@kolisachint/hoocode-agent 0.5.59 → 0.5.60

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

@@ -80,6 +80,15 @@ export class EmbsearchService {
         return this.state;
     }
     /** Semantic search is usable (index ready, or still building with partial data). */
+    /**
+     * Model id reported by the running daemon, once it is up.
+     *
+     * This — not the binary's version — identifies which model produced the
+     * vectors in the store, because `--model` decouples the two.
+     */
+    modelId() {
+        return this.meta?.modelId;
+    }
     isAvailable() {
         return this.state.phase === "ready" || this.state.phase === "indexing";
     }
@@ -163,6 +172,32 @@ export class EmbsearchService {
             return undefined;
         }
     }
+    /**
+     * What the store on disk says about itself, read straight from its manifest.
+     *
+     * `store-info` exists precisely for the case where the daemon will not open
+     * the store: a `serve` pairs a store with an embedder and refuses the pair
+     * when their models disagree, so at that moment nothing else can tell us
+     * what built it. Returns undefined when there is no readable store — which
+     * includes a binary too old to have the subcommand, and so degrades to the
+     * previous behaviour rather than guessing.
+     */
+    probeStore(binary, storeDir) {
+        try {
+            const out = execFileSync(binary, ["store-info", "--path", getVectorStoreDir(storeDir), "--json"], {
+                encoding: "utf-8",
+                timeout: 10_000,
+                stdio: ["ignore", "pipe", "ignore"],
+            });
+            const parsed = JSON.parse(out);
+            if (typeof parsed.model_id !== "string")
+                return undefined;
+            return { modelId: parsed.model_id, live: parsed.live ?? 0 };
+        }
+        catch {
+            return undefined;
+        }
+    }
     async resolveBinary() {
         if (this.options.binaryPath) {
             return this.options.binaryPath;
@@ -220,11 +255,41 @@ export class EmbsearchService {
                 binaryPath: binary,
                 storePath: getVectorStoreDir(storeDir),
                 hybrid: wantHybrid,
+                modelDir: this.options.modelDir,
             });
             await this.client.ready();
             return await this.client.info();
         };
-        let info = await openClient();
+        /** Discard the store and start clean. The only recovery from a store the
+         *  current binary cannot use — and the only way to be rid of vectors that
+         *  outlived the metadata describing them. */
+        const rebuildFrom = async (why) => {
+            console.error(`embsearch: rebuilding the index (${why})`);
+            this.setState({ phase: "indexing", done: 0, total: 0 });
+            await this.closeClient();
+            rmSync(storeDir, { recursive: true, force: true });
+            return await openClient();
+        };
+        let info;
+        try {
+            info = await openClient();
+        }
+        catch (err) {
+            // The daemon refuses to open a store whose recorded model disagrees
+            // with its own — correctly, since vectors from different models are
+            // not comparable. But refusing is where it stopped: the store stayed
+            // on disk, the daemon never came up, and this service went
+            // permanently unavailable with a rebuild one directory-removal away.
+            //
+            // Only a store that is present and *readable* is treated this way. If
+            // `store-info` cannot read it either, the problem is not a model
+            // mismatch and destroying an index would be the wrong response, so
+            // the original failure stands.
+            const store = this.probeStore(binary, storeDir);
+            if (!store)
+                throw err;
+            info = await rebuildFrom(`built by model '${store.modelId}', which this binary cannot read`);
+        }
         // Hybrid-ness is fixed when a store is created and `--hybrid` against an
         // existing plain store only warns, so an index built before this was the
         // default would silently stay dense-only and every BM25 query against it
@@ -232,10 +297,7 @@ export class EmbsearchService {
         // rebuild once when it disagrees. `info.hybrid` is undefined on daemons
         // too old to report it — those cannot serve BM25 anyway, so leave them be.
         if (wantHybrid && info.hybrid === false) {
-            this.setState({ phase: "indexing", done: 0, total: 0 });
-            await this.closeClient();
-            rmSync(storeDir, { recursive: true, force: true });
-            info = await openClient();
+            info = await rebuildFrom("the existing store carries no BM25 index");
         }
         this.hybridStore = info.hybrid === true;
         // A daemon old enough to omit the field is left on the version verdict —
@@ -246,7 +308,24 @@ export class EmbsearchService {
             throw new Error("embsearch binary uses the mock embedder (not semantic); install an onnx build");
         }
         // Missing/stale sidecar (format, chunker, or model changed) → clean rebuild.
-        this.meta = loadIndexMeta(storeDir, info.modelId) ?? emptyIndexMeta(this.options.cwd, info.modelId);
+        //
+        // Resetting the sidecar alone is not enough, and used to be all this did.
+        // The sidecar is the only record of how many chunks each file produced,
+        // so an empty one reports zero for every file — and `indexChangedFiles`
+        // removes stale chunks by counting down from that number. Re-chunking a
+        // file into *fewer* pieces then leaves its tail vectors (`path#N`,
+        // `path#N+1`, …) in the store, holding text that no longer exists
+        // anywhere, retrievable forever. Upserts hide it: chunk counts look
+        // right, the sidecar looks right, and only search results are wrong.
+        //
+        // So when the sidecar cannot be trusted and the store is not already
+        // empty, the store goes too.
+        let meta = loadIndexMeta(storeDir, info.modelId);
+        if (!meta && info.count > 0) {
+            info = await rebuildFrom("index metadata is missing or was written by a different chunker or model");
+            meta = loadIndexMeta(storeDir, info.modelId);
+        }
+        this.meta = meta ?? emptyIndexMeta(this.options.cwd, info.modelId);
         this.meta.lastUsedMs = Date.now();
         await this.indexChangedFiles(scan.files, storeDir, signal);
     }
@@ -287,7 +366,7 @@ export class EmbsearchService {
         const work = [];
         let totalChunks = 0;
         for (const { file, content, hash } of toIndex) {
-            const chunks = chunkFile(file.rel, content);
+            const chunks = chunkFile(file.rel, content, this.options.chunkMaxChars);
             work.push({
                 rel: file.rel,
                 fileMeta: {

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,EAAE,MAAM,IAAI,CAAC;AAC1C,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,GAIf,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;AA+CD,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,8DAA8D;IACtD,WAAW,GAAG,KAAK,CAAC;IAC5B,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;;;OAGG;IACH,wBAAwB,GAAY;QACnC,OAAO,IAAI,CAAC,gBAAgB,IAAI,IAAI,CAAC,WAAW,CAAC;IAAA,CACjD;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,yEAAyE;QACzE,0EAA0E;QAC1E,wEAAwE;QACxE,MAAM,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,WAAW,IAAI,IAAI,CAAC,gBAAgB,CAAC;QAErE,MAAM,UAAU,GAAG,KAAK,IAAkC,EAAE,CAAC;YAC5D,IAAI,CAAC,MAAM,GAAG,IAAI,eAAe,CAAC;gBACjC,UAAU,EAAE,MAAM;gBAClB,SAAS,EAAE,iBAAiB,CAAC,QAAQ,CAAC;gBACtC,MAAM,EAAE,UAAU;aAClB,CAAC,CAAC;YACH,MAAM,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,CAAC;YAC1B,OAAO,MAAM,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QAAA,CAChC,CAAC;QAEF,IAAI,IAAI,GAAG,MAAM,UAAU,EAAE,CAAC;QAE9B,yEAAyE;QACzE,yEAAyE;QACzE,yEAAyE;QACzE,yEAAyE;QACzE,wEAAwE;QACxE,6EAA2E;QAC3E,IAAI,UAAU,IAAI,IAAI,CAAC,MAAM,KAAK,KAAK,EAAE,CAAC;YACzC,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;YACxD,MAAM,IAAI,CAAC,WAAW,EAAE,CAAC;YACzB,MAAM,CAAC,QAAQ,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;YACnD,IAAI,GAAG,MAAM,UAAU,EAAE,CAAC;QAC3B,CAAC;QACD,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC;QACxC,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-semantic-index):\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, rmSync } 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 EmbSearchDaemonInfo,\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.\n\t *\n\t * Defaults to whatever the daemon can serve. Fixed at store creation, so an\n\t * existing store that disagrees is rebuilt once; when overriding this to\n\t * hold two different stores for one repo, pair it with a distinct\n\t * `storeDir` so they do not fight over the same directory.\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 store actually opened carries a BM25 index. */\n\tprivate hybridStore = 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/**\n\t * Whether a BM25-only query will work: the daemon has to understand the\n\t * `lexical` retriever *and* the open store has to carry a BM25 index.\n\t */\n\tsupportsLexicalRetriever(): boolean {\n\t\treturn this.lexicalRetriever && this.hybridStore;\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\t// A hybrid store carries a BM25 index next to its vectors, which is what\n\t\t// lets search use BM25 as its lexical leg instead of ripgrep. Callers may\n\t\t// force it either way; by default it follows what the daemon can serve.\n\t\tconst wantHybrid = this.options.hybridStore ?? this.lexicalRetriever;\n\n\t\tconst openClient = async (): Promise<EmbSearchDaemonInfo> => {\n\t\t\tthis.client = new EmbSearchClient({\n\t\t\t\tbinaryPath: binary,\n\t\t\t\tstorePath: getVectorStoreDir(storeDir),\n\t\t\t\thybrid: wantHybrid,\n\t\t\t});\n\t\t\tawait this.client.ready();\n\t\t\treturn await this.client.info();\n\t\t};\n\n\t\tlet info = await openClient();\n\n\t\t// Hybrid-ness is fixed when a store is created and `--hybrid` against an\n\t\t// existing plain store only warns, so an index built before this was the\n\t\t// default would silently stay dense-only and every BM25 query against it\n\t\t// would fail. Ask the store itself rather than trusting the sidecar, and\n\t\t// rebuild once when it disagrees. `info.hybrid` is undefined on daemons\n\t\t// too old to report it — those cannot serve BM25 anyway, so leave them be.\n\t\tif (wantHybrid && info.hybrid === false) {\n\t\t\tthis.setState({ phase: \"indexing\", done: 0, total: 0 });\n\t\t\tawait this.closeClient();\n\t\t\trmSync(storeDir, { recursive: true, force: true });\n\t\t\tinfo = await openClient();\n\t\t}\n\t\tthis.hybridStore = info.hybrid === true;\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"]}
+{"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,EAAE,MAAM,IAAI,CAAC;AAC1C,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,GAIf,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;AAkED,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,8DAA8D;IACtD,WAAW,GAAG,KAAK,CAAC;IAC5B,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;;;;;OAKG;IACH,OAAO,GAAuB;QAC7B,OAAO,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC;IAAA,CAC1B;IAED,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;;;OAGG;IACH,wBAAwB,GAAY;QACnC,OAAO,IAAI,CAAC,gBAAgB,IAAI,IAAI,CAAC,WAAW,CAAC;IAAA,CACjD;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;IAED;;;;;;;;;OASG;IACK,UAAU,CAAC,MAAc,EAAE,QAAgB,EAAiD;QACnG,IAAI,CAAC;YACJ,MAAM,GAAG,GAAG,YAAY,CAAC,MAAM,EAAE,CAAC,YAAY,EAAE,QAAQ,EAAE,iBAAiB,CAAC,QAAQ,CAAC,EAAE,QAAQ,CAAC,EAAE;gBACjG,QAAQ,EAAE,OAAO;gBACjB,OAAO,EAAE,MAAM;gBACf,KAAK,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,QAAQ,CAAC;aACnC,CAAC,CAAC;YACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAyC,CAAC;YACvE,IAAI,OAAO,MAAM,CAAC,QAAQ,KAAK,QAAQ;gBAAE,OAAO,SAAS,CAAC;YAC1D,OAAO,EAAE,OAAO,EAAE,MAAM,CAAC,QAAQ,EAAE,IAAI,EAAE,MAAM,CAAC,IAAI,IAAI,CAAC,EAAE,CAAC;QAC7D,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,yEAAyE;QACzE,0EAA0E;QAC1E,wEAAwE;QACxE,MAAM,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,WAAW,IAAI,IAAI,CAAC,gBAAgB,CAAC;QAErE,MAAM,UAAU,GAAG,KAAK,IAAkC,EAAE,CAAC;YAC5D,IAAI,CAAC,MAAM,GAAG,IAAI,eAAe,CAAC;gBACjC,UAAU,EAAE,MAAM;gBAClB,SAAS,EAAE,iBAAiB,CAAC,QAAQ,CAAC;gBACtC,MAAM,EAAE,UAAU;gBAClB,QAAQ,EAAE,IAAI,CAAC,OAAO,CAAC,QAAQ;aAC/B,CAAC,CAAC;YACH,MAAM,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,CAAC;YAC1B,OAAO,MAAM,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QAAA,CAChC,CAAC;QAEF;;qDAE6C;QAC7C,MAAM,WAAW,GAAG,KAAK,EAAE,GAAW,EAAgC,EAAE,CAAC;YACxE,OAAO,CAAC,KAAK,CAAC,oCAAoC,GAAG,GAAG,CAAC,CAAC;YAC1D,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;YACxD,MAAM,IAAI,CAAC,WAAW,EAAE,CAAC;YACzB,MAAM,CAAC,QAAQ,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;YACnD,OAAO,MAAM,UAAU,EAAE,CAAC;QAAA,CAC1B,CAAC;QAEF,IAAI,IAAyB,CAAC;QAC9B,IAAI,CAAC;YACJ,IAAI,GAAG,MAAM,UAAU,EAAE,CAAC;QAC3B,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACd,oEAAoE;YACpE,sEAAoE;YACpE,qEAAqE;YACrE,2DAA2D;YAC3D,qEAAqE;YACrE,EAAE;YACF,sEAAsE;YACtE,iEAAiE;YACjE,mEAAmE;YACnE,+BAA+B;YAC/B,MAAM,KAAK,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;YAChD,IAAI,CAAC,KAAK;gBAAE,MAAM,GAAG,CAAC;YACtB,IAAI,GAAG,MAAM,WAAW,CAAC,mBAAmB,KAAK,CAAC,OAAO,kCAAkC,CAAC,CAAC;QAC9F,CAAC;QAED,yEAAyE;QACzE,yEAAyE;QACzE,yEAAyE;QACzE,yEAAyE;QACzE,wEAAwE;QACxE,6EAA2E;QAC3E,IAAI,UAAU,IAAI,IAAI,CAAC,MAAM,KAAK,KAAK,EAAE,CAAC;YACzC,IAAI,GAAG,MAAM,WAAW,CAAC,0CAA0C,CAAC,CAAC;QACtE,CAAC;QACD,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC;QACxC,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,EAAE;QACF,0EAA0E;QAC1E,wEAAwE;QACxE,0EAAwE;QACxE,wEAAwE;QACxE,mEAAmE;QACnE,oEAAkE;QAClE,oEAAoE;QACpE,qEAAqE;QACrE,EAAE;QACF,qEAAqE;QACrE,6BAA6B;QAC7B,IAAI,IAAI,GAAG,aAAa,CAAC,QAAQ,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;QACjD,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC;YAC7B,IAAI,GAAG,MAAM,WAAW,CAAC,0EAA0E,CAAC,CAAC;YACrG,IAAI,GAAG,aAAa,CAAC,QAAQ,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;QAC9C,CAAC;QACD,IAAI,CAAC,IAAI,GAAG,IAAI,IAAI,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;QACnE,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,EAAE,IAAI,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC;YACxE,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-semantic-index):\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, rmSync } 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 EmbSearchDaemonInfo,\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/**\n\t * Model directory handed to the daemon as `--model`, overriding the model\n\t * bundled in the binary.\n\t *\n\t * Only the eval harness sets this, to score two embedding models from one\n\t * binary. Pair it with a distinct `storeDir`: vectors from different models\n\t * are incompatible, and the daemon refuses to open a store built by another\n\t * model rather than mixing them.\n\t */\n\tmodelDir?: string;\n\t/**\n\t * Override the chunker's character cap.\n\t *\n\t * Only the eval harness sets this, to sweep the chunk window. It changes\n\t * what every vector in the store *is*, and nothing stored records it, so it\n\t * must be paired with a distinct `storeDir` exactly as `modelDir` is —\n\t * otherwise a run silently scores an index built at another cap.\n\t */\n\tchunkMaxChars?: number;\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.\n\t *\n\t * Defaults to whatever the daemon can serve. Fixed at store creation, so an\n\t * existing store that disagrees is rebuilt once; when overriding this to\n\t * hold two different stores for one repo, pair it with a distinct\n\t * `storeDir` so they do not fight over the same directory.\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 store actually opened carries a BM25 index. */\n\tprivate hybridStore = 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\t/**\n\t * Model id reported by the running daemon, once it is up.\n\t *\n\t * This — not the binary's version — identifies which model produced the\n\t * vectors in the store, because `--model` decouples the two.\n\t */\n\tmodelId(): string | undefined {\n\t\treturn this.meta?.modelId;\n\t}\n\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/**\n\t * Whether a BM25-only query will work: the daemon has to understand the\n\t * `lexical` retriever *and* the open store has to carry a BM25 index.\n\t */\n\tsupportsLexicalRetriever(): boolean {\n\t\treturn this.lexicalRetriever && this.hybridStore;\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\t/**\n\t * What the store on disk says about itself, read straight from its manifest.\n\t *\n\t * `store-info` exists precisely for the case where the daemon will not open\n\t * the store: a `serve` pairs a store with an embedder and refuses the pair\n\t * when their models disagree, so at that moment nothing else can tell us\n\t * what built it. Returns undefined when there is no readable store — which\n\t * includes a binary too old to have the subcommand, and so degrades to the\n\t * previous behaviour rather than guessing.\n\t */\n\tprivate probeStore(binary: string, storeDir: string): { modelId: string; live: number } | undefined {\n\t\ttry {\n\t\t\tconst out = execFileSync(binary, [\"store-info\", \"--path\", getVectorStoreDir(storeDir), \"--json\"], {\n\t\t\t\tencoding: \"utf-8\",\n\t\t\t\ttimeout: 10_000,\n\t\t\t\tstdio: [\"ignore\", \"pipe\", \"ignore\"],\n\t\t\t});\n\t\t\tconst parsed = JSON.parse(out) as { model_id?: string; live?: number };\n\t\t\tif (typeof parsed.model_id !== \"string\") return undefined;\n\t\t\treturn { modelId: parsed.model_id, live: parsed.live ?? 0 };\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} < ${this.options.thresholdBytes} bytes)`,\n\t\t\t});\n\t\t\treturn;\n\t\t}\n\n\t\tconst binaryVersion = this.probeBinaryVersion(binary);\n\t\tthis.lexicalRetriever = binaryVersion !== undefined && atLeast(binaryVersion, MIN_LEXICAL_RETRIEVER_VERSION);\n\t\t// Provisional: the version says the daemon understands `rerank`. Whether\n\t\t// it can serve one is answered by `info` below, once the client is up.\n\t\tthis.crossEncoder = binaryVersion !== undefined && atLeast(binaryVersion, MIN_RERANK_VERSION);\n\n\t\tconst storeDir = this.options.storeDir ?? getEmbsearchStoreDir(this.options.cwd);\n\t\t// A hybrid store carries a BM25 index next to its vectors, which is what\n\t\t// lets search use BM25 as its lexical leg instead of ripgrep. Callers may\n\t\t// force it either way; by default it follows what the daemon can serve.\n\t\tconst wantHybrid = this.options.hybridStore ?? this.lexicalRetriever;\n\n\t\tconst openClient = async (): Promise<EmbSearchDaemonInfo> => {\n\t\t\tthis.client = new EmbSearchClient({\n\t\t\t\tbinaryPath: binary,\n\t\t\t\tstorePath: getVectorStoreDir(storeDir),\n\t\t\t\thybrid: wantHybrid,\n\t\t\t\tmodelDir: this.options.modelDir,\n\t\t\t});\n\t\t\tawait this.client.ready();\n\t\t\treturn await this.client.info();\n\t\t};\n\n\t\t/** Discard the store and start clean. The only recovery from a store the\n\t\t *  current binary cannot use — and the only way to be rid of vectors that\n\t\t *  outlived the metadata describing them. */\n\t\tconst rebuildFrom = async (why: string): Promise<EmbSearchDaemonInfo> => {\n\t\t\tconsole.error(`embsearch: rebuilding the index (${why})`);\n\t\t\tthis.setState({ phase: \"indexing\", done: 0, total: 0 });\n\t\t\tawait this.closeClient();\n\t\t\trmSync(storeDir, { recursive: true, force: true });\n\t\t\treturn await openClient();\n\t\t};\n\n\t\tlet info: EmbSearchDaemonInfo;\n\t\ttry {\n\t\t\tinfo = await openClient();\n\t\t} catch (err) {\n\t\t\t// The daemon refuses to open a store whose recorded model disagrees\n\t\t\t// with its own — correctly, since vectors from different models are\n\t\t\t// not comparable. But refusing is where it stopped: the store stayed\n\t\t\t// on disk, the daemon never came up, and this service went\n\t\t\t// permanently unavailable with a rebuild one directory-removal away.\n\t\t\t//\n\t\t\t// Only a store that is present and *readable* is treated this way. If\n\t\t\t// `store-info` cannot read it either, the problem is not a model\n\t\t\t// mismatch and destroying an index would be the wrong response, so\n\t\t\t// the original failure stands.\n\t\t\tconst store = this.probeStore(binary, storeDir);\n\t\t\tif (!store) throw err;\n\t\t\tinfo = await rebuildFrom(`built by model '${store.modelId}', which this binary cannot read`);\n\t\t}\n\n\t\t// Hybrid-ness is fixed when a store is created and `--hybrid` against an\n\t\t// existing plain store only warns, so an index built before this was the\n\t\t// default would silently stay dense-only and every BM25 query against it\n\t\t// would fail. Ask the store itself rather than trusting the sidecar, and\n\t\t// rebuild once when it disagrees. `info.hybrid` is undefined on daemons\n\t\t// too old to report it — those cannot serve BM25 anyway, so leave them be.\n\t\tif (wantHybrid && info.hybrid === false) {\n\t\t\tinfo = await rebuildFrom(\"the existing store carries no BM25 index\");\n\t\t}\n\t\tthis.hybridStore = info.hybrid === true;\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\t//\n\t\t// Resetting the sidecar alone is not enough, and used to be all this did.\n\t\t// The sidecar is the only record of how many chunks each file produced,\n\t\t// so an empty one reports zero for every file — and `indexChangedFiles`\n\t\t// removes stale chunks by counting down from that number. Re-chunking a\n\t\t// file into *fewer* pieces then leaves its tail vectors (`path#N`,\n\t\t// `path#N+1`, …) in the store, holding text that no longer exists\n\t\t// anywhere, retrievable forever. Upserts hide it: chunk counts look\n\t\t// right, the sidecar looks right, and only search results are wrong.\n\t\t//\n\t\t// So when the sidecar cannot be trusted and the store is not already\n\t\t// empty, the store goes too.\n\t\tlet meta = loadIndexMeta(storeDir, info.modelId);\n\t\tif (!meta && info.count > 0) {\n\t\t\tinfo = await rebuildFrom(\"index metadata is missing or was written by a different chunker or model\");\n\t\t\tmeta = loadIndexMeta(storeDir, info.modelId);\n\t\t}\n\t\tthis.meta = meta ?? 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, this.options.chunkMaxChars);\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-harness.d.ts

@@ -52,6 +52,15 @@ export interface EvalProvenance {
      *  {@link CORPUS_EXCLUSIONS}. A score against a different exclusion list is
      *  a score against a different corpus, so it is recorded, not assumed. */
     corpusExcluded: string[];
+    /**
+     * Set only when the run scored a deliberately shrunk corpus.
+     *
+     * A smaller distractor pool makes every query easier, so these metrics are
+     * higher than a full-corpus run's and are **not** comparable to one. They
+     * are comparable to another subsampled run with the same target and seed,
+     * which is what makes this useful for screening model arms.
+     */
+    corpusSubsample?: CorpusSubsampleInfo;
     /** SHA of the tree whose retrieval code ran. Usually equals `corpusSha`,
      *  but differs when pinning an old corpus with today's code. */
     harnessSha: string;
@@ -71,11 +80,25 @@ export interface EvalProvenance {
         /** Indexed chunk count when the index reached `ready`. */
         chunkCount?: number;
         phase: string;
-        /** Binary that served the embeddings, and its self-reported version.
-         *  The embedding model is baked into the binary at build time, so this
-         *  is the only thing that identifies which model produced a score. */
+        /** Binary that served the embeddings, and its self-reported version. */
         binaryPath?: string;
         binaryVersion?: string;
+        /**
+         * Model id the daemon reported — the thing that actually identifies
+         * which model produced these scores.
+         *
+         * This used to be inferred from `binaryVersion`, on the reasoning that
+         * the model was baked into the binary at build time. `--model <dir>`
+         * ends that: one binary now serves any number of models, so two arms of
+         * a model comparison would have carried identical provenance and been
+         * indistinguishable in the record. The id is a hash over the model's
+         * whole spec (pooling, token limit, prefixes), so a change to any of
+         * them shows up here.
+         */
+        modelId?: string;
+        /** Model directory passed as `--model`, when the run overrode the
+         *  bundled model. Absent means the binary's own model was used. */
+        modelDir?: string;
     };
     /** Daemon-side BM25 hybrid store, when the run included one. Absent means
      *  the record has no `daemon-hybrid` rows. */
@@ -83,6 +106,34 @@ export interface EvalProvenance {
         available: boolean;
         phase: string;
     };
+    /**
+     * Wall time, split at the seam between building the index and scoring the
+     * gold set.
+     *
+     * Recorded because the cost side of a model comparison is almost entirely
+     * indexing, and a single total cannot show it: the first such comparison
+     * could only report "17 -> 60 min" for whole runs and had to note that the
+     * figure was "not isolated from query work", which left the headline cost
+     * of the change unmeasured. These two numbers are machine- and
+     * load-dependent and say nothing about retrieval quality; they are a budget,
+     * not a metric.
+     */
+    /**
+     * Chunker character cap, when an arm overrode it. Absent means the shipped
+     * `CHUNK_MAX_CHARS`. Records differing here are not comparable: the chunks
+     * are different text, so every id, span and vector differs.
+     */
+    chunkMaxChars?: number;
+    timing?: {
+        /** Seconds spent bringing the index(es) to `ready`, model load included. */
+        indexSeconds: number;
+        /** Seconds spent running every config over every gold query. */
+        querySeconds: number;
+        /** True when one hybrid store served both the dense and BM25 roles
+         *  rather than the corpus being embedded twice. Runs with this false
+         *  paid roughly double the indexing time. */
+        sharedStore: boolean;
+    };
     runtime: {
         node: string;
         platform: string;
@@ -116,9 +167,32 @@ export interface PinnedCorpus {
     /** Files removed from the corpus before indexing. Empty when the corpus is
      *  the live working tree, which is never mutated. */
     excluded: string[];
+    /** Present only on a subsampled run. Its presence is what marks a record as
+     *  incomparable to a full-corpus one. */
+    subsample?: CorpusSubsampleInfo;
     /** Removes the worktree, if one was created. */
     dispose: () => void;
 }
+/** Request to shrink the corpus to a chunk budget. See {@link pinCorpus}. */
+export interface CorpusSubsampleRequest {
+    /** Approximate chunk budget. Gold-bearing files are kept past it. */
+    targetChunks: number;
+    /** Files that must survive regardless of budget — the gold-bearing ones. */
+    keepRelPaths: readonly string[];
+    /** Seed for the distractor draw, so a budget reproduces exactly. */
+    seed: number;
+}
+/** What a subsampled run did, recorded so it can never be read as a full one. */
+export interface CorpusSubsampleInfo {
+    targetChunks: number;
+    /** Chunks actually kept. Exceeds the target when gold files alone do. */
+    chunkCount: number;
+    filesKept: number;
+    filesDropped: number;
+    /** Gold-bearing files, all of which are kept unconditionally. */
+    goldFilesKept: number;
+    seed: number;
+}
 /**
  * Files that describe this eval rather than being searched by it.
  *
@@ -145,9 +219,16 @@ export declare const CORPUS_EXCLUSIONS: readonly string[];
  * With a `ref`, checks out a detached worktree at that commit so the corpus is
  * byte-identical on every rerun. Without one, falls back to the live working
  * tree and reports `dirty` so the record shows the run was not reproducible.
+ *
+ * `subsample` shrinks the corpus to a chunk budget, for screening runs where a
+ * full arm costs too much to iterate on. It keeps every gold-bearing file and
+ * draws distractors deterministically. This is a real change to what is being
+ * measured — a smaller distractor pool makes retrieval easier and inflates
+ * every metric — so it is recorded in the record and folded into the worktree
+ * path, and it needs a `ref`.
  */
-export declare function pinCorpus(repoRoot: string, ref: string | undefined): PinnedCorpus;
-export declare function collectProvenance(repoRoot: string, corpus: PinnedCorpus, corpusRef: string, service: EmbsearchService | undefined, embsearchBinary?: string, hybridService?: EmbsearchService): EvalProvenance;
+export declare function pinCorpus(repoRoot: string, ref: string | undefined, subsample?: CorpusSubsampleRequest): PinnedCorpus;
+export declare function collectProvenance(repoRoot: string, corpus: PinnedCorpus, corpusRef: string, service: EmbsearchService | undefined, embsearchBinary?: string, hybridService?: EmbsearchService, modelDir?: string, timing?: EvalProvenance["timing"], chunkMaxChars?: number): EvalProvenance;
 export declare function summarizeGoldSet(dataset: readonly EvalQuery[]): EvalRunRecord["goldSet"];
 export interface RunEvalSuiteOptions {
     cwd: string;

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

@@ -1 +1 @@
-{"version":3,"file":"eval-harness.d.ts","sourceRoot":"","sources":["../../../src/core/search/eval-harness.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;GAqBG;AAQH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAC1E,OAAO,EAAE,KAAK,UAAU,EAAE,KAAK,SAAS,EAAE,KAAK,eAAe,EAAiB,MAAM,WAAW,CAAC;AAEjG,+DAA+D;AAC/D,MAAM,WAAW,aAAa;IAC7B,KAAK,EAAE,MAAM,CAAC;IACd,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;IACnB,GAAG,EAAE,MAAM,CAAC;IACZ,wCAAwC;IACxC,CAAC,EAAE,MAAM,CAAC;IACV,uEAAuE;IACvE,QAAQ,EAAE,MAAM,CAAC;CACjB;AAED,2EAA2E;AAC3E,MAAM,WAAW,cAAc;IAC9B,WAAW,EAAE,MAAM,CAAC;IACpB,uDAAuD;IACvD,SAAS,EAAE,MAAM,CAAC;IAClB,iEAAiE;IACjE,SAAS,EAAE,MAAM,CAAC;IAClB;iEAC2D;IAC3D,qBAAqB,EAAE,OAAO,CAAC;IAC/B;0DACsD;IACtD,WAAW,EAAE,OAAO,CAAC;IACrB;;8EAE0E;IAC1E,cAAc,EAAE,MAAM,EAAE,CAAC;IACzB;oEACgE;IAChE,UAAU,EAAE,MAAM,CAAC;IACnB;;;;;;OAMG;IACH,mBAAmB,EAAE,MAAM,CAAC;IAC5B;yDACqD;IACrD,QAAQ,EAAE;QACT,SAAS,EAAE,OAAO,CAAC;QACnB,MAAM,CAAC,EAAE,MAAM,CAAC;QAChB,0DAA0D;QAC1D,UAAU,CAAC,EAAE,MAAM,CAAC;QACpB,KAAK,EAAE,MAAM,CAAC;QACd;;8EAEsE;QACtE,UAAU,CAAC,EAAE,MAAM,CAAC;QACpB,aAAa,CAAC,EAAE,MAAM,CAAC;KACvB,CAAC;IACF;kDAC8C;IAC9C,YAAY,CAAC,EAAE;QAAE,SAAS,EAAE,OAAO,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;IACrD,OAAO,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,QAAQ,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC;CAC1D;AAED,MAAM,WAAW,aAAa;IAC7B,UAAU,EAAE,cAAc,CAAC;IAC3B,OAAO,EAAE;QAAE,UAAU,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;QAAC,aAAa,EAAE,MAAM,CAAA;KAAE,CAAC;IACxF,OAAO,EAAE,SAAS,UAAU,EAAE,CAAC;IAC/B,UAAU,EAAE,aAAa,EAAE,CAAC;IAC5B,QAAQ,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,eAAe,EAAE,CAAA;KAAE,CAAC,CAAC;CAC3E;AAMD;0EAC0E;AAC1E,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,MAAM,GAAG,MAAM,CA+B5D;AAED,MAAM,WAAW,YAAY;IAC5B,qCAAqC;IACrC,GAAG,EAAE,MAAM,CAAC;IACZ,GAAG,EAAE,MAAM,CAAC;IACZ,eAAe,EAAE,OAAO,CAAC;IACzB,KAAK,EAAE,OAAO,CAAC;IACf;yDACqD;IACrD,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB,gDAAgD;IAChD,OAAO,EAAE,MAAM,IAAI,CAAC;CACpB;AAED;;;;;;;;;;;;;;;;;;GAkBG;AACH,eAAO,MAAM,iBAAiB,EAAE,SAAS,MAAM,EAK9C,CAAC;AAEF;;;;;;GAMG;AACH,wBAAgB,SAAS,CAAC,QAAQ,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,SAAS,GAAG,YAAY,CAwDjF;AAYD,wBAAgB,iBAAiB,CAChC,QAAQ,EAAE,MAAM,EAChB,MAAM,EAAE,YAAY,EACpB,SAAS,EAAE,MAAM,EACjB,OAAO,EAAE,gBAAgB,GAAG,SAAS,EACrC,eAAe,CAAC,EAAE,MAAM,EACxB,aAAa,CAAC,EAAE,gBAAgB,GAC9B,cAAc,CA4BhB;AAED,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,SAAS,SAAS,EAAE,GAAG,aAAa,CAAC,SAAS,CAAC,CAQxF;AAED,MAAM,WAAW,mBAAmB;IACnC,GAAG,EAAE,MAAM,CAAC;IACZ,OAAO,EAAE,SAAS,SAAS,EAAE,CAAC;IAC9B,OAAO,EAAE,SAAS,UAAU,EAAE,CAAC;IAC/B,OAAO,CAAC,EAAE,gBAAgB,CAAC;IAC3B;wEACoE;IACpE,aAAa,CAAC,EAAE,gBAAgB,CAAC;IACjC,OAAO,CAAC,EAAE,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,SAAS,KAAK,IAAI,CAAC;CACpD;AAED,wBAAsB,YAAY,CAAC,OAAO,EAAE,mBAAmB,GAAG,OAAO,CAAC;IACzE,UAAU,EAAE,aAAa,EAAE,CAAC;IAC5B,QAAQ,EAAE,aAAa,CAAC,UAAU,CAAC,CAAC;CACpC,CAAC,CA4CD;AAED,wBAAgB,oBAAoB,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,GAAG,MAAM,CAcjF","sourcesContent":["/**\n * Eval harness: corpus pinning, provenance capture, and run records.\n *\n * The scoring math lives in `eval.ts`; this module is everything around it\n * that makes a number *comparable to a later number*. Three problems it\n * exists to solve, all of which bit the first eval round\n * (docs/hybrid-retrieval-design.md, \"Eval results\"):\n *\n *  1. **The corpus is the repo.** Retrieval is measured over hoocode itself,\n *     so every commit moves the thing being measured. A baseline taken today\n *     and a rerun taken after a retrieval change differ by both the change\n *     and the intervening commits, and nothing in the output says so. Fix:\n *     run against a detached git worktree pinned to an explicit SHA, and put\n *     that SHA in the record.\n *  2. **Nothing was recorded.** Results were printed to a terminal and\n *     hand-copied into a markdown table with no repo SHA, no embedder\n *     identity, and no index state. Fix: emit a machine-readable run record.\n *  3. **A degraded run looks like a real one.** With no embsearch binary the\n *     semantic and hybrid rows silently degrade to lexical, producing a table\n *     that is all-lexical but reads like a full sweep. Fix: `embedder` in the\n *     record, plus a per-row degraded count that the writer refuses to hide.\n */\n\nimport { createHash } from \"node:crypto\";\nimport { existsSync, readdirSync, readFileSync, statSync } from \"node:fs\";\nimport { execFileSync } from \"child_process\";\nimport { rmSync } from \"fs\";\nimport { tmpdir } from \"os\";\nimport path from \"path\";\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { type EvalConfig, type EvalQuery, type EvalQueryResult, evaluateQuery } from \"./eval.js\";\n\n/** Metrics aggregated per config across the whole gold set. */\nexport interface EvalAggregate {\n\tlabel: string;\n\trecallAt1: number;\n\trecallAt5: number;\n\trecallAt10: number;\n\trecallAt50: number;\n\tmrr: number;\n\t/** Queries scored under this config. */\n\tn: number;\n\t/** How many of them ran degraded (requested retriever unavailable). */\n\tdegraded: number;\n}\n\n/** Everything needed to decide whether two run records may be compared. */\nexport interface EvalProvenance {\n\ttimestampMs: number;\n\t/** SHA of the corpus actually indexed and searched. */\n\tcorpusSha: string;\n\t/** Ref the caller asked for, before resolution (e.g. \"HEAD\"). */\n\tcorpusRef: string;\n\t/** True when the corpus came from the live working tree rather than a\n\t *  pinned worktree — results are then not reproducible. */\n\tcorpusFromWorkingTree: boolean;\n\t/** Uncommitted changes present at run time. Only meaningful (and only\n\t *  possible) when `corpusFromWorkingTree` is true. */\n\tcorpusDirty: boolean;\n\t/** Files removed from the corpus before indexing — see\n\t *  {@link CORPUS_EXCLUSIONS}. A score against a different exclusion list is\n\t *  a score against a different corpus, so it is recorded, not assumed. */\n\tcorpusExcluded: string[];\n\t/** SHA of the tree whose retrieval code ran. Usually equals `corpusSha`,\n\t *  but differs when pinning an old corpus with today's code. */\n\tharnessSha: string;\n\t/**\n\t * Content hash of `src/core/search` + the chunker. Every tuning constant\n\t * that shapes a result — the fusion cap, top-k depths, rerank weights,\n\t * chunk sizing — lives in those files, so a changed hash means the\n\t * numbers are not comparable, without this module having to maintain a\n\t * hand-copied (and inevitably stale) list of constants.\n\t */\n\tretrievalSourceHash: string;\n\t/** Embedding backend state. `available: false` means every semantic and\n\t *  hybrid row in this record degraded to lexical. */\n\tembedder: {\n\t\tavailable: boolean;\n\t\treason?: string;\n\t\t/** Indexed chunk count when the index reached `ready`. */\n\t\tchunkCount?: number;\n\t\tphase: string;\n\t\t/** Binary that served the embeddings, and its self-reported version.\n\t\t *  The embedding model is baked into the binary at build time, so this\n\t\t *  is the only thing that identifies which model produced a score. */\n\t\tbinaryPath?: string;\n\t\tbinaryVersion?: string;\n\t};\n\t/** Daemon-side BM25 hybrid store, when the run included one. Absent means\n\t *  the record has no `daemon-hybrid` rows. */\n\tdaemonHybrid?: { available: boolean; phase: string };\n\truntime: { node: string; platform: string; arch: string };\n}\n\nexport interface EvalRunRecord {\n\tprovenance: EvalProvenance;\n\tgoldSet: { queryCount: number; byClass: Record<string, number>; goldSpanCount: number };\n\tconfigs: readonly EvalConfig[];\n\taggregates: EvalAggregate[];\n\tperQuery: Array<{ id: string; class: string; results: EvalQueryResult[] }>;\n}\n\nfunction git(repoRoot: string, args: string[]): string {\n\treturn execFileSync(\"git\", [\"-C\", repoRoot, ...args], { encoding: \"utf-8\" }).trim();\n}\n\n/** Hash every retrieval-shaping source file, so a tuning change is visible as\n *  a changed provenance field rather than an unexplained metric shift. */\nexport function hashRetrievalSource(repoRoot: string): string {\n\tconst roots = [\n\t\tpath.join(repoRoot, \"packages/coding-agent/src/core/search\"),\n\t\tpath.join(repoRoot, \"packages/coding-agent/src/core/embsearch/chunker.ts\"),\n\t];\n\tconst files: string[] = [];\n\tconst walk = (target: string): void => {\n\t\tlet stat: ReturnType<typeof statSync>;\n\t\ttry {\n\t\t\tstat = statSync(target);\n\t\t} catch {\n\t\t\treturn;\n\t\t}\n\t\tif (stat.isDirectory()) {\n\t\t\tfor (const entry of readdirSync(target).sort()) walk(path.join(target, entry));\n\t\t} else if (target.endsWith(\".ts\")) {\n\t\t\tfiles.push(target);\n\t\t}\n\t};\n\tfor (const root of roots) walk(root);\n\n\tconst hash = createHash(\"sha256\");\n\tfor (const file of files) {\n\t\t// Eval-only modules are excluded: changing how we measure must not look\n\t\t// like changing what we measure.\n\t\tconst base = path.basename(file);\n\t\tif (base.startsWith(\"eval\")) continue;\n\t\thash.update(path.relative(repoRoot, file).replace(/\\\\/g, \"/\"));\n\t\thash.update(readFileSync(file));\n\t}\n\treturn hash.digest(\"hex\").slice(0, 16);\n}\n\nexport interface PinnedCorpus {\n\t/** Directory to index and search. */\n\tcwd: string;\n\tsha: string;\n\tfromWorkingTree: boolean;\n\tdirty: boolean;\n\t/** Files removed from the corpus before indexing. Empty when the corpus is\n\t *  the live working tree, which is never mutated. */\n\texcluded: string[];\n\t/** Removes the worktree, if one was created. */\n\tdispose: () => void;\n}\n\n/**\n * Files that describe this eval rather than being searched by it.\n *\n * The fixtures hold all 62 query strings verbatim, so every query is a perfect\n * lexical match against its own entry, and the design note quotes the same\n * queries while discussing the classes they belong to. Measured before this\n * exclusion existed: **56 of 62 queries had one of these files in the top 10,\n * 27 of 62 had one as the #1 result, and they consumed 133 of the 620\n * top-10 slots** — a fifth of the window, spent on the eval reading itself.\n *\n * That is not a ranking artifact a reranker can fix: it displaces real answers\n * out of the window entirely, which is why two boundary-class queries were\n * absent from the top *50* rather than merely buried. Retrieving your own\n * question is not retrieval, so the corpus is scored without them.\n *\n * Removed from the pinned worktree before indexing, never from the repo — and\n * recorded in the run's provenance so a score is never silently taken against\n * a different corpus than it claims.\n */\nexport const CORPUS_EXCLUSIONS: readonly string[] = [\n\t\"packages/coding-agent/test/fixtures/search-eval.json\",\n\t\"packages/coding-agent/test/fixtures/search-eval-live.json\",\n\t\"packages/coding-agent/test/fixtures/search-eval-baseline.json\",\n\t\"docs/hybrid-retrieval-design.md\",\n];\n\n/**\n * Materialize the corpus to evaluate.\n *\n * With a `ref`, checks out a detached worktree at that commit so the corpus is\n * byte-identical on every rerun. Without one, falls back to the live working\n * tree and reports `dirty` so the record shows the run was not reproducible.\n */\nexport function pinCorpus(repoRoot: string, ref: string | undefined): PinnedCorpus {\n\tconst dirty = git(repoRoot, [\"status\", \"--porcelain\"]).length > 0;\n\tif (!ref) {\n\t\t// The live working tree is the user's checkout; deleting files from it to\n\t\t// tidy a measurement would be an unforgivable trade. Working-tree runs\n\t\t// are already stamped non-reproducible, so they carry the contamination.\n\t\treturn {\n\t\t\tcwd: repoRoot,\n\t\t\tsha: git(repoRoot, [\"rev-parse\", \"HEAD\"]),\n\t\t\tfromWorkingTree: true,\n\t\t\tdirty,\n\t\t\texcluded: [],\n\t\t\tdispose: () => {},\n\t\t};\n\t}\n\n\tconst sha = git(repoRoot, [\"rev-parse\", ref]);\n\t// Deterministic path, not mkdtemp: the embedding store is keyed by a hash of\n\t// the corpus directory, so a fresh temp path every run would re-embed all\n\t// ~17k chunks (minutes) instead of reusing the store built for this exact\n\t// SHA. The worktree is still removed afterwards; only the store persists.\n\tconst dir = path.join(tmpdir(), `hoocode-search-eval-${sha.slice(0, 12)}`);\n\tif (existsSync(dir)) {\n\t\t// Left behind by an interrupted run — drop it so `worktree add` succeeds.\n\t\ttry {\n\t\t\tgit(repoRoot, [\"worktree\", \"remove\", \"--force\", dir]);\n\t\t} catch {\n\t\t\trmSync(dir, { recursive: true, force: true });\n\t\t\tgit(repoRoot, [\"worktree\", \"prune\"]);\n\t\t}\n\t}\n\tgit(repoRoot, [\"worktree\", \"add\", \"--detach\", dir, sha]);\n\n\tconst excluded: string[] = [];\n\tfor (const rel of CORPUS_EXCLUSIONS) {\n\t\tconst target = path.join(dir, rel);\n\t\tif (existsSync(target)) {\n\t\t\trmSync(target, { force: true });\n\t\t\texcluded.push(rel);\n\t\t}\n\t}\n\n\treturn {\n\t\tcwd: dir,\n\t\tsha,\n\t\tfromWorkingTree: false,\n\t\tdirty: false,\n\t\texcluded,\n\t\tdispose: () => {\n\t\t\ttry {\n\t\t\t\tgit(repoRoot, [\"worktree\", \"remove\", \"--force\", dir]);\n\t\t\t} catch {\n\t\t\t\trmSync(dir, { recursive: true, force: true });\n\t\t\t}\n\t\t},\n\t};\n}\n\n/** `<binary> --version`, or undefined when it cannot be run. */\nfunction probeBinaryVersion(binaryPath: string | undefined): string | undefined {\n\tif (!binaryPath) return undefined;\n\ttry {\n\t\treturn execFileSync(binaryPath, [\"--version\"], { encoding: \"utf-8\" }).trim();\n\t} catch {\n\t\treturn undefined;\n\t}\n}\n\nexport function collectProvenance(\n\trepoRoot: string,\n\tcorpus: PinnedCorpus,\n\tcorpusRef: string,\n\tservice: EmbsearchService | undefined,\n\tembsearchBinary?: string,\n\thybridService?: EmbsearchService,\n): EvalProvenance {\n\tconst state = service?.getState();\n\tconst phase = state?.phase ?? \"absent\";\n\treturn {\n\t\ttimestampMs: Date.now(),\n\t\tcorpusSha: corpus.sha,\n\t\tcorpusRef,\n\t\tcorpusFromWorkingTree: corpus.fromWorkingTree,\n\t\tcorpusDirty: corpus.dirty,\n\t\tcorpusExcluded: corpus.excluded,\n\t\tharnessSha: git(repoRoot, [\"rev-parse\", \"HEAD\"]),\n\t\tretrievalSourceHash: hashRetrievalSource(repoRoot),\n\t\tembedder: {\n\t\t\t// `ready` is the only phase the service reaches with a real embedder:\n\t\t\t// it rejects the mock backend at startup, so availability here also\n\t\t\t// certifies the numbers came from a genuine ONNX build.\n\t\t\tavailable: service?.isAvailable() ?? false,\n\t\t\treason: state && \"reason\" in state ? state.reason : undefined,\n\t\t\tchunkCount: state?.phase === \"ready\" ? state.chunkCount : undefined,\n\t\t\tphase,\n\t\t\tbinaryPath: embsearchBinary,\n\t\t\tbinaryVersion: probeBinaryVersion(embsearchBinary),\n\t\t},\n\t\tdaemonHybrid: hybridService\n\t\t\t? { available: hybridService.isAvailable(), phase: hybridService.getState().phase }\n\t\t\t: undefined,\n\t\truntime: { node: process.version, platform: process.platform, arch: process.arch },\n\t};\n}\n\nexport function summarizeGoldSet(dataset: readonly EvalQuery[]): EvalRunRecord[\"goldSet\"] {\n\tconst byClass: Record<string, number> = {};\n\tlet goldSpanCount = 0;\n\tfor (const query of dataset) {\n\t\tbyClass[query.class] = (byClass[query.class] ?? 0) + 1;\n\t\tgoldSpanCount += query.gold.length;\n\t}\n\treturn { queryCount: dataset.length, byClass, goldSpanCount };\n}\n\nexport interface RunEvalSuiteOptions {\n\tcwd: string;\n\tdataset: readonly EvalQuery[];\n\tconfigs: readonly EvalConfig[];\n\tservice?: EmbsearchService;\n\t/** Second service backed by a daemon-side BM25 hybrid store, for the\n\t *  `daemon-hybrid` configs. Absent means those rows are omitted. */\n\thybridService?: EmbsearchService;\n\tonQuery?: (index: number, query: EvalQuery) => void;\n}\n\nexport async function runEvalSuite(options: RunEvalSuiteOptions): Promise<{\n\taggregates: EvalAggregate[];\n\tperQuery: EvalRunRecord[\"perQuery\"];\n}> {\n\tconst { cwd, dataset, configs, service } = options;\n\tconst totals = new Map<string, EvalAggregate>();\n\tconst perQuery: EvalRunRecord[\"perQuery\"] = [];\n\n\tfor (const [index, evalQuery] of dataset.entries()) {\n\t\toptions.onQuery?.(index, evalQuery);\n\t\tconst results = await evaluateQuery(cwd, evalQuery, configs, service, options.hybridService);\n\t\tperQuery.push({ id: evalQuery.id, class: evalQuery.class, results });\n\t\tfor (const result of results) {\n\t\t\tconst total = totals.get(result.label) ?? {\n\t\t\t\tlabel: result.label,\n\t\t\t\trecallAt1: 0,\n\t\t\t\trecallAt5: 0,\n\t\t\t\trecallAt10: 0,\n\t\t\t\trecallAt50: 0,\n\t\t\t\tmrr: 0,\n\t\t\t\tn: 0,\n\t\t\t\tdegraded: 0,\n\t\t\t};\n\t\t\ttotal.recallAt1 += result.recallAt1;\n\t\t\ttotal.recallAt5 += result.recallAt5;\n\t\t\ttotal.recallAt10 += result.recallAt10;\n\t\t\ttotal.recallAt50 += result.recallAt50;\n\t\t\ttotal.mrr += result.mrr;\n\t\t\ttotal.n++;\n\t\t\tif (result.degraded) total.degraded++;\n\t\t\ttotals.set(result.label, total);\n\t\t}\n\t}\n\n\tconst aggregates = configs\n\t\t.map((config) => totals.get(config.label))\n\t\t.filter((total): total is EvalAggregate => total !== undefined)\n\t\t.map((total) => ({\n\t\t\t...total,\n\t\t\trecallAt1: total.recallAt1 / total.n,\n\t\t\trecallAt5: total.recallAt5 / total.n,\n\t\t\trecallAt10: total.recallAt10 / total.n,\n\t\t\trecallAt50: total.recallAt50 / total.n,\n\t\t\tmrr: total.mrr / total.n,\n\t\t}));\n\n\treturn { aggregates, perQuery };\n}\n\nexport function formatAggregateTable(aggregates: readonly EvalAggregate[]): string {\n\tconst pct = (x: number) => `${Math.round(x * 100)}%`.padStart(5);\n\tconst lines = [\n\t\t\"config           |  R@1  |  R@5  | R@10  | R@50  |  MRR  | notes\",\n\t\t\"-----------------|-------|-------|-------|-------|-------|------\",\n\t];\n\tfor (const a of aggregates) {\n\t\tconst notes = a.degraded === a.n ? \"degraded to lexical\" : a.degraded > 0 ? `${a.degraded}/${a.n} degraded` : \"\";\n\t\tlines.push(\n\t\t\t`${a.label.padEnd(16)} | ${pct(a.recallAt1)} | ${pct(a.recallAt5)} | ${pct(a.recallAt10)} | ` +\n\t\t\t\t`${pct(a.recallAt50)} | ${a.mrr.toFixed(3)} | ${notes}`,\n\t\t);\n\t}\n\treturn lines.join(\"\\n\");\n}\n"]}
+{"version":3,"file":"eval-harness.d.ts","sourceRoot":"","sources":["../../../src/core/search/eval-harness.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;GAqBG;AASH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAE1E,OAAO,EAAE,KAAK,UAAU,EAAE,KAAK,SAAS,EAAE,KAAK,eAAe,EAAiB,MAAM,WAAW,CAAC;AAEjG,+DAA+D;AAC/D,MAAM,WAAW,aAAa;IAC7B,KAAK,EAAE,MAAM,CAAC;IACd,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;IACnB,GAAG,EAAE,MAAM,CAAC;IACZ,wCAAwC;IACxC,CAAC,EAAE,MAAM,CAAC;IACV,uEAAuE;IACvE,QAAQ,EAAE,MAAM,CAAC;CACjB;AAED,2EAA2E;AAC3E,MAAM,WAAW,cAAc;IAC9B,WAAW,EAAE,MAAM,CAAC;IACpB,uDAAuD;IACvD,SAAS,EAAE,MAAM,CAAC;IAClB,iEAAiE;IACjE,SAAS,EAAE,MAAM,CAAC;IAClB;iEAC2D;IAC3D,qBAAqB,EAAE,OAAO,CAAC;IAC/B;0DACsD;IACtD,WAAW,EAAE,OAAO,CAAC;IACrB;;8EAE0E;IAC1E,cAAc,EAAE,MAAM,EAAE,CAAC;IACzB;;;;;;;OAOG;IACH,eAAe,CAAC,EAAE,mBAAmB,CAAC;IACtC;oEACgE;IAChE,UAAU,EAAE,MAAM,CAAC;IACnB;;;;;;OAMG;IACH,mBAAmB,EAAE,MAAM,CAAC;IAC5B;yDACqD;IACrD,QAAQ,EAAE;QACT,SAAS,EAAE,OAAO,CAAC;QACnB,MAAM,CAAC,EAAE,MAAM,CAAC;QAChB,0DAA0D;QAC1D,UAAU,CAAC,EAAE,MAAM,CAAC;QACpB,KAAK,EAAE,MAAM,CAAC;QACd,wEAAwE;QACxE,UAAU,CAAC,EAAE,MAAM,CAAC;QACpB,aAAa,CAAC,EAAE,MAAM,CAAC;QACvB;;;;;;;;;;;WAWG;QACH,OAAO,CAAC,EAAE,MAAM,CAAC;QACjB;2EACmE;QACnE,QAAQ,CAAC,EAAE,MAAM,CAAC;KAClB,CAAC;IACF;kDAC8C;IAC9C,YAAY,CAAC,EAAE;QAAE,SAAS,EAAE,OAAO,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;IACrD;;;;;;;;;;;OAWG;IACH;;;;OAIG;IACH,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,MAAM,CAAC,EAAE;QACR,4EAA4E;QAC5E,YAAY,EAAE,MAAM,CAAC;QACrB,gEAAgE;QAChE,YAAY,EAAE,MAAM,CAAC;QACrB;;qDAE6C;QAC7C,WAAW,EAAE,OAAO,CAAC;KACrB,CAAC;IACF,OAAO,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,QAAQ,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC;CAC1D;AAED,MAAM,WAAW,aAAa;IAC7B,UAAU,EAAE,cAAc,CAAC;IAC3B,OAAO,EAAE;QAAE,UAAU,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;QAAC,aAAa,EAAE,MAAM,CAAA;KAAE,CAAC;IACxF,OAAO,EAAE,SAAS,UAAU,EAAE,CAAC;IAC/B,UAAU,EAAE,aAAa,EAAE,CAAC;IAC5B,QAAQ,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,eAAe,EAAE,CAAA;KAAE,CAAC,CAAC;CAC3E;AAMD;0EAC0E;AAC1E,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,MAAM,GAAG,MAAM,CA+B5D;AAED,MAAM,WAAW,YAAY;IAC5B,qCAAqC;IACrC,GAAG,EAAE,MAAM,CAAC;IACZ,GAAG,EAAE,MAAM,CAAC;IACZ,eAAe,EAAE,OAAO,CAAC;IACzB,KAAK,EAAE,OAAO,CAAC;IACf;yDACqD;IACrD,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB;6CACyC;IACzC,SAAS,CAAC,EAAE,mBAAmB,CAAC;IAChC,gDAAgD;IAChD,OAAO,EAAE,MAAM,IAAI,CAAC;CACpB;AAED,6EAA6E;AAC7E,MAAM,WAAW,sBAAsB;IACtC,qEAAqE;IACrE,YAAY,EAAE,MAAM,CAAC;IACrB,8EAA4E;IAC5E,YAAY,EAAE,SAAS,MAAM,EAAE,CAAC;IAChC,oEAAoE;IACpE,IAAI,EAAE,MAAM,CAAC;CACb;AAED,iFAAiF;AACjF,MAAM,WAAW,mBAAmB;IACnC,YAAY,EAAE,MAAM,CAAC;IACrB,yEAAyE;IACzE,UAAU,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,MAAM,CAAC;IAClB,YAAY,EAAE,MAAM,CAAC;IACrB,iEAAiE;IACjE,aAAa,EAAE,MAAM,CAAC;IACtB,IAAI,EAAE,MAAM,CAAC;CACb;AAgHD;;;;;;;;;;;;;;;;;;GAkBG;AACH,eAAO,MAAM,iBAAiB,EAAE,SAAS,MAAM,EAK9C,CAAC;AAEF;;;;;;;;;;;;;GAaG;AACH,wBAAgB,SAAS,CAAC,QAAQ,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,SAAS,EAAE,SAAS,CAAC,EAAE,sBAAsB,GAAG,YAAY,CA2ErH;AAYD,wBAAgB,iBAAiB,CAChC,QAAQ,EAAE,MAAM,EAChB,MAAM,EAAE,YAAY,EACpB,SAAS,EAAE,MAAM,EACjB,OAAO,EAAE,gBAAgB,GAAG,SAAS,EACrC,eAAe,CAAC,EAAE,MAAM,EACxB,aAAa,CAAC,EAAE,gBAAgB,EAChC,QAAQ,CAAC,EAAE,MAAM,EACjB,MAAM,CAAC,EAAE,cAAc,CAAC,QAAQ,CAAC,EACjC,aAAa,CAAC,EAAE,MAAM,GACpB,cAAc,CAiChB;AAED,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,SAAS,SAAS,EAAE,GAAG,aAAa,CAAC,SAAS,CAAC,CAQxF;AAED,MAAM,WAAW,mBAAmB;IACnC,GAAG,EAAE,MAAM,CAAC;IACZ,OAAO,EAAE,SAAS,SAAS,EAAE,CAAC;IAC9B,OAAO,EAAE,SAAS,UAAU,EAAE,CAAC;IAC/B,OAAO,CAAC,EAAE,gBAAgB,CAAC;IAC3B;wEACoE;IACpE,aAAa,CAAC,EAAE,gBAAgB,CAAC;IACjC,OAAO,CAAC,EAAE,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,SAAS,KAAK,IAAI,CAAC;CACpD;AAED,wBAAsB,YAAY,CAAC,OAAO,EAAE,mBAAmB,GAAG,OAAO,CAAC;IACzE,UAAU,EAAE,aAAa,EAAE,CAAC;IAC5B,QAAQ,EAAE,aAAa,CAAC,UAAU,CAAC,CAAC;CACpC,CAAC,CA4CD;AAED,wBAAgB,oBAAoB,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,GAAG,MAAM,CAcjF","sourcesContent":["/**\n * Eval harness: corpus pinning, provenance capture, and run records.\n *\n * The scoring math lives in `eval.ts`; this module is everything around it\n * that makes a number *comparable to a later number*. Three problems it\n * exists to solve, all of which bit the first eval round\n * (docs/hybrid-retrieval-design.md, \"Eval results\"):\n *\n *  1. **The corpus is the repo.** Retrieval is measured over hoocode itself,\n *     so every commit moves the thing being measured. A baseline taken today\n *     and a rerun taken after a retrieval change differ by both the change\n *     and the intervening commits, and nothing in the output says so. Fix:\n *     run against a detached git worktree pinned to an explicit SHA, and put\n *     that SHA in the record.\n *  2. **Nothing was recorded.** Results were printed to a terminal and\n *     hand-copied into a markdown table with no repo SHA, no embedder\n *     identity, and no index state. Fix: emit a machine-readable run record.\n *  3. **A degraded run looks like a real one.** With no embsearch binary the\n *     semantic and hybrid rows silently degrade to lexical, producing a table\n *     that is all-lexical but reads like a full sweep. Fix: `embedder` in the\n *     record, plus a per-row degraded count that the writer refuses to hide.\n */\n\nimport { createHash } from \"node:crypto\";\nimport { existsSync, readdirSync, readFileSync, statSync } from \"node:fs\";\nimport { execFileSync } from \"child_process\";\nimport { rmSync } from \"fs\";\nimport { tmpdir } from \"os\";\nimport path from \"path\";\nimport { chunkFile } from \"../embsearch/chunker.js\";\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { scanRepo } from \"../embsearch/repo-scan.js\";\nimport { type EvalConfig, type EvalQuery, type EvalQueryResult, evaluateQuery } from \"./eval.js\";\n\n/** Metrics aggregated per config across the whole gold set. */\nexport interface EvalAggregate {\n\tlabel: string;\n\trecallAt1: number;\n\trecallAt5: number;\n\trecallAt10: number;\n\trecallAt50: number;\n\tmrr: number;\n\t/** Queries scored under this config. */\n\tn: number;\n\t/** How many of them ran degraded (requested retriever unavailable). */\n\tdegraded: number;\n}\n\n/** Everything needed to decide whether two run records may be compared. */\nexport interface EvalProvenance {\n\ttimestampMs: number;\n\t/** SHA of the corpus actually indexed and searched. */\n\tcorpusSha: string;\n\t/** Ref the caller asked for, before resolution (e.g. \"HEAD\"). */\n\tcorpusRef: string;\n\t/** True when the corpus came from the live working tree rather than a\n\t *  pinned worktree — results are then not reproducible. */\n\tcorpusFromWorkingTree: boolean;\n\t/** Uncommitted changes present at run time. Only meaningful (and only\n\t *  possible) when `corpusFromWorkingTree` is true. */\n\tcorpusDirty: boolean;\n\t/** Files removed from the corpus before indexing — see\n\t *  {@link CORPUS_EXCLUSIONS}. A score against a different exclusion list is\n\t *  a score against a different corpus, so it is recorded, not assumed. */\n\tcorpusExcluded: string[];\n\t/**\n\t * Set only when the run scored a deliberately shrunk corpus.\n\t *\n\t * A smaller distractor pool makes every query easier, so these metrics are\n\t * higher than a full-corpus run's and are **not** comparable to one. They\n\t * are comparable to another subsampled run with the same target and seed,\n\t * which is what makes this useful for screening model arms.\n\t */\n\tcorpusSubsample?: CorpusSubsampleInfo;\n\t/** SHA of the tree whose retrieval code ran. Usually equals `corpusSha`,\n\t *  but differs when pinning an old corpus with today's code. */\n\tharnessSha: string;\n\t/**\n\t * Content hash of `src/core/search` + the chunker. Every tuning constant\n\t * that shapes a result — the fusion cap, top-k depths, rerank weights,\n\t * chunk sizing — lives in those files, so a changed hash means the\n\t * numbers are not comparable, without this module having to maintain a\n\t * hand-copied (and inevitably stale) list of constants.\n\t */\n\tretrievalSourceHash: string;\n\t/** Embedding backend state. `available: false` means every semantic and\n\t *  hybrid row in this record degraded to lexical. */\n\tembedder: {\n\t\tavailable: boolean;\n\t\treason?: string;\n\t\t/** Indexed chunk count when the index reached `ready`. */\n\t\tchunkCount?: number;\n\t\tphase: string;\n\t\t/** Binary that served the embeddings, and its self-reported version. */\n\t\tbinaryPath?: string;\n\t\tbinaryVersion?: string;\n\t\t/**\n\t\t * Model id the daemon reported — the thing that actually identifies\n\t\t * which model produced these scores.\n\t\t *\n\t\t * This used to be inferred from `binaryVersion`, on the reasoning that\n\t\t * the model was baked into the binary at build time. `--model <dir>`\n\t\t * ends that: one binary now serves any number of models, so two arms of\n\t\t * a model comparison would have carried identical provenance and been\n\t\t * indistinguishable in the record. The id is a hash over the model's\n\t\t * whole spec (pooling, token limit, prefixes), so a change to any of\n\t\t * them shows up here.\n\t\t */\n\t\tmodelId?: string;\n\t\t/** Model directory passed as `--model`, when the run overrode the\n\t\t *  bundled model. Absent means the binary's own model was used. */\n\t\tmodelDir?: string;\n\t};\n\t/** Daemon-side BM25 hybrid store, when the run included one. Absent means\n\t *  the record has no `daemon-hybrid` rows. */\n\tdaemonHybrid?: { available: boolean; phase: string };\n\t/**\n\t * Wall time, split at the seam between building the index and scoring the\n\t * gold set.\n\t *\n\t * Recorded because the cost side of a model comparison is almost entirely\n\t * indexing, and a single total cannot show it: the first such comparison\n\t * could only report \"17 -> 60 min\" for whole runs and had to note that the\n\t * figure was \"not isolated from query work\", which left the headline cost\n\t * of the change unmeasured. These two numbers are machine- and\n\t * load-dependent and say nothing about retrieval quality; they are a budget,\n\t * not a metric.\n\t */\n\t/**\n\t * Chunker character cap, when an arm overrode it. Absent means the shipped\n\t * `CHUNK_MAX_CHARS`. Records differing here are not comparable: the chunks\n\t * are different text, so every id, span and vector differs.\n\t */\n\tchunkMaxChars?: number;\n\ttiming?: {\n\t\t/** Seconds spent bringing the index(es) to `ready`, model load included. */\n\t\tindexSeconds: number;\n\t\t/** Seconds spent running every config over every gold query. */\n\t\tquerySeconds: number;\n\t\t/** True when one hybrid store served both the dense and BM25 roles\n\t\t *  rather than the corpus being embedded twice. Runs with this false\n\t\t *  paid roughly double the indexing time. */\n\t\tsharedStore: boolean;\n\t};\n\truntime: { node: string; platform: string; arch: string };\n}\n\nexport interface EvalRunRecord {\n\tprovenance: EvalProvenance;\n\tgoldSet: { queryCount: number; byClass: Record<string, number>; goldSpanCount: number };\n\tconfigs: readonly EvalConfig[];\n\taggregates: EvalAggregate[];\n\tperQuery: Array<{ id: string; class: string; results: EvalQueryResult[] }>;\n}\n\nfunction git(repoRoot: string, args: string[]): string {\n\treturn execFileSync(\"git\", [\"-C\", repoRoot, ...args], { encoding: \"utf-8\" }).trim();\n}\n\n/** Hash every retrieval-shaping source file, so a tuning change is visible as\n *  a changed provenance field rather than an unexplained metric shift. */\nexport function hashRetrievalSource(repoRoot: string): string {\n\tconst roots = [\n\t\tpath.join(repoRoot, \"packages/coding-agent/src/core/search\"),\n\t\tpath.join(repoRoot, \"packages/coding-agent/src/core/embsearch/chunker.ts\"),\n\t];\n\tconst files: string[] = [];\n\tconst walk = (target: string): void => {\n\t\tlet stat: ReturnType<typeof statSync>;\n\t\ttry {\n\t\t\tstat = statSync(target);\n\t\t} catch {\n\t\t\treturn;\n\t\t}\n\t\tif (stat.isDirectory()) {\n\t\t\tfor (const entry of readdirSync(target).sort()) walk(path.join(target, entry));\n\t\t} else if (target.endsWith(\".ts\")) {\n\t\t\tfiles.push(target);\n\t\t}\n\t};\n\tfor (const root of roots) walk(root);\n\n\tconst hash = createHash(\"sha256\");\n\tfor (const file of files) {\n\t\t// Eval-only modules are excluded: changing how we measure must not look\n\t\t// like changing what we measure.\n\t\tconst base = path.basename(file);\n\t\tif (base.startsWith(\"eval\")) continue;\n\t\thash.update(path.relative(repoRoot, file).replace(/\\\\/g, \"/\"));\n\t\thash.update(readFileSync(file));\n\t}\n\treturn hash.digest(\"hex\").slice(0, 16);\n}\n\nexport interface PinnedCorpus {\n\t/** Directory to index and search. */\n\tcwd: string;\n\tsha: string;\n\tfromWorkingTree: boolean;\n\tdirty: boolean;\n\t/** Files removed from the corpus before indexing. Empty when the corpus is\n\t *  the live working tree, which is never mutated. */\n\texcluded: string[];\n\t/** Present only on a subsampled run. Its presence is what marks a record as\n\t *  incomparable to a full-corpus one. */\n\tsubsample?: CorpusSubsampleInfo;\n\t/** Removes the worktree, if one was created. */\n\tdispose: () => void;\n}\n\n/** Request to shrink the corpus to a chunk budget. See {@link pinCorpus}. */\nexport interface CorpusSubsampleRequest {\n\t/** Approximate chunk budget. Gold-bearing files are kept past it. */\n\ttargetChunks: number;\n\t/** Files that must survive regardless of budget — the gold-bearing ones. */\n\tkeepRelPaths: readonly string[];\n\t/** Seed for the distractor draw, so a budget reproduces exactly. */\n\tseed: number;\n}\n\n/** What a subsampled run did, recorded so it can never be read as a full one. */\nexport interface CorpusSubsampleInfo {\n\ttargetChunks: number;\n\t/** Chunks actually kept. Exceeds the target when gold files alone do. */\n\tchunkCount: number;\n\tfilesKept: number;\n\tfilesDropped: number;\n\t/** Gold-bearing files, all of which are kept unconditionally. */\n\tgoldFilesKept: number;\n\tseed: number;\n}\n\n/**\n * Cut `dir` down to a chunk budget, in place.\n *\n * Counts chunks with the indexer's own chunker rather than estimating from\n * file size, because the budget is meant to predict indexing time and\n * indexing time is per chunk. Gold-bearing files are never candidates for\n * removal: dropping one would make its queries unanswerable and score the\n * arm on a corpus that cannot contain the answer.\n *\n * Deletion is what makes this apply to every leg at once. Filtering the\n * indexer's file list instead would shrink the dense and BM25 legs while grep\n * still walked the full tree, and the legs would then be answering about\n * different corpora.\n */\nfunction applySubsample(dir: string, request: CorpusSubsampleRequest): CorpusSubsampleInfo {\n\tconst gold = new Set(request.keepRelPaths);\n\tconst counts = new Map<string, number>();\n\tfor (const file of scanRepo(dir).files) {\n\t\t// The scanner skips `.git` as a *directory*, but a linked worktree's\n\t\t// `.git` is a file holding the path to the real gitdir — so it comes back\n\t\t// as an ordinary indexable file. Deleting it detaches the worktree from\n\t\t// the repo, and `worktree remove` then fails on a tree git can no longer\n\t\t// validate. Never a candidate.\n\t\tif (file.rel === \".git\" || file.rel.startsWith(`.git${path.sep}`) || file.rel.startsWith(\".git/\")) {\n\t\t\tcontinue;\n\t\t}\n\t\tlet content: string;\n\t\ttry {\n\t\t\tcontent = readFileSync(path.join(dir, file.rel), \"utf8\");\n\t\t} catch {\n\t\t\tcontinue;\n\t\t}\n\t\t// Deliberately the *default* cap, never the arm's.\n\t\t//\n\t\t// The budget picks which files survive, and a chunk-cap sweep must\n\t\t// compare caps over identical source text. Counting at the arm's cap\n\t\t// would let a cap-2000 arm — whose chunks are bigger, so fewer fit the\n\t\t// budget — keep far more of the repo than a cap-1000 arm, and the two\n\t\t// would then differ by corpus as well as by cap. Sizing is a secondary\n\t\t// concern to that: bigger caps produce fewer chunks and index faster\n\t\t// anyway, so the budget only ever overestimates their cost.\n\t\tconst n = chunkFile(file.rel, content).length;\n\t\tif (n > 0) counts.set(file.rel, n);\n\t}\n\n\tconst keep = new Set<string>();\n\tlet chunks = 0;\n\tlet goldFilesKept = 0;\n\tfor (const rel of counts.keys()) {\n\t\tif (!gold.has(rel)) continue;\n\t\tkeep.add(rel);\n\t\tchunks += counts.get(rel) ?? 0;\n\t\tgoldFilesKept++;\n\t}\n\n\t// Shuffle the distractors rather than taking the scan's order, which is\n\t// directory order — that would keep a few whole subtrees and drop the rest,\n\t// making the sample a slice of the repo instead of a sample of it.\n\tconst rng = seededRandom(request.seed);\n\tconst others = [...counts.keys()].filter((rel) => !gold.has(rel));\n\tfor (let i = others.length - 1; i > 0; i--) {\n\t\tconst j = Math.floor(rng() * (i + 1));\n\t\t[others[i], others[j]] = [others[j], others[i]];\n\t}\n\tfor (const rel of others) {\n\t\tif (chunks >= request.targetChunks) break;\n\t\tkeep.add(rel);\n\t\tchunks += counts.get(rel) ?? 0;\n\t}\n\n\tlet filesDropped = 0;\n\tfor (const rel of counts.keys()) {\n\t\tif (keep.has(rel)) continue;\n\t\ttry {\n\t\t\trmSync(path.join(dir, rel), { force: true });\n\t\t\tfilesDropped++;\n\t\t} catch {\n\t\t\t// A file the scanner listed but cannot be removed stays in the\n\t\t\t// corpus; it inflates the sample slightly and is not worth failing\n\t\t\t// the run over.\n\t\t}\n\t}\n\n\treturn {\n\t\ttargetChunks: request.targetChunks,\n\t\tchunkCount: chunks,\n\t\tfilesKept: keep.size,\n\t\tfilesDropped,\n\t\tgoldFilesKept,\n\t\tseed: request.seed,\n\t};\n}\n\n/**\n * Deterministic PRNG (mulberry32).\n *\n * `Math.random()` would make a \"reproducible\" subsample a different corpus on\n * every run, which is the one property this must not have.\n */\nfunction seededRandom(seed: number): () => number {\n\tlet a = seed >>> 0;\n\treturn () => {\n\t\ta = (a + 0x6d2b79f5) >>> 0;\n\t\tlet t = a;\n\t\tt = Math.imul(t ^ (t >>> 15), t | 1);\n\t\tt ^= t + Math.imul(t ^ (t >>> 7), t | 61);\n\t\treturn ((t ^ (t >>> 14)) >>> 0) / 4294967296;\n\t};\n}\n\n/**\n * Files that describe this eval rather than being searched by it.\n *\n * The fixtures hold all 62 query strings verbatim, so every query is a perfect\n * lexical match against its own entry, and the design note quotes the same\n * queries while discussing the classes they belong to. Measured before this\n * exclusion existed: **56 of 62 queries had one of these files in the top 10,\n * 27 of 62 had one as the #1 result, and they consumed 133 of the 620\n * top-10 slots** — a fifth of the window, spent on the eval reading itself.\n *\n * That is not a ranking artifact a reranker can fix: it displaces real answers\n * out of the window entirely, which is why two boundary-class queries were\n * absent from the top *50* rather than merely buried. Retrieving your own\n * question is not retrieval, so the corpus is scored without them.\n *\n * Removed from the pinned worktree before indexing, never from the repo — and\n * recorded in the run's provenance so a score is never silently taken against\n * a different corpus than it claims.\n */\nexport const CORPUS_EXCLUSIONS: readonly string[] = [\n\t\"packages/coding-agent/test/fixtures/search-eval.json\",\n\t\"packages/coding-agent/test/fixtures/search-eval-live.json\",\n\t\"packages/coding-agent/test/fixtures/search-eval-baseline.json\",\n\t\"docs/hybrid-retrieval-design.md\",\n];\n\n/**\n * Materialize the corpus to evaluate.\n *\n * With a `ref`, checks out a detached worktree at that commit so the corpus is\n * byte-identical on every rerun. Without one, falls back to the live working\n * tree and reports `dirty` so the record shows the run was not reproducible.\n *\n * `subsample` shrinks the corpus to a chunk budget, for screening runs where a\n * full arm costs too much to iterate on. It keeps every gold-bearing file and\n * draws distractors deterministically. This is a real change to what is being\n * measured — a smaller distractor pool makes retrieval easier and inflates\n * every metric — so it is recorded in the record and folded into the worktree\n * path, and it needs a `ref`.\n */\nexport function pinCorpus(repoRoot: string, ref: string | undefined, subsample?: CorpusSubsampleRequest): PinnedCorpus {\n\tconst dirty = git(repoRoot, [\"status\", \"--porcelain\"]).length > 0;\n\tif (!ref) {\n\t\tif (subsample) {\n\t\t\t// Subsampling deletes files. Against the live checkout that is the\n\t\t\t// user's source tree, so this refuses rather than asks.\n\t\t\tthrow new Error(\n\t\t\t\t\"subsampling requires --corpus-ref: it deletes files, and the working tree is not ours to cut\",\n\t\t\t);\n\t\t}\n\t\t// The live working tree is the user's checkout; deleting files from it to\n\t\t// tidy a measurement would be an unforgivable trade. Working-tree runs\n\t\t// are already stamped non-reproducible, so they carry the contamination.\n\t\treturn {\n\t\t\tcwd: repoRoot,\n\t\t\tsha: git(repoRoot, [\"rev-parse\", \"HEAD\"]),\n\t\t\tfromWorkingTree: true,\n\t\t\tdirty,\n\t\t\texcluded: [],\n\t\t\tdispose: () => {},\n\t\t};\n\t}\n\n\tconst sha = git(repoRoot, [\"rev-parse\", ref]);\n\t// Deterministic path, not mkdtemp: the embedding store is keyed by a hash of\n\t// the corpus directory, so a fresh temp path every run would re-embed all\n\t// ~17k chunks (minutes) instead of reusing the store built for this exact\n\t// SHA. The worktree is still removed afterwards; only the store persists.\n\t//\n\t// The subsample is part of the key. Without it a screening run and a full\n\t// run at the same SHA would share this path *and* the store derived from it,\n\t// so the second would silently score the first's index — a wrong number that\n\t// looks entirely normal.\n\t// No chunk cap in the key: the file set is cap-independent by construction\n\t// (see `applySubsample`), so cap arms share one worktree. The *store* key\n\t// does carry the cap, because the vectors differ.\n\tconst subsampleKey = subsample ? `-fast${subsample.targetChunks}s${subsample.seed}` : \"\";\n\tconst dir = path.join(tmpdir(), `hoocode-search-eval-${sha.slice(0, 12)}${subsampleKey}`);\n\tif (existsSync(dir)) {\n\t\t// Left behind by an interrupted run — drop it so `worktree add` succeeds.\n\t\ttry {\n\t\t\tgit(repoRoot, [\"worktree\", \"remove\", \"--force\", dir]);\n\t\t} catch {\n\t\t\trmSync(dir, { recursive: true, force: true });\n\t\t\tgit(repoRoot, [\"worktree\", \"prune\"]);\n\t\t}\n\t}\n\tgit(repoRoot, [\"worktree\", \"add\", \"--detach\", dir, sha]);\n\n\tconst excluded: string[] = [];\n\tfor (const rel of CORPUS_EXCLUSIONS) {\n\t\tconst target = path.join(dir, rel);\n\t\tif (existsSync(target)) {\n\t\t\trmSync(target, { force: true });\n\t\t\texcluded.push(rel);\n\t\t}\n\t}\n\n\tconst subsampleInfo = subsample ? applySubsample(dir, subsample) : undefined;\n\n\treturn {\n\t\tcwd: dir,\n\t\tsha,\n\t\tfromWorkingTree: false,\n\t\tdirty: false,\n\t\texcluded,\n\t\tsubsample: subsampleInfo,\n\t\tdispose: () => {\n\t\t\ttry {\n\t\t\t\tgit(repoRoot, [\"worktree\", \"remove\", \"--force\", dir]);\n\t\t\t} catch {\n\t\t\t\trmSync(dir, { recursive: true, force: true });\n\t\t\t}\n\t\t},\n\t};\n}\n\n/** `<binary> --version`, or undefined when it cannot be run. */\nfunction probeBinaryVersion(binaryPath: string | undefined): string | undefined {\n\tif (!binaryPath) return undefined;\n\ttry {\n\t\treturn execFileSync(binaryPath, [\"--version\"], { encoding: \"utf-8\" }).trim();\n\t} catch {\n\t\treturn undefined;\n\t}\n}\n\nexport function collectProvenance(\n\trepoRoot: string,\n\tcorpus: PinnedCorpus,\n\tcorpusRef: string,\n\tservice: EmbsearchService | undefined,\n\tembsearchBinary?: string,\n\thybridService?: EmbsearchService,\n\tmodelDir?: string,\n\ttiming?: EvalProvenance[\"timing\"],\n\tchunkMaxChars?: number,\n): EvalProvenance {\n\tconst state = service?.getState();\n\tconst phase = state?.phase ?? \"absent\";\n\treturn {\n\t\ttimestampMs: Date.now(),\n\t\tcorpusSha: corpus.sha,\n\t\tcorpusRef,\n\t\tcorpusFromWorkingTree: corpus.fromWorkingTree,\n\t\tcorpusDirty: corpus.dirty,\n\t\tcorpusExcluded: corpus.excluded,\n\t\tcorpusSubsample: corpus.subsample,\n\t\tharnessSha: git(repoRoot, [\"rev-parse\", \"HEAD\"]),\n\t\tretrievalSourceHash: hashRetrievalSource(repoRoot),\n\t\tembedder: {\n\t\t\t// `ready` is the only phase the service reaches with a real embedder:\n\t\t\t// it rejects the mock backend at startup, so availability here also\n\t\t\t// certifies the numbers came from a genuine ONNX build.\n\t\t\tavailable: service?.isAvailable() ?? false,\n\t\t\treason: state && \"reason\" in state ? state.reason : undefined,\n\t\t\tchunkCount: state?.phase === \"ready\" ? state.chunkCount : undefined,\n\t\t\tphase,\n\t\t\tbinaryPath: embsearchBinary,\n\t\t\tbinaryVersion: probeBinaryVersion(embsearchBinary),\n\t\t\tmodelId: service?.modelId(),\n\t\t\tmodelDir,\n\t\t},\n\t\tdaemonHybrid: hybridService\n\t\t\t? { available: hybridService.isAvailable(), phase: hybridService.getState().phase }\n\t\t\t: undefined,\n\t\ttiming,\n\t\tchunkMaxChars,\n\t\truntime: { node: process.version, platform: process.platform, arch: process.arch },\n\t};\n}\n\nexport function summarizeGoldSet(dataset: readonly EvalQuery[]): EvalRunRecord[\"goldSet\"] {\n\tconst byClass: Record<string, number> = {};\n\tlet goldSpanCount = 0;\n\tfor (const query of dataset) {\n\t\tbyClass[query.class] = (byClass[query.class] ?? 0) + 1;\n\t\tgoldSpanCount += query.gold.length;\n\t}\n\treturn { queryCount: dataset.length, byClass, goldSpanCount };\n}\n\nexport interface RunEvalSuiteOptions {\n\tcwd: string;\n\tdataset: readonly EvalQuery[];\n\tconfigs: readonly EvalConfig[];\n\tservice?: EmbsearchService;\n\t/** Second service backed by a daemon-side BM25 hybrid store, for the\n\t *  `daemon-hybrid` configs. Absent means those rows are omitted. */\n\thybridService?: EmbsearchService;\n\tonQuery?: (index: number, query: EvalQuery) => void;\n}\n\nexport async function runEvalSuite(options: RunEvalSuiteOptions): Promise<{\n\taggregates: EvalAggregate[];\n\tperQuery: EvalRunRecord[\"perQuery\"];\n}> {\n\tconst { cwd, dataset, configs, service } = options;\n\tconst totals = new Map<string, EvalAggregate>();\n\tconst perQuery: EvalRunRecord[\"perQuery\"] = [];\n\n\tfor (const [index, evalQuery] of dataset.entries()) {\n\t\toptions.onQuery?.(index, evalQuery);\n\t\tconst results = await evaluateQuery(cwd, evalQuery, configs, service, options.hybridService);\n\t\tperQuery.push({ id: evalQuery.id, class: evalQuery.class, results });\n\t\tfor (const result of results) {\n\t\t\tconst total = totals.get(result.label) ?? {\n\t\t\t\tlabel: result.label,\n\t\t\t\trecallAt1: 0,\n\t\t\t\trecallAt5: 0,\n\t\t\t\trecallAt10: 0,\n\t\t\t\trecallAt50: 0,\n\t\t\t\tmrr: 0,\n\t\t\t\tn: 0,\n\t\t\t\tdegraded: 0,\n\t\t\t};\n\t\t\ttotal.recallAt1 += result.recallAt1;\n\t\t\ttotal.recallAt5 += result.recallAt5;\n\t\t\ttotal.recallAt10 += result.recallAt10;\n\t\t\ttotal.recallAt50 += result.recallAt50;\n\t\t\ttotal.mrr += result.mrr;\n\t\t\ttotal.n++;\n\t\t\tif (result.degraded) total.degraded++;\n\t\t\ttotals.set(result.label, total);\n\t\t}\n\t}\n\n\tconst aggregates = configs\n\t\t.map((config) => totals.get(config.label))\n\t\t.filter((total): total is EvalAggregate => total !== undefined)\n\t\t.map((total) => ({\n\t\t\t...total,\n\t\t\trecallAt1: total.recallAt1 / total.n,\n\t\t\trecallAt5: total.recallAt5 / total.n,\n\t\t\trecallAt10: total.recallAt10 / total.n,\n\t\t\trecallAt50: total.recallAt50 / total.n,\n\t\t\tmrr: total.mrr / total.n,\n\t\t}));\n\n\treturn { aggregates, perQuery };\n}\n\nexport function formatAggregateTable(aggregates: readonly EvalAggregate[]): string {\n\tconst pct = (x: number) => `${Math.round(x * 100)}%`.padStart(5);\n\tconst lines = [\n\t\t\"config           |  R@1  |  R@5  | R@10  | R@50  |  MRR  | notes\",\n\t\t\"-----------------|-------|-------|-------|-------|-------|------\",\n\t];\n\tfor (const a of aggregates) {\n\t\tconst notes = a.degraded === a.n ? \"degraded to lexical\" : a.degraded > 0 ? `${a.degraded}/${a.n} degraded` : \"\";\n\t\tlines.push(\n\t\t\t`${a.label.padEnd(16)} | ${pct(a.recallAt1)} | ${pct(a.recallAt5)} | ${pct(a.recallAt10)} | ` +\n\t\t\t\t`${pct(a.recallAt50)} | ${a.mrr.toFixed(3)} | ${notes}`,\n\t\t);\n\t}\n\treturn lines.join(\"\\n\");\n}\n"]}