t\t\t\"an older daemon ignores the field and answers with dense results\",\n\t\t\t);\n\t\t}\n\t\tconst results = await this.client.query(query, k, retriever);\n\t\tconst hits: SemanticChunkHit[] = [];\n\t\tconst matchGlob = (rel: string): boolean => {\n\t\t\tif (!glob) return true;\n\t\t\treturn minimatch(rel, glob, { dot: true, matchBase: !glob.includes(\"/\") });\n\t\t};\n\t\tfor (const result of results) {\n\t\t\tconst sep = result.id.lastIndexOf(\"#\");\n\t\t\tif (sep === -1) continue;\n\t\t\tconst rel = result.id.slice(0, sep);\n\t\t\tif (!matchGlob(rel)) continue;\n\t\t\tconst chunkIndex = Number.parseInt(result.id.slice(sep + 1), 10);\n\t\t\tconst range = this.meta.files[rel]?.chunks[chunkIndex];\n\t\t\tif (!range) continue;\n\t\t\thits.push({ id: result.id, path: rel, startLine: range[0], endLine: range[1], score: result.score });\n\t\t}\n\t\treturn hits;\n\t}\n\n\t/**\n\t * Resolve a repo-relative path + line to its enclosing indexed chunk, or\n\t * undefined when the file/line is not covered by the index. Chunks overlap\n\t * by a few lines; the first (lowest-index) containing chunk wins so the\n\t * mapping is deterministic.\n\t */\n\tfindEnclosingChunk(\n\t\trel: string,\n\t\tline: number,\n\t): { id: string; path: string; startLine: number; endLine: number } | undefined {\n\t\tconst file = this.meta?.files[rel];\n\t\tif (!file) return undefined;\n\t\tfor (let i = 0; i < file.chunks.length; i++) {\n\t\t\tconst [startLine, endLine] = file.chunks[i];\n\t\t\tif (line >= startLine && line <= endLine) {\n\t\t\t\treturn { id: `${rel}#${i}`, path: rel, startLine, endLine };\n\t\t\t}\n\t\t}\n\t\treturn undefined;\n\t}\n\n\tprivate async closeClient(): Promise<void> {\n\t\tconst client = this.client;\n\t\tthis.client = undefined;\n\t\tif (client && !client.isClosed) {\n\t\t\ttry {\n\t\t\t\tawait client.close();\n\t\t\t} catch {\n\t\t\t\t// already dead\n\t\t\t}\n\t\t}\n\t}\n\n\t/** Persist state and shut the daemon down. Safe to call twice. */\n\tasync dispose(): Promise<void> {\n\t\tif (this.disposed) return;\n\t\tthis.disposed = true;\n\t\tif (this.client && !this.client.isClosed) {\n\t\t\ttry {\n\t\t\t\tawait this.client.save();\n\t\t\t} catch {\n\t\t\t\t// daemon may have exited; nothing to save\n\t\t\t}\n\t\t}\n\t\tawait this.closeClient();\n\t}\n}\n\n// --- Per-cwd service registry ---\n//\n// The search tool is constructed by the generic tool factory table and\n// only receives `cwd`; the service is created later during session init (flag\n// gated). This registry connects the two without threading a service instance\n// through every layer between main.ts and the tool factories.\n\nconst services = new Map<string, EmbsearchService>();\n\nexport function registerEmbsearchService(cwd: string, service: EmbsearchService): void {\n\tconst old = services.get(cwd);\n\tif (old && old !== service) {\n\t\told.dispose().catch(() => {});\n\t}\n\tservices.set(cwd, service);\n}\n\nexport function getEmbsearchService(cwd: string): EmbsearchService | undefined {\n\treturn services.get(cwd);\n}\n\nexport function unregisterEmbsearchService(cwd: string): void {\n\tservices.delete(cwd);\n}\n"]}
+{"version":3,"file":"embsearch-service.js","sourceRoot":"","sources":["../../../src/core/embsearch/embsearch-service.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,eAAe,CAAC;AAC7C,OAAO,EAAE,YAAY,EAAE,MAAM,IAAI,CAAC;AAClC,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AACtC,OAAO,EAAE,UAAU,EAAE,MAAM,8BAA8B,CAAC;AAC1D,OAAO,EAAE,SAAS,EAAE,MAAM,cAAc,CAAC;AACzC,OAAO,EAEN,eAAe,GAGf,MAAM,aAAa,CAAC;AACrB,OAAO,EACN,cAAc,EAEd,oBAAoB,EACpB,iBAAiB,EACjB,WAAW,EAEX,aAAa,EACb,aAAa,GACb,MAAM,iBAAiB,CAAC;AACzB,OAAO,EAAqB,QAAQ,EAAE,MAAM,gBAAgB,CAAC;AAE7D;oDACoD;AACpD,MAAM,eAAe,GAAG,EAAE,CAAC;AAC3B,+EAA+E;AAC/E,MAAM,cAAc,GAAG,EAAE,CAAC;AAC1B,0FAAwF;AACxF,MAAM,aAAa,GAAG,cAAc,CAAC;AACrC;;;;;;;GAOG;AACH,MAAM,6BAA6B,GAAG,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAU,CAAC;AACzD;;;;;;;;;GASG;AACH,MAAM,kBAAkB,GAAG,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAU,CAAC;AAE9C,0EAA0E;AAC1E,SAAS,kBAAkB,CAAC,MAAc,EAAwB;IACjE,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,qBAAqB,CAAC,CAAC;IAClD,OAAO,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;AAAA,CAClF;AAED,SAAS,OAAO,CAAC,OAA0B,EAAE,OAA0B,EAAW;IACjF,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACzC,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QAC7B,IAAI,IAAI,KAAK,OAAO,CAAC,CAAC,CAAC;YAAE,OAAO,IAAI,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;IACnD,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AA2CD,MAAM,OAAO,gBAAgB;IACX,OAAO,CAA0B;IAC1C,MAAM,CAA8B;IACpC,IAAI,CAAwB;IAC5B,KAAK,GAAmB,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;IAC1C,QAAQ,GAAG,KAAK,CAAC;IACzB,iEAAiE;IACzD,gBAAgB,GAAG,KAAK,CAAC;IACjC,wEAAwE;IAChE,YAAY,GAAG,KAAK,CAAC;IAE7B,YAAY,OAAgC,EAAE;QAC7C,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;IAAA,CACvB;IAED,QAAQ,GAAmB;QAC1B,OAAO,IAAI,CAAC,KAAK,CAAC;IAAA,CAClB;IAED,oFAAoF;IACpF,WAAW,GAAY;QACtB,OAAO,IAAI,CAAC,KAAK,CAAC,KAAK,KAAK,OAAO,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,KAAK,UAAU,CAAC;IAAA,CACvE;IAEO,QAAQ,CAAC,KAAqB,EAAQ;QAC7C,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,OAAO,CAAC,UAAU,EAAE,CAAC,KAAK,CAAC,CAAC;IAAA,CACjC;IAED,8DAA8D;IAC9D,wBAAwB,GAAY;QACnC,OAAO,IAAI,CAAC,gBAAgB,CAAC;IAAA,CAC7B;IAED;;;;;;;;;;;;;;;;;;;OAmBG;IACH,UAAU,CAAC,MAAoB,EAAY;QAC1C,IAAI,CAAC,IAAI,CAAC,IAAI;YAAE,OAAO,EAAE,CAAC;QAC1B,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC;QACvB,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,IAAI,CAAC;YACJ,KAAK,MAAM,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC,KAAK,EAAE,CAAC;gBAC7D,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBACnC,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO,IAAI,KAAK,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YAChG,CAAC;QACF,CAAC;QAAC,MAAM,CAAC;YACR,kEAAkE;YAClE,uDAAuD;YACvD,OAAO,EAAE,CAAC;QACX,CAAC;QACD,OAAO,KAAK,CAAC;IAAA,CACb;IAED,iEAAiE;IACjE,oBAAoB,GAAY;QAC/B,OAAO,IAAI,CAAC,YAAY,CAAC;IAAA,CACzB;IAED;;;;;;OAMG;IACH,KAAK,CAAC,MAAM,CAAC,KAAa,EAAE,QAAkC,EAAE,CAAS,EAAoC;QAC5G,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,CAAC,QAAQ,EAAE,CAAC;YAC1C,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAC;QACpD,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,CAAC;YACxB,MAAM,IAAI,KAAK,CACd,iDAAiD,kBAAkB,CAAC,IAAI,CAAC,GAAG,CAAC,aAAa;gBACzF,6FAA2F;gBAC3F,6BAA6B,CAC9B,CAAC;QACH,CAAC;QACD,OAAO,MAAM,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC;IAAA,CACpD;IAEO,kBAAkB,CAAC,MAAc,EAAwB;QAChE,IAAI,CAAC;YACJ,OAAO,kBAAkB,CAAC,YAAY,CAAC,MAAM,EAAE,CAAC,WAAW,CAAC,EAAE,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC;QACxG,CAAC;QAAC,MAAM,CAAC;YACR,OAAO,SAAS,CAAC;QAClB,CAAC;IAAA,CACD;IAEO,KAAK,CAAC,aAAa,GAAgC;QAC1D,IAAI,IAAI,CAAC,OAAO,CAAC,UAAU,EAAE,CAAC;YAC7B,OAAO,IAAI,CAAC,OAAO,CAAC,UAAU,CAAC;QAChC,CAAC;QACD,6EAA6E;QAC7E,8EAA8E;QAC9E,gEAAgE;QAChE,OAAO,MAAM,UAAU,CAAC,WAAW,EAAE,IAAI,EAAE,CAAC,aAAa,EAAE,UAAU,EAAE,EAAE,CAAC;YACzE,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,aAAa,EAAE,aAAa,EAAE,UAAU,EAAE,CAAC,CAAC;QAAA,CACnE,CAAC,CAAC;IAAA,CACH;IAED;;;OAGG;IACH,KAAK,CAAC,KAAK,CAAC,MAAoB,EAAiB;QAChD,IAAI,CAAC;YACJ,MAAM,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;QACxB,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,MAAM,GAAG,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;YAC1D,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,aAAa,EAAE,MAAM,EAAE,CAAC,CAAC;YAChD,MAAM,IAAI,CAAC,WAAW,EAAE,CAAC;QAC1B,CAAC;IAAA,CACD;IAEO,KAAK,CAAC,GAAG,CAAC,MAAoB,EAAiB;QACtD,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,aAAa,EAAE,CAAC;QAC1C,IAAI,CAAC,MAAM,EAAE,CAAC;YACb,IAAI,CAAC,QAAQ,CAAC;gBACb,KAAK,EAAE,aAAa;gBACpB,MAAM,EAAE,kEAAkE;aAC1E,CAAC,CAAC;YACH,OAAO;QACR,CAAC;QAED,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;QAChD,IAAI,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,cAAc,EAAE,CAAC;YACnD,IAAI,CAAC,QAAQ,CAAC;gBACb,KAAK,EAAE,SAAS;gBAChB,MAAM,EAAE,yBAAyB,IAAI,CAAC,UAAU,MAAM,IAAI,CAAC,OAAO,CAAC,cAAc,SAAS;aAC1F,CAAC,CAAC;YACH,OAAO;QACR,CAAC;QAED,MAAM,aAAa,GAAG,IAAI,CAAC,kBAAkB,CAAC,MAAM,CAAC,CAAC;QACtD,IAAI,CAAC,gBAAgB,GAAG,aAAa,KAAK,SAAS,IAAI,OAAO,CAAC,aAAa,EAAE,6BAA6B,CAAC,CAAC;QAC7G,yEAAyE;QACzE,uEAAuE;QACvE,IAAI,CAAC,YAAY,GAAG,aAAa,KAAK,SAAS,IAAI,OAAO,CAAC,aAAa,EAAE,kBAAkB,CAAC,CAAC;QAE9F,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,QAAQ,IAAI,oBAAoB,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;QACjF,IAAI,CAAC,MAAM,GAAG,IAAI,eAAe,CAAC;YACjC,UAAU,EAAE,MAAM;YAClB,SAAS,EAAE,iBAAiB,CAAC,QAAQ,CAAC;YACtC,MAAM,EAAE,IAAI,CAAC,OAAO,CAAC,WAAW;SAChC,CAAC,CAAC;QACH,MAAM,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,CAAC;QAE1B,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QACtC,2EAAyE;QACzE,uEAAuE;QACvE,IAAI,IAAI,CAAC,MAAM,KAAK,KAAK;YAAE,IAAI,CAAC,YAAY,GAAG,KAAK,CAAC;QACrD,IAAI,IAAI,CAAC,OAAO,KAAK,aAAa,EAAE,CAAC;YACpC,MAAM,IAAI,KAAK,CAAC,+EAA+E,CAAC,CAAC;QAClG,CAAC;QAED,+EAA6E;QAC7E,IAAI,CAAC,IAAI,GAAG,aAAa,CAAC,QAAQ,EAAE,IAAI,CAAC,OAAO,CAAC,IAAI,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;QACpG,IAAI,CAAC,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAElC,MAAM,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,KAAK,EAAE,QAAQ,EAAE,MAAM,CAAC,CAAC;IAAA,CAC3D;IAEO,KAAK,CAAC,iBAAiB,CAAC,KAAqB,EAAE,QAAgB,EAAE,MAAoB,EAAiB;QAC7G,MAAM,IAAI,GAAG,IAAI,CAAC,IAAK,CAAC;QACxB,MAAM,MAAM,GAAG,IAAI,CAAC,MAAO,CAAC;QAE5B,0EAA0E;QAC1E,MAAM,OAAO,GAAiE,EAAE,CAAC;QACjF,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;QAC/B,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YAC1B,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACnB,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACnC,IAAI,KAAK,IAAI,KAAK,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO,IAAI,KAAK,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI;gBAAE,SAAS;YAClF,IAAI,OAAe,CAAC;YACpB,IAAI,CAAC;gBACJ,OAAO,GAAG,YAAY,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;YAC3C,CAAC;YAAC,MAAM,CAAC;gBACR,SAAS;YACV,CAAC;YACD,MAAM,IAAI,GAAG,WAAW,CAAC,OAAO,CAAC,CAAC;YAClC,IAAI,KAAK,IAAI,KAAK,CAAC,IAAI,KAAK,IAAI,EAAE,CAAC;gBAClC,oDAAkD;gBAClD,KAAK,CAAC,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC;gBAC7B,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC;gBACvB,SAAS;YACV,CAAC;YACD,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC;QACvC,CAAC;QACD,MAAM,QAAQ,GAAG,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;QAEzE,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACnD,aAAa,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;YAC9B,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,UAAU,EAAE,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YACtE,OAAO;QACR,CAAC;QAED,+DAA+D;QAC/D,MAAM,IAAI,GAA4F,EAAE,CAAC;QACzG,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,KAAK,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,OAAO,EAAE,CAAC;YAC/C,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;YAC5C,IAAI,CAAC,IAAI,CAAC;gBACT,GAAG,EAAE,IAAI,CAAC,GAAG;gBACb,QAAQ,EAAE;oBACT,OAAO,EAAE,IAAI,CAAC,OAAO;oBACrB,IAAI,EAAE,IAAI,CAAC,IAAI;oBACf,IAAI;oBACJ,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC;iBACnD;gBACD,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;aACvD,CAAC,CAAC;YACH,WAAW,IAAI,MAAM,CAAC,MAAM,CAAC;QAC9B,CAAC;QAED,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,CAAC,CAAC;QAElE,4DAA4D;QAC5D,KAAK,MAAM,GAAG,IAAI,QAAQ,EAAE,CAAC;YAC5B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE;gBAAE,MAAM,MAAM,CAAC,MAAM,CAAC,GAAG,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC;YAC3F,OAAO,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QACxB,CAAC;QAED,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;YACzB,IAAI,MAAM,EAAE,OAAO,IAAI,IAAI,CAAC,QAAQ;gBAAE,OAAO;YAC7C,MAAM,aAAa,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,IAAI,CAAC,CAAC;YAC/D,mEAAmE;YACnE,KAAK,IAAI,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,GAAG,aAAa,EAAE,CAAC,EAAE;gBAAE,MAAM,MAAM,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC;YAEjG,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,IAAI,eAAe,EAAE,CAAC;gBAC7E,IAAI,MAAM,EAAE,OAAO,IAAI,IAAI,CAAC,QAAQ;oBAAE,OAAO;gBAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,EAAE,MAAM,GAAG,eAAe,CAAC,CAAC;gBAClE,MAAM,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;gBACzB,IAAI,IAAI,KAAK,CAAC,MAAM,CAAC;gBACrB,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,KAAK,EAAE,WAAW,EAAE,CAAC,CAAC;gBAC/D,uDAAuD;gBACvD,MAAM,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,UAAU,CAAC,OAAO,EAAE,cAAc,CAAC,CAAC,CAAC;YACrE,CAAC;YACD,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC;QACtC,CAAC;QAED,MAAM,MAAM,CAAC,OAAO,EAAE,CAAC;QACvB,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACpB,aAAa,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;QAC9B,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,UAAU,EAAE,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAAA,CACtE;IAEO,WAAW,CAAC,IAAe,EAAU;QAC5C,IAAI,CAAC,GAAG,CAAC,CAAC;QACV,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC;YAAE,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC;QAC9E,OAAO,CAAC,CAAC;IAAA,CACT;IAED,4DAA4D;IAC5D,KAAK,CAAC,MAAM,CAAC,KAAa,EAAE,CAAC,GAAG,EAAE,EAA0B;QAC3D,OAAO,MAAM,IAAI,CAAC,YAAY,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;IAAA,CACzC;IAED,wEAAwE;IACxE,KAAK,CAAC,YAAY,CACjB,KAAa,EACb,CAAC,GAAG,EAAE,EACN,IAAa,EACb,SAAS,GAAoB,OAAO,EACN;QAC9B,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,CAAC,QAAQ,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;YACxD,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAC;QACpD,CAAC;QACD,IAAI,SAAS,KAAK,SAAS,IAAI,CAAC,IAAI,CAAC,gBAAgB,EAAE,CAAC;YACvD,MAAM,IAAI,KAAK,CACd,0DAA0D,6BAA6B,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK;gBACrG,kEAAkE,CACnE,CAAC;QACH,CAAC;QACD,MAAM,OAAO,GAAG,MAAM,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC,EAAE,SAAS,CAAC,CAAC;QAC7D,MAAM,IAAI,GAAuB,EAAE,CAAC;QACpC,MAAM,SAAS,GAAG,CAAC,GAAW,EAAW,EAAE,CAAC;YAC3C,IAAI,CAAC,IAAI;gBAAE,OAAO,IAAI,CAAC;YACvB,OAAO,SAAS,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,GAAG,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAAA,CAC3E,CAAC;QACF,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC9B,MAAM,GAAG,GAAG,MAAM,CAAC,EAAE,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;YACvC,IAAI,GAAG,KAAK,CAAC,CAAC;gBAAE,SAAS;YACzB,MAAM,GAAG,GAAG,MAAM,CAAC,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;YACpC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC;gBAAE,SAAS;YAC9B,MAAM,UAAU,GAAG,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC,KAAK,CAAC,GAAG,GAAG,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YACjE,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,MAAM,CAAC,UAAU,CAAC,CAAC;YACvD,IAAI,CAAC,KAAK;gBAAE,SAAS;YACrB,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,CAAC,EAAE,EAAE,IAAI,EAAE,GAAG,EAAE,SAAS,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,EAAE,CAAC,CAAC;QACtG,CAAC;QACD,OAAO,IAAI,CAAC;IAAA,CACZ;IAED;;;;;OAKG;IACH,kBAAkB,CACjB,GAAW,EACX,IAAY,EACmE;QAC/E,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC;QACnC,IAAI,CAAC,IAAI;YAAE,OAAO,SAAS,CAAC;QAC5B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAC7C,MAAM,CAAC,SAAS,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;YAC5C,IAAI,IAAI,IAAI,SAAS,IAAI,IAAI,IAAI,OAAO,EAAE,CAAC;gBAC1C,OAAO,EAAE,EAAE,EAAE,GAAG,GAAG,IAAI,CAAC,EAAE,EAAE,IAAI,EAAE,GAAG,EAAE,SAAS,EAAE,OAAO,EAAE,CAAC;YAC7D,CAAC;QACF,CAAC;QACD,OAAO,SAAS,CAAC;IAAA,CACjB;IAEO,KAAK,CAAC,WAAW,GAAkB;QAC1C,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC;QAC3B,IAAI,CAAC,MAAM,GAAG,SAAS,CAAC;QACxB,IAAI,MAAM,IAAI,CAAC,MAAM,CAAC,QAAQ,EAAE,CAAC;YAChC,IAAI,CAAC;gBACJ,MAAM,MAAM,CAAC,KAAK,EAAE,CAAC;YACtB,CAAC;YAAC,MAAM,CAAC;gBACR,eAAe;YAChB,CAAC;QACF,CAAC;IAAA,CACD;IAED,kEAAkE;IAClE,KAAK,CAAC,OAAO,GAAkB;QAC9B,IAAI,IAAI,CAAC,QAAQ;YAAE,OAAO;QAC1B,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC;QACrB,IAAI,IAAI,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,EAAE,CAAC;YAC1C,IAAI,CAAC;gBACJ,MAAM,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;YAC1B,CAAC;YAAC,MAAM,CAAC;gBACR,0CAA0C;YAC3C,CAAC;QACF,CAAC;QACD,MAAM,IAAI,CAAC,WAAW,EAAE,CAAC;IAAA,CACzB;CACD;AAED,mCAAmC;AACnC,EAAE;AACF,uEAAuE;AACvE,8EAA8E;AAC9E,8EAA8E;AAC9E,8DAA8D;AAE9D,MAAM,QAAQ,GAAG,IAAI,GAAG,EAA4B,CAAC;AAErD,MAAM,UAAU,wBAAwB,CAAC,GAAW,EAAE,OAAyB,EAAQ;IACtF,MAAM,GAAG,GAAG,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;IAC9B,IAAI,GAAG,IAAI,GAAG,KAAK,OAAO,EAAE,CAAC;QAC5B,GAAG,CAAC,OAAO,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAC,CAAC,CAAC,CAAC;IAC/B,CAAC;IACD,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;AAAA,CAC3B;AAED,MAAM,UAAU,mBAAmB,CAAC,GAAW,EAAgC;IAC9E,OAAO,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;AAAA,CACzB;AAED,MAAM,UAAU,0BAA0B,CAAC,GAAW,EAAQ;IAC7D,QAAQ,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;AAAA,CACrB","sourcesContent":["/**\n * Orchestrates semantic indexing and search for a repository.\n *\n * Lifecycle (all behind --enable-embsearchtools):\n *  1. `start()` — resolve the embsearch binary, scan the repo (ignore-aware),\n *     apply the byte threshold. Under threshold → dormant. Over → spawn the\n *     daemon, verify the backend is not the mock embedder, then index changed\n *     files in the background in small batches, reporting progress.\n *  2. `search()` — top-k semantic query, mapping chunk ids back to\n *     `path:start-end` via the sidecar metadata.\n *  3. `dispose()` — save + close the daemon.\n *\n * Every failure degrades to `unavailable` with a reason; nothing here ever\n * blocks session startup or affects grep/find.\n */\n\nimport { execFileSync } from \"child_process\";\nimport { readFileSync } from \"fs\";\nimport { minimatch } from \"minimatch\";\nimport { ensureTool } from \"../../utils/tools-manager.js\";\nimport { chunkFile } from \"./chunker.js\";\nimport {\n\ttype DaemonRetriever,\n\tEmbSearchClient,\n\ttype EmbSearchRerankPassage,\n\ttype EmbSearchRerankResult,\n} from \"./client.js\";\nimport {\n\temptyIndexMeta,\n\ttype FileMeta,\n\tgetEmbsearchStoreDir,\n\tgetVectorStoreDir,\n\thashContent,\n\ttype IndexMeta,\n\tloadIndexMeta,\n\tsaveIndexMeta,\n} from \"./index-meta.js\";\nimport { type RepoScanFile, scanRepo } from \"./repo-scan.js\";\n\n/** Chunks per bulk request. Small enough that a concurrent query is never\n *  stuck long behind one padded batch inference. */\nconst BULK_BATCH_SIZE = 48;\n/** Yield between batches so background indexing doesn't starve the session. */\nconst BATCH_YIELD_MS = 15;\n/** The Rust mock backend's model id — semantically meaningless, never index with it. */\nconst MOCK_MODEL_ID = \"mock-hash-v1\";\n/**\n * First embsearch release serving `retriever: \"lexical\"`.\n *\n * The guard matters because the daemon does not reject unknown request fields:\n * an older binary silently ignores `retriever` and answers with dense results.\n * Fusing that list a second time as a \"bm25\" leg would double-count it and\n * corrupt the ranking with no error anywhere — so refuse instead of degrading.\n */\nconst MIN_LEXICAL_RETRIEVER_VERSION = [0, 2, 0] as const;\n/**\n * First embsearch release serving the `rerank` op.\n *\n * Necessary but no longer sufficient. Releases from 0.3.1 carry the op without\n * the ~23 MB cross-encoder weights — they measured worse than the\n * deterministic reranker on five of six query classes, so they are no longer\n * bundled — and such a daemon answers `rerank` with an error. The version is\n * therefore only the floor for *asking*; `info.rerank` is the answer, and\n * {@link EmbsearchService.supportsCrossEncoder} needs both.\n */\nconst MIN_RERANK_VERSION = [0, 3, 0] as const;\n\n/** `embsearch 0.2.0` -> [0, 2, 0]; undefined when it cannot be parsed. */\nfunction parseBinaryVersion(output: string): number[] | undefined {\n\tconst match = output.match(/(\\d+)\\.(\\d+)\\.(\\d+)/);\n\treturn match ? [Number(match[1]), Number(match[2]), Number(match[3])] : undefined;\n}\n\nfunction atLeast(version: readonly number[], minimum: readonly number[]): boolean {\n\tfor (let i = 0; i < minimum.length; i++) {\n\t\tconst part = version[i] ?? 0;\n\t\tif (part !== minimum[i]) return part > minimum[i];\n\t}\n\treturn true;\n}\n\nexport type EmbsearchState =\n\t| { phase: \"idle\" }\n\t| { phase: \"skipped\"; reason: string }\n\t| { phase: \"downloading\"; receivedBytes: number; totalBytes: number | null }\n\t| { phase: \"indexing\"; done: number; total: number }\n\t| { phase: \"ready\"; chunkCount: number }\n\t| { phase: \"unavailable\"; reason: string };\n\nexport interface EmbsearchServiceOptions {\n\tcwd: string;\n\t/** Explicit binary path (settings override). Default: \"embsearch\" from PATH. */\n\tbinaryPath?: string;\n\t/** Minimum indexable bytes before indexing kicks in. */\n\tthresholdBytes: number;\n\t/**\n\t * Override the store location. Only the eval harness sets this, so a\n\t * second index (e.g. a BM25-hybrid store) can exist for the same repo\n\t * without colliding with the primary one.\n\t */\n\tstoreDir?: string;\n\t/**\n\t * Create the store with the daemon's BM25 lexical index. Fixed at store\n\t * creation, so this must pair with a distinct `storeDir`.\n\t */\n\thybridStore?: boolean;\n\t/** Progress callback for UI (footer / stderr lines). */\n\tonProgress?: (state: EmbsearchState) => void;\n}\n\nexport interface SemanticHit {\n\tpath: string;\n\tstartLine: number;\n\tendLine: number;\n\tscore: number;\n}\n\nexport interface SemanticChunkHit extends SemanticHit {\n\t/** Per-build chunk id (`relpath#index`) — the fusion identity for hybrid search. */\n\tid: string;\n}\n\nexport class EmbsearchService {\n\tprivate readonly options: EmbsearchServiceOptions;\n\tprivate client: EmbSearchClient | undefined;\n\tprivate meta: IndexMeta | undefined;\n\tprivate state: EmbsearchState = { phase: \"idle\" };\n\tprivate disposed = false;\n\t/** Whether the resolved binary serves `retriever: \"lexical\"`. */\n\tprivate lexicalRetriever = false;\n\t/** Whether the resolved binary serves the cross-encoder `rerank` op. */\n\tprivate crossEncoder = false;\n\n\tconstructor(options: EmbsearchServiceOptions) {\n\t\tthis.options = options;\n\t}\n\n\tgetState(): EmbsearchState {\n\t\treturn this.state;\n\t}\n\n\t/** Semantic search is usable (index ready, or still building with partial data). */\n\tisAvailable(): boolean {\n\t\treturn this.state.phase === \"ready\" || this.state.phase === \"indexing\";\n\t}\n\n\tprivate setState(state: EmbsearchState): void {\n\t\tthis.state = state;\n\t\tthis.options.onProgress?.(state);\n\t}\n\n\t/** Whether the running daemon can serve a BM25-only query. */\n\tsupportsLexicalRetriever(): boolean {\n\t\treturn this.lexicalRetriever;\n\t}\n\n\t/**\n\t * Repo files whose on-disk content the index does not have — unknown to it,\n\t * or changed since it last read them.\n\t *\n\t * This is the set BM25 is structurally blind to, and the only place the\n\t * grep leg still earns its keep once BM25 is available. An agent that edits\n\t * a file and immediately searches for what it wrote is asking about exactly\n\t * these files; the index cannot answer until the next pass.\n\t *\n\t * Compares mtime and size only, never hashing: the check runs per query, and\n\t * a false positive merely lets grep cover a file BM25 already covers, while\n\t * a false negative would lose the edit.\n\t *\n\t * Deliberately uncached. A cache here caches the *absence* of an edit, which\n\t * is the one thing this must never do — an agent writes a file and searches\n\t * for it in the same breath. A 1s TTL was tried and cost the live-edit set\n\t * 75% to 100% of its score depending on how the timing fell, which is worse\n\t * than wrong: it was non-deterministic. One scan is ~25ms over ~1k files and\n\t * happens once per search, against retrieval that already costs more.\n\t */\n\tstaleFiles(signal?: AbortSignal): string[] {\n\t\tif (!this.meta) return [];\n\t\tconst meta = this.meta;\n\t\tconst files: string[] = [];\n\t\ttry {\n\t\t\tfor (const file of scanRepo(this.options.cwd, signal).files) {\n\t\t\t\tconst known = meta.files[file.rel];\n\t\t\t\tif (!known || known.mtimeMs !== file.mtimeMs || known.size !== file.size) files.push(file.rel);\n\t\t\t}\n\t\t} catch {\n\t\t\t// A failed scan must not silently narrow the grep leg to nothing;\n\t\t\t// report no staleness and let the indexed legs answer.\n\t\t\treturn [];\n\t\t}\n\t\treturn files;\n\t}\n\n\t/** Whether the running daemon can score with a cross-encoder. */\n\tsupportsCrossEncoder(): boolean {\n\t\treturn this.crossEncoder;\n\t}\n\n\t/**\n\t * Cross-encoder rerank of caller-supplied passages.\n\t *\n\t * Unlike the retrievers this does not consult the index at all — it scores\n\t * exactly the text passed in, which is why the caller sends its expanded\n\t * windows rather than chunk ids.\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tif (!this.client || this.client.isClosed) {\n\t\t\tthrow new Error(\"semantic index is not available\");\n\t\t}\n\t\tif (!this.crossEncoder) {\n\t\t\tthrow new Error(\n\t\t\t\t`this embsearch daemon cannot rerank (needs >= ${MIN_RERANK_VERSION.join(\".\")} reporting ` +\n\t\t\t\t\t\"`rerank: true`); released binaries ship without cross-encoder weights — start the daemon \" +\n\t\t\t\t\t\"with --reranker-model <dir>\",\n\t\t\t);\n\t\t}\n\t\treturn await this.client.rerank(query, passages, k);\n\t}\n\n\tprivate probeBinaryVersion(binary: string): number[] | undefined {\n\t\ttry {\n\t\t\treturn parseBinaryVersion(execFileSync(binary, [\"--version\"], { encoding: \"utf-8\", timeout: 10_000 }));\n\t\t} catch {\n\t\t\treturn undefined;\n\t\t}\n\t}\n\n\tprivate async resolveBinary(): Promise<string | undefined> {\n\t\tif (this.options.binaryPath) {\n\t\t\treturn this.options.binaryPath;\n\t\t}\n\t\t// Surface the on-demand binary download through the same progress channel as\n\t\t// indexing, so the first-run fetch renders a progress bar instead of a stall.\n\t\t// A cached binary resolves without ever invoking this callback.\n\t\treturn await ensureTool(\"embsearch\", true, (receivedBytes, totalBytes) => {\n\t\t\tthis.setState({ phase: \"downloading\", receivedBytes, totalBytes });\n\t\t});\n\t}\n\n\t/**\n\t * Scan, threshold-check, and (when needed) index in the background.\n\t * Resolves when indexing completes or the feature settles dormant.\n\t */\n\tasync start(signal?: AbortSignal): Promise<void> {\n\t\ttry {\n\t\t\tawait this.run(signal);\n\t\t} catch (e) {\n\t\t\tconst reason = e instanceof Error ? e.message : String(e);\n\t\t\tthis.setState({ phase: \"unavailable\", reason });\n\t\t\tawait this.closeClient();\n\t\t}\n\t}\n\n\tprivate async run(signal?: AbortSignal): Promise<void> {\n\t\tconst binary = await this.resolveBinary();\n\t\tif (!binary) {\n\t\t\tthis.setState({\n\t\t\t\tphase: \"unavailable\",\n\t\t\t\treason: \"embsearch binary not found (PATH or embsearchBinaryPath setting)\",\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst scan = scanRepo(this.options.cwd, signal);\n\t\tif (scan.totalBytes < this.options.thresholdBytes) {\n\t\t\tthis.setState({\n\t\t\t\tphase: \"skipped\",\n\t\t\t\treason: `repo under threshold (${scan.totalBytes} < ${this.options.thresholdBytes} bytes)`,\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst binaryVersion = this.probeBinaryVersion(binary);\n\t\tthis.lexicalRetriever = binaryVersion !== undefined && atLeast(binaryVersion, MIN_LEXICAL_RETRIEVER_VERSION);\n\t\t// Provisional: the version says the daemon understands `rerank`. Whether\n\t\t// it can serve one is answered by `info` below, once the client is up.\n\t\tthis.crossEncoder = binaryVersion !== undefined && atLeast(binaryVersion, MIN_RERANK_VERSION);\n\n\t\tconst storeDir = this.options.storeDir ?? getEmbsearchStoreDir(this.options.cwd);\n\t\tthis.client = new EmbSearchClient({\n\t\t\tbinaryPath: binary,\n\t\t\tstorePath: getVectorStoreDir(storeDir),\n\t\t\thybrid: this.options.hybridStore,\n\t\t});\n\t\tawait this.client.ready();\n\n\t\tconst info = await this.client.info();\n\t\t// A daemon old enough to omit the field is left on the version verdict —\n\t\t// back then the weights were bundled, so version did imply capability.\n\t\tif (info.rerank === false) this.crossEncoder = false;\n\t\tif (info.modelId === MOCK_MODEL_ID) {\n\t\t\tthrow new Error(\"embsearch binary uses the mock embedder (not semantic); install an onnx build\");\n\t\t}\n\n\t\t// Missing/stale sidecar (format, chunker, or model changed) → clean rebuild.\n\t\tthis.meta = loadIndexMeta(storeDir, info.modelId) ?? emptyIndexMeta(this.options.cwd, info.modelId);\n\t\tthis.meta.lastUsedMs = Date.now();\n\n\t\tawait this.indexChangedFiles(scan.files, storeDir, signal);\n\t}\n\n\tprivate async indexChangedFiles(files: RepoScanFile[], storeDir: string, signal?: AbortSignal): Promise<void> {\n\t\tconst meta = this.meta!;\n\t\tconst client = this.client!;\n\n\t\t// Diff scan vs sidecar: cheap mtime+size check first, hash only on delta.\n\t\tconst toIndex: Array<{ file: RepoScanFile; content: string; hash: string }> = [];\n\t\tconst seen = new Set<string>();\n\t\tfor (const file of files) {\n\t\t\tseen.add(file.rel);\n\t\t\tconst known = meta.files[file.rel];\n\t\t\tif (known && known.mtimeMs === file.mtimeMs && known.size === file.size) continue;\n\t\t\tlet content: string;\n\t\t\ttry {\n\t\t\t\tcontent = readFileSync(file.abs, \"utf-8\");\n\t\t\t} catch {\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tconst hash = hashContent(content);\n\t\t\tif (known && known.hash === hash) {\n\t\t\t\t// Touched but unchanged — refresh stat info only.\n\t\t\t\tknown.mtimeMs = file.mtimeMs;\n\t\t\t\tknown.size = file.size;\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\ttoIndex.push({ file, content, hash });\n\t\t}\n\t\tconst toRemove = Object.keys(meta.files).filter((rel) => !seen.has(rel));\n\n\t\tif (toIndex.length === 0 && toRemove.length === 0) {\n\t\t\tsaveIndexMeta(storeDir, meta);\n\t\t\tthis.setState({ phase: \"ready\", chunkCount: this.countChunks(meta) });\n\t\t\treturn;\n\t\t}\n\n\t\t// Chunk changed files; count total upserts for exact progress.\n\t\tconst work: Array<{ rel: string; fileMeta: FileMeta; chunks: Array<{ id: string; text: string }> }> = [];\n\t\tlet totalChunks = 0;\n\t\tfor (const { file, content, hash } of toIndex) {\n\t\t\tconst chunks = chunkFile(file.rel, content);\n\t\t\twork.push({\n\t\t\t\trel: file.rel,\n\t\t\t\tfileMeta: {\n\t\t\t\t\tmtimeMs: file.mtimeMs,\n\t\t\t\t\tsize: file.size,\n\t\t\t\t\thash,\n\t\t\t\t\tchunks: chunks.map((c) => [c.startLine, c.endLine]),\n\t\t\t\t},\n\t\t\t\tchunks: chunks.map((c) => ({ id: c.id, text: c.text })),\n\t\t\t});\n\t\t\ttotalChunks += chunks.length;\n\t\t}\n\n\t\tthis.setState({ phase: \"indexing\", done: 0, total: totalChunks });\n\n\t\t// Drop vectors of deleted files and superseded chunk tails.\n\t\tfor (const rel of toRemove) {\n\t\t\tfor (let i = 0; i < meta.files[rel].chunks.length; i++) await client.remove(`${rel}#${i}`);\n\t\t\tdelete meta.files[rel];\n\t\t}\n\n\t\tlet done = 0;\n\t\tfor (const item of work) {\n\t\t\tif (signal?.aborted || this.disposed) return;\n\t\t\tconst oldChunkCount = meta.files[item.rel]?.chunks.length ?? 0;\n\t\t\t// Remove old chunks beyond the new count (upsert covers the rest).\n\t\t\tfor (let i = item.chunks.length; i < oldChunkCount; i++) await client.remove(`${item.rel}#${i}`);\n\n\t\t\tfor (let offset = 0; offset < item.chunks.length; offset += BULK_BATCH_SIZE) {\n\t\t\t\tif (signal?.aborted || this.disposed) return;\n\t\t\t\tconst batch = item.chunks.slice(offset, offset + BULK_BATCH_SIZE);\n\t\t\t\tawait client.bulk(batch);\n\t\t\t\tdone += batch.length;\n\t\t\t\tthis.setState({ phase: \"indexing\", done, total: totalChunks });\n\t\t\t\t// Yield so queries and the event loop stay responsive.\n\t\t\t\tawait new Promise((resolve) => setTimeout(resolve, BATCH_YIELD_MS));\n\t\t\t}\n\t\t\tmeta.files[item.rel] = item.fileMeta;\n\t\t}\n\n\t\tawait client.compact();\n\t\tawait client.save();\n\t\tsaveIndexMeta(storeDir, meta);\n\t\tthis.setState({ phase: \"ready\", chunkCount: this.countChunks(meta) });\n\t}\n\n\tprivate countChunks(meta: IndexMeta): number {\n\t\tlet n = 0;\n\t\tfor (const rel of Object.keys(meta.files)) n += meta.files[rel].chunks.length;\n\t\treturn n;\n\t}\n\n\t/** Top-`k` semantic hits as `path` + line range + score. */\n\tasync search(query: string, k = 10): Promise<SemanticHit[]> {\n\t\treturn await this.searchChunks(query, k);\n\t}\n\n\t/** Top-`k` semantic hits including their chunk ids, for rank fusion. */\n\tasync searchChunks(\n\t\tquery: string,\n\t\tk = 10,\n\t\tglob?: string,\n\t\tretriever: DaemonRetriever = \"dense\",\n\t): Promise<SemanticChunkHit[]> {\n\t\tif (!this.client || this.client.isClosed || !this.meta) {\n\t\t\tthrow new Error(\"semantic index is not available\");\n\t\t}\n\t\tif (retriever === \"lexical\" && !this.lexicalRetriever) {\n\t\t\tthrow new Error(\n\t\t\t\t`embsearch is too old for retriever \"lexical\" (needs >= ${MIN_LEXICAL_RETRIEVER_VERSION.join(\".\")}); ` +\n\t\t\t\t\t\"an older daemon ignores the field and answers with dense results\",\n\t\t\t);\n\t\t}\n\t\tconst results = await this.client.query(query, k, retriever);\n\t\tconst hits: SemanticChunkHit[] = [];\n\t\tconst matchGlob = (rel: string): boolean => {\n\t\t\tif (!glob) return true;\n\t\t\treturn minimatch(rel, glob, { dot: true, matchBase: !glob.includes(\"/\") });\n\t\t};\n\t\tfor (const result of results) {\n\t\t\tconst sep = result.id.lastIndexOf(\"#\");\n\t\t\tif (sep === -1) continue;\n\t\t\tconst rel = result.id.slice(0, sep);\n\t\t\tif (!matchGlob(rel)) continue;\n\t\t\tconst chunkIndex = Number.parseInt(result.id.slice(sep + 1), 10);\n\t\t\tconst range = this.meta.files[rel]?.chunks[chunkIndex];\n\t\t\tif (!range) continue;\n\t\t\thits.push({ id: result.id, path: rel, startLine: range[0], endLine: range[1], score: result.score });\n\t\t}\n\t\treturn hits;\n\t}\n\n\t/**\n\t * Resolve a repo-relative path + line to its enclosing indexed chunk, or\n\t * undefined when the file/line is not covered by the index. Chunks overlap\n\t * by a few lines; the first (lowest-index) containing chunk wins so the\n\t * mapping is deterministic.\n\t */\n\tfindEnclosingChunk(\n\t\trel: string,\n\t\tline: number,\n\t): { id: string; path: string; startLine: number; endLine: number } | undefined {\n\t\tconst file = this.meta?.files[rel];\n\t\tif (!file) return undefined;\n\t\tfor (let i = 0; i < file.chunks.length; i++) {\n\t\t\tconst [startLine, endLine] = file.chunks[i];\n\t\t\tif (line >= startLine && line <= endLine) {\n\t\t\t\treturn { id: `${rel}#${i}`, path: rel, startLine, endLine };\n\t\t\t}\n\t\t}\n\t\treturn undefined;\n\t}\n\n\tprivate async closeClient(): Promise<void> {\n\t\tconst client = this.client;\n\t\tthis.client = undefined;\n\t\tif (client && !client.isClosed) {\n\t\t\ttry {\n\t\t\t\tawait client.close();\n\t\t\t} catch {\n\t\t\t\t// already dead\n\t\t\t}\n\t\t}\n\t}\n\n\t/** Persist state and shut the daemon down. Safe to call twice. */\n\tasync dispose(): Promise<void> {\n\t\tif (this.disposed) return;\n\t\tthis.disposed = true;\n\t\tif (this.client && !this.client.isClosed) {\n\t\t\ttry {\n\t\t\t\tawait this.client.save();\n\t\t\t} catch {\n\t\t\t\t// daemon may have exited; nothing to save\n\t\t\t}\n\t\t}\n\t\tawait this.closeClient();\n\t}\n}\n\n// --- Per-cwd service registry ---\n//\n// The search tool is constructed by the generic tool factory table and\n// only receives `cwd`; the service is created later during session init (flag\n// gated). This registry connects the two without threading a service instance\n// through every layer between main.ts and the tool factories.\n\nconst services = new Map<string, EmbsearchService>();\n\nexport function registerEmbsearchService(cwd: string, service: EmbsearchService): void {\n\tconst old = services.get(cwd);\n\tif (old && old !== service) {\n\t\told.dispose().catch(() => {});\n\t}\n\tservices.set(cwd, service);\n}\n\nexport function getEmbsearchService(cwd: string): EmbsearchService | undefined {\n\treturn services.get(cwd);\n}\n\nexport function unregisterEmbsearchService(cwd: string): void {\n\tservices.delete(cwd);\n}\n"]}

package/dist/core/search/eval.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"eval.d.ts","sourceRoot":"","sources":["../../../src/core/search/eval.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAE1E,OAAO,KAAK,EAAE,aAAa,EAAE,kBAAkB,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AAEhF,MAAM,WAAW,YAAY;IAC5B,gCAAgC;IAChC,IAAI,EAAE,MAAM,CAAC;IACb,mEAAmE;IACnE,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB;;;;;;OAMG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB;;;;;;OAMG;IACH,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CACxB;AAED,MAAM,WAAW,SAAS;IACzB,EAAE,EAAE,MAAM,CAAC;IACX;6DACyD;IACzD,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,YAAY,EAAE,CAAC;CACrB;AAED,MAAM,WAAW,UAAU;IAC1B,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,UAAU,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;OAKG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;;;OAIG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;iEAC6D;IAC7D,YAAY,CAAC,EAAE,OAAO,CAAC;CACvB;AAED;;;;;;;;;;GAUG;AACH,eAAO,MAAM,YAAY,EAAE,SAAS,UAAU,EA6B7C,CAAC;AAKF,wBAAgB,eAAe,CAAC,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,YAAY,GAAG,OAAO,CAKhF;AAED,gFAAgF;AAChF,wBAAgB,SAAS,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,EAAE,IAAI,EAAE,SAAS,YAAY,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,MAAM,CAQhH;AAED;;;;;;;;GAQG;AACH,wBAAgB,GAAG,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,EAAE,IAAI,EAAE,SAAS,YAAY,EAAE,GAAG,MAAM,CAQ/F;AAED,MAAM,WAAW,eAAe;IAC/B,KAAK,EAAE,MAAM,CAAC;IACd,YAAY,EAAE,kBAAkB,CAAC;IACjC,QAAQ,EAAE,OAAO,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;IACnB,GAAG,EAAE,MAAM,CAAC;CACZ;AAED,wBAAsB,aAAa,CAClC,GAAG,EAAE,MAAM,EACX,SAAS,EAAE,SAAS,EACpB,OAAO,GAAE,SAAS,UAAU,EAAiB,EAC7C,OAAO,CAAC,EAAE,gBAAgB,EAC1B,aAAa,CAAC,EAAE,gBAAgB,GAC9B,OAAO,CAAC,eAAe,EAAE,CAAC,CAiC5B","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.d.ts","sourceRoot":"","sources":["../../../src/core/search/eval.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAE1E,OAAO,KAAK,EAAE,aAAa,EAAE,kBAAkB,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AAEhF,MAAM,WAAW,YAAY;IAC5B,gCAAgC;IAChC,IAAI,EAAE,MAAM,CAAC;IACb,mEAAmE;IACnE,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB;;;;;;OAMG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB;;;;;;OAMG;IACH,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CACxB;AAED,MAAM,WAAW,SAAS;IACzB,EAAE,EAAE,MAAM,CAAC;IACX;6DACyD;IACzD,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,YAAY,EAAE,CAAC;CACrB;AAED,MAAM,WAAW,UAAU;IAC1B,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,UAAU,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;OAKG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;;;OAIG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;iEAC6D;IAC7D,YAAY,CAAC,EAAE,OAAO,CAAC;CACvB;AAED;;;;;;;;;;GAUG;AACH,eAAO,MAAM,YAAY,EAAE,SAAS,UAAU,EAoC7C,CAAC;AAKF,wBAAgB,eAAe,CAAC,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,YAAY,GAAG,OAAO,CAKhF;AAED,gFAAgF;AAChF,wBAAgB,SAAS,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,EAAE,IAAI,EAAE,SAAS,YAAY,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,MAAM,CAQhH;AAED;;;;;;;;GAQG;AACH,wBAAgB,GAAG,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,EAAE,IAAI,EAAE,SAAS,YAAY,EAAE,GAAG,MAAM,CAQ/F;AAED,MAAM,WAAW,eAAe;IAC/B,KAAK,EAAE,MAAM,CAAC;IACd,YAAY,EAAE,kBAAkB,CAAC;IACjC,QAAQ,EAAE,OAAO,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;IACnB,GAAG,EAAE,MAAM,CAAC;CACZ;AAED,wBAAsB,aAAa,CAClC,GAAG,EAAE,MAAM,EACX,SAAS,EAAE,SAAS,EACpB,OAAO,GAAE,SAAS,UAAU,EAAiB,EAC7C,OAAO,CAAC,EAAE,gBAAgB,EAC1B,aAAa,CAAC,EAAE,gBAAgB,GAC9B,OAAO,CAAC,eAAe,EAAE,CAAC,CAiC5B","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"]}