@kolisachint/hoocode-agent 0.4.168 → 0.5.0

package/CHANGELOG.md

@@ -1,5 +1,7 @@
 # Changelog
 
+## [0.5.0] - 2026-08-08
+
 ## [0.4.168] - 2026-08-08
 
 ## [0.4.167] - 2026-08-08

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

@@ -10,7 +10,7 @@
  * Bump CHUNKER_VERSION when the strategy changes — a version mismatch in the
  * sidecar triggers a clean rebuild of the store.
  */
-export declare const CHUNKER_VERSION = 1;
+export declare const CHUNKER_VERSION = 2;
 export interface Chunk {
     /** `relpath#index` — the id stored in the vector index. */
     id: string;

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

@@ -1 +1 @@
-{"version":3,"file":"chunker.d.ts","sourceRoot":"","sources":["../../../src/core/embsearch/chunker.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,eAAO,MAAM,eAAe,IAAI,CAAC;AASjC,MAAM,WAAW,KAAK;IACrB,6DAA2D;IAC3D,EAAE,EAAE,MAAM,CAAC;IACX,iCAAiC;IACjC,IAAI,EAAE,MAAM,CAAC;IACb,oCAAoC;IACpC,SAAS,EAAE,MAAM,CAAC;IAClB,kCAAkC;IAClC,OAAO,EAAE,MAAM,CAAC;CAChB;AAQD;;;GAGG;AACH,wBAAgB,SAAS,CAAC,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,KAAK,EAAE,CAoCnE","sourcesContent":["/**\n * File chunking for semantic indexing.\n *\n * Splits a file into overlapping line windows, each capped by characters so a\n * chunk stays within the embedding model's effective token window (MiniLM\n * truncates around 256 tokens ≈ 1000 chars). Chunk ids are `relpath#index`;\n * the id → line-range mapping is kept in the sidecar metadata (index-meta.ts)\n * so search hits can be rendered as `path:start-end`.\n *\n * Bump CHUNKER_VERSION when the strategy changes — a version mismatch in the\n * sidecar triggers a clean rebuild of the store.\n */\n\nexport const CHUNKER_VERSION = 1;\n\n/** Target lines per chunk. */\nconst CHUNK_LINES = 60;\n/** Overlapping lines between consecutive chunks, for context continuity. */\nconst CHUNK_OVERLAP_LINES = 10;\n/** Hard character cap per chunk (MiniLM truncates ~256 tokens ≈ 1000 chars). */\nconst CHUNK_MAX_CHARS = 1000;\n\nexport interface Chunk {\n\t/** `relpath#index` — the id stored in the vector index. */\n\tid: string;\n\t/** Text sent to the embedder. */\n\ttext: string;\n\t/** 1-based inclusive start line. */\n\tstartLine: number;\n\t/** 1-based inclusive end line. */\n\tendLine: number;\n}\n\n/** Heuristic binary sniff: NUL byte in the first 8KB. */\nfunction looksBinary(content: string): boolean {\n\tconst probe = content.slice(0, 8192);\n\treturn probe.includes(\"\\u0000\");\n}\n\n/**\n * Split `content` into chunks. `relPath` becomes the id prefix. Returns an\n * empty array for empty or binary-looking content.\n */\nexport function chunkFile(relPath: string, content: string): Chunk[] {\n\tif (!content.trim() || looksBinary(content)) return [];\n\tconst lines = content.replace(/\\r\\n/g, \"\\n\").replace(/\\r/g, \"\\n\").split(\"\\n\");\n\tconst chunks: Chunk[] = [];\n\tlet start = 0; // 0-based\n\tlet index = 0;\n\n\twhile (start < lines.length) {\n\t\tlet end = start; // exclusive\n\t\tlet chars = 0;\n\t\twhile (end < lines.length && end - start < CHUNK_LINES) {\n\t\t\tconst lineLen = lines[end].length + 1;\n\t\t\tif (chars + lineLen > CHUNK_MAX_CHARS && end > start) break;\n\t\t\tchars += lineLen;\n\t\t\tend++;\n\t\t}\n\t\tlet text = lines.slice(start, end).join(\"\\n\").trim();\n\t\tif (text.length > CHUNK_MAX_CHARS) {\n\t\t\t// Oversized chunk (e.g. long minified line): keep the prefix. The\n\t\t\t// underlying model would truncate anyway, so this stays bounded.\n\t\t\ttext = text.slice(0, CHUNK_MAX_CHARS);\n\t\t}\n\t\tif (text) {\n\t\t\tchunks.push({\n\t\t\t\tid: `${relPath}#${index}`,\n\t\t\t\ttext,\n\t\t\t\tstartLine: start + 1,\n\t\t\t\tendLine: end,\n\t\t\t});\n\t\t\tindex++;\n\t\t}\n\t\tif (end >= lines.length) break;\n\t\t// Step forward with overlap, but always make progress.\n\t\tstart = Math.max(end - CHUNK_OVERLAP_LINES, start + 1);\n\t}\n\treturn chunks;\n}\n"]}
+{"version":3,"file":"chunker.d.ts","sourceRoot":"","sources":["../../../src/core/embsearch/chunker.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,eAAO,MAAM,eAAe,IAAI,CAAC;AAmBjC,MAAM,WAAW,KAAK;IACrB,6DAA2D;IAC3D,EAAE,EAAE,MAAM,CAAC;IACX,iCAAiC;IACjC,IAAI,EAAE,MAAM,CAAC;IACb,oCAAoC;IACpC,SAAS,EAAE,MAAM,CAAC;IAClB,kCAAkC;IAClC,OAAO,EAAE,MAAM,CAAC;CAChB;AAQD;;;GAGG;AACH,wBAAgB,SAAS,CAAC,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,KAAK,EAAE,CA8CnE","sourcesContent":["/**\n * File chunking for semantic indexing.\n *\n * Splits a file into overlapping line windows, each capped by characters so a\n * chunk stays within the embedding model's effective token window (MiniLM\n * truncates around 256 tokens ≈ 1000 chars). Chunk ids are `relpath#index`;\n * the id → line-range mapping is kept in the sidecar metadata (index-meta.ts)\n * so search hits can be rendered as `path:start-end`.\n *\n * Bump CHUNKER_VERSION when the strategy changes — a version mismatch in the\n * sidecar triggers a clean rebuild of the store.\n */\n\nexport const CHUNKER_VERSION = 2;\n\n/**\n * Target lines per chunk.\n *\n * Halving this (30 lines / 500 chars) was measured and rejected. The theory\n * was dilution — a four-line answer sharing one vector with five neighbouring\n * methods — but sharper chunks cost reach: Recall@50 fell 79% -> 69% on\n * `auto +rr` and the boundary class went from 2 of 4 findable to 0 of 4.\n * A fixed top-k over smaller chunks retrieves less *content*: 50 chunks at\n * ~460 chars sees half the corpus that 50 at ~930 does. Recall@1 ticked up\n * (53% -> 55%), which is the sharpening, and it did not pay for the loss.\n */\nconst CHUNK_LINES = 60;\n/** Overlapping lines between consecutive chunks, for context continuity. */\nconst CHUNK_OVERLAP_LINES = 10;\n/** Hard character cap per chunk (MiniLM truncates ~256 tokens ≈ 1000 chars). */\nconst CHUNK_MAX_CHARS = 1000;\n\nexport interface Chunk {\n\t/** `relpath#index` — the id stored in the vector index. */\n\tid: string;\n\t/** Text sent to the embedder. */\n\ttext: string;\n\t/** 1-based inclusive start line. */\n\tstartLine: number;\n\t/** 1-based inclusive end line. */\n\tendLine: number;\n}\n\n/** Heuristic binary sniff: NUL byte in the first 8KB. */\nfunction looksBinary(content: string): boolean {\n\tconst probe = content.slice(0, 8192);\n\treturn probe.includes(\"\\u0000\");\n}\n\n/**\n * Split `content` into chunks. `relPath` becomes the id prefix. Returns an\n * empty array for empty or binary-looking content.\n */\nexport function chunkFile(relPath: string, content: string): Chunk[] {\n\tif (!content.trim() || looksBinary(content)) return [];\n\tconst lines = content.replace(/\\r\\n/g, \"\\n\").replace(/\\r/g, \"\\n\").split(\"\\n\");\n\tconst chunks: Chunk[] = [];\n\tlet start = 0; // 0-based\n\tlet index = 0;\n\n\twhile (start < lines.length) {\n\t\tlet end = start; // exclusive\n\t\tlet chars = 0;\n\t\twhile (end < lines.length && end - start < CHUNK_LINES) {\n\t\t\tconst lineLen = lines[end].length + 1;\n\t\t\tif (chars + lineLen > CHUNK_MAX_CHARS && end > start) break;\n\t\t\tchars += lineLen;\n\t\t\tend++;\n\t\t}\n\t\t// Trim whole blank lines off both ends, and narrow the recorded range to\n\t\t// match. Trimming the joined string instead left `startLine`/`endLine`\n\t\t// claiming lines whose text was never embedded — 31% of chunks in this\n\t\t// repo, up to 3 lines each. Those lines were unfindable by the vector\n\t\t// leg while still being handed to the context assembler as if they were\n\t\t// part of the chunk.\n\t\tlet from = start;\n\t\tlet to = end; // exclusive\n\t\twhile (from < to && lines[from].trim() === \"\") from++;\n\t\twhile (to > from && lines[to - 1].trim() === \"\") to--;\n\t\tlet text = lines.slice(from, to).join(\"\\n\").trim();\n\t\tif (text.length > CHUNK_MAX_CHARS) {\n\t\t\t// Oversized chunk (e.g. long minified line): keep the prefix. The\n\t\t\t// underlying model would truncate anyway, so this stays bounded.\n\t\t\ttext = text.slice(0, CHUNK_MAX_CHARS);\n\t\t}\n\t\tif (text) {\n\t\t\tchunks.push({\n\t\t\t\tid: `${relPath}#${index}`,\n\t\t\t\ttext,\n\t\t\t\tstartLine: from + 1,\n\t\t\t\tendLine: to,\n\t\t\t});\n\t\t\tindex++;\n\t\t}\n\t\tif (end >= lines.length) break;\n\t\t// Step forward with overlap, but always make progress.\n\t\tstart = Math.max(end - CHUNK_OVERLAP_LINES, start + 1);\n\t}\n\treturn chunks;\n}\n"]}

package/dist/core/embsearch/chunker.js

@@ -10,8 +10,18 @@
  * Bump CHUNKER_VERSION when the strategy changes — a version mismatch in the
  * sidecar triggers a clean rebuild of the store.
  */
-export const CHUNKER_VERSION = 1;
-/** Target lines per chunk. */
+export const CHUNKER_VERSION = 2;
+/**
+ * Target lines per chunk.
+ *
+ * Halving this (30 lines / 500 chars) was measured and rejected. The theory
+ * was dilution — a four-line answer sharing one vector with five neighbouring
+ * methods — but sharper chunks cost reach: Recall@50 fell 79% -> 69% on
+ * `auto +rr` and the boundary class went from 2 of 4 findable to 0 of 4.
+ * A fixed top-k over smaller chunks retrieves less *content*: 50 chunks at
+ * ~460 chars sees half the corpus that 50 at ~930 does. Recall@1 ticked up
+ * (53% -> 55%), which is the sharpening, and it did not pay for the loss.
+ */
 const CHUNK_LINES = 60;
 /** Overlapping lines between consecutive chunks, for context continuity. */
 const CHUNK_OVERLAP_LINES = 10;
@@ -43,7 +53,19 @@ export function chunkFile(relPath, content) {
             chars += lineLen;
             end++;
         }
-        let text = lines.slice(start, end).join("\n").trim();
+        // Trim whole blank lines off both ends, and narrow the recorded range to
+        // match. Trimming the joined string instead left `startLine`/`endLine`
+        // claiming lines whose text was never embedded — 31% of chunks in this
+        // repo, up to 3 lines each. Those lines were unfindable by the vector
+        // leg while still being handed to the context assembler as if they were
+        // part of the chunk.
+        let from = start;
+        let to = end; // exclusive
+        while (from < to && lines[from].trim() === "")
+            from++;
+        while (to > from && lines[to - 1].trim() === "")
+            to--;
+        let text = lines.slice(from, to).join("\n").trim();
         if (text.length > CHUNK_MAX_CHARS) {
             // Oversized chunk (e.g. long minified line): keep the prefix. The
             // underlying model would truncate anyway, so this stays bounded.
@@ -53,8 +75,8 @@ export function chunkFile(relPath, content) {
             chunks.push({
                 id: `${relPath}#${index}`,
                 text,
-                startLine: start + 1,
-                endLine: end,
+                startLine: from + 1,
+                endLine: to,
             });
             index++;
         }

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

@@ -1 +1 @@
-{"version":3,"file":"chunker.js","sourceRoot":"","sources":["../../../src/core/embsearch/chunker.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,CAAC;AAEjC,8BAA8B;AAC9B,MAAM,WAAW,GAAG,EAAE,CAAC;AACvB,4EAA4E;AAC5E,MAAM,mBAAmB,GAAG,EAAE,CAAC;AAC/B,kFAAgF;AAChF,MAAM,eAAe,GAAG,IAAI,CAAC;AAa7B,yDAAyD;AACzD,SAAS,WAAW,CAAC,OAAe,EAAW;IAC9C,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC;IACrC,OAAO,KAAK,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;AAAA,CAChC;AAED;;;GAGG;AACH,MAAM,UAAU,SAAS,CAAC,OAAe,EAAE,OAAe,EAAW;IACpE,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,WAAW,CAAC,OAAO,CAAC;QAAE,OAAO,EAAE,CAAC;IACvD,MAAM,KAAK,GAAG,OAAO,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC9E,MAAM,MAAM,GAAY,EAAE,CAAC;IAC3B,IAAI,KAAK,GAAG,CAAC,CAAC,CAAC,UAAU;IACzB,IAAI,KAAK,GAAG,CAAC,CAAC;IAEd,OAAO,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;QAC7B,IAAI,GAAG,GAAG,KAAK,CAAC,CAAC,YAAY;QAC7B,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,OAAO,GAAG,GAAG,KAAK,CAAC,MAAM,IAAI,GAAG,GAAG,KAAK,GAAG,WAAW,EAAE,CAAC;YACxD,MAAM,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;YACtC,IAAI,KAAK,GAAG,OAAO,GAAG,eAAe,IAAI,GAAG,GAAG,KAAK;gBAAE,MAAM;YAC5D,KAAK,IAAI,OAAO,CAAC;YACjB,GAAG,EAAE,CAAC;QACP,CAAC;QACD,IAAI,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;QACrD,IAAI,IAAI,CAAC,MAAM,GAAG,eAAe,EAAE,CAAC;YACnC,kEAAkE;YAClE,iEAAiE;YACjE,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,eAAe,CAAC,CAAC;QACvC,CAAC;QACD,IAAI,IAAI,EAAE,CAAC;YACV,MAAM,CAAC,IAAI,CAAC;gBACX,EAAE,EAAE,GAAG,OAAO,IAAI,KAAK,EAAE;gBACzB,IAAI;gBACJ,SAAS,EAAE,KAAK,GAAG,CAAC;gBACpB,OAAO,EAAE,GAAG;aACZ,CAAC,CAAC;YACH,KAAK,EAAE,CAAC;QACT,CAAC;QACD,IAAI,GAAG,IAAI,KAAK,CAAC,MAAM;YAAE,MAAM;QAC/B,uDAAuD;QACvD,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,mBAAmB,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC;IACxD,CAAC;IACD,OAAO,MAAM,CAAC;AAAA,CACd","sourcesContent":["/**\n * File chunking for semantic indexing.\n *\n * Splits a file into overlapping line windows, each capped by characters so a\n * chunk stays within the embedding model's effective token window (MiniLM\n * truncates around 256 tokens ≈ 1000 chars). Chunk ids are `relpath#index`;\n * the id → line-range mapping is kept in the sidecar metadata (index-meta.ts)\n * so search hits can be rendered as `path:start-end`.\n *\n * Bump CHUNKER_VERSION when the strategy changes — a version mismatch in the\n * sidecar triggers a clean rebuild of the store.\n */\n\nexport const CHUNKER_VERSION = 1;\n\n/** Target lines per chunk. */\nconst CHUNK_LINES = 60;\n/** Overlapping lines between consecutive chunks, for context continuity. */\nconst CHUNK_OVERLAP_LINES = 10;\n/** Hard character cap per chunk (MiniLM truncates ~256 tokens ≈ 1000 chars). */\nconst CHUNK_MAX_CHARS = 1000;\n\nexport interface Chunk {\n\t/** `relpath#index` — the id stored in the vector index. */\n\tid: string;\n\t/** Text sent to the embedder. */\n\ttext: string;\n\t/** 1-based inclusive start line. */\n\tstartLine: number;\n\t/** 1-based inclusive end line. */\n\tendLine: number;\n}\n\n/** Heuristic binary sniff: NUL byte in the first 8KB. */\nfunction looksBinary(content: string): boolean {\n\tconst probe = content.slice(0, 8192);\n\treturn probe.includes(\"\\u0000\");\n}\n\n/**\n * Split `content` into chunks. `relPath` becomes the id prefix. Returns an\n * empty array for empty or binary-looking content.\n */\nexport function chunkFile(relPath: string, content: string): Chunk[] {\n\tif (!content.trim() || looksBinary(content)) return [];\n\tconst lines = content.replace(/\\r\\n/g, \"\\n\").replace(/\\r/g, \"\\n\").split(\"\\n\");\n\tconst chunks: Chunk[] = [];\n\tlet start = 0; // 0-based\n\tlet index = 0;\n\n\twhile (start < lines.length) {\n\t\tlet end = start; // exclusive\n\t\tlet chars = 0;\n\t\twhile (end < lines.length && end - start < CHUNK_LINES) {\n\t\t\tconst lineLen = lines[end].length + 1;\n\t\t\tif (chars + lineLen > CHUNK_MAX_CHARS && end > start) break;\n\t\t\tchars += lineLen;\n\t\t\tend++;\n\t\t}\n\t\tlet text = lines.slice(start, end).join(\"\\n\").trim();\n\t\tif (text.length > CHUNK_MAX_CHARS) {\n\t\t\t// Oversized chunk (e.g. long minified line): keep the prefix. The\n\t\t\t// underlying model would truncate anyway, so this stays bounded.\n\t\t\ttext = text.slice(0, CHUNK_MAX_CHARS);\n\t\t}\n\t\tif (text) {\n\t\t\tchunks.push({\n\t\t\t\tid: `${relPath}#${index}`,\n\t\t\t\ttext,\n\t\t\t\tstartLine: start + 1,\n\t\t\t\tendLine: end,\n\t\t\t});\n\t\t\tindex++;\n\t\t}\n\t\tif (end >= lines.length) break;\n\t\t// Step forward with overlap, but always make progress.\n\t\tstart = Math.max(end - CHUNK_OVERLAP_LINES, start + 1);\n\t}\n\treturn chunks;\n}\n"]}
+{"version":3,"file":"chunker.js","sourceRoot":"","sources":["../../../src/core/embsearch/chunker.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,CAAC;AAEjC;;;;;;;;;;GAUG;AACH,MAAM,WAAW,GAAG,EAAE,CAAC;AACvB,4EAA4E;AAC5E,MAAM,mBAAmB,GAAG,EAAE,CAAC;AAC/B,kFAAgF;AAChF,MAAM,eAAe,GAAG,IAAI,CAAC;AAa7B,yDAAyD;AACzD,SAAS,WAAW,CAAC,OAAe,EAAW;IAC9C,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC;IACrC,OAAO,KAAK,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;AAAA,CAChC;AAED;;;GAGG;AACH,MAAM,UAAU,SAAS,CAAC,OAAe,EAAE,OAAe,EAAW;IACpE,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,WAAW,CAAC,OAAO,CAAC;QAAE,OAAO,EAAE,CAAC;IACvD,MAAM,KAAK,GAAG,OAAO,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC9E,MAAM,MAAM,GAAY,EAAE,CAAC;IAC3B,IAAI,KAAK,GAAG,CAAC,CAAC,CAAC,UAAU;IACzB,IAAI,KAAK,GAAG,CAAC,CAAC;IAEd,OAAO,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;QAC7B,IAAI,GAAG,GAAG,KAAK,CAAC,CAAC,YAAY;QAC7B,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,OAAO,GAAG,GAAG,KAAK,CAAC,MAAM,IAAI,GAAG,GAAG,KAAK,GAAG,WAAW,EAAE,CAAC;YACxD,MAAM,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;YACtC,IAAI,KAAK,GAAG,OAAO,GAAG,eAAe,IAAI,GAAG,GAAG,KAAK;gBAAE,MAAM;YAC5D,KAAK,IAAI,OAAO,CAAC;YACjB,GAAG,EAAE,CAAC;QACP,CAAC;QACD,yEAAyE;QACzE,uEAAuE;QACvE,yEAAuE;QACvE,sEAAsE;QACtE,wEAAwE;QACxE,qBAAqB;QACrB,IAAI,IAAI,GAAG,KAAK,CAAC;QACjB,IAAI,EAAE,GAAG,GAAG,CAAC,CAAC,YAAY;QAC1B,OAAO,IAAI,GAAG,EAAE,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE;YAAE,IAAI,EAAE,CAAC;QACtD,OAAO,EAAE,GAAG,IAAI,IAAI,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE;YAAE,EAAE,EAAE,CAAC;QACtD,IAAI,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;QACnD,IAAI,IAAI,CAAC,MAAM,GAAG,eAAe,EAAE,CAAC;YACnC,kEAAkE;YAClE,iEAAiE;YACjE,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,eAAe,CAAC,CAAC;QACvC,CAAC;QACD,IAAI,IAAI,EAAE,CAAC;YACV,MAAM,CAAC,IAAI,CAAC;gBACX,EAAE,EAAE,GAAG,OAAO,IAAI,KAAK,EAAE;gBACzB,IAAI;gBACJ,SAAS,EAAE,IAAI,GAAG,CAAC;gBACnB,OAAO,EAAE,EAAE;aACX,CAAC,CAAC;YACH,KAAK,EAAE,CAAC;QACT,CAAC;QACD,IAAI,GAAG,IAAI,KAAK,CAAC,MAAM;YAAE,MAAM;QAC/B,uDAAuD;QACvD,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,mBAAmB,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC;IACxD,CAAC;IACD,OAAO,MAAM,CAAC;AAAA,CACd","sourcesContent":["/**\n * File chunking for semantic indexing.\n *\n * Splits a file into overlapping line windows, each capped by characters so a\n * chunk stays within the embedding model's effective token window (MiniLM\n * truncates around 256 tokens ≈ 1000 chars). Chunk ids are `relpath#index`;\n * the id → line-range mapping is kept in the sidecar metadata (index-meta.ts)\n * so search hits can be rendered as `path:start-end`.\n *\n * Bump CHUNKER_VERSION when the strategy changes — a version mismatch in the\n * sidecar triggers a clean rebuild of the store.\n */\n\nexport const CHUNKER_VERSION = 2;\n\n/**\n * Target lines per chunk.\n *\n * Halving this (30 lines / 500 chars) was measured and rejected. The theory\n * was dilution — a four-line answer sharing one vector with five neighbouring\n * methods — but sharper chunks cost reach: Recall@50 fell 79% -> 69% on\n * `auto +rr` and the boundary class went from 2 of 4 findable to 0 of 4.\n * A fixed top-k over smaller chunks retrieves less *content*: 50 chunks at\n * ~460 chars sees half the corpus that 50 at ~930 does. Recall@1 ticked up\n * (53% -> 55%), which is the sharpening, and it did not pay for the loss.\n */\nconst CHUNK_LINES = 60;\n/** Overlapping lines between consecutive chunks, for context continuity. */\nconst CHUNK_OVERLAP_LINES = 10;\n/** Hard character cap per chunk (MiniLM truncates ~256 tokens ≈ 1000 chars). */\nconst CHUNK_MAX_CHARS = 1000;\n\nexport interface Chunk {\n\t/** `relpath#index` — the id stored in the vector index. */\n\tid: string;\n\t/** Text sent to the embedder. */\n\ttext: string;\n\t/** 1-based inclusive start line. */\n\tstartLine: number;\n\t/** 1-based inclusive end line. */\n\tendLine: number;\n}\n\n/** Heuristic binary sniff: NUL byte in the first 8KB. */\nfunction looksBinary(content: string): boolean {\n\tconst probe = content.slice(0, 8192);\n\treturn probe.includes(\"\\u0000\");\n}\n\n/**\n * Split `content` into chunks. `relPath` becomes the id prefix. Returns an\n * empty array for empty or binary-looking content.\n */\nexport function chunkFile(relPath: string, content: string): Chunk[] {\n\tif (!content.trim() || looksBinary(content)) return [];\n\tconst lines = content.replace(/\\r\\n/g, \"\\n\").replace(/\\r/g, \"\\n\").split(\"\\n\");\n\tconst chunks: Chunk[] = [];\n\tlet start = 0; // 0-based\n\tlet index = 0;\n\n\twhile (start < lines.length) {\n\t\tlet end = start; // exclusive\n\t\tlet chars = 0;\n\t\twhile (end < lines.length && end - start < CHUNK_LINES) {\n\t\t\tconst lineLen = lines[end].length + 1;\n\t\t\tif (chars + lineLen > CHUNK_MAX_CHARS && end > start) break;\n\t\t\tchars += lineLen;\n\t\t\tend++;\n\t\t}\n\t\t// Trim whole blank lines off both ends, and narrow the recorded range to\n\t\t// match. Trimming the joined string instead left `startLine`/`endLine`\n\t\t// claiming lines whose text was never embedded — 31% of chunks in this\n\t\t// repo, up to 3 lines each. Those lines were unfindable by the vector\n\t\t// leg while still being handed to the context assembler as if they were\n\t\t// part of the chunk.\n\t\tlet from = start;\n\t\tlet to = end; // exclusive\n\t\twhile (from < to && lines[from].trim() === \"\") from++;\n\t\twhile (to > from && lines[to - 1].trim() === \"\") to--;\n\t\tlet text = lines.slice(from, to).join(\"\\n\").trim();\n\t\tif (text.length > CHUNK_MAX_CHARS) {\n\t\t\t// Oversized chunk (e.g. long minified line): keep the prefix. The\n\t\t\t// underlying model would truncate anyway, so this stays bounded.\n\t\t\ttext = text.slice(0, CHUNK_MAX_CHARS);\n\t\t}\n\t\tif (text) {\n\t\t\tchunks.push({\n\t\t\t\tid: `${relPath}#${index}`,\n\t\t\t\ttext,\n\t\t\t\tstartLine: from + 1,\n\t\t\t\tendLine: to,\n\t\t\t});\n\t\t\tindex++;\n\t\t}\n\t\tif (end >= lines.length) break;\n\t\t// Step forward with overlap, but always make progress.\n\t\tstart = Math.max(end - CHUNK_OVERLAP_LINES, start + 1);\n\t}\n\treturn chunks;\n}\n"]}

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

@@ -14,6 +14,14 @@ export interface EmbSearchDaemonInfo {
     modelId: string;
     dim: number;
     count: number;
+    /**
+     * Whether the open store carries a BM25 index alongside its vectors.
+     *
+     * Fixed when the store is created — passing `--hybrid` at a non-hybrid
+     * store only warns — so this is the only reliable way to find out, and the
+     * signal that an existing store has to be rebuilt rather than reused.
+     */
+    hybrid?: boolean;
     /**
      * Whether this daemon can actually serve `rerank`, or `undefined` from a
      * daemon too old to say.

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

@@ -1 +1 @@
-{"version":3,"file":"client.d.ts","sourceRoot":"","sources":["../../../src/core/embsearch/client.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAIH,MAAM,WAAW,eAAe;IAC/B,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,mBAAmB;IACnC,OAAO,EAAE,MAAM,CAAC;IAChB,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd;;;;;;;;OAQG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;CACjB;AAED,oDAAoD;AACpD,MAAM,WAAW,sBAAsB;IACtC,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;CACb;AAED;wDACwD;AACxD,MAAM,WAAW,qBAAqB;IACrC,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,mBAAmB;IACnC,QAAQ,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,MAAM,CAAC;CAChB;AAED,wEAAwE;AACxE,MAAM,MAAM,eAAe,GAAG,OAAO,GAAG,SAAS,GAAG,QAAQ,CAAC;AAE7D,MAAM,WAAW,sBAAsB;IACtC,6EAA6E;IAC7E,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,sCAAsC;IACtC,UAAU,EAAE,MAAM,CAAC;IACnB,0CAA0C;IAC1C,SAAS,EAAE,MAAM,CAAC;IAClB,6DAA6D;IAC7D,MAAM,CAAC,EAAE,QAAQ,GAAG,KAAK,GAAG,WAAW,CAAC;CACxC;AA0BD,qBAAa,eAAe;IAC3B,OAAO,CAAC,IAAI,CAAiC;IAC7C,OAAO,CAAC,KAAK,CAAiB;IAC9B,OAAO,CAAC,MAAM,CAAM;IACpB,OAAO,CAAC,MAAM,CAAS;IACvB,OAAO,CAAC,YAAY,CAAgB;IAEpC,YAAY,IAAI,EAAE,sBAAsB,EA+BvC;IAED,6DAA6D;IAC7D,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAErB;IAED,IAAI,QAAQ,IAAI,OAAO,CAEtB;IAED,OAAO,CAAC,QAAQ;IAqBhB,OAAO,CAAC,IAAI;IAQZ;;;;;;;;;;;;OAYG;IACG,KAAK,CAAC,IAAI,EAAE,MAAM,EAAE,CAAC,SAAK,EAAE,SAAS,GAAE,eAAyB,GAAG,OAAO,CAAC,eAAe,EAAE,CAAC,CAKlG;IAED;;;;;;;;OAQG;IACG,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,sBAAsB,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,qBAAqB,EAAE,CAAC,CAG3G;IAED;;;;OAIG;IACG,IAAI,CAAC,KAAK,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAGnF;IAED,qEAAqE;IAC/D,IAAI,IAAI,OAAO,CAAC,mBAAmB,CAAC,CAGzC;IAED,yDAAyD;IACnD,MAAM,CAAC,EAAE,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,CAGzC;IAED,uDAAuD;IACjD,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAE7B;IAED,gDAAgD;IAC1C,IAAI,IAAI,OAAO,CAAC,IAAI,CAAC,CAE1B;IAED,+DAA+D;IACzD,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAI3B;CACD","sourcesContent":["/**\n * Client for the `embsearch` stdio daemon (vendored from\n * github.com/kolisachint/embeddingsearchtools ts/client.ts).\n *\n * Spawns `embsearch serve` once and talks newline-delimited JSON over its\n * stdin/stdout. The process stays alive so the model and index load a single\n * time; every query after startup is hot.\n */\n\nimport { type ChildProcessWithoutNullStreams, spawn } from \"child_process\";\n\nexport interface EmbSearchResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchDaemonInfo {\n\tmodelId: string;\n\tdim: number;\n\tcount: number;\n\t/**\n\t * Whether this daemon can actually serve `rerank`, or `undefined` from a\n\t * daemon too old to say.\n\t *\n\t * Reported separately from the version because the two came apart:\n\t * released binaries carry the `rerank` op but no longer bundle the ~23 MB\n\t * cross-encoder weights, so a version check alone would advertise a\n\t * reranker that fails on first call.\n\t */\n\trerank?: boolean;\n}\n\n/** One candidate sent for cross-encoder scoring. */\nexport interface EmbSearchRerankPassage {\n\tid: string;\n\ttext: string;\n}\n\n/** A cross-encoder relevance logit. Higher is more relevant, but the scale is\n *  unnormalized and comparable only within one call. */\nexport interface EmbSearchRerankResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchBulkResult {\n\tinserted: number;\n\tupdated: number;\n}\n\n/** Which daemon-side retriever answers a query (embsearch >= 0.2.0). */\nexport type DaemonRetriever = \"dense\" | \"lexical\" | \"hybrid\";\n\nexport interface EmbSearchClientOptions {\n\t/** Open/create the store with a BM25 lexical index alongside the vectors. */\n\thybrid?: boolean;\n\t/** Path to the `embsearch` binary. */\n\tbinaryPath: string;\n\t/** Store directory passed as `--path`. */\n\tstorePath: string;\n\t/** Metric for a freshly created store. Default: \"cosine\". */\n\tmetric?: \"cosine\" | \"dot\" | \"euclidean\";\n}\n\ninterface Pending {\n\tresolve: (value: EmbSearchRawResponse) => void;\n\treject: (err: Error) => void;\n}\n\ninterface EmbSearchRawResponse {\n\tok: boolean;\n\terror?: string;\n\tresults?: EmbSearchResult[];\n\tinserted?: boolean;\n\tremoved?: boolean;\n\tcount?: number;\n\tinserted_count?: number;\n\tupdated_count?: number;\n\tmodel_id?: string;\n\tdim?: number;\n\t/** Cross-encoder scores. Separate from `results` because the daemon's\n\t *  logits are not comparable to retrieval scores. */\n\treranked?: EmbSearchRerankResult[];\n\t/** `info` only: whether `rerank` will work. Absent on daemons too old to\n\t *  report it. */\n\trerank?: boolean;\n}\n\nexport class EmbSearchClient {\n\tprivate proc: ChildProcessWithoutNullStreams;\n\tprivate queue: Pending[] = [];\n\tprivate buffer = \"\";\n\tprivate closed = false;\n\tprivate readyPromise: Promise<void>;\n\n\tconstructor(opts: EmbSearchClientOptions) {\n\t\tconst args = [\"serve\", \"--path\", opts.storePath];\n\t\tif (opts.metric) args.push(\"--metric\", opts.metric);\n\t\t// Hybrid-ness is fixed when a store is created: passing --hybrid against\n\t\t// an existing non-hybrid store warns and is ignored daemon-side, so a\n\t\t// hybrid store needs its own directory.\n\t\tif (opts.hybrid) args.push(\"--hybrid\");\n\n\t\tthis.proc = spawn(opts.binaryPath, args, { stdio: [\"pipe\", \"pipe\", \"pipe\"] });\n\n\t\tthis.proc.stdout.setEncoding(\"utf8\");\n\t\tthis.proc.stdout.on(\"data\", (chunk: string) => this.onStdout(chunk));\n\t\t// Swallow the informational stderr banner; errors surface via responses/exit.\n\t\tthis.proc.stderr.resume();\n\n\t\t// Readiness = the daemon answering a ping (probes the actual request loop).\n\t\tthis.readyPromise = this.send({ op: \"ping\" }).then(() => undefined);\n\t\t// Spawn failures reject the pending ping via the exit handler; mark the\n\t\t// promise handled so a failed spawn doesn't raise an unhandled rejection\n\t\t// before ready() is awaited.\n\t\tthis.readyPromise.catch(() => {});\n\n\t\tthis.proc.on(\"error\", (err) => {\n\t\t\tthis.closed = true;\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t\tthis.proc.on(\"exit\", (code) => {\n\t\t\tthis.closed = true;\n\t\t\tconst err = new Error(`embsearch daemon exited (code ${code})`);\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t}\n\n\t/** Resolves once the daemon has loaded the model + index. */\n\tready(): Promise<void> {\n\t\treturn this.readyPromise;\n\t}\n\n\tget isClosed(): boolean {\n\t\treturn this.closed;\n\t}\n\n\tprivate onStdout(chunk: string): void {\n\t\tthis.buffer += chunk;\n\t\t// Each response is one line; dispatch FIFO against the pending queue.\n\t\tfor (let nl = this.buffer.indexOf(\"\\n\"); nl !== -1; nl = this.buffer.indexOf(\"\\n\")) {\n\t\t\tconst line = this.buffer.slice(0, nl).trim();\n\t\t\tthis.buffer = this.buffer.slice(nl + 1);\n\t\t\tif (!line) continue;\n\t\t\tconst pending = this.queue.shift();\n\t\t\tif (!pending) continue;\n\t\t\tlet msg: EmbSearchRawResponse;\n\t\t\ttry {\n\t\t\t\tmsg = JSON.parse(line) as EmbSearchRawResponse;\n\t\t\t} catch {\n\t\t\t\tpending.reject(new Error(`bad response: ${line}`));\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tif (msg.ok) pending.resolve(msg);\n\t\t\telse pending.reject(new Error(msg.error ?? \"unknown error\"));\n\t\t}\n\t}\n\n\tprivate send(req: Record<string, unknown>): Promise<EmbSearchRawResponse> {\n\t\tif (this.closed) return Promise.reject(new Error(\"embsearch client is closed\"));\n\t\treturn new Promise<EmbSearchRawResponse>((resolve, reject) => {\n\t\t\tthis.queue.push({ resolve, reject });\n\t\t\tthis.proc.stdin.write(`${JSON.stringify(req)}\\n`);\n\t\t});\n\t}\n\n\t/**\n\t * Search for the top-`k` matches for `text`.\n\t *\n\t * `retriever` selects which leg answers:\n\t *  - `dense` (default) — vector search, the historical behaviour;\n\t *  - `lexical` — BM25 only, raw scores, no embedding computed;\n\t *  - `hybrid` — both, pre-fused by the daemon's own RRF constant.\n\t *\n\t * `lexical` and `hybrid` need a store created with `--hybrid`. Prefer\n\t * `lexical` over `hybrid` when fusing here: `hybrid` collapses both legs\n\t * into one RRF score, discarding the per-retriever ranks the trace records\n\t * and preventing n-way fusion with the grep leg.\n\t */\n\tasync query(text: string, k = 10, retriever: DaemonRetriever = \"dense\"): Promise<EmbSearchResult[]> {\n\t\tconst res = await this.send(\n\t\t\tretriever === \"dense\" ? { op: \"query\", text, k } : { op: \"query\", text, k, retriever },\n\t\t);\n\t\treturn res.results ?? [];\n\t}\n\n\t/**\n\t * Score `passages` against `query` with the daemon's cross-encoder and\n\t * return the best `k`, best first.\n\t *\n\t * Passages are sent inline rather than referenced by id: the caller has the\n\t * exact spans it intends to show the model, and a cross-encoder scores the\n\t * text it is given, so sending anything else would score the wrong thing.\n\t * Requires an onnx build with reranker weights (embsearch >= 0.3.0).\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tconst res = await this.send({ op: \"rerank\", query, passages, k });\n\t\treturn res.reranked ?? [];\n\t}\n\n\t/**\n\t * Batched insert-or-replace. One embedding inference for the whole batch —\n\t * the fast path for bulk indexing. Keep batches modest (e.g. 32–64) so a\n\t * concurrent query is not stuck behind a huge inference.\n\t */\n\tasync bulk(items: Array<{ id: string; text: string }>): Promise<EmbSearchBulkResult> {\n\t\tconst res = await this.send({ op: \"bulk\", items });\n\t\treturn { inserted: res.inserted_count ?? 0, updated: res.updated_count ?? 0 };\n\t}\n\n\t/** Model id, dimensionality, and live vector count of the daemon. */\n\tasync info(): Promise<EmbSearchDaemonInfo> {\n\t\tconst res = await this.send({ op: \"info\" });\n\t\treturn { modelId: res.model_id ?? \"\", dim: res.dim ?? 0, count: res.count ?? 0, rerank: res.rerank };\n\t}\n\n\t/** Remove a record. Resolves to `true` if it existed. */\n\tasync remove(id: string): Promise<boolean> {\n\t\tconst res = await this.send({ op: \"remove\", id });\n\t\treturn res.removed === true;\n\t}\n\n\t/** Reclaim tombstoned rows left behind by `remove`. */\n\tasync compact(): Promise<void> {\n\t\tawait this.send({ op: \"compact\" });\n\t}\n\n\t/** Persist the index to the store directory. */\n\tasync save(): Promise<void> {\n\t\tawait this.send({ op: \"save\" });\n\t}\n\n\t/** Shut the daemon down, closing stdin so it exits cleanly. */\n\tasync close(): Promise<void> {\n\t\tif (this.closed) return;\n\t\tthis.proc.stdin.end();\n\t\tawait new Promise<void>((resolve) => this.proc.on(\"exit\", () => resolve()));\n\t}\n}\n"]}
+{"version":3,"file":"client.d.ts","sourceRoot":"","sources":["../../../src/core/embsearch/client.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAIH,MAAM,WAAW,eAAe;IAC/B,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,mBAAmB;IACnC,OAAO,EAAE,MAAM,CAAC;IAChB,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd;;;;;;OAMG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;;;;OAQG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;CACjB;AAED,oDAAoD;AACpD,MAAM,WAAW,sBAAsB;IACtC,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;CACb;AAED;wDACwD;AACxD,MAAM,WAAW,qBAAqB;IACrC,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;CACd;AAED,MAAM,WAAW,mBAAmB;IACnC,QAAQ,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,MAAM,CAAC;CAChB;AAED,wEAAwE;AACxE,MAAM,MAAM,eAAe,GAAG,OAAO,GAAG,SAAS,GAAG,QAAQ,CAAC;AAE7D,MAAM,WAAW,sBAAsB;IACtC,6EAA6E;IAC7E,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,sCAAsC;IACtC,UAAU,EAAE,MAAM,CAAC;IACnB,0CAA0C;IAC1C,SAAS,EAAE,MAAM,CAAC;IAClB,6DAA6D;IAC7D,MAAM,CAAC,EAAE,QAAQ,GAAG,KAAK,GAAG,WAAW,CAAC;CACxC;AA4BD,qBAAa,eAAe;IAC3B,OAAO,CAAC,IAAI,CAAiC;IAC7C,OAAO,CAAC,KAAK,CAAiB;IAC9B,OAAO,CAAC,MAAM,CAAM;IACpB,OAAO,CAAC,MAAM,CAAS;IACvB,OAAO,CAAC,YAAY,CAAgB;IAEpC,YAAY,IAAI,EAAE,sBAAsB,EA+BvC;IAED,6DAA6D;IAC7D,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAErB;IAED,IAAI,QAAQ,IAAI,OAAO,CAEtB;IAED,OAAO,CAAC,QAAQ;IAqBhB,OAAO,CAAC,IAAI;IAQZ;;;;;;;;;;;;OAYG;IACG,KAAK,CAAC,IAAI,EAAE,MAAM,EAAE,CAAC,SAAK,EAAE,SAAS,GAAE,eAAyB,GAAG,OAAO,CAAC,eAAe,EAAE,CAAC,CAKlG;IAED;;;;;;;;OAQG;IACG,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,sBAAsB,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,qBAAqB,EAAE,CAAC,CAG3G;IAED;;;;OAIG;IACG,IAAI,CAAC,KAAK,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAGnF;IAED,qEAAqE;IAC/D,IAAI,IAAI,OAAO,CAAC,mBAAmB,CAAC,CASzC;IAED,yDAAyD;IACnD,MAAM,CAAC,EAAE,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,CAGzC;IAED,uDAAuD;IACjD,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAE7B;IAED,gDAAgD;IAC1C,IAAI,IAAI,OAAO,CAAC,IAAI,CAAC,CAE1B;IAED,+DAA+D;IACzD,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAI3B;CACD","sourcesContent":["/**\n * Client for the `embsearch` stdio daemon (vendored from\n * github.com/kolisachint/embeddingsearchtools ts/client.ts).\n *\n * Spawns `embsearch serve` once and talks newline-delimited JSON over its\n * stdin/stdout. The process stays alive so the model and index load a single\n * time; every query after startup is hot.\n */\n\nimport { type ChildProcessWithoutNullStreams, spawn } from \"child_process\";\n\nexport interface EmbSearchResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchDaemonInfo {\n\tmodelId: string;\n\tdim: number;\n\tcount: number;\n\t/**\n\t * Whether the open store carries a BM25 index alongside its vectors.\n\t *\n\t * Fixed when the store is created — passing `--hybrid` at a non-hybrid\n\t * store only warns — so this is the only reliable way to find out, and the\n\t * signal that an existing store has to be rebuilt rather than reused.\n\t */\n\thybrid?: boolean;\n\t/**\n\t * Whether this daemon can actually serve `rerank`, or `undefined` from a\n\t * daemon too old to say.\n\t *\n\t * Reported separately from the version because the two came apart:\n\t * released binaries carry the `rerank` op but no longer bundle the ~23 MB\n\t * cross-encoder weights, so a version check alone would advertise a\n\t * reranker that fails on first call.\n\t */\n\trerank?: boolean;\n}\n\n/** One candidate sent for cross-encoder scoring. */\nexport interface EmbSearchRerankPassage {\n\tid: string;\n\ttext: string;\n}\n\n/** A cross-encoder relevance logit. Higher is more relevant, but the scale is\n *  unnormalized and comparable only within one call. */\nexport interface EmbSearchRerankResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchBulkResult {\n\tinserted: number;\n\tupdated: number;\n}\n\n/** Which daemon-side retriever answers a query (embsearch >= 0.2.0). */\nexport type DaemonRetriever = \"dense\" | \"lexical\" | \"hybrid\";\n\nexport interface EmbSearchClientOptions {\n\t/** Open/create the store with a BM25 lexical index alongside the vectors. */\n\thybrid?: boolean;\n\t/** Path to the `embsearch` binary. */\n\tbinaryPath: string;\n\t/** Store directory passed as `--path`. */\n\tstorePath: string;\n\t/** Metric for a freshly created store. Default: \"cosine\". */\n\tmetric?: \"cosine\" | \"dot\" | \"euclidean\";\n}\n\ninterface Pending {\n\tresolve: (value: EmbSearchRawResponse) => void;\n\treject: (err: Error) => void;\n}\n\ninterface EmbSearchRawResponse {\n\tok: boolean;\n\terror?: string;\n\tresults?: EmbSearchResult[];\n\tinserted?: boolean;\n\tremoved?: boolean;\n\tcount?: number;\n\tinserted_count?: number;\n\tupdated_count?: number;\n\tmodel_id?: string;\n\tdim?: number;\n\t/** Cross-encoder scores. Separate from `results` because the daemon's\n\t *  logits are not comparable to retrieval scores. */\n\treranked?: EmbSearchRerankResult[];\n\t/** `info` only: whether the open store has a BM25 index. */\n\thybrid?: boolean;\n\t/** `info` only: whether `rerank` will work. Absent on daemons too old to\n\t *  report it. */\n\trerank?: boolean;\n}\n\nexport class EmbSearchClient {\n\tprivate proc: ChildProcessWithoutNullStreams;\n\tprivate queue: Pending[] = [];\n\tprivate buffer = \"\";\n\tprivate closed = false;\n\tprivate readyPromise: Promise<void>;\n\n\tconstructor(opts: EmbSearchClientOptions) {\n\t\tconst args = [\"serve\", \"--path\", opts.storePath];\n\t\tif (opts.metric) args.push(\"--metric\", opts.metric);\n\t\t// Hybrid-ness is fixed when a store is created: passing --hybrid against\n\t\t// an existing non-hybrid store warns and is ignored daemon-side, so a\n\t\t// hybrid store needs its own directory.\n\t\tif (opts.hybrid) args.push(\"--hybrid\");\n\n\t\tthis.proc = spawn(opts.binaryPath, args, { stdio: [\"pipe\", \"pipe\", \"pipe\"] });\n\n\t\tthis.proc.stdout.setEncoding(\"utf8\");\n\t\tthis.proc.stdout.on(\"data\", (chunk: string) => this.onStdout(chunk));\n\t\t// Swallow the informational stderr banner; errors surface via responses/exit.\n\t\tthis.proc.stderr.resume();\n\n\t\t// Readiness = the daemon answering a ping (probes the actual request loop).\n\t\tthis.readyPromise = this.send({ op: \"ping\" }).then(() => undefined);\n\t\t// Spawn failures reject the pending ping via the exit handler; mark the\n\t\t// promise handled so a failed spawn doesn't raise an unhandled rejection\n\t\t// before ready() is awaited.\n\t\tthis.readyPromise.catch(() => {});\n\n\t\tthis.proc.on(\"error\", (err) => {\n\t\t\tthis.closed = true;\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t\tthis.proc.on(\"exit\", (code) => {\n\t\t\tthis.closed = true;\n\t\t\tconst err = new Error(`embsearch daemon exited (code ${code})`);\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t}\n\n\t/** Resolves once the daemon has loaded the model + index. */\n\tready(): Promise<void> {\n\t\treturn this.readyPromise;\n\t}\n\n\tget isClosed(): boolean {\n\t\treturn this.closed;\n\t}\n\n\tprivate onStdout(chunk: string): void {\n\t\tthis.buffer += chunk;\n\t\t// Each response is one line; dispatch FIFO against the pending queue.\n\t\tfor (let nl = this.buffer.indexOf(\"\\n\"); nl !== -1; nl = this.buffer.indexOf(\"\\n\")) {\n\t\t\tconst line = this.buffer.slice(0, nl).trim();\n\t\t\tthis.buffer = this.buffer.slice(nl + 1);\n\t\t\tif (!line) continue;\n\t\t\tconst pending = this.queue.shift();\n\t\t\tif (!pending) continue;\n\t\t\tlet msg: EmbSearchRawResponse;\n\t\t\ttry {\n\t\t\t\tmsg = JSON.parse(line) as EmbSearchRawResponse;\n\t\t\t} catch {\n\t\t\t\tpending.reject(new Error(`bad response: ${line}`));\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tif (msg.ok) pending.resolve(msg);\n\t\t\telse pending.reject(new Error(msg.error ?? \"unknown error\"));\n\t\t}\n\t}\n\n\tprivate send(req: Record<string, unknown>): Promise<EmbSearchRawResponse> {\n\t\tif (this.closed) return Promise.reject(new Error(\"embsearch client is closed\"));\n\t\treturn new Promise<EmbSearchRawResponse>((resolve, reject) => {\n\t\t\tthis.queue.push({ resolve, reject });\n\t\t\tthis.proc.stdin.write(`${JSON.stringify(req)}\\n`);\n\t\t});\n\t}\n\n\t/**\n\t * Search for the top-`k` matches for `text`.\n\t *\n\t * `retriever` selects which leg answers:\n\t *  - `dense` (default) — vector search, the historical behaviour;\n\t *  - `lexical` — BM25 only, raw scores, no embedding computed;\n\t *  - `hybrid` — both, pre-fused by the daemon's own RRF constant.\n\t *\n\t * `lexical` and `hybrid` need a store created with `--hybrid`. Prefer\n\t * `lexical` over `hybrid` when fusing here: `hybrid` collapses both legs\n\t * into one RRF score, discarding the per-retriever ranks the trace records\n\t * and preventing n-way fusion with the grep leg.\n\t */\n\tasync query(text: string, k = 10, retriever: DaemonRetriever = \"dense\"): Promise<EmbSearchResult[]> {\n\t\tconst res = await this.send(\n\t\t\tretriever === \"dense\" ? { op: \"query\", text, k } : { op: \"query\", text, k, retriever },\n\t\t);\n\t\treturn res.results ?? [];\n\t}\n\n\t/**\n\t * Score `passages` against `query` with the daemon's cross-encoder and\n\t * return the best `k`, best first.\n\t *\n\t * Passages are sent inline rather than referenced by id: the caller has the\n\t * exact spans it intends to show the model, and a cross-encoder scores the\n\t * text it is given, so sending anything else would score the wrong thing.\n\t * Requires an onnx build with reranker weights (embsearch >= 0.3.0).\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tconst res = await this.send({ op: \"rerank\", query, passages, k });\n\t\treturn res.reranked ?? [];\n\t}\n\n\t/**\n\t * Batched insert-or-replace. One embedding inference for the whole batch —\n\t * the fast path for bulk indexing. Keep batches modest (e.g. 32–64) so a\n\t * concurrent query is not stuck behind a huge inference.\n\t */\n\tasync bulk(items: Array<{ id: string; text: string }>): Promise<EmbSearchBulkResult> {\n\t\tconst res = await this.send({ op: \"bulk\", items });\n\t\treturn { inserted: res.inserted_count ?? 0, updated: res.updated_count ?? 0 };\n\t}\n\n\t/** Model id, dimensionality, and live vector count of the daemon. */\n\tasync info(): Promise<EmbSearchDaemonInfo> {\n\t\tconst res = await this.send({ op: \"info\" });\n\t\treturn {\n\t\t\tmodelId: res.model_id ?? \"\",\n\t\t\tdim: res.dim ?? 0,\n\t\t\tcount: res.count ?? 0,\n\t\t\thybrid: res.hybrid,\n\t\t\trerank: res.rerank,\n\t\t};\n\t}\n\n\t/** Remove a record. Resolves to `true` if it existed. */\n\tasync remove(id: string): Promise<boolean> {\n\t\tconst res = await this.send({ op: \"remove\", id });\n\t\treturn res.removed === true;\n\t}\n\n\t/** Reclaim tombstoned rows left behind by `remove`. */\n\tasync compact(): Promise<void> {\n\t\tawait this.send({ op: \"compact\" });\n\t}\n\n\t/** Persist the index to the store directory. */\n\tasync save(): Promise<void> {\n\t\tawait this.send({ op: \"save\" });\n\t}\n\n\t/** Shut the daemon down, closing stdin so it exits cleanly. */\n\tasync close(): Promise<void> {\n\t\tif (this.closed) return;\n\t\tthis.proc.stdin.end();\n\t\tawait new Promise<void>((resolve) => this.proc.on(\"exit\", () => resolve()));\n\t}\n}\n"]}

package/dist/core/embsearch/client.js

@@ -127,7 +127,13 @@ export class EmbSearchClient {
     /** Model id, dimensionality, and live vector count of the daemon. */
     async info() {
         const res = await this.send({ op: "info" });
-        return { modelId: res.model_id ?? "", dim: res.dim ?? 0, count: res.count ?? 0, rerank: res.rerank };
+        return {
+            modelId: res.model_id ?? "",
+            dim: res.dim ?? 0,
+            count: res.count ?? 0,
+            hybrid: res.hybrid,
+            rerank: res.rerank,
+        };
     }
     /** Remove a record. Resolves to `true` if it existed. */
     async remove(id) {

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

@@ -1 +1 @@
-{"version":3,"file":"client.js","sourceRoot":"","sources":["../../../src/core/embsearch/client.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAuC,KAAK,EAAE,MAAM,eAAe,CAAC;AA+E3E,MAAM,OAAO,eAAe;IACnB,IAAI,CAAiC;IACrC,KAAK,GAAc,EAAE,CAAC;IACtB,MAAM,GAAG,EAAE,CAAC;IACZ,MAAM,GAAG,KAAK,CAAC;IACf,YAAY,CAAgB;IAEpC,YAAY,IAA4B,EAAE;QACzC,MAAM,IAAI,GAAG,CAAC,OAAO,EAAE,QAAQ,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;QACjD,IAAI,IAAI,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;QACpD,yEAAyE;QACzE,sEAAsE;QACtE,wCAAwC;QACxC,IAAI,IAAI,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAEvC,IAAI,CAAC,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,KAAK,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;QAE9E,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;QACrC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,KAAa,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;QACrE,8EAA8E;QAC9E,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;QAE1B,4EAA4E;QAC5E,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;QACpE,wEAAwE;QACxE,yEAAyE;QACzE,6BAA6B;QAC7B,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAC,CAAC,CAAC,CAAC;QAElC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;YAC9B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;YACnB,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAA,CACpD,CAAC,CAAC;QACH,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC;YAC9B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;YACnB,MAAM,GAAG,GAAG,IAAI,KAAK,CAAC,iCAAiC,IAAI,GAAG,CAAC,CAAC;YAChE,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAA,CACpD,CAAC,CAAC;IAAA,CACH;IAED,6DAA6D;IAC7D,KAAK,GAAkB;QACtB,OAAO,IAAI,CAAC,YAAY,CAAC;IAAA,CACzB;IAED,IAAI,QAAQ,GAAY;QACvB,OAAO,IAAI,CAAC,MAAM,CAAC;IAAA,CACnB;IAEO,QAAQ,CAAC,KAAa,EAAQ;QACrC,IAAI,CAAC,MAAM,IAAI,KAAK,CAAC;QACrB,sEAAsE;QACtE,KAAK,IAAI,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,EAAE,KAAK,CAAC,CAAC,EAAE,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;YACpF,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;YAC7C,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;YACxC,IAAI,CAAC,IAAI;gBAAE,SAAS;YACpB,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;YACnC,IAAI,CAAC,OAAO;gBAAE,SAAS;YACvB,IAAI,GAAyB,CAAC;YAC9B,IAAI,CAAC;gBACJ,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAyB,CAAC;YAChD,CAAC;YAAC,MAAM,CAAC;gBACR,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,iBAAiB,IAAI,EAAE,CAAC,CAAC,CAAC;gBACnD,SAAS;YACV,CAAC;YACD,IAAI,GAAG,CAAC,EAAE;gBAAE,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;;gBAC5B,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,GAAG,CAAC,KAAK,IAAI,eAAe,CAAC,CAAC,CAAC;QAC9D,CAAC;IAAA,CACD;IAEO,IAAI,CAAC,GAA4B,EAAiC;QACzE,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC,CAAC;QAChF,OAAO,IAAI,OAAO,CAAuB,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,CAAC;YAC7D,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;YACrC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAAA,CAClD,CAAC,CAAC;IAAA,CACH;IAED;;;;;;;;;;;;OAYG;IACH,KAAK,CAAC,KAAK,CAAC,IAAY,EAAE,CAAC,GAAG,EAAE,EAAE,SAAS,GAAoB,OAAO,EAA8B;QACnG,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAC1B,SAAS,KAAK,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,SAAS,EAAE,CACtF,CAAC;QACF,OAAO,GAAG,CAAC,OAAO,IAAI,EAAE,CAAC;IAAA,CACzB;IAED;;;;;;;;OAQG;IACH,KAAK,CAAC,MAAM,CAAC,KAAa,EAAE,QAAkC,EAAE,CAAS,EAAoC;QAC5G,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QAClE,OAAO,GAAG,CAAC,QAAQ,IAAI,EAAE,CAAC;IAAA,CAC1B;IAED;;;;OAIG;IACH,KAAK,CAAC,IAAI,CAAC,KAA0C,EAAgC;QACpF,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;QACnD,OAAO,EAAE,QAAQ,EAAE,GAAG,CAAC,cAAc,IAAI,CAAC,EAAE,OAAO,EAAE,GAAG,CAAC,aAAa,IAAI,CAAC,EAAE,CAAC;IAAA,CAC9E;IAED,qEAAqE;IACrE,KAAK,CAAC,IAAI,GAAiC;QAC1C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;QAC5C,OAAO,EAAE,OAAO,EAAE,GAAG,CAAC,QAAQ,IAAI,EAAE,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,CAAC,KAAK,IAAI,CAAC,EAAE,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,CAAC;IAAA,CACrG;IAED,yDAAyD;IACzD,KAAK,CAAC,MAAM,CAAC,EAAU,EAAoB;QAC1C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;QAClD,OAAO,GAAG,CAAC,OAAO,KAAK,IAAI,CAAC;IAAA,CAC5B;IAED,uDAAuD;IACvD,KAAK,CAAC,OAAO,GAAkB;QAC9B,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,CAAC;IAAA,CACnC;IAED,gDAAgD;IAChD,KAAK,CAAC,IAAI,GAAkB;QAC3B,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;IAAA,CAChC;IAED,+DAA+D;IAC/D,KAAK,CAAC,KAAK,GAAkB;QAC5B,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO;QACxB,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC;QACtB,MAAM,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;IAAA,CAC5E;CACD","sourcesContent":["/**\n * Client for the `embsearch` stdio daemon (vendored from\n * github.com/kolisachint/embeddingsearchtools ts/client.ts).\n *\n * Spawns `embsearch serve` once and talks newline-delimited JSON over its\n * stdin/stdout. The process stays alive so the model and index load a single\n * time; every query after startup is hot.\n */\n\nimport { type ChildProcessWithoutNullStreams, spawn } from \"child_process\";\n\nexport interface EmbSearchResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchDaemonInfo {\n\tmodelId: string;\n\tdim: number;\n\tcount: number;\n\t/**\n\t * Whether this daemon can actually serve `rerank`, or `undefined` from a\n\t * daemon too old to say.\n\t *\n\t * Reported separately from the version because the two came apart:\n\t * released binaries carry the `rerank` op but no longer bundle the ~23 MB\n\t * cross-encoder weights, so a version check alone would advertise a\n\t * reranker that fails on first call.\n\t */\n\trerank?: boolean;\n}\n\n/** One candidate sent for cross-encoder scoring. */\nexport interface EmbSearchRerankPassage {\n\tid: string;\n\ttext: string;\n}\n\n/** A cross-encoder relevance logit. Higher is more relevant, but the scale is\n *  unnormalized and comparable only within one call. */\nexport interface EmbSearchRerankResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchBulkResult {\n\tinserted: number;\n\tupdated: number;\n}\n\n/** Which daemon-side retriever answers a query (embsearch >= 0.2.0). */\nexport type DaemonRetriever = \"dense\" | \"lexical\" | \"hybrid\";\n\nexport interface EmbSearchClientOptions {\n\t/** Open/create the store with a BM25 lexical index alongside the vectors. */\n\thybrid?: boolean;\n\t/** Path to the `embsearch` binary. */\n\tbinaryPath: string;\n\t/** Store directory passed as `--path`. */\n\tstorePath: string;\n\t/** Metric for a freshly created store. Default: \"cosine\". */\n\tmetric?: \"cosine\" | \"dot\" | \"euclidean\";\n}\n\ninterface Pending {\n\tresolve: (value: EmbSearchRawResponse) => void;\n\treject: (err: Error) => void;\n}\n\ninterface EmbSearchRawResponse {\n\tok: boolean;\n\terror?: string;\n\tresults?: EmbSearchResult[];\n\tinserted?: boolean;\n\tremoved?: boolean;\n\tcount?: number;\n\tinserted_count?: number;\n\tupdated_count?: number;\n\tmodel_id?: string;\n\tdim?: number;\n\t/** Cross-encoder scores. Separate from `results` because the daemon's\n\t *  logits are not comparable to retrieval scores. */\n\treranked?: EmbSearchRerankResult[];\n\t/** `info` only: whether `rerank` will work. Absent on daemons too old to\n\t *  report it. */\n\trerank?: boolean;\n}\n\nexport class EmbSearchClient {\n\tprivate proc: ChildProcessWithoutNullStreams;\n\tprivate queue: Pending[] = [];\n\tprivate buffer = \"\";\n\tprivate closed = false;\n\tprivate readyPromise: Promise<void>;\n\n\tconstructor(opts: EmbSearchClientOptions) {\n\t\tconst args = [\"serve\", \"--path\", opts.storePath];\n\t\tif (opts.metric) args.push(\"--metric\", opts.metric);\n\t\t// Hybrid-ness is fixed when a store is created: passing --hybrid against\n\t\t// an existing non-hybrid store warns and is ignored daemon-side, so a\n\t\t// hybrid store needs its own directory.\n\t\tif (opts.hybrid) args.push(\"--hybrid\");\n\n\t\tthis.proc = spawn(opts.binaryPath, args, { stdio: [\"pipe\", \"pipe\", \"pipe\"] });\n\n\t\tthis.proc.stdout.setEncoding(\"utf8\");\n\t\tthis.proc.stdout.on(\"data\", (chunk: string) => this.onStdout(chunk));\n\t\t// Swallow the informational stderr banner; errors surface via responses/exit.\n\t\tthis.proc.stderr.resume();\n\n\t\t// Readiness = the daemon answering a ping (probes the actual request loop).\n\t\tthis.readyPromise = this.send({ op: \"ping\" }).then(() => undefined);\n\t\t// Spawn failures reject the pending ping via the exit handler; mark the\n\t\t// promise handled so a failed spawn doesn't raise an unhandled rejection\n\t\t// before ready() is awaited.\n\t\tthis.readyPromise.catch(() => {});\n\n\t\tthis.proc.on(\"error\", (err) => {\n\t\t\tthis.closed = true;\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t\tthis.proc.on(\"exit\", (code) => {\n\t\t\tthis.closed = true;\n\t\t\tconst err = new Error(`embsearch daemon exited (code ${code})`);\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t}\n\n\t/** Resolves once the daemon has loaded the model + index. */\n\tready(): Promise<void> {\n\t\treturn this.readyPromise;\n\t}\n\n\tget isClosed(): boolean {\n\t\treturn this.closed;\n\t}\n\n\tprivate onStdout(chunk: string): void {\n\t\tthis.buffer += chunk;\n\t\t// Each response is one line; dispatch FIFO against the pending queue.\n\t\tfor (let nl = this.buffer.indexOf(\"\\n\"); nl !== -1; nl = this.buffer.indexOf(\"\\n\")) {\n\t\t\tconst line = this.buffer.slice(0, nl).trim();\n\t\t\tthis.buffer = this.buffer.slice(nl + 1);\n\t\t\tif (!line) continue;\n\t\t\tconst pending = this.queue.shift();\n\t\t\tif (!pending) continue;\n\t\t\tlet msg: EmbSearchRawResponse;\n\t\t\ttry {\n\t\t\t\tmsg = JSON.parse(line) as EmbSearchRawResponse;\n\t\t\t} catch {\n\t\t\t\tpending.reject(new Error(`bad response: ${line}`));\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tif (msg.ok) pending.resolve(msg);\n\t\t\telse pending.reject(new Error(msg.error ?? \"unknown error\"));\n\t\t}\n\t}\n\n\tprivate send(req: Record<string, unknown>): Promise<EmbSearchRawResponse> {\n\t\tif (this.closed) return Promise.reject(new Error(\"embsearch client is closed\"));\n\t\treturn new Promise<EmbSearchRawResponse>((resolve, reject) => {\n\t\t\tthis.queue.push({ resolve, reject });\n\t\t\tthis.proc.stdin.write(`${JSON.stringify(req)}\\n`);\n\t\t});\n\t}\n\n\t/**\n\t * Search for the top-`k` matches for `text`.\n\t *\n\t * `retriever` selects which leg answers:\n\t *  - `dense` (default) — vector search, the historical behaviour;\n\t *  - `lexical` — BM25 only, raw scores, no embedding computed;\n\t *  - `hybrid` — both, pre-fused by the daemon's own RRF constant.\n\t *\n\t * `lexical` and `hybrid` need a store created with `--hybrid`. Prefer\n\t * `lexical` over `hybrid` when fusing here: `hybrid` collapses both legs\n\t * into one RRF score, discarding the per-retriever ranks the trace records\n\t * and preventing n-way fusion with the grep leg.\n\t */\n\tasync query(text: string, k = 10, retriever: DaemonRetriever = \"dense\"): Promise<EmbSearchResult[]> {\n\t\tconst res = await this.send(\n\t\t\tretriever === \"dense\" ? { op: \"query\", text, k } : { op: \"query\", text, k, retriever },\n\t\t);\n\t\treturn res.results ?? [];\n\t}\n\n\t/**\n\t * Score `passages` against `query` with the daemon's cross-encoder and\n\t * return the best `k`, best first.\n\t *\n\t * Passages are sent inline rather than referenced by id: the caller has the\n\t * exact spans it intends to show the model, and a cross-encoder scores the\n\t * text it is given, so sending anything else would score the wrong thing.\n\t * Requires an onnx build with reranker weights (embsearch >= 0.3.0).\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tconst res = await this.send({ op: \"rerank\", query, passages, k });\n\t\treturn res.reranked ?? [];\n\t}\n\n\t/**\n\t * Batched insert-or-replace. One embedding inference for the whole batch —\n\t * the fast path for bulk indexing. Keep batches modest (e.g. 32–64) so a\n\t * concurrent query is not stuck behind a huge inference.\n\t */\n\tasync bulk(items: Array<{ id: string; text: string }>): Promise<EmbSearchBulkResult> {\n\t\tconst res = await this.send({ op: \"bulk\", items });\n\t\treturn { inserted: res.inserted_count ?? 0, updated: res.updated_count ?? 0 };\n\t}\n\n\t/** Model id, dimensionality, and live vector count of the daemon. */\n\tasync info(): Promise<EmbSearchDaemonInfo> {\n\t\tconst res = await this.send({ op: \"info\" });\n\t\treturn { modelId: res.model_id ?? \"\", dim: res.dim ?? 0, count: res.count ?? 0, rerank: res.rerank };\n\t}\n\n\t/** Remove a record. Resolves to `true` if it existed. */\n\tasync remove(id: string): Promise<boolean> {\n\t\tconst res = await this.send({ op: \"remove\", id });\n\t\treturn res.removed === true;\n\t}\n\n\t/** Reclaim tombstoned rows left behind by `remove`. */\n\tasync compact(): Promise<void> {\n\t\tawait this.send({ op: \"compact\" });\n\t}\n\n\t/** Persist the index to the store directory. */\n\tasync save(): Promise<void> {\n\t\tawait this.send({ op: \"save\" });\n\t}\n\n\t/** Shut the daemon down, closing stdin so it exits cleanly. */\n\tasync close(): Promise<void> {\n\t\tif (this.closed) return;\n\t\tthis.proc.stdin.end();\n\t\tawait new Promise<void>((resolve) => this.proc.on(\"exit\", () => resolve()));\n\t}\n}\n"]}
+{"version":3,"file":"client.js","sourceRoot":"","sources":["../../../src/core/embsearch/client.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAuC,KAAK,EAAE,MAAM,eAAe,CAAC;AAyF3E,MAAM,OAAO,eAAe;IACnB,IAAI,CAAiC;IACrC,KAAK,GAAc,EAAE,CAAC;IACtB,MAAM,GAAG,EAAE,CAAC;IACZ,MAAM,GAAG,KAAK,CAAC;IACf,YAAY,CAAgB;IAEpC,YAAY,IAA4B,EAAE;QACzC,MAAM,IAAI,GAAG,CAAC,OAAO,EAAE,QAAQ,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;QACjD,IAAI,IAAI,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;QACpD,yEAAyE;QACzE,sEAAsE;QACtE,wCAAwC;QACxC,IAAI,IAAI,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAEvC,IAAI,CAAC,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,KAAK,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;QAE9E,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;QACrC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,KAAa,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;QACrE,8EAA8E;QAC9E,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;QAE1B,4EAA4E;QAC5E,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;QACpE,wEAAwE;QACxE,yEAAyE;QACzE,6BAA6B;QAC7B,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAC,CAAC,CAAC,CAAC;QAElC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;YAC9B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;YACnB,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAA,CACpD,CAAC,CAAC;QACH,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC;YAC9B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;YACnB,MAAM,GAAG,GAAG,IAAI,KAAK,CAAC,iCAAiC,IAAI,GAAG,CAAC,CAAC;YAChE,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;gBAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAA,CACpD,CAAC,CAAC;IAAA,CACH;IAED,6DAA6D;IAC7D,KAAK,GAAkB;QACtB,OAAO,IAAI,CAAC,YAAY,CAAC;IAAA,CACzB;IAED,IAAI,QAAQ,GAAY;QACvB,OAAO,IAAI,CAAC,MAAM,CAAC;IAAA,CACnB;IAEO,QAAQ,CAAC,KAAa,EAAQ;QACrC,IAAI,CAAC,MAAM,IAAI,KAAK,CAAC;QACrB,sEAAsE;QACtE,KAAK,IAAI,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,EAAE,KAAK,CAAC,CAAC,EAAE,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;YACpF,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;YAC7C,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;YACxC,IAAI,CAAC,IAAI;gBAAE,SAAS;YACpB,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;YACnC,IAAI,CAAC,OAAO;gBAAE,SAAS;YACvB,IAAI,GAAyB,CAAC;YAC9B,IAAI,CAAC;gBACJ,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAyB,CAAC;YAChD,CAAC;YAAC,MAAM,CAAC;gBACR,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,iBAAiB,IAAI,EAAE,CAAC,CAAC,CAAC;gBACnD,SAAS;YACV,CAAC;YACD,IAAI,GAAG,CAAC,EAAE;gBAAE,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;;gBAC5B,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,GAAG,CAAC,KAAK,IAAI,eAAe,CAAC,CAAC,CAAC;QAC9D,CAAC;IAAA,CACD;IAEO,IAAI,CAAC,GAA4B,EAAiC;QACzE,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC,CAAC;QAChF,OAAO,IAAI,OAAO,CAAuB,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,CAAC;YAC7D,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;YACrC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAAA,CAClD,CAAC,CAAC;IAAA,CACH;IAED;;;;;;;;;;;;OAYG;IACH,KAAK,CAAC,KAAK,CAAC,IAAY,EAAE,CAAC,GAAG,EAAE,EAAE,SAAS,GAAoB,OAAO,EAA8B;QACnG,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAC1B,SAAS,KAAK,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,SAAS,EAAE,CACtF,CAAC;QACF,OAAO,GAAG,CAAC,OAAO,IAAI,EAAE,CAAC;IAAA,CACzB;IAED;;;;;;;;OAQG;IACH,KAAK,CAAC,MAAM,CAAC,KAAa,EAAE,QAAkC,EAAE,CAAS,EAAoC;QAC5G,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QAClE,OAAO,GAAG,CAAC,QAAQ,IAAI,EAAE,CAAC;IAAA,CAC1B;IAED;;;;OAIG;IACH,KAAK,CAAC,IAAI,CAAC,KAA0C,EAAgC;QACpF,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;QACnD,OAAO,EAAE,QAAQ,EAAE,GAAG,CAAC,cAAc,IAAI,CAAC,EAAE,OAAO,EAAE,GAAG,CAAC,aAAa,IAAI,CAAC,EAAE,CAAC;IAAA,CAC9E;IAED,qEAAqE;IACrE,KAAK,CAAC,IAAI,GAAiC;QAC1C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;QAC5C,OAAO;YACN,OAAO,EAAE,GAAG,CAAC,QAAQ,IAAI,EAAE;YAC3B,GAAG,EAAE,GAAG,CAAC,GAAG,IAAI,CAAC;YACjB,KAAK,EAAE,GAAG,CAAC,KAAK,IAAI,CAAC;YACrB,MAAM,EAAE,GAAG,CAAC,MAAM;YAClB,MAAM,EAAE,GAAG,CAAC,MAAM;SAClB,CAAC;IAAA,CACF;IAED,yDAAyD;IACzD,KAAK,CAAC,MAAM,CAAC,EAAU,EAAoB;QAC1C,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;QAClD,OAAO,GAAG,CAAC,OAAO,KAAK,IAAI,CAAC;IAAA,CAC5B;IAED,uDAAuD;IACvD,KAAK,CAAC,OAAO,GAAkB;QAC9B,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,CAAC;IAAA,CACnC;IAED,gDAAgD;IAChD,KAAK,CAAC,IAAI,GAAkB;QAC3B,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;IAAA,CAChC;IAED,+DAA+D;IAC/D,KAAK,CAAC,KAAK,GAAkB;QAC5B,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO;QACxB,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC;QACtB,MAAM,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;IAAA,CAC5E;CACD","sourcesContent":["/**\n * Client for the `embsearch` stdio daemon (vendored from\n * github.com/kolisachint/embeddingsearchtools ts/client.ts).\n *\n * Spawns `embsearch serve` once and talks newline-delimited JSON over its\n * stdin/stdout. The process stays alive so the model and index load a single\n * time; every query after startup is hot.\n */\n\nimport { type ChildProcessWithoutNullStreams, spawn } from \"child_process\";\n\nexport interface EmbSearchResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchDaemonInfo {\n\tmodelId: string;\n\tdim: number;\n\tcount: number;\n\t/**\n\t * Whether the open store carries a BM25 index alongside its vectors.\n\t *\n\t * Fixed when the store is created — passing `--hybrid` at a non-hybrid\n\t * store only warns — so this is the only reliable way to find out, and the\n\t * signal that an existing store has to be rebuilt rather than reused.\n\t */\n\thybrid?: boolean;\n\t/**\n\t * Whether this daemon can actually serve `rerank`, or `undefined` from a\n\t * daemon too old to say.\n\t *\n\t * Reported separately from the version because the two came apart:\n\t * released binaries carry the `rerank` op but no longer bundle the ~23 MB\n\t * cross-encoder weights, so a version check alone would advertise a\n\t * reranker that fails on first call.\n\t */\n\trerank?: boolean;\n}\n\n/** One candidate sent for cross-encoder scoring. */\nexport interface EmbSearchRerankPassage {\n\tid: string;\n\ttext: string;\n}\n\n/** A cross-encoder relevance logit. Higher is more relevant, but the scale is\n *  unnormalized and comparable only within one call. */\nexport interface EmbSearchRerankResult {\n\tid: string;\n\tscore: number;\n}\n\nexport interface EmbSearchBulkResult {\n\tinserted: number;\n\tupdated: number;\n}\n\n/** Which daemon-side retriever answers a query (embsearch >= 0.2.0). */\nexport type DaemonRetriever = \"dense\" | \"lexical\" | \"hybrid\";\n\nexport interface EmbSearchClientOptions {\n\t/** Open/create the store with a BM25 lexical index alongside the vectors. */\n\thybrid?: boolean;\n\t/** Path to the `embsearch` binary. */\n\tbinaryPath: string;\n\t/** Store directory passed as `--path`. */\n\tstorePath: string;\n\t/** Metric for a freshly created store. Default: \"cosine\". */\n\tmetric?: \"cosine\" | \"dot\" | \"euclidean\";\n}\n\ninterface Pending {\n\tresolve: (value: EmbSearchRawResponse) => void;\n\treject: (err: Error) => void;\n}\n\ninterface EmbSearchRawResponse {\n\tok: boolean;\n\terror?: string;\n\tresults?: EmbSearchResult[];\n\tinserted?: boolean;\n\tremoved?: boolean;\n\tcount?: number;\n\tinserted_count?: number;\n\tupdated_count?: number;\n\tmodel_id?: string;\n\tdim?: number;\n\t/** Cross-encoder scores. Separate from `results` because the daemon's\n\t *  logits are not comparable to retrieval scores. */\n\treranked?: EmbSearchRerankResult[];\n\t/** `info` only: whether the open store has a BM25 index. */\n\thybrid?: boolean;\n\t/** `info` only: whether `rerank` will work. Absent on daemons too old to\n\t *  report it. */\n\trerank?: boolean;\n}\n\nexport class EmbSearchClient {\n\tprivate proc: ChildProcessWithoutNullStreams;\n\tprivate queue: Pending[] = [];\n\tprivate buffer = \"\";\n\tprivate closed = false;\n\tprivate readyPromise: Promise<void>;\n\n\tconstructor(opts: EmbSearchClientOptions) {\n\t\tconst args = [\"serve\", \"--path\", opts.storePath];\n\t\tif (opts.metric) args.push(\"--metric\", opts.metric);\n\t\t// Hybrid-ness is fixed when a store is created: passing --hybrid against\n\t\t// an existing non-hybrid store warns and is ignored daemon-side, so a\n\t\t// hybrid store needs its own directory.\n\t\tif (opts.hybrid) args.push(\"--hybrid\");\n\n\t\tthis.proc = spawn(opts.binaryPath, args, { stdio: [\"pipe\", \"pipe\", \"pipe\"] });\n\n\t\tthis.proc.stdout.setEncoding(\"utf8\");\n\t\tthis.proc.stdout.on(\"data\", (chunk: string) => this.onStdout(chunk));\n\t\t// Swallow the informational stderr banner; errors surface via responses/exit.\n\t\tthis.proc.stderr.resume();\n\n\t\t// Readiness = the daemon answering a ping (probes the actual request loop).\n\t\tthis.readyPromise = this.send({ op: \"ping\" }).then(() => undefined);\n\t\t// Spawn failures reject the pending ping via the exit handler; mark the\n\t\t// promise handled so a failed spawn doesn't raise an unhandled rejection\n\t\t// before ready() is awaited.\n\t\tthis.readyPromise.catch(() => {});\n\n\t\tthis.proc.on(\"error\", (err) => {\n\t\t\tthis.closed = true;\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t\tthis.proc.on(\"exit\", (code) => {\n\t\t\tthis.closed = true;\n\t\t\tconst err = new Error(`embsearch daemon exited (code ${code})`);\n\t\t\tfor (const p of this.queue.splice(0)) p.reject(err);\n\t\t});\n\t}\n\n\t/** Resolves once the daemon has loaded the model + index. */\n\tready(): Promise<void> {\n\t\treturn this.readyPromise;\n\t}\n\n\tget isClosed(): boolean {\n\t\treturn this.closed;\n\t}\n\n\tprivate onStdout(chunk: string): void {\n\t\tthis.buffer += chunk;\n\t\t// Each response is one line; dispatch FIFO against the pending queue.\n\t\tfor (let nl = this.buffer.indexOf(\"\\n\"); nl !== -1; nl = this.buffer.indexOf(\"\\n\")) {\n\t\t\tconst line = this.buffer.slice(0, nl).trim();\n\t\t\tthis.buffer = this.buffer.slice(nl + 1);\n\t\t\tif (!line) continue;\n\t\t\tconst pending = this.queue.shift();\n\t\t\tif (!pending) continue;\n\t\t\tlet msg: EmbSearchRawResponse;\n\t\t\ttry {\n\t\t\t\tmsg = JSON.parse(line) as EmbSearchRawResponse;\n\t\t\t} catch {\n\t\t\t\tpending.reject(new Error(`bad response: ${line}`));\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tif (msg.ok) pending.resolve(msg);\n\t\t\telse pending.reject(new Error(msg.error ?? \"unknown error\"));\n\t\t}\n\t}\n\n\tprivate send(req: Record<string, unknown>): Promise<EmbSearchRawResponse> {\n\t\tif (this.closed) return Promise.reject(new Error(\"embsearch client is closed\"));\n\t\treturn new Promise<EmbSearchRawResponse>((resolve, reject) => {\n\t\t\tthis.queue.push({ resolve, reject });\n\t\t\tthis.proc.stdin.write(`${JSON.stringify(req)}\\n`);\n\t\t});\n\t}\n\n\t/**\n\t * Search for the top-`k` matches for `text`.\n\t *\n\t * `retriever` selects which leg answers:\n\t *  - `dense` (default) — vector search, the historical behaviour;\n\t *  - `lexical` — BM25 only, raw scores, no embedding computed;\n\t *  - `hybrid` — both, pre-fused by the daemon's own RRF constant.\n\t *\n\t * `lexical` and `hybrid` need a store created with `--hybrid`. Prefer\n\t * `lexical` over `hybrid` when fusing here: `hybrid` collapses both legs\n\t * into one RRF score, discarding the per-retriever ranks the trace records\n\t * and preventing n-way fusion with the grep leg.\n\t */\n\tasync query(text: string, k = 10, retriever: DaemonRetriever = \"dense\"): Promise<EmbSearchResult[]> {\n\t\tconst res = await this.send(\n\t\t\tretriever === \"dense\" ? { op: \"query\", text, k } : { op: \"query\", text, k, retriever },\n\t\t);\n\t\treturn res.results ?? [];\n\t}\n\n\t/**\n\t * Score `passages` against `query` with the daemon's cross-encoder and\n\t * return the best `k`, best first.\n\t *\n\t * Passages are sent inline rather than referenced by id: the caller has the\n\t * exact spans it intends to show the model, and a cross-encoder scores the\n\t * text it is given, so sending anything else would score the wrong thing.\n\t * Requires an onnx build with reranker weights (embsearch >= 0.3.0).\n\t */\n\tasync rerank(query: string, passages: EmbSearchRerankPassage[], k: number): Promise<EmbSearchRerankResult[]> {\n\t\tconst res = await this.send({ op: \"rerank\", query, passages, k });\n\t\treturn res.reranked ?? [];\n\t}\n\n\t/**\n\t * Batched insert-or-replace. One embedding inference for the whole batch —\n\t * the fast path for bulk indexing. Keep batches modest (e.g. 32–64) so a\n\t * concurrent query is not stuck behind a huge inference.\n\t */\n\tasync bulk(items: Array<{ id: string; text: string }>): Promise<EmbSearchBulkResult> {\n\t\tconst res = await this.send({ op: \"bulk\", items });\n\t\treturn { inserted: res.inserted_count ?? 0, updated: res.updated_count ?? 0 };\n\t}\n\n\t/** Model id, dimensionality, and live vector count of the daemon. */\n\tasync info(): Promise<EmbSearchDaemonInfo> {\n\t\tconst res = await this.send({ op: \"info\" });\n\t\treturn {\n\t\t\tmodelId: res.model_id ?? \"\",\n\t\t\tdim: res.dim ?? 0,\n\t\t\tcount: res.count ?? 0,\n\t\t\thybrid: res.hybrid,\n\t\t\trerank: res.rerank,\n\t\t};\n\t}\n\n\t/** Remove a record. Resolves to `true` if it existed. */\n\tasync remove(id: string): Promise<boolean> {\n\t\tconst res = await this.send({ op: \"remove\", id });\n\t\treturn res.removed === true;\n\t}\n\n\t/** Reclaim tombstoned rows left behind by `remove`. */\n\tasync compact(): Promise<void> {\n\t\tawait this.send({ op: \"compact\" });\n\t}\n\n\t/** Persist the index to the store directory. */\n\tasync save(): Promise<void> {\n\t\tawait this.send({ op: \"save\" });\n\t}\n\n\t/** Shut the daemon down, closing stdin so it exits cleanly. */\n\tasync close(): Promise<void> {\n\t\tif (this.closed) return;\n\t\tthis.proc.stdin.end();\n\t\tawait new Promise<void>((resolve) => this.proc.on(\"exit\", () => resolve()));\n\t}\n}\n"]}

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

@@ -47,8 +47,12 @@ export interface EmbsearchServiceOptions {
      */
     storeDir?: string;
     /**
-     * Create the store with the daemon's BM25 lexical index. Fixed at store
-     * creation, so this must pair with a distinct `storeDir`.
+     * Create the store with the daemon's BM25 lexical index.
+     *
+     * Defaults to whatever the daemon can serve. Fixed at store creation, so an
+     * existing store that disagrees is rebuilt once; when overriding this to
+     * hold two different stores for one repo, pair it with a distinct
+     * `storeDir` so they do not fight over the same directory.
      */
     hybridStore?: boolean;
     /** Progress callback for UI (footer / stderr lines). */
@@ -72,6 +76,8 @@ export declare class EmbsearchService {
     private disposed;
     /** Whether the resolved binary serves `retriever: "lexical"`. */
     private lexicalRetriever;
+    /** Whether the store actually opened carries a BM25 index. */
+    private hybridStore;
     /** Whether the resolved binary serves the cross-encoder `rerank` op. */
     private crossEncoder;
     constructor(options: EmbsearchServiceOptions);
@@ -79,7 +85,10 @@ export declare class EmbsearchService {
     /** Semantic search is usable (index ready, or still building with partial data). */
     isAvailable(): boolean;
     private setState;
-    /** Whether the running daemon can serve a BM25-only query. */
+    /**
+     * Whether a BM25-only query will work: the daemon has to understand the
+     * `lexical` retriever *and* the open store has to carry a BM25 index.
+     */
     supportsLexicalRetriever(): boolean;
     /**
      * Repo files whose on-disk content the index does not have — unknown to it,

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

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

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

@@ -14,7 +14,7 @@
  * blocks session startup or affects grep/find.
  */
 import { execFileSync } from "child_process";
-import { readFileSync } from "fs";
+import { readFileSync, rmSync } from "fs";
 import { minimatch } from "minimatch";
 import { ensureTool } from "../../utils/tools-manager.js";
 import { chunkFile } from "./chunker.js";
@@ -69,6 +69,8 @@ export class EmbsearchService {
     disposed = false;
     /** Whether the resolved binary serves `retriever: "lexical"`. */
     lexicalRetriever = false;
+    /** Whether the store actually opened carries a BM25 index. */
+    hybridStore = false;
     /** Whether the resolved binary serves the cross-encoder `rerank` op. */
     crossEncoder = false;
     constructor(options) {
@@ -85,9 +87,12 @@ export class EmbsearchService {
         this.state = state;
         this.options.onProgress?.(state);
     }
-    /** Whether the running daemon can serve a BM25-only query. */
+    /**
+     * Whether a BM25-only query will work: the daemon has to understand the
+     * `lexical` retriever *and* the open store has to carry a BM25 index.
+     */
     supportsLexicalRetriever() {
-        return this.lexicalRetriever;
+        return this.lexicalRetriever && this.hybridStore;
     }
     /**
      * Repo files whose on-disk content the index does not have — unknown to it,
@@ -206,13 +211,33 @@ export class EmbsearchService {
         // it can serve one is answered by `info` below, once the client is up.
         this.crossEncoder = binaryVersion !== undefined && atLeast(binaryVersion, MIN_RERANK_VERSION);
         const storeDir = this.options.storeDir ?? getEmbsearchStoreDir(this.options.cwd);
-        this.client = new EmbSearchClient({
-            binaryPath: binary,
-            storePath: getVectorStoreDir(storeDir),
-            hybrid: this.options.hybridStore,
-        });
-        await this.client.ready();
-        const info = await this.client.info();
+        // A hybrid store carries a BM25 index next to its vectors, which is what
+        // lets search use BM25 as its lexical leg instead of ripgrep. Callers may
+        // force it either way; by default it follows what the daemon can serve.
+        const wantHybrid = this.options.hybridStore ?? this.lexicalRetriever;
+        const openClient = async () => {
+            this.client = new EmbSearchClient({
+                binaryPath: binary,
+                storePath: getVectorStoreDir(storeDir),
+                hybrid: wantHybrid,
+            });
+            await this.client.ready();
+            return await this.client.info();
+        };
+        let info = await openClient();
+        // 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
+        // would fail. Ask the store itself rather than trusting the sidecar, and
+        // 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();
+        }
+        this.hybridStore = info.hybrid === true;
         // A daemon old enough to omit the field is left on the version verdict —
         // back then the weights were bundled, so version did imply capability.
         if (info.rerank === false)

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

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

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

@@ -1 +1 @@
-{"version":3,"file":"eval.d.ts","sourceRoot":"","sources":["../../../src/core/search/eval.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAE1E,OAAO,KAAK,EAAE,aAAa,EAAE,kBAAkB,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AAEhF,MAAM,WAAW,YAAY;IAC5B,gCAAgC;IAChC,IAAI,EAAE,MAAM,CAAC;IACb,mEAAmE;IACnE,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB;;;;;;OAMG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB;;;;;;OAMG;IACH,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CACxB;AAED,MAAM,WAAW,SAAS;IACzB,EAAE,EAAE,MAAM,CAAC;IACX;6DACyD;IACzD,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,YAAY,EAAE,CAAC;CACrB;AAED,MAAM,WAAW,UAAU;IAC1B,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,UAAU,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;OAKG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;;;OAIG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;iEAC6D;IAC7D,YAAY,CAAC,EAAE,OAAO,CAAC;CACvB;AAED;;;;;;;;;;GAUG;AACH,eAAO,MAAM,YAAY,EAAE,SAAS,UAAU,EAoC7C,CAAC;AAKF,wBAAgB,eAAe,CAAC,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,YAAY,GAAG,OAAO,CAKhF;AAED,gFAAgF;AAChF,wBAAgB,SAAS,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,EAAE,IAAI,EAAE,SAAS,YAAY,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,MAAM,CAQhH;AAED;;;;;;;;GAQG;AACH,wBAAgB,GAAG,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,EAAE,IAAI,EAAE,SAAS,YAAY,EAAE,GAAG,MAAM,CAQ/F;AAED,MAAM,WAAW,eAAe;IAC/B,KAAK,EAAE,MAAM,CAAC;IACd,YAAY,EAAE,kBAAkB,CAAC;IACjC,QAAQ,EAAE,OAAO,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;IACnB,GAAG,EAAE,MAAM,CAAC;CACZ;AAED,wBAAsB,aAAa,CAClC,GAAG,EAAE,MAAM,EACX,SAAS,EAAE,SAAS,EACpB,OAAO,GAAE,SAAS,UAAU,EAAiB,EAC7C,OAAO,CAAC,EAAE,gBAAgB,EAC1B,aAAa,CAAC,EAAE,gBAAgB,GAC9B,OAAO,CAAC,eAAe,EAAE,CAAC,CAiC5B","sourcesContent":["/**\n * Retrieval evaluation gate (docs/hybrid-retrieval-design.md, step 6 of the\n * shipping order).\n *\n * Measures Recall@K and MRR for lexical, semantic, and hybrid retrieval across\n * an RRF `k` sweep, against a gold set keyed by **path + line range matched by\n * span overlap** — never by chunkId, which is only stable per index build.\n * Recall@50 doubles as the reranker gate: a gold span that never reaches the\n * fused top-50 cannot be rescued by any reranker.\n *\n * Rank-sensitive metrics are the point of Recall@1 and MRR: an agent reads the\n * top result or two, so \"the gold span is somewhere in the top 5\" understates\n * how much ordering matters. Recall@5/10/50 are kept for continuity with the\n * numbers already published in the design note.\n *\n * Driven by `scripts/search-eval.ts` via `eval-harness.ts`; the scoring logic\n * lives here so it is unit-testable without an embedding index.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { retrieveCandidates } from \"./hybrid-search.js\";\nimport type { CandidateSpan, ResolvedSearchMode, SearchMode } from \"./types.js\";\n\nexport interface EvalGoldSpan {\n\t/** Repo-relative POSIX path. */\n\tpath: string;\n\t/** 1-based inclusive; omit both to accept any span in the file. */\n\tstartLine?: number;\n\tendLine?: number;\n\t/**\n\t * Literal source text that must occur inside `[startLine, endLine]`. Not\n\t * used for scoring — it is how the gold set survives the corpus moving\n\t * underneath it: `scripts/search-eval-gold.ts` re-resolves the line range\n\t * from this anchor, and the fixture test fails when an anchor no longer\n\t * sits inside its recorded range. Omitted for `scope: \"file\"` entries.\n\t */\n\tanchor?: string;\n\t/**\n\t * `\"span\"` (default) scores by overlap with the recorded line range;\n\t * `\"file\"` accepts any span in the file. File scope is deliberate for\n\t * path-class queries, where the whole file *is* the answer — it is\n\t * recorded explicitly so a file-level score is never mistaken for a\n\t * span-level one.\n\t */\n\tscope?: \"span\" | \"file\";\n}\n\nexport interface EvalQuery {\n\tid: string;\n\t/** Query class from the design doc (exact-symbol, path, error-fragment,\n\t *  conceptual, cross-file, boundary). Reporting only. */\n\tclass: string;\n\tquery: string;\n\tgold: EvalGoldSpan[];\n}\n\nexport interface EvalConfig {\n\tlabel: string;\n\tmode: SearchMode;\n\trrfK?: number;\n\trerank?: boolean;\n\t/**\n\t * Score this config against a daemon-side hybrid store (BM25 fused with\n\t * vectors inside the Rust daemon) instead of the dense-only index. Skipped\n\t * when the harness has no hybrid service, so records without one simply\n\t * omit the row rather than silently scoring it as plain semantic.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as a separate leg and fuse it here with\n\t * dense and grep. Like `daemonHybrid` it needs a hybrid store, so it is\n\t * skipped when the harness has no hybrid service.\n\t */\n\tbm25Leg?: boolean;\n\t/** Rerank with the daemon's cross-encoder rather than the deterministic\n\t *  scorer. Needs a service, so it is skipped without one. */\n\tcrossEncoder?: boolean;\n}\n\n/**\n * The sweep from the design doc — single retrievers, hybrid across k, the\n * routed auto mode — plus reranked (`+rr`) variants for the step 7 gate.\n *\n * Exported because `scripts/search-eval.ts` and `test/search-eval.test.ts`\n * both consume it. It previously lost its `export` to a knip-driven\n * dead-export sweep (02efaab): the only importer was a `.mjs` script reading\n * from `dist/`, which static analysis cannot see, and the eval gate silently\n * stopped running. The TypeScript importers are the fix — do not \"clean up\"\n * this export without checking them.\n */\nexport const EVAL_CONFIGS: readonly EvalConfig[] = [\n\t{ label: \"lexical\", mode: \"lexical\" },\n\t{ label: \"semantic\", mode: \"semantic\" },\n\t{ label: \"hybrid k=0\", mode: \"hybrid\", rrfK: 0 },\n\t{ label: \"hybrid k=2\", mode: \"hybrid\", rrfK: 2 },\n\t{ label: \"hybrid k=10\", mode: \"hybrid\", rrfK: 10 },\n\t{ label: \"hybrid k=60\", mode: \"hybrid\", rrfK: 60 },\n\t{ label: \"auto\", mode: \"auto\" },\n\t{ label: \"lexical +rr\", mode: \"lexical\", rerank: true },\n\t{ label: \"semantic +rr\", mode: \"semantic\", rerank: true },\n\t{ label: \"hybrid k=2 +rr\", mode: \"hybrid\", rrfK: 2, rerank: true },\n\t{ label: \"hybrid k=60 +rr\", mode: \"hybrid\", rrfK: 60, rerank: true },\n\t{ label: \"auto +rr\", mode: \"auto\", rerank: true },\n\t// Is BM25 a better lexical leg than ripgrep? These two run the daemon's own\n\t// vector+BM25 fusion and no ripgrep at all, so comparing them against\n\t// \"semantic\" isolates what BM25 adds, and against \"hybrid k=2\" compares the\n\t// two lexical legs at the system level.\n\t{ label: \"daemon-hybrid\", mode: \"semantic\", daemonHybrid: true },\n\t{ label: \"daemon-hybrid +rr\", mode: \"semantic\", rerank: true, daemonHybrid: true },\n\t// Three-way fusion: dense + BM25 + grep, all fused here so `k` is ours and\n\t// the per-leg ranks survive into the trace. `bm25+dense` isolates what the\n\t// grep leg still contributes once BM25 is present.\n\t//\n\t// `3-way` used to lose badly to `bm25+dense` (MRR 0.608 -> 0.552, 18 of 62\n\t// queries worse) because two lexical views of the same corpus dilute each\n\t// other in fusion. Its grep leg is now scoped to files the index has not\n\t// read, so the two rows are identical on this corpus by construction —\n\t// the difference only appears under `--live-edits`, where `3-way` scores\n\t// 1.000 and `bm25+dense` scores 0.000.\n\t{ label: \"bm25+dense +rr\", mode: \"semantic\", rerank: true, bm25Leg: true },\n\t{ label: \"3-way +rr\", mode: \"hybrid\", rerank: true, bm25Leg: true },\n\t// Cross-encoder reranking, against the same retrieval each deterministic\n\t// `+rr` row uses — so the delta is the reranker and nothing else.\n\t{ label: \"semantic +ce\", mode: \"semantic\", crossEncoder: true },\n\t{ label: \"auto +ce\", mode: \"auto\", crossEncoder: true },\n\t{ label: \"bm25+dense +ce\", mode: \"semantic\", bm25Leg: true, crossEncoder: true },\n];\n\n/** Candidates fetched per eval query — deep enough for the reranker gate. */\nconst EVAL_FETCH_LIMIT = 50;\n\nexport function spanMatchesGold(span: CandidateSpan, gold: EvalGoldSpan): boolean {\n\tif (span.path !== gold.path) return false;\n\tif (gold.scope === \"file\") return true;\n\tif (gold.startLine === undefined || gold.endLine === undefined) return true;\n\treturn span.startLine <= gold.endLine && span.endLine >= gold.startLine;\n}\n\n/** Fraction of gold spans matched by at least one of the top-`k` candidates. */\nexport function recallAtK(candidates: readonly CandidateSpan[], gold: readonly EvalGoldSpan[], k: number): number {\n\tif (gold.length === 0) return 0;\n\tconst top = candidates.slice(0, k);\n\tlet matched = 0;\n\tfor (const g of gold) {\n\t\tif (top.some((c) => spanMatchesGold(c, g))) matched++;\n\t}\n\treturn matched / gold.length;\n}\n\n/**\n * Mean reciprocal rank, averaged over gold spans: for each gold span, `1/rank`\n * of the first candidate matching it (0 when it never appears).\n *\n * Averaging per gold span rather than taking the single first hit keeps\n * multi-span queries (the cross-file class) honest — finding one of two\n * required sites should not score like finding both. For single-span queries\n * this reduces to textbook MRR.\n */\nexport function mrr(candidates: readonly CandidateSpan[], gold: readonly EvalGoldSpan[]): number {\n\tif (gold.length === 0) return 0;\n\tlet total = 0;\n\tfor (const g of gold) {\n\t\tconst index = candidates.findIndex((c) => spanMatchesGold(c, g));\n\t\tif (index >= 0) total += 1 / (index + 1);\n\t}\n\treturn total / gold.length;\n}\n\nexport interface EvalQueryResult {\n\tlabel: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegraded: boolean;\n\trecallAt1: number;\n\trecallAt5: number;\n\trecallAt10: number;\n\trecallAt50: number;\n\tmrr: number;\n}\n\nexport async function evaluateQuery(\n\tcwd: string,\n\tevalQuery: EvalQuery,\n\tconfigs: readonly EvalConfig[] = EVAL_CONFIGS,\n\tservice?: EmbsearchService,\n\thybridService?: EmbsearchService,\n): Promise<EvalQueryResult[]> {\n\tconst results: EvalQueryResult[] = [];\n\tfor (const config of configs) {\n\t\t// A daemon-hybrid config against the plain store would error in the\n\t\t// daemon (`query_hybrid requires a hybrid store`); omit the row instead.\n\t\tif ((config.daemonHybrid || config.bm25Leg) && !hybridService) continue;\n\t\t// Cross-encoder rows need a live daemon; omit rather than silently\n\t\t// falling back to the deterministic reranker under a \"+ce\" label.\n\t\tif (config.crossEncoder && !(config.bm25Leg ? hybridService : service)?.supportsCrossEncoder()) continue;\n\t\tconst retrieved = await retrieveCandidates({\n\t\t\tcwd,\n\t\t\tquery: evalQuery.query,\n\t\t\tmode: config.mode,\n\t\t\trrfK: config.rrfK,\n\t\t\trerank: config.rerank ?? false,\n\t\t\tlimit: EVAL_FETCH_LIMIT,\n\t\t\tservice: config.daemonHybrid || config.bm25Leg ? hybridService : service,\n\t\t\tdaemonHybrid: config.daemonHybrid,\n\t\t\tbm25Leg: config.bm25Leg,\n\t\t\tcrossEncoder: config.crossEncoder,\n\t\t});\n\t\tresults.push({\n\t\t\tlabel: config.label,\n\t\t\tresolvedMode: retrieved.resolvedMode,\n\t\t\tdegraded: retrieved.degradedReason !== undefined,\n\t\t\trecallAt1: recallAtK(retrieved.candidates, evalQuery.gold, 1),\n\t\t\trecallAt5: recallAtK(retrieved.candidates, evalQuery.gold, 5),\n\t\t\trecallAt10: recallAtK(retrieved.candidates, evalQuery.gold, 10),\n\t\t\trecallAt50: recallAtK(retrieved.candidates, evalQuery.gold, 50),\n\t\t\tmrr: mrr(retrieved.candidates, evalQuery.gold),\n\t\t});\n\t}\n\treturn results;\n}\n"]}
+{"version":3,"file":"eval.d.ts","sourceRoot":"","sources":["../../../src/core/search/eval.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAE1E,OAAO,KAAK,EAAE,aAAa,EAAE,kBAAkB,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AAEhF,MAAM,WAAW,YAAY;IAC5B,gCAAgC;IAChC,IAAI,EAAE,MAAM,CAAC;IACb,mEAAmE;IACnE,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB;;;;;;OAMG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB;;;;;;OAMG;IACH,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CACxB;AAED,MAAM,WAAW,SAAS;IACzB,EAAE,EAAE,MAAM,CAAC;IACX;6DACyD;IACzD,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,YAAY,EAAE,CAAC;CACrB;AAED,MAAM,WAAW,UAAU;IAC1B,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,UAAU,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;OAKG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;;;OAIG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;iEAC6D;IAC7D,YAAY,CAAC,EAAE,OAAO,CAAC;CACvB;AAED;;;;;;;;;;GAUG;AACH,eAAO,MAAM,YAAY,EAAE,SAAS,UAAU,EAsC7C,CAAC;AAKF,wBAAgB,eAAe,CAAC,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,YAAY,GAAG,OAAO,CAKhF;AAED,gFAAgF;AAChF,wBAAgB,SAAS,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,EAAE,IAAI,EAAE,SAAS,YAAY,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,MAAM,CAQhH;AAED;;;;;;;;GAQG;AACH,wBAAgB,GAAG,CAAC,UAAU,EAAE,SAAS,aAAa,EAAE,EAAE,IAAI,EAAE,SAAS,YAAY,EAAE,GAAG,MAAM,CAQ/F;AAED,MAAM,WAAW,eAAe;IAC/B,KAAK,EAAE,MAAM,CAAC;IACd,YAAY,EAAE,kBAAkB,CAAC;IACjC,QAAQ,EAAE,OAAO,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;IACnB,GAAG,EAAE,MAAM,CAAC;CACZ;AAED,wBAAsB,aAAa,CAClC,GAAG,EAAE,MAAM,EACX,SAAS,EAAE,SAAS,EACpB,OAAO,GAAE,SAAS,UAAU,EAAiB,EAC7C,OAAO,CAAC,EAAE,gBAAgB,EAC1B,aAAa,CAAC,EAAE,gBAAgB,GAC9B,OAAO,CAAC,eAAe,EAAE,CAAC,CAqC5B","sourcesContent":["/**\n * Retrieval evaluation gate (docs/hybrid-retrieval-design.md, step 6 of the\n * shipping order).\n *\n * Measures Recall@K and MRR for lexical, semantic, and hybrid retrieval across\n * an RRF `k` sweep, against a gold set keyed by **path + line range matched by\n * span overlap** — never by chunkId, which is only stable per index build.\n * Recall@50 doubles as the reranker gate: a gold span that never reaches the\n * fused top-50 cannot be rescued by any reranker.\n *\n * Rank-sensitive metrics are the point of Recall@1 and MRR: an agent reads the\n * top result or two, so \"the gold span is somewhere in the top 5\" understates\n * how much ordering matters. Recall@5/10/50 are kept for continuity with the\n * numbers already published in the design note.\n *\n * Driven by `scripts/search-eval.ts` via `eval-harness.ts`; the scoring logic\n * lives here so it is unit-testable without an embedding index.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { retrieveCandidates } from \"./hybrid-search.js\";\nimport type { CandidateSpan, ResolvedSearchMode, SearchMode } from \"./types.js\";\n\nexport interface EvalGoldSpan {\n\t/** Repo-relative POSIX path. */\n\tpath: string;\n\t/** 1-based inclusive; omit both to accept any span in the file. */\n\tstartLine?: number;\n\tendLine?: number;\n\t/**\n\t * Literal source text that must occur inside `[startLine, endLine]`. Not\n\t * used for scoring — it is how the gold set survives the corpus moving\n\t * underneath it: `scripts/search-eval-gold.ts` re-resolves the line range\n\t * from this anchor, and the fixture test fails when an anchor no longer\n\t * sits inside its recorded range. Omitted for `scope: \"file\"` entries.\n\t */\n\tanchor?: string;\n\t/**\n\t * `\"span\"` (default) scores by overlap with the recorded line range;\n\t * `\"file\"` accepts any span in the file. File scope is deliberate for\n\t * path-class queries, where the whole file *is* the answer — it is\n\t * recorded explicitly so a file-level score is never mistaken for a\n\t * span-level one.\n\t */\n\tscope?: \"span\" | \"file\";\n}\n\nexport interface EvalQuery {\n\tid: string;\n\t/** Query class from the design doc (exact-symbol, path, error-fragment,\n\t *  conceptual, cross-file, boundary). Reporting only. */\n\tclass: string;\n\tquery: string;\n\tgold: EvalGoldSpan[];\n}\n\nexport interface EvalConfig {\n\tlabel: string;\n\tmode: SearchMode;\n\trrfK?: number;\n\trerank?: boolean;\n\t/**\n\t * Score this config against a daemon-side hybrid store (BM25 fused with\n\t * vectors inside the Rust daemon) instead of the dense-only index. Skipped\n\t * when the harness has no hybrid service, so records without one simply\n\t * omit the row rather than silently scoring it as plain semantic.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as a separate leg and fuse it here with\n\t * dense and grep. Like `daemonHybrid` it needs a hybrid store, so it is\n\t * skipped when the harness has no hybrid service.\n\t */\n\tbm25Leg?: boolean;\n\t/** Rerank with the daemon's cross-encoder rather than the deterministic\n\t *  scorer. Needs a service, so it is skipped without one. */\n\tcrossEncoder?: boolean;\n}\n\n/**\n * The sweep from the design doc — single retrievers, hybrid across k, the\n * routed auto mode — plus reranked (`+rr`) variants for the step 7 gate.\n *\n * Exported because `scripts/search-eval.ts` and `test/search-eval.test.ts`\n * both consume it. It previously lost its `export` to a knip-driven\n * dead-export sweep (02efaab): the only importer was a `.mjs` script reading\n * from `dist/`, which static analysis cannot see, and the eval gate silently\n * stopped running. The TypeScript importers are the fix — do not \"clean up\"\n * this export without checking them.\n */\nexport const EVAL_CONFIGS: readonly EvalConfig[] = [\n\t{ label: \"lexical\", mode: \"lexical\" },\n\t{ label: \"semantic\", mode: \"semantic\" },\n\t{ label: \"hybrid k=0\", mode: \"hybrid\", rrfK: 0 },\n\t{ label: \"hybrid k=2\", mode: \"hybrid\", rrfK: 2 },\n\t{ label: \"hybrid k=10\", mode: \"hybrid\", rrfK: 10 },\n\t{ label: \"hybrid k=60\", mode: \"hybrid\", rrfK: 60 },\n\t{ label: \"auto\", mode: \"auto\" },\n\t{ label: \"lexical +rr\", mode: \"lexical\", rerank: true },\n\t{ label: \"semantic +rr\", mode: \"semantic\", rerank: true },\n\t{ label: \"hybrid k=2 +rr\", mode: \"hybrid\", rrfK: 2, rerank: true },\n\t{ label: \"hybrid k=60 +rr\", mode: \"hybrid\", rrfK: 60, rerank: true },\n\t{ label: \"auto +rr\", mode: \"auto\", rerank: true },\n\t// Is BM25 a better lexical leg than ripgrep? These two run the daemon's own\n\t// vector+BM25 fusion and no ripgrep at all, so comparing them against\n\t// \"semantic\" isolates what BM25 adds, and against \"hybrid k=2\" compares the\n\t// two lexical legs at the system level.\n\t{ label: \"daemon-hybrid\", mode: \"semantic\", daemonHybrid: true },\n\t{ label: \"daemon-hybrid +rr\", mode: \"semantic\", rerank: true, daemonHybrid: true },\n\t// Three-way fusion: dense + BM25 + grep, all fused here so `k` is ours and\n\t// the per-leg ranks survive into the trace. `bm25+dense` isolates what the\n\t// grep leg still contributes once BM25 is present.\n\t//\n\t// `3-way` used to lose badly to `bm25+dense` (MRR 0.608 -> 0.552, 18 of 62\n\t// queries worse) because two lexical views of the same corpus dilute each\n\t// other in fusion. Its grep leg is now scoped to files the index has not\n\t// read, so the two rows are identical on this corpus by construction —\n\t// the difference only appears under `--live-edits`, where `3-way` scores\n\t// 1.000 and `bm25+dense` scores 0.000.\n\t{ label: \"bm25+dense +rr\", mode: \"semantic\", rerank: true, bm25Leg: true },\n\t// The shipped default: BM25 over the indexed corpus, grep narrowed to what\n\t// the index has not read, dense alongside both.\n\t{ label: \"3-way +rr\", mode: \"hybrid\", rerank: true, bm25Leg: true },\n\t// Cross-encoder reranking, against the same retrieval each deterministic\n\t// `+rr` row uses — so the delta is the reranker and nothing else.\n\t{ label: \"semantic +ce\", mode: \"semantic\", crossEncoder: true },\n\t{ label: \"auto +ce\", mode: \"auto\", crossEncoder: true },\n\t{ label: \"bm25+dense +ce\", mode: \"semantic\", bm25Leg: true, crossEncoder: true },\n];\n\n/** Candidates fetched per eval query — deep enough for the reranker gate. */\nconst EVAL_FETCH_LIMIT = 50;\n\nexport function spanMatchesGold(span: CandidateSpan, gold: EvalGoldSpan): boolean {\n\tif (span.path !== gold.path) return false;\n\tif (gold.scope === \"file\") return true;\n\tif (gold.startLine === undefined || gold.endLine === undefined) return true;\n\treturn span.startLine <= gold.endLine && span.endLine >= gold.startLine;\n}\n\n/** Fraction of gold spans matched by at least one of the top-`k` candidates. */\nexport function recallAtK(candidates: readonly CandidateSpan[], gold: readonly EvalGoldSpan[], k: number): number {\n\tif (gold.length === 0) return 0;\n\tconst top = candidates.slice(0, k);\n\tlet matched = 0;\n\tfor (const g of gold) {\n\t\tif (top.some((c) => spanMatchesGold(c, g))) matched++;\n\t}\n\treturn matched / gold.length;\n}\n\n/**\n * Mean reciprocal rank, averaged over gold spans: for each gold span, `1/rank`\n * of the first candidate matching it (0 when it never appears).\n *\n * Averaging per gold span rather than taking the single first hit keeps\n * multi-span queries (the cross-file class) honest — finding one of two\n * required sites should not score like finding both. For single-span queries\n * this reduces to textbook MRR.\n */\nexport function mrr(candidates: readonly CandidateSpan[], gold: readonly EvalGoldSpan[]): number {\n\tif (gold.length === 0) return 0;\n\tlet total = 0;\n\tfor (const g of gold) {\n\t\tconst index = candidates.findIndex((c) => spanMatchesGold(c, g));\n\t\tif (index >= 0) total += 1 / (index + 1);\n\t}\n\treturn total / gold.length;\n}\n\nexport interface EvalQueryResult {\n\tlabel: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegraded: boolean;\n\trecallAt1: number;\n\trecallAt5: number;\n\trecallAt10: number;\n\trecallAt50: number;\n\tmrr: number;\n}\n\nexport async function evaluateQuery(\n\tcwd: string,\n\tevalQuery: EvalQuery,\n\tconfigs: readonly EvalConfig[] = EVAL_CONFIGS,\n\tservice?: EmbsearchService,\n\thybridService?: EmbsearchService,\n): Promise<EvalQueryResult[]> {\n\tconst results: EvalQueryResult[] = [];\n\tfor (const config of configs) {\n\t\t// A daemon-hybrid config against the plain store would error in the\n\t\t// daemon (`query_hybrid requires a hybrid store`); omit the row instead.\n\t\tif ((config.daemonHybrid || config.bm25Leg) && !hybridService) continue;\n\t\t// Cross-encoder rows need a live daemon; omit rather than silently\n\t\t// falling back to the deterministic reranker under a \"+ce\" label.\n\t\tif (config.crossEncoder && !(config.bm25Leg ? hybridService : service)?.supportsCrossEncoder()) continue;\n\t\tconst retrieved = await retrieveCandidates({\n\t\t\tcwd,\n\t\t\tquery: evalQuery.query,\n\t\t\tmode: config.mode,\n\t\t\trrfK: config.rrfK,\n\t\t\trerank: config.rerank ?? false,\n\t\t\tlimit: EVAL_FETCH_LIMIT,\n\t\t\tservice: config.daemonHybrid || config.bm25Leg ? hybridService : service,\n\t\t\tdaemonHybrid: config.daemonHybrid,\n\t\t\t// Explicit, never inherited: `bm25Leg` now defaults to on in production,\n\t\t\t// and letting that leak in would silently give every historical row a\n\t\t\t// BM25 leg it never had — the numbers would stop meaning what their\n\t\t\t// labels say.\n\t\t\tbm25Leg: config.bm25Leg ?? false,\n\t\t\tcrossEncoder: config.crossEncoder,\n\t\t});\n\t\tresults.push({\n\t\t\tlabel: config.label,\n\t\t\tresolvedMode: retrieved.resolvedMode,\n\t\t\tdegraded: retrieved.degradedReason !== undefined,\n\t\t\trecallAt1: recallAtK(retrieved.candidates, evalQuery.gold, 1),\n\t\t\trecallAt5: recallAtK(retrieved.candidates, evalQuery.gold, 5),\n\t\t\trecallAt10: recallAtK(retrieved.candidates, evalQuery.gold, 10),\n\t\t\trecallAt50: recallAtK(retrieved.candidates, evalQuery.gold, 50),\n\t\t\tmrr: mrr(retrieved.candidates, evalQuery.gold),\n\t\t});\n\t}\n\treturn results;\n}\n"]}

package/dist/core/search/eval.js

@@ -58,6 +58,8 @@ export const EVAL_CONFIGS = [
     // the difference only appears under `--live-edits`, where `3-way` scores
     // 1.000 and `bm25+dense` scores 0.000.
     { label: "bm25+dense +rr", mode: "semantic", rerank: true, bm25Leg: true },
+    // The shipped default: BM25 over the indexed corpus, grep narrowed to what
+    // the index has not read, dense alongside both.
     { label: "3-way +rr", mode: "hybrid", rerank: true, bm25Leg: true },
     // Cross-encoder reranking, against the same retrieval each deterministic
     // `+rr` row uses — so the delta is the reranker and nothing else.
@@ -128,7 +130,11 @@ export async function evaluateQuery(cwd, evalQuery, configs = EVAL_CONFIGS, serv
             limit: EVAL_FETCH_LIMIT,
             service: config.daemonHybrid || config.bm25Leg ? hybridService : service,
             daemonHybrid: config.daemonHybrid,
-            bm25Leg: config.bm25Leg,
+            // Explicit, never inherited: `bm25Leg` now defaults to on in production,
+            // and letting that leak in would silently give every historical row a
+            // BM25 leg it never had — the numbers would stop meaning what their
+            // labels say.
+            bm25Leg: config.bm25Leg ?? false,
             crossEncoder: config.crossEncoder,
         });
         results.push({

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

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

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

@@ -38,8 +38,15 @@ export interface RetrieveOptions {
     daemonHybrid?: boolean;
     /**
      * Fetch the daemon's BM25 index as its own ranked list and fuse it here,
-     * alongside dense and grep. Needs a store built with `--hybrid` and an
-     * embsearch new enough to serve `retriever: "lexical"`.
+     * alongside dense and grep.
+     *
+     * Defaults to on wherever the daemon can serve it, because BM25 is the
+     * better lexical leg on the indexed corpus: Recall@50 0.790 -> 0.879, 6 of
+     * 62 queries better and 0 worse (p <= 0.05). When it is on, the grep leg
+     * narrows to files the index has not read — see `staleFiles`.
+     *
+     * Set `false` to force ripgrep as the only lexical leg; the eval harness
+     * does this to keep measuring what the old rows measured.
      */
     bm25Leg?: boolean;
     /**

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

@@ -1 +1 @@
-{"version":3,"file":"hybrid-search.d.ts","sourceRoot":"","sources":["../../../src/core/search/hybrid-search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAS1E,OAAO,KAAK,EAAiB,cAAc,EAAa,kBAAkB,EAAE,UAAU,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAuCxH,MAAM,WAAW,eAAe;IAC/B,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,CAAC,EAAE,UAAU,CAAC;IAClB,kDAAkD;IAClD,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,yCAAyC;IACzC,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,wFAAwF;IACxF,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,wEAAwE;IACxE,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;;;;OAQG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;;;OAIG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;;;;OAIG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB,OAAO,CAAC,EAAE,gBAAgB,CAAC;IAC3B,MAAM,CAAC,EAAE,WAAW,CAAC;CACrB;AAED,MAAM,WAAW,cAAc;IAC9B,UAAU,EAAE,cAAc,EAAE,CAAC;IAC7B,YAAY,EAAE,kBAAkB,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,UAAU,EAAE,WAAW,CAAC,YAAY,CAAC,CAAC;IACtC,UAAU,EAAE,WAAW,CAAC,YAAY,CAAC,CAAC;IACtC,uDAAuD;IACvD,QAAQ,CAAC,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;IAC3C,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,WAAW,CAAC,QAAQ,CAAC,CAAC;CAC/B;AAED,MAAM,WAAW,gBAAiB,SAAQ,eAAe;IACxD,oDAAoD;IACpD,WAAW,CAAC,EAAE,MAAM,CAAC;CACrB;AAED,MAAM,WAAW,eAAe;IAC/B,IAAI,EAAE,MAAM,CAAC;IACb,YAAY,EAAE,kBAAkB,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,WAAW,EAAE,MAAM,CAAC;IACpB,uDAAuD;IACvD,QAAQ,CAAC,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;CAC3C;AAYD;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,oBAAoB,CACnC,UAAU,EAAE,SAAS,cAAc,EAAE,EACrC,UAAU,EAAE,WAAW,CAAC,MAAM,CAAC,GAC7B,cAAc,EAAE,CAQlB;AAED,wBAAsB,kBAAkB,CAAC,OAAO,EAAE,eAAe,GAAG,OAAO,CAAC,cAAc,CAAC,CAsK1F;AAED,wBAAsB,SAAS,CAAC,OAAO,EAAE,gBAAgB,GAAG,OAAO,CAAC,eAAe,CAAC,CAyBnF","sourcesContent":["/**\n * Hybrid search orchestrator: resolve mode, run retrievers in parallel, fuse\n * by rank, expand within budget, trace everything\n * (docs/hybrid-retrieval-design.md).\n *\n * Single-retriever modes flow through the same pipeline — rrfFuse over one\n * list preserves its order — so lexical, semantic, and hybrid all produce the\n * same result shape and the same trace record.\n *\n * `retrieveCandidates` is the candidate-level core (also used by the eval\n * harness, which needs forced modes, a configurable `k`, and no trace\n * pollution); `runSearch` wraps it with span expansion and tracing for the\n * tool.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { adaptGrepHits, type ChunkLookup } from \"./adapter.js\";\nimport { assembleContext } from \"./context-assembler.js\";\nimport { crossEncoderRerank } from \"./cross-rerank.js\";\nimport { runLexicalRetriever } from \"./lexical-retriever.js\";\nimport { resolveSearchMode } from \"./mode.js\";\nimport { rerankCandidates } from \"./rerank.js\";\nimport { DEFAULT_RRF_K, rrfFuse } from \"./rrf.js\";\nimport { writeSearchTrace } from \"./trace.js\";\nimport type { CandidateSpan, FusedCandidate, RankedHit, ResolvedSearchMode, SearchMode, SearchTrace } from \"./types.js\";\n\n/** Raw grep line-hits fetched per query (pre-collapse). */\nconst LEXICAL_MATCH_LIMIT = 200;\n/** Adapted lexical candidates entering fusion. The eval gate showed the\n *  uncapped lexical tail diluting hybrid below plain semantic: lexical\n *  precision is front-loaded by the adapter's term-evidence ranking, while\n *  RRF weighs a rank-30 lexical candidate like a rank-30 embedding hit. */\nconst LEXICAL_FUSION_CAP = 20;\n/**\n * Embedding hits fetched per query when fusing with the grep leg.\n *\n * Deliberately shallow. The grep list is capped at {@link LEXICAL_FUSION_CAP}\n * because its tail is unranked noise, so a deep dense pool here simply\n * outnumbers it: raising this to 200 alongside a 20-candidate grep list cost\n * `auto +rr` 0.472 -> 0.450 MRR.\n */\nconst EMBED_TOP_K = 50;\n/**\n * Per-leg depth when both retrievers are ranked ones (dense + BM25).\n *\n * Four times {@link FUSED_WINDOW}, matching what the daemon's `query_hybrid`\n * does internally (`pool = 4·k`). Fusing at the same depth as the window loses\n * any candidate ranked well by one retriever but just outside the other's\n * top-50 — worth 12pp R@10 and 8pp R@50 for `bm25+dense`, which is exactly\n * what closed the gap to daemon-side fusion.\n */\nconst FUSION_POOL_TOP_K = 200;\n/** BM25 depth, matching {@link FUSION_POOL_TOP_K} so neither ranked leg is\n *  structurally advantaged by pool size. The daemon returns only documents\n *  sharing a query term, so this is an upper bound, not a fill. */\nconst BM25_TOP_K = FUSION_POOL_TOP_K;\n/** Candidates from unindexed/stale files moved ahead of the fused window.\n *  Small on purpose: this answers \"what did I just write\", not \"search the\n *  working tree\". */\nconst STALE_HOIST_CAP = 5;\n/** Fused candidates kept for reranking / final slicing. */\nconst FUSED_WINDOW = 50;\n\nexport interface RetrieveOptions {\n\tcwd: string;\n\tquery: string;\n\tmode?: SearchMode;\n\t/** Optional glob filter applied to file paths. */\n\tglob?: string;\n\t/** Maximum fused candidates returned. */\n\tlimit?: number;\n\t/** RRF constant override (eval harness sweeps this). Default: {@link DEFAULT_RRF_K}. */\n\trrfK?: number;\n\t/** Rerank the fused top-50 before slicing to `limit`. Default: true. */\n\trerank?: boolean;\n\t/**\n\t * Ask the daemon to fuse its own BM25 index with the vectors and return one\n\t * already-fused ranking, instead of taking a dense-only list. Needs a store\n\t * built with `--hybrid`. The fused list arrives as a single \"embed\" leg,\n\t * because a pre-fused ranking has no per-retriever structure left to record.\n\t *\n\t * Prefer {@link bm25Leg}: fusing here keeps the legs separable in the trace\n\t * and lets the grep leg participate.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as its own ranked list and fuse it here,\n\t * alongside dense and grep. Needs a store built with `--hybrid` and an\n\t * embsearch new enough to serve `retriever: \"lexical\"`.\n\t */\n\tbm25Leg?: boolean;\n\t/**\n\t * Reorder the fused shortlist with the daemon's cross-encoder instead of\n\t * the deterministic reranker. Needs embsearch >= 0.3.0; costs one model\n\t * pass per scored candidate.\n\t */\n\tcrossEncoder?: boolean;\n\tservice?: EmbsearchService;\n\tsignal?: AbortSignal;\n}\n\nexport interface RetrieveResult {\n\tcandidates: FusedCandidate[];\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tindexPhase: SearchTrace[\"indexPhase\"];\n\tretrievers: SearchTrace[\"retrievers\"];\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n\trrfK: number;\n\trerank?: SearchTrace[\"rerank\"];\n}\n\nexport interface RunSearchOptions extends RetrieveOptions {\n\t/** Approximate token budget for the result text. */\n\ttokenBudget?: number;\n}\n\nexport interface RunSearchResult {\n\ttext: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tresultCount: number;\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n}\n\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\n/**\n * Move candidates from files the index has not read to the front.\n *\n * They are there because grep found them and nothing else could: the index is\n * ranking a stale copy of the file, or has never seen it. Left to fuse, they\n * lose — RRF rewards agreement, and one leg reporting a single document is\n * outvoted by two legs agreeing on hundreds. Measured: scoping grep to stale\n * files without this hoist scored 25% on the live-edit set where unscoped grep\n * scored 100%, because the fused window filled with consensus hits about the\n * indexed copy.\n *\n * Capped, because \"stale\" scales with how far behind the index is. A few\n * edited files is the case this exists for; a fresh checkout makes everything\n * stale, and hoisting all of it would quietly turn hybrid search back into\n * grep. Past the cap the remainder keeps its fused position.\n */\nexport function hoistStaleCandidates(\n\tcandidates: readonly FusedCandidate[],\n\tstaleFiles: ReadonlySet<string>,\n): FusedCandidate[] {\n\tif (staleFiles.size === 0) return [...candidates];\n\tconst hoisted: FusedCandidate[] = [];\n\tconst rest: FusedCandidate[] = [];\n\tfor (const candidate of candidates) {\n\t\t(staleFiles.has(candidate.path) && hoisted.length < STALE_HOIST_CAP ? hoisted : rest).push(candidate);\n\t}\n\treturn [...hoisted, ...rest];\n}\n\nexport async function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult> {\n\tconst { cwd, query, service, signal } = options;\n\tconst glob = normalizeSearchGlob(options.glob);\n\tconst requestedMode = options.mode ?? \"auto\";\n\tconst limit = Math.max(1, options.limit ?? 10);\n\tconst rrfK = options.rrfK ?? DEFAULT_RRF_K;\n\n\tconst state = service?.getState();\n\tconst embedAvailable = service?.isAvailable() ?? false;\n\tconst embedUnavailableReason =\n\t\tstate === undefined\n\t\t\t? \"semantic index is not enabled\"\n\t\t\t: state.phase === \"unavailable\" || state.phase === \"skipped\"\n\t\t\t\t? state.reason\n\t\t\t\t: state.phase === \"idle\"\n\t\t\t\t\t? \"semantic index has not started\"\n\t\t\t\t\t: undefined;\n\n\tconst resolution = resolveSearchMode(query, requestedMode, embedAvailable, embedUnavailableReason);\n\tconst mode = resolution.mode;\n\n\t// Map lexical hits onto indexed chunk ids whenever the sidecar is usable,\n\t// even in lexical-only mode, so identities line up across modes.\n\tconst lookupChunk: ChunkLookup | undefined = embedAvailable\n\t\t? (rel, line) => service!.findEnclosingChunk(rel, line)\n\t\t: undefined;\n\n\tconst spans = new Map<string, CandidateSpan>();\n\t/** Files the index does not have current content for, when grep is scoped\n\t *  to them. Read again after fusion — see the hoist below. */\n\tconst staleSet = new Set<string>();\n\tconst lists: RankedHit[][] = [];\n\tconst retrieverStats: SearchTrace[\"retrievers\"] = {};\n\tconst errors: Error[] = [];\n\n\tconst runLexical = async (scopeToStale = false): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Scoped run: grep covers only what the index has not read yet, so it\n\t\t\t// adds the one thing BM25 cannot see without re-voting on documents\n\t\t\t// BM25 already ranked. Fusing two lexical views of the same corpus is\n\t\t\t// what made the three-leg configuration lose (18 of 62 queries worse).\n\t\t\tconst paths = scopeToStale ? service?.staleFiles(signal) : undefined;\n\t\t\tif (paths) for (const rel of paths) staleSet.add(rel);\n\t\t\tconst lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal, paths });\n\t\t\tconst adapted = adaptGrepHits(lineHits, lookupChunk);\n\t\t\t// In single-retriever lexical mode the full list is the result; in\n\t\t\t// hybrid, only the front-loaded head is trustworthy enough to vote.\n\t\t\tconst hits = mode === \"hybrid\" ? adapted.hits.slice(0, LEXICAL_FUSION_CAP) : adapted.hits;\n\t\t\tfor (const [id, span] of adapted.spans) if (!spans.has(id)) spans.set(id, span);\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runEmbed = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// The flat index pads top-k with whatever exists; with the cosine\n\t\t\t// metric the store uses, score <= 0 means \"no relation at all\", so\n\t\t\t// those padding hits would cast RRF votes on pure noise.\n\t\t\t// A BM25 leg is itself ranked, so the pair can afford — and needs — the\n\t\t\t// deeper pool; the grep leg cannot (see EMBED_TOP_K).\n\t\t\tconst topK = options.bm25Leg ? FUSION_POOL_TOP_K : EMBED_TOP_K;\n\t\t\tconst chunkHits = (\n\t\t\t\tawait service!.searchChunks(query, topK, glob, options.daemonHybrid ? \"hybrid\" : \"dense\")\n\t\t\t).filter((hit) => hit.score > 0);\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"embed\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runBm25 = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Raw BM25 sums; only the ordering enters fusion, the score is a\n\t\t\t// diagnostic. Unlike the dense leg there is no zero-score padding to\n\t\t\t// filter — the daemon omits documents sharing no query term.\n\t\t\tconst chunkHits = await service!.searchChunks(query, BM25_TOP_K, glob, \"lexical\");\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"bm25\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tif (!spans.has(hit.id)) {\n\t\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t\t}\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\t// BM25 is the better lexical leg where the index is current: it beats grep\n\t// on Recall@50 (0.790 -> 0.879, 6 queries better and 0 worse, p <= 0.05)\n\t// because it ranks the whole corpus rather than truncating a match stream.\n\t// It is also blind to anything indexed later than it was written, which is\n\t// precisely where grep still wins — so grep runs scoped to that set instead\n\t// of being dropped or left to duplicate BM25 over the whole tree.\n\tconst bm25AsLexicalLeg = options.bm25Leg && embedAvailable && mode !== \"lexical\";\n\n\tconst runs: Promise<void>[] = [];\n\tif (mode === \"lexical\" || mode === \"hybrid\") runs.push(runLexical(bm25AsLexicalLeg));\n\tif (mode === \"semantic\" || mode === \"hybrid\") runs.push(runEmbed());\n\t// `semantic` gets the BM25 leg too: the caller asked for the index, and\n\t// BM25 is part of it. Only an explicit `lexical` request excludes it.\n\tif (bm25AsLexicalLeg) runs.push(runBm25());\n\tawait Promise.all(runs);\n\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t// A partial failure in hybrid degrades to whichever retriever survived;\n\t// only a total loss is an error.\n\tif (lists.length === 0) throw errors[0] ?? new Error(\"search produced no retriever results\");\n\n\tconst fused = rrfFuse(lists, rrfK).slice(0, FUSED_WINDOW);\n\tlet candidates: FusedCandidate[] = [];\n\tfor (const hit of fused) {\n\t\tconst span = spans.get(hit.id);\n\t\tif (span) candidates.push({ ...hit, ...span });\n\t}\n\n\tlet rerankInfo: SearchTrace[\"rerank\"];\n\tif (options.crossEncoder && service?.supportsCrossEncoder()) {\n\t\tconst reranked = await crossEncoderRerank(query, candidates, cwd, service);\n\t\trerankInfo = { applied: true, candidateCount: reranked.scored, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t} else if (options.rerank !== false) {\n\t\tconst reranked = rerankCandidates(query, candidates, cwd);\n\t\trerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t}\n\t// After reranking, not before: both rerankers weight the fused prior, and a\n\t// lone grep hit from an unindexed file has the lowest prior there is, so\n\t// hoisting first would simply be undone. The reranker still orders the\n\t// hoisted set against itself.\n\tcandidates = hoistStaleCandidates(candidates, staleSet).slice(0, limit);\n\n\treturn {\n\t\tcandidates,\n\t\tresolvedMode: mode,\n\t\tdegradedReason: resolution.degradedReason,\n\t\tindexPhase: state?.phase === \"ready\" ? \"ready\" : state?.phase === \"indexing\" ? \"indexing\" : \"unavailable\",\n\t\tretrievers: retrieverStats,\n\t\tindexing: state?.phase === \"indexing\" ? { done: state.done, total: state.total } : undefined,\n\t\trrfK,\n\t\trerank: rerankInfo,\n\t};\n}\n\nexport async function runSearch(options: RunSearchOptions): Promise<RunSearchResult> {\n\tconst retrieved = await retrieveCandidates(options);\n\n\tconst assembled = assembleContext(retrieved.candidates, { cwd: options.cwd, tokenBudget: options.tokenBudget });\n\n\twriteSearchTrace(options.cwd, {\n\t\ttimestampMs: Date.now(),\n\t\tquery: options.query,\n\t\trequestedMode: options.mode ?? \"auto\",\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tindexPhase: retrieved.indexPhase,\n\t\trrfK: retrieved.resolvedMode === \"hybrid\" ? retrieved.rrfK : undefined,\n\t\tretrievers: retrieved.retrievers,\n\t\tfused: retrieved.candidates.map(({ id, rrfScore, ranks, rawScores }) => ({ id, rrfScore, ranks, rawScores })),\n\t\trerank: retrieved.rerank,\n\t});\n\n\treturn {\n\t\ttext: assembled.text,\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tresultCount: retrieved.candidates.length,\n\t\tindexing: retrieved.indexing,\n\t};\n}\n"]}
+{"version":3,"file":"hybrid-search.d.ts","sourceRoot":"","sources":["../../../src/core/search/hybrid-search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAS1E,OAAO,KAAK,EAAiB,cAAc,EAAa,kBAAkB,EAAE,UAAU,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAuCxH,MAAM,WAAW,eAAe;IAC/B,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,CAAC,EAAE,UAAU,CAAC;IAClB,kDAAkD;IAClD,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,yCAAyC;IACzC,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,wFAAwF;IACxF,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,wEAAwE;IACxE,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB;;;;;;;;OAQG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;;;;;;;;;;OAWG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB;;;;OAIG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB,OAAO,CAAC,EAAE,gBAAgB,CAAC;IAC3B,MAAM,CAAC,EAAE,WAAW,CAAC;CACrB;AAED,MAAM,WAAW,cAAc;IAC9B,UAAU,EAAE,cAAc,EAAE,CAAC;IAC7B,YAAY,EAAE,kBAAkB,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,UAAU,EAAE,WAAW,CAAC,YAAY,CAAC,CAAC;IACtC,UAAU,EAAE,WAAW,CAAC,YAAY,CAAC,CAAC;IACtC,uDAAuD;IACvD,QAAQ,CAAC,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;IAC3C,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,WAAW,CAAC,QAAQ,CAAC,CAAC;CAC/B;AAED,MAAM,WAAW,gBAAiB,SAAQ,eAAe;IACxD,oDAAoD;IACpD,WAAW,CAAC,EAAE,MAAM,CAAC;CACrB;AAED,MAAM,WAAW,eAAe;IAC/B,IAAI,EAAE,MAAM,CAAC;IACb,YAAY,EAAE,kBAAkB,CAAC;IACjC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,WAAW,EAAE,MAAM,CAAC;IACpB,uDAAuD;IACvD,QAAQ,CAAC,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;CAC3C;AAYD;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,oBAAoB,CACnC,UAAU,EAAE,SAAS,cAAc,EAAE,EACrC,UAAU,EAAE,WAAW,CAAC,MAAM,CAAC,GAC7B,cAAc,EAAE,CAQlB;AAED,wBAAsB,kBAAkB,CAAC,OAAO,EAAE,eAAe,GAAG,OAAO,CAAC,cAAc,CAAC,CAuK1F;AAED,wBAAsB,SAAS,CAAC,OAAO,EAAE,gBAAgB,GAAG,OAAO,CAAC,eAAe,CAAC,CAyBnF","sourcesContent":["/**\n * Hybrid search orchestrator: resolve mode, run retrievers in parallel, fuse\n * by rank, expand within budget, trace everything\n * (docs/hybrid-retrieval-design.md).\n *\n * Single-retriever modes flow through the same pipeline — rrfFuse over one\n * list preserves its order — so lexical, semantic, and hybrid all produce the\n * same result shape and the same trace record.\n *\n * `retrieveCandidates` is the candidate-level core (also used by the eval\n * harness, which needs forced modes, a configurable `k`, and no trace\n * pollution); `runSearch` wraps it with span expansion and tracing for the\n * tool.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { adaptGrepHits, type ChunkLookup } from \"./adapter.js\";\nimport { assembleContext } from \"./context-assembler.js\";\nimport { crossEncoderRerank } from \"./cross-rerank.js\";\nimport { runLexicalRetriever } from \"./lexical-retriever.js\";\nimport { resolveSearchMode } from \"./mode.js\";\nimport { rerankCandidates } from \"./rerank.js\";\nimport { DEFAULT_RRF_K, rrfFuse } from \"./rrf.js\";\nimport { writeSearchTrace } from \"./trace.js\";\nimport type { CandidateSpan, FusedCandidate, RankedHit, ResolvedSearchMode, SearchMode, SearchTrace } from \"./types.js\";\n\n/** Raw grep line-hits fetched per query (pre-collapse). */\nconst LEXICAL_MATCH_LIMIT = 200;\n/** Adapted lexical candidates entering fusion. The eval gate showed the\n *  uncapped lexical tail diluting hybrid below plain semantic: lexical\n *  precision is front-loaded by the adapter's term-evidence ranking, while\n *  RRF weighs a rank-30 lexical candidate like a rank-30 embedding hit. */\nconst LEXICAL_FUSION_CAP = 20;\n/**\n * Embedding hits fetched per query when fusing with the grep leg.\n *\n * Deliberately shallow. The grep list is capped at {@link LEXICAL_FUSION_CAP}\n * because its tail is unranked noise, so a deep dense pool here simply\n * outnumbers it: raising this to 200 alongside a 20-candidate grep list cost\n * `auto +rr` 0.472 -> 0.450 MRR.\n */\nconst EMBED_TOP_K = 50;\n/**\n * Per-leg depth when both retrievers are ranked ones (dense + BM25).\n *\n * Four times {@link FUSED_WINDOW}, matching what the daemon's `query_hybrid`\n * does internally (`pool = 4·k`). Fusing at the same depth as the window loses\n * any candidate ranked well by one retriever but just outside the other's\n * top-50 — worth 12pp R@10 and 8pp R@50 for `bm25+dense`, which is exactly\n * what closed the gap to daemon-side fusion.\n */\nconst FUSION_POOL_TOP_K = 200;\n/** BM25 depth, matching {@link FUSION_POOL_TOP_K} so neither ranked leg is\n *  structurally advantaged by pool size. The daemon returns only documents\n *  sharing a query term, so this is an upper bound, not a fill. */\nconst BM25_TOP_K = FUSION_POOL_TOP_K;\n/** Candidates from unindexed/stale files moved ahead of the fused window.\n *  Small on purpose: this answers \"what did I just write\", not \"search the\n *  working tree\". */\nconst STALE_HOIST_CAP = 5;\n/** Fused candidates kept for reranking / final slicing. */\nconst FUSED_WINDOW = 50;\n\nexport interface RetrieveOptions {\n\tcwd: string;\n\tquery: string;\n\tmode?: SearchMode;\n\t/** Optional glob filter applied to file paths. */\n\tglob?: string;\n\t/** Maximum fused candidates returned. */\n\tlimit?: number;\n\t/** RRF constant override (eval harness sweeps this). Default: {@link DEFAULT_RRF_K}. */\n\trrfK?: number;\n\t/** Rerank the fused top-50 before slicing to `limit`. Default: true. */\n\trerank?: boolean;\n\t/**\n\t * Ask the daemon to fuse its own BM25 index with the vectors and return one\n\t * already-fused ranking, instead of taking a dense-only list. Needs a store\n\t * built with `--hybrid`. The fused list arrives as a single \"embed\" leg,\n\t * because a pre-fused ranking has no per-retriever structure left to record.\n\t *\n\t * Prefer {@link bm25Leg}: fusing here keeps the legs separable in the trace\n\t * and lets the grep leg participate.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as its own ranked list and fuse it here,\n\t * alongside dense and grep.\n\t *\n\t * Defaults to on wherever the daemon can serve it, because BM25 is the\n\t * better lexical leg on the indexed corpus: Recall@50 0.790 -> 0.879, 6 of\n\t * 62 queries better and 0 worse (p <= 0.05). When it is on, the grep leg\n\t * narrows to files the index has not read — see `staleFiles`.\n\t *\n\t * Set `false` to force ripgrep as the only lexical leg; the eval harness\n\t * does this to keep measuring what the old rows measured.\n\t */\n\tbm25Leg?: boolean;\n\t/**\n\t * Reorder the fused shortlist with the daemon's cross-encoder instead of\n\t * the deterministic reranker. Needs embsearch >= 0.3.0; costs one model\n\t * pass per scored candidate.\n\t */\n\tcrossEncoder?: boolean;\n\tservice?: EmbsearchService;\n\tsignal?: AbortSignal;\n}\n\nexport interface RetrieveResult {\n\tcandidates: FusedCandidate[];\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tindexPhase: SearchTrace[\"indexPhase\"];\n\tretrievers: SearchTrace[\"retrievers\"];\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n\trrfK: number;\n\trerank?: SearchTrace[\"rerank\"];\n}\n\nexport interface RunSearchOptions extends RetrieveOptions {\n\t/** Approximate token budget for the result text. */\n\ttokenBudget?: number;\n}\n\nexport interface RunSearchResult {\n\ttext: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tresultCount: number;\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n}\n\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\n/**\n * Move candidates from files the index has not read to the front.\n *\n * They are there because grep found them and nothing else could: the index is\n * ranking a stale copy of the file, or has never seen it. Left to fuse, they\n * lose — RRF rewards agreement, and one leg reporting a single document is\n * outvoted by two legs agreeing on hundreds. Measured: scoping grep to stale\n * files without this hoist scored 25% on the live-edit set where unscoped grep\n * scored 100%, because the fused window filled with consensus hits about the\n * indexed copy.\n *\n * Capped, because \"stale\" scales with how far behind the index is. A few\n * edited files is the case this exists for; a fresh checkout makes everything\n * stale, and hoisting all of it would quietly turn hybrid search back into\n * grep. Past the cap the remainder keeps its fused position.\n */\nexport function hoistStaleCandidates(\n\tcandidates: readonly FusedCandidate[],\n\tstaleFiles: ReadonlySet<string>,\n): FusedCandidate[] {\n\tif (staleFiles.size === 0) return [...candidates];\n\tconst hoisted: FusedCandidate[] = [];\n\tconst rest: FusedCandidate[] = [];\n\tfor (const candidate of candidates) {\n\t\t(staleFiles.has(candidate.path) && hoisted.length < STALE_HOIST_CAP ? hoisted : rest).push(candidate);\n\t}\n\treturn [...hoisted, ...rest];\n}\n\nexport async function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult> {\n\tconst { cwd, query, service, signal } = options;\n\tconst glob = normalizeSearchGlob(options.glob);\n\tconst requestedMode = options.mode ?? \"auto\";\n\tconst limit = Math.max(1, options.limit ?? 10);\n\tconst rrfK = options.rrfK ?? DEFAULT_RRF_K;\n\n\tconst state = service?.getState();\n\tconst embedAvailable = service?.isAvailable() ?? false;\n\tconst embedUnavailableReason =\n\t\tstate === undefined\n\t\t\t? \"semantic index is not enabled\"\n\t\t\t: state.phase === \"unavailable\" || state.phase === \"skipped\"\n\t\t\t\t? state.reason\n\t\t\t\t: state.phase === \"idle\"\n\t\t\t\t\t? \"semantic index has not started\"\n\t\t\t\t\t: undefined;\n\n\tconst resolution = resolveSearchMode(query, requestedMode, embedAvailable, embedUnavailableReason);\n\tconst mode = resolution.mode;\n\n\t// Map lexical hits onto indexed chunk ids whenever the sidecar is usable,\n\t// even in lexical-only mode, so identities line up across modes.\n\tconst lookupChunk: ChunkLookup | undefined = embedAvailable\n\t\t? (rel, line) => service!.findEnclosingChunk(rel, line)\n\t\t: undefined;\n\n\tconst spans = new Map<string, CandidateSpan>();\n\t/** Files the index does not have current content for, when grep is scoped\n\t *  to them. Read again after fusion — see the hoist below. */\n\tconst staleSet = new Set<string>();\n\tconst lists: RankedHit[][] = [];\n\tconst retrieverStats: SearchTrace[\"retrievers\"] = {};\n\tconst errors: Error[] = [];\n\n\tconst runLexical = async (scopeToStale = false): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Scoped run: grep covers only what the index has not read yet, so it\n\t\t\t// adds the one thing BM25 cannot see without re-voting on documents\n\t\t\t// BM25 already ranked. Fusing two lexical views of the same corpus is\n\t\t\t// what made the three-leg configuration lose (18 of 62 queries worse).\n\t\t\tconst paths = scopeToStale ? service?.staleFiles(signal) : undefined;\n\t\t\tif (paths) for (const rel of paths) staleSet.add(rel);\n\t\t\tconst lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal, paths });\n\t\t\tconst adapted = adaptGrepHits(lineHits, lookupChunk);\n\t\t\t// In single-retriever lexical mode the full list is the result; in\n\t\t\t// hybrid, only the front-loaded head is trustworthy enough to vote.\n\t\t\tconst hits = mode === \"hybrid\" ? adapted.hits.slice(0, LEXICAL_FUSION_CAP) : adapted.hits;\n\t\t\tfor (const [id, span] of adapted.spans) if (!spans.has(id)) spans.set(id, span);\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runEmbed = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// The flat index pads top-k with whatever exists; with the cosine\n\t\t\t// metric the store uses, score <= 0 means \"no relation at all\", so\n\t\t\t// those padding hits would cast RRF votes on pure noise.\n\t\t\t// A BM25 leg is itself ranked, so the pair can afford — and needs — the\n\t\t\t// deeper pool; the grep leg cannot (see EMBED_TOP_K).\n\t\t\tconst topK = options.bm25Leg ? FUSION_POOL_TOP_K : EMBED_TOP_K;\n\t\t\tconst chunkHits = (\n\t\t\t\tawait service!.searchChunks(query, topK, glob, options.daemonHybrid ? \"hybrid\" : \"dense\")\n\t\t\t).filter((hit) => hit.score > 0);\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"embed\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runBm25 = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Raw BM25 sums; only the ordering enters fusion, the score is a\n\t\t\t// diagnostic. Unlike the dense leg there is no zero-score padding to\n\t\t\t// filter — the daemon omits documents sharing no query term.\n\t\t\tconst chunkHits = await service!.searchChunks(query, BM25_TOP_K, glob, \"lexical\");\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"bm25\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tif (!spans.has(hit.id)) {\n\t\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t\t}\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\t// BM25 is the better lexical leg where the index is current: it beats grep\n\t// on Recall@50 (0.790 -> 0.879, 6 queries better and 0 worse, p <= 0.05)\n\t// because it ranks the whole corpus rather than truncating a match stream.\n\t// It is also blind to anything indexed later than it was written, which is\n\t// precisely where grep still wins — so grep runs scoped to that set instead\n\t// of being dropped or left to duplicate BM25 over the whole tree.\n\tconst bm25AsLexicalLeg =\n\t\t(options.bm25Leg ?? service?.supportsLexicalRetriever() ?? false) && embedAvailable && mode !== \"lexical\";\n\n\tconst runs: Promise<void>[] = [];\n\tif (mode === \"lexical\" || mode === \"hybrid\") runs.push(runLexical(bm25AsLexicalLeg));\n\tif (mode === \"semantic\" || mode === \"hybrid\") runs.push(runEmbed());\n\t// `semantic` gets the BM25 leg too: the caller asked for the index, and\n\t// BM25 is part of it. Only an explicit `lexical` request excludes it.\n\tif (bm25AsLexicalLeg) runs.push(runBm25());\n\tawait Promise.all(runs);\n\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t// A partial failure in hybrid degrades to whichever retriever survived;\n\t// only a total loss is an error.\n\tif (lists.length === 0) throw errors[0] ?? new Error(\"search produced no retriever results\");\n\n\tconst fused = rrfFuse(lists, rrfK).slice(0, FUSED_WINDOW);\n\tlet candidates: FusedCandidate[] = [];\n\tfor (const hit of fused) {\n\t\tconst span = spans.get(hit.id);\n\t\tif (span) candidates.push({ ...hit, ...span });\n\t}\n\n\tlet rerankInfo: SearchTrace[\"rerank\"];\n\tif (options.crossEncoder && service?.supportsCrossEncoder()) {\n\t\tconst reranked = await crossEncoderRerank(query, candidates, cwd, service);\n\t\trerankInfo = { applied: true, candidateCount: reranked.scored, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t} else if (options.rerank !== false) {\n\t\tconst reranked = rerankCandidates(query, candidates, cwd);\n\t\trerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t}\n\t// After reranking, not before: both rerankers weight the fused prior, and a\n\t// lone grep hit from an unindexed file has the lowest prior there is, so\n\t// hoisting first would simply be undone. The reranker still orders the\n\t// hoisted set against itself.\n\tcandidates = hoistStaleCandidates(candidates, staleSet).slice(0, limit);\n\n\treturn {\n\t\tcandidates,\n\t\tresolvedMode: mode,\n\t\tdegradedReason: resolution.degradedReason,\n\t\tindexPhase: state?.phase === \"ready\" ? \"ready\" : state?.phase === \"indexing\" ? \"indexing\" : \"unavailable\",\n\t\tretrievers: retrieverStats,\n\t\tindexing: state?.phase === \"indexing\" ? { done: state.done, total: state.total } : undefined,\n\t\trrfK,\n\t\trerank: rerankInfo,\n\t};\n}\n\nexport async function runSearch(options: RunSearchOptions): Promise<RunSearchResult> {\n\tconst retrieved = await retrieveCandidates(options);\n\n\tconst assembled = assembleContext(retrieved.candidates, { cwd: options.cwd, tokenBudget: options.tokenBudget });\n\n\twriteSearchTrace(options.cwd, {\n\t\ttimestampMs: Date.now(),\n\t\tquery: options.query,\n\t\trequestedMode: options.mode ?? \"auto\",\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tindexPhase: retrieved.indexPhase,\n\t\trrfK: retrieved.resolvedMode === \"hybrid\" ? retrieved.rrfK : undefined,\n\t\tretrievers: retrieved.retrievers,\n\t\tfused: retrieved.candidates.map(({ id, rrfScore, ranks, rawScores }) => ({ id, rrfScore, ranks, rawScores })),\n\t\trerank: retrieved.rerank,\n\t});\n\n\treturn {\n\t\ttext: assembled.text,\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tresultCount: retrieved.candidates.length,\n\t\tindexing: retrieved.indexing,\n\t};\n}\n"]}

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

@@ -207,7 +207,7 @@ export async function retrieveCandidates(options) {
     // It is also blind to anything indexed later than it was written, which is
     // precisely where grep still wins — so grep runs scoped to that set instead
     // of being dropped or left to duplicate BM25 over the whole tree.
-    const bm25AsLexicalLeg = options.bm25Leg && embedAvailable && mode !== "lexical";
+    const bm25AsLexicalLeg = (options.bm25Leg ?? service?.supportsLexicalRetriever() ?? false) && embedAvailable && mode !== "lexical";
     const runs = [];
     if (mode === "lexical" || mode === "hybrid")
         runs.push(runLexical(bm25AsLexicalLeg));

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

@@ -1 +1 @@
-{"version":3,"file":"hybrid-search.js","sourceRoot":"","sources":["../../../src/core/search/hybrid-search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAGH,OAAO,EAAE,aAAa,EAAoB,MAAM,cAAc,CAAC;AAC/D,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AACvD,OAAO,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAC;AAC7D,OAAO,EAAE,iBAAiB,EAAE,MAAM,WAAW,CAAC;AAC9C,OAAO,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAC/C,OAAO,EAAE,aAAa,EAAE,OAAO,EAAE,MAAM,UAAU,CAAC;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAG9C,2DAA2D;AAC3D,MAAM,mBAAmB,GAAG,GAAG,CAAC;AAChC;;;2EAG2E;AAC3E,MAAM,kBAAkB,GAAG,EAAE,CAAC;AAC9B;;;;;;;GAOG;AACH,MAAM,WAAW,GAAG,EAAE,CAAC;AACvB;;;;;;;;GAQG;AACH,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAC9B;;mEAEmE;AACnE,MAAM,UAAU,GAAG,iBAAiB,CAAC;AACrC;;qBAEqB;AACrB,MAAM,eAAe,GAAG,CAAC,CAAC;AAC1B,2DAA2D;AAC3D,MAAM,YAAY,GAAG,EAAE,CAAC;AAkExB,SAAS,mBAAmB,CAAC,IAAwB,EAAsB;IAC1E,IAAI,CAAC,IAAI;QAAE,OAAO,SAAS,CAAC;IAC5B,yEAAyE;IACzE,6EAA6E;IAC7E,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;QAC5E,OAAO,MAAM,IAAI,EAAE,CAAC;IACrB,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,oBAAoB,CACnC,UAAqC,EACrC,UAA+B,EACZ;IACnB,IAAI,UAAU,CAAC,IAAI,KAAK,CAAC;QAAE,OAAO,CAAC,GAAG,UAAU,CAAC,CAAC;IAClD,MAAM,OAAO,GAAqB,EAAE,CAAC;IACrC,MAAM,IAAI,GAAqB,EAAE,CAAC;IAClC,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACpC,CAAC,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,OAAO,CAAC,MAAM,GAAG,eAAe,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IACvG,CAAC;IACD,OAAO,CAAC,GAAG,OAAO,EAAE,GAAG,IAAI,CAAC,CAAC;AAAA,CAC7B;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,OAAwB,EAA2B;IAC3F,MAAM,EAAE,GAAG,EAAE,KAAK,EAAE,OAAO,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IAChD,MAAM,IAAI,GAAG,mBAAmB,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IAC/C,MAAM,aAAa,GAAG,OAAO,CAAC,IAAI,IAAI,MAAM,CAAC;IAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;IAC/C,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,IAAI,aAAa,CAAC;IAE3C,MAAM,KAAK,GAAG,OAAO,EAAE,QAAQ,EAAE,CAAC;IAClC,MAAM,cAAc,GAAG,OAAO,EAAE,WAAW,EAAE,IAAI,KAAK,CAAC;IACvD,MAAM,sBAAsB,GAC3B,KAAK,KAAK,SAAS;QAClB,CAAC,CAAC,+BAA+B;QACjC,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,aAAa,IAAI,KAAK,CAAC,KAAK,KAAK,SAAS;YAC3D,CAAC,CAAC,KAAK,CAAC,MAAM;YACd,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,MAAM;gBACvB,CAAC,CAAC,gCAAgC;gBAClC,CAAC,CAAC,SAAS,CAAC;IAEhB,MAAM,UAAU,GAAG,iBAAiB,CAAC,KAAK,EAAE,aAAa,EAAE,cAAc,EAAE,sBAAsB,CAAC,CAAC;IACnG,MAAM,IAAI,GAAG,UAAU,CAAC,IAAI,CAAC;IAE7B,0EAA0E;IAC1E,iEAAiE;IACjE,MAAM,WAAW,GAA4B,cAAc;QAC1D,CAAC,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,CAAC,OAAQ,CAAC,kBAAkB,CAAC,GAAG,EAAE,IAAI,CAAC;QACvD,CAAC,CAAC,SAAS,CAAC;IAEb,MAAM,KAAK,GAAG,IAAI,GAAG,EAAyB,CAAC;IAC/C;oEAC8D;IAC9D,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAC;IACnC,MAAM,KAAK,GAAkB,EAAE,CAAC;IAChC,MAAM,cAAc,GAA8B,EAAE,CAAC;IACrD,MAAM,MAAM,GAAY,EAAE,CAAC;IAE3B,MAAM,UAAU,GAAG,KAAK,EAAE,YAAY,GAAG,KAAK,EAAiB,EAAE,CAAC;QACjE,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,sEAAsE;YACtE,oEAAoE;YACpE,sEAAsE;YACtE,uEAAuE;YACvE,MAAM,KAAK,GAAG,YAAY,CAAC,CAAC,CAAC,OAAO,EAAE,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;YACrE,IAAI,KAAK;gBAAE,KAAK,MAAM,GAAG,IAAI,KAAK;oBAAE,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACtD,MAAM,QAAQ,GAAG,MAAM,mBAAmB,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,KAAK,EAAE,mBAAmB,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;YAC5G,MAAM,OAAO,GAAG,aAAa,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;YACrD,mEAAmE;YACnE,oEAAoE;YACpE,MAAM,IAAI,GAAG,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC;YAC1F,KAAK,MAAM,CAAC,EAAE,EAAE,IAAI,CAAC,IAAI,OAAO,CAAC,KAAK;gBAAE,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;oBAAE,KAAK,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;YAChF,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC1E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,QAAQ,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC3C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,kEAAkE;YAClE,mEAAmE;YACnE,yDAAyD;YACzD,4EAAwE;YACxE,sDAAsD;YACtD,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,WAAW,CAAC;YAC/D,MAAM,SAAS,GAAG,CACjB,MAAM,OAAQ,CAAC,YAAY,CAAC,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,CACzF,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;YACjC,MAAM,IAAI,GAAgB,SAAS,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;gBACpD,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC;gBACX,KAAK,EAAE,GAAG,CAAC,KAAK;gBAChB,MAAM,EAAE,OAAO;aACf,CAAC,CAAC,CAAC;YACJ,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;gBAC7B,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;YACvF,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,KAAK,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACrF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,KAAK,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC3E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,OAAO,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC1C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,iEAAiE;YACjE,qEAAqE;YACrE,+DAA6D;YAC7D,MAAM,SAAS,GAAG,MAAM,OAAQ,CAAC,YAAY,CAAC,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,CAAC,CAAC;YAClF,MAAM,IAAI,GAAgB,SAAS,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;gBACpD,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC;gBACX,KAAK,EAAE,GAAG,CAAC,KAAK;gBAChB,MAAM,EAAE,MAAM;aACd,CAAC,CAAC,CAAC;YACJ,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;gBAC7B,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,CAAC;oBACxB,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;gBACvF,CAAC;YACF,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC1E,CAAC;IAAA,CACD,CAAC;IAEF,2EAA2E;IAC3E,yEAAyE;IACzE,2EAA2E;IAC3E,2EAA2E;IAC3E,8EAA4E;IAC5E,kEAAkE;IAClE,MAAM,gBAAgB,GAAG,OAAO,CAAC,OAAO,IAAI,cAAc,IAAI,IAAI,KAAK,SAAS,CAAC;IAEjF,MAAM,IAAI,GAAoB,EAAE,CAAC;IACjC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,QAAQ;QAAE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,gBAAgB,CAAC,CAAC,CAAC;IACrF,IAAI,IAAI,KAAK,UAAU,IAAI,IAAI,KAAK,QAAQ;QAAE,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC;IACpE,wEAAwE;IACxE,sEAAsE;IACtE,IAAI,gBAAgB;QAAE,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC;IAC3C,MAAM,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACxB,IAAI,MAAM,EAAE,OAAO;QAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;IAC1D,wEAAwE;IACxE,iCAAiC;IACjC,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,MAAM,MAAM,CAAC,CAAC,CAAC,IAAI,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC;IAE7F,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,YAAY,CAAC,CAAC;IAC1D,IAAI,UAAU,GAAqB,EAAE,CAAC;IACtC,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAC/B,IAAI,IAAI;YAAE,UAAU,CAAC,IAAI,CAAC,EAAE,GAAG,GAAG,EAAE,GAAG,IAAI,EAAE,CAAC,CAAC;IAChD,CAAC;IAED,IAAI,UAAiC,CAAC;IACtC,IAAI,OAAO,CAAC,YAAY,IAAI,OAAO,EAAE,oBAAoB,EAAE,EAAE,CAAC;QAC7D,MAAM,QAAQ,GAAG,MAAM,kBAAkB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,EAAE,OAAO,CAAC,CAAC;QAC3E,UAAU,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,QAAQ,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,SAAS,EAAE,CAAC;QAC/F,UAAU,GAAG,QAAQ,CAAC,UAAU,CAAC;IAClC,CAAC;SAAM,IAAI,OAAO,CAAC,MAAM,KAAK,KAAK,EAAE,CAAC;QACrC,MAAM,QAAQ,GAAG,gBAAgB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,CAAC,CAAC;QAC1D,UAAU,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,UAAU,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,SAAS,EAAE,CAAC;QACjG,UAAU,GAAG,QAAQ,CAAC,UAAU,CAAC;IAClC,CAAC;IACD,4EAA4E;IAC5E,yEAAyE;IACzE,uEAAuE;IACvE,8BAA8B;IAC9B,UAAU,GAAG,oBAAoB,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAExE,OAAO;QACN,UAAU;QACV,YAAY,EAAE,IAAI;QAClB,cAAc,EAAE,UAAU,CAAC,cAAc;QACzC,UAAU,EAAE,KAAK,EAAE,KAAK,KAAK,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,EAAE,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,aAAa;QACzG,UAAU,EAAE,cAAc;QAC1B,QAAQ,EAAE,KAAK,EAAE,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS;QAC5F,IAAI;QACJ,MAAM,EAAE,UAAU;KAClB,CAAC;AAAA,CACF;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,OAAyB,EAA4B;IACpF,MAAM,SAAS,GAAG,MAAM,kBAAkB,CAAC,OAAO,CAAC,CAAC;IAEpD,MAAM,SAAS,GAAG,eAAe,CAAC,SAAS,CAAC,UAAU,EAAE,EAAE,GAAG,EAAE,OAAO,CAAC,GAAG,EAAE,WAAW,EAAE,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC;IAEhH,gBAAgB,CAAC,OAAO,CAAC,GAAG,EAAE;QAC7B,WAAW,EAAE,IAAI,CAAC,GAAG,EAAE;QACvB,KAAK,EAAE,OAAO,CAAC,KAAK;QACpB,aAAa,EAAE,OAAO,CAAC,IAAI,IAAI,MAAM;QACrC,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;QACxC,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,IAAI,EAAE,SAAS,CAAC,YAAY,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS;QACtE,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,KAAK,EAAE,SAAS,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC,CAAC;QAC7G,MAAM,EAAE,SAAS,CAAC,MAAM;KACxB,CAAC,CAAC;IAEH,OAAO;QACN,IAAI,EAAE,SAAS,CAAC,IAAI;QACpB,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;QACxC,WAAW,EAAE,SAAS,CAAC,UAAU,CAAC,MAAM;QACxC,QAAQ,EAAE,SAAS,CAAC,QAAQ;KAC5B,CAAC;AAAA,CACF","sourcesContent":["/**\n * Hybrid search orchestrator: resolve mode, run retrievers in parallel, fuse\n * by rank, expand within budget, trace everything\n * (docs/hybrid-retrieval-design.md).\n *\n * Single-retriever modes flow through the same pipeline — rrfFuse over one\n * list preserves its order — so lexical, semantic, and hybrid all produce the\n * same result shape and the same trace record.\n *\n * `retrieveCandidates` is the candidate-level core (also used by the eval\n * harness, which needs forced modes, a configurable `k`, and no trace\n * pollution); `runSearch` wraps it with span expansion and tracing for the\n * tool.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { adaptGrepHits, type ChunkLookup } from \"./adapter.js\";\nimport { assembleContext } from \"./context-assembler.js\";\nimport { crossEncoderRerank } from \"./cross-rerank.js\";\nimport { runLexicalRetriever } from \"./lexical-retriever.js\";\nimport { resolveSearchMode } from \"./mode.js\";\nimport { rerankCandidates } from \"./rerank.js\";\nimport { DEFAULT_RRF_K, rrfFuse } from \"./rrf.js\";\nimport { writeSearchTrace } from \"./trace.js\";\nimport type { CandidateSpan, FusedCandidate, RankedHit, ResolvedSearchMode, SearchMode, SearchTrace } from \"./types.js\";\n\n/** Raw grep line-hits fetched per query (pre-collapse). */\nconst LEXICAL_MATCH_LIMIT = 200;\n/** Adapted lexical candidates entering fusion. The eval gate showed the\n *  uncapped lexical tail diluting hybrid below plain semantic: lexical\n *  precision is front-loaded by the adapter's term-evidence ranking, while\n *  RRF weighs a rank-30 lexical candidate like a rank-30 embedding hit. */\nconst LEXICAL_FUSION_CAP = 20;\n/**\n * Embedding hits fetched per query when fusing with the grep leg.\n *\n * Deliberately shallow. The grep list is capped at {@link LEXICAL_FUSION_CAP}\n * because its tail is unranked noise, so a deep dense pool here simply\n * outnumbers it: raising this to 200 alongside a 20-candidate grep list cost\n * `auto +rr` 0.472 -> 0.450 MRR.\n */\nconst EMBED_TOP_K = 50;\n/**\n * Per-leg depth when both retrievers are ranked ones (dense + BM25).\n *\n * Four times {@link FUSED_WINDOW}, matching what the daemon's `query_hybrid`\n * does internally (`pool = 4·k`). Fusing at the same depth as the window loses\n * any candidate ranked well by one retriever but just outside the other's\n * top-50 — worth 12pp R@10 and 8pp R@50 for `bm25+dense`, which is exactly\n * what closed the gap to daemon-side fusion.\n */\nconst FUSION_POOL_TOP_K = 200;\n/** BM25 depth, matching {@link FUSION_POOL_TOP_K} so neither ranked leg is\n *  structurally advantaged by pool size. The daemon returns only documents\n *  sharing a query term, so this is an upper bound, not a fill. */\nconst BM25_TOP_K = FUSION_POOL_TOP_K;\n/** Candidates from unindexed/stale files moved ahead of the fused window.\n *  Small on purpose: this answers \"what did I just write\", not \"search the\n *  working tree\". */\nconst STALE_HOIST_CAP = 5;\n/** Fused candidates kept for reranking / final slicing. */\nconst FUSED_WINDOW = 50;\n\nexport interface RetrieveOptions {\n\tcwd: string;\n\tquery: string;\n\tmode?: SearchMode;\n\t/** Optional glob filter applied to file paths. */\n\tglob?: string;\n\t/** Maximum fused candidates returned. */\n\tlimit?: number;\n\t/** RRF constant override (eval harness sweeps this). Default: {@link DEFAULT_RRF_K}. */\n\trrfK?: number;\n\t/** Rerank the fused top-50 before slicing to `limit`. Default: true. */\n\trerank?: boolean;\n\t/**\n\t * Ask the daemon to fuse its own BM25 index with the vectors and return one\n\t * already-fused ranking, instead of taking a dense-only list. Needs a store\n\t * built with `--hybrid`. The fused list arrives as a single \"embed\" leg,\n\t * because a pre-fused ranking has no per-retriever structure left to record.\n\t *\n\t * Prefer {@link bm25Leg}: fusing here keeps the legs separable in the trace\n\t * and lets the grep leg participate.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as its own ranked list and fuse it here,\n\t * alongside dense and grep. Needs a store built with `--hybrid` and an\n\t * embsearch new enough to serve `retriever: \"lexical\"`.\n\t */\n\tbm25Leg?: boolean;\n\t/**\n\t * Reorder the fused shortlist with the daemon's cross-encoder instead of\n\t * the deterministic reranker. Needs embsearch >= 0.3.0; costs one model\n\t * pass per scored candidate.\n\t */\n\tcrossEncoder?: boolean;\n\tservice?: EmbsearchService;\n\tsignal?: AbortSignal;\n}\n\nexport interface RetrieveResult {\n\tcandidates: FusedCandidate[];\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tindexPhase: SearchTrace[\"indexPhase\"];\n\tretrievers: SearchTrace[\"retrievers\"];\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n\trrfK: number;\n\trerank?: SearchTrace[\"rerank\"];\n}\n\nexport interface RunSearchOptions extends RetrieveOptions {\n\t/** Approximate token budget for the result text. */\n\ttokenBudget?: number;\n}\n\nexport interface RunSearchResult {\n\ttext: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tresultCount: number;\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n}\n\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\n/**\n * Move candidates from files the index has not read to the front.\n *\n * They are there because grep found them and nothing else could: the index is\n * ranking a stale copy of the file, or has never seen it. Left to fuse, they\n * lose — RRF rewards agreement, and one leg reporting a single document is\n * outvoted by two legs agreeing on hundreds. Measured: scoping grep to stale\n * files without this hoist scored 25% on the live-edit set where unscoped grep\n * scored 100%, because the fused window filled with consensus hits about the\n * indexed copy.\n *\n * Capped, because \"stale\" scales with how far behind the index is. A few\n * edited files is the case this exists for; a fresh checkout makes everything\n * stale, and hoisting all of it would quietly turn hybrid search back into\n * grep. Past the cap the remainder keeps its fused position.\n */\nexport function hoistStaleCandidates(\n\tcandidates: readonly FusedCandidate[],\n\tstaleFiles: ReadonlySet<string>,\n): FusedCandidate[] {\n\tif (staleFiles.size === 0) return [...candidates];\n\tconst hoisted: FusedCandidate[] = [];\n\tconst rest: FusedCandidate[] = [];\n\tfor (const candidate of candidates) {\n\t\t(staleFiles.has(candidate.path) && hoisted.length < STALE_HOIST_CAP ? hoisted : rest).push(candidate);\n\t}\n\treturn [...hoisted, ...rest];\n}\n\nexport async function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult> {\n\tconst { cwd, query, service, signal } = options;\n\tconst glob = normalizeSearchGlob(options.glob);\n\tconst requestedMode = options.mode ?? \"auto\";\n\tconst limit = Math.max(1, options.limit ?? 10);\n\tconst rrfK = options.rrfK ?? DEFAULT_RRF_K;\n\n\tconst state = service?.getState();\n\tconst embedAvailable = service?.isAvailable() ?? false;\n\tconst embedUnavailableReason =\n\t\tstate === undefined\n\t\t\t? \"semantic index is not enabled\"\n\t\t\t: state.phase === \"unavailable\" || state.phase === \"skipped\"\n\t\t\t\t? state.reason\n\t\t\t\t: state.phase === \"idle\"\n\t\t\t\t\t? \"semantic index has not started\"\n\t\t\t\t\t: undefined;\n\n\tconst resolution = resolveSearchMode(query, requestedMode, embedAvailable, embedUnavailableReason);\n\tconst mode = resolution.mode;\n\n\t// Map lexical hits onto indexed chunk ids whenever the sidecar is usable,\n\t// even in lexical-only mode, so identities line up across modes.\n\tconst lookupChunk: ChunkLookup | undefined = embedAvailable\n\t\t? (rel, line) => service!.findEnclosingChunk(rel, line)\n\t\t: undefined;\n\n\tconst spans = new Map<string, CandidateSpan>();\n\t/** Files the index does not have current content for, when grep is scoped\n\t *  to them. Read again after fusion — see the hoist below. */\n\tconst staleSet = new Set<string>();\n\tconst lists: RankedHit[][] = [];\n\tconst retrieverStats: SearchTrace[\"retrievers\"] = {};\n\tconst errors: Error[] = [];\n\n\tconst runLexical = async (scopeToStale = false): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Scoped run: grep covers only what the index has not read yet, so it\n\t\t\t// adds the one thing BM25 cannot see without re-voting on documents\n\t\t\t// BM25 already ranked. Fusing two lexical views of the same corpus is\n\t\t\t// what made the three-leg configuration lose (18 of 62 queries worse).\n\t\t\tconst paths = scopeToStale ? service?.staleFiles(signal) : undefined;\n\t\t\tif (paths) for (const rel of paths) staleSet.add(rel);\n\t\t\tconst lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal, paths });\n\t\t\tconst adapted = adaptGrepHits(lineHits, lookupChunk);\n\t\t\t// In single-retriever lexical mode the full list is the result; in\n\t\t\t// hybrid, only the front-loaded head is trustworthy enough to vote.\n\t\t\tconst hits = mode === \"hybrid\" ? adapted.hits.slice(0, LEXICAL_FUSION_CAP) : adapted.hits;\n\t\t\tfor (const [id, span] of adapted.spans) if (!spans.has(id)) spans.set(id, span);\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runEmbed = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// The flat index pads top-k with whatever exists; with the cosine\n\t\t\t// metric the store uses, score <= 0 means \"no relation at all\", so\n\t\t\t// those padding hits would cast RRF votes on pure noise.\n\t\t\t// A BM25 leg is itself ranked, so the pair can afford — and needs — the\n\t\t\t// deeper pool; the grep leg cannot (see EMBED_TOP_K).\n\t\t\tconst topK = options.bm25Leg ? FUSION_POOL_TOP_K : EMBED_TOP_K;\n\t\t\tconst chunkHits = (\n\t\t\t\tawait service!.searchChunks(query, topK, glob, options.daemonHybrid ? \"hybrid\" : \"dense\")\n\t\t\t).filter((hit) => hit.score > 0);\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"embed\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runBm25 = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Raw BM25 sums; only the ordering enters fusion, the score is a\n\t\t\t// diagnostic. Unlike the dense leg there is no zero-score padding to\n\t\t\t// filter — the daemon omits documents sharing no query term.\n\t\t\tconst chunkHits = await service!.searchChunks(query, BM25_TOP_K, glob, \"lexical\");\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"bm25\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tif (!spans.has(hit.id)) {\n\t\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t\t}\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\t// BM25 is the better lexical leg where the index is current: it beats grep\n\t// on Recall@50 (0.790 -> 0.879, 6 queries better and 0 worse, p <= 0.05)\n\t// because it ranks the whole corpus rather than truncating a match stream.\n\t// It is also blind to anything indexed later than it was written, which is\n\t// precisely where grep still wins — so grep runs scoped to that set instead\n\t// of being dropped or left to duplicate BM25 over the whole tree.\n\tconst bm25AsLexicalLeg = options.bm25Leg && embedAvailable && mode !== \"lexical\";\n\n\tconst runs: Promise<void>[] = [];\n\tif (mode === \"lexical\" || mode === \"hybrid\") runs.push(runLexical(bm25AsLexicalLeg));\n\tif (mode === \"semantic\" || mode === \"hybrid\") runs.push(runEmbed());\n\t// `semantic` gets the BM25 leg too: the caller asked for the index, and\n\t// BM25 is part of it. Only an explicit `lexical` request excludes it.\n\tif (bm25AsLexicalLeg) runs.push(runBm25());\n\tawait Promise.all(runs);\n\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t// A partial failure in hybrid degrades to whichever retriever survived;\n\t// only a total loss is an error.\n\tif (lists.length === 0) throw errors[0] ?? new Error(\"search produced no retriever results\");\n\n\tconst fused = rrfFuse(lists, rrfK).slice(0, FUSED_WINDOW);\n\tlet candidates: FusedCandidate[] = [];\n\tfor (const hit of fused) {\n\t\tconst span = spans.get(hit.id);\n\t\tif (span) candidates.push({ ...hit, ...span });\n\t}\n\n\tlet rerankInfo: SearchTrace[\"rerank\"];\n\tif (options.crossEncoder && service?.supportsCrossEncoder()) {\n\t\tconst reranked = await crossEncoderRerank(query, candidates, cwd, service);\n\t\trerankInfo = { applied: true, candidateCount: reranked.scored, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t} else if (options.rerank !== false) {\n\t\tconst reranked = rerankCandidates(query, candidates, cwd);\n\t\trerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t}\n\t// After reranking, not before: both rerankers weight the fused prior, and a\n\t// lone grep hit from an unindexed file has the lowest prior there is, so\n\t// hoisting first would simply be undone. The reranker still orders the\n\t// hoisted set against itself.\n\tcandidates = hoistStaleCandidates(candidates, staleSet).slice(0, limit);\n\n\treturn {\n\t\tcandidates,\n\t\tresolvedMode: mode,\n\t\tdegradedReason: resolution.degradedReason,\n\t\tindexPhase: state?.phase === \"ready\" ? \"ready\" : state?.phase === \"indexing\" ? \"indexing\" : \"unavailable\",\n\t\tretrievers: retrieverStats,\n\t\tindexing: state?.phase === \"indexing\" ? { done: state.done, total: state.total } : undefined,\n\t\trrfK,\n\t\trerank: rerankInfo,\n\t};\n}\n\nexport async function runSearch(options: RunSearchOptions): Promise<RunSearchResult> {\n\tconst retrieved = await retrieveCandidates(options);\n\n\tconst assembled = assembleContext(retrieved.candidates, { cwd: options.cwd, tokenBudget: options.tokenBudget });\n\n\twriteSearchTrace(options.cwd, {\n\t\ttimestampMs: Date.now(),\n\t\tquery: options.query,\n\t\trequestedMode: options.mode ?? \"auto\",\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tindexPhase: retrieved.indexPhase,\n\t\trrfK: retrieved.resolvedMode === \"hybrid\" ? retrieved.rrfK : undefined,\n\t\tretrievers: retrieved.retrievers,\n\t\tfused: retrieved.candidates.map(({ id, rrfScore, ranks, rawScores }) => ({ id, rrfScore, ranks, rawScores })),\n\t\trerank: retrieved.rerank,\n\t});\n\n\treturn {\n\t\ttext: assembled.text,\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tresultCount: retrieved.candidates.length,\n\t\tindexing: retrieved.indexing,\n\t};\n}\n"]}
+{"version":3,"file":"hybrid-search.js","sourceRoot":"","sources":["../../../src/core/search/hybrid-search.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAGH,OAAO,EAAE,aAAa,EAAoB,MAAM,cAAc,CAAC;AAC/D,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AACvD,OAAO,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAC;AAC7D,OAAO,EAAE,iBAAiB,EAAE,MAAM,WAAW,CAAC;AAC9C,OAAO,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAC/C,OAAO,EAAE,aAAa,EAAE,OAAO,EAAE,MAAM,UAAU,CAAC;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAG9C,2DAA2D;AAC3D,MAAM,mBAAmB,GAAG,GAAG,CAAC;AAChC;;;2EAG2E;AAC3E,MAAM,kBAAkB,GAAG,EAAE,CAAC;AAC9B;;;;;;;GAOG;AACH,MAAM,WAAW,GAAG,EAAE,CAAC;AACvB;;;;;;;;GAQG;AACH,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAC9B;;mEAEmE;AACnE,MAAM,UAAU,GAAG,iBAAiB,CAAC;AACrC;;qBAEqB;AACrB,MAAM,eAAe,GAAG,CAAC,CAAC;AAC1B,2DAA2D;AAC3D,MAAM,YAAY,GAAG,EAAE,CAAC;AAyExB,SAAS,mBAAmB,CAAC,IAAwB,EAAsB;IAC1E,IAAI,CAAC,IAAI;QAAE,OAAO,SAAS,CAAC;IAC5B,yEAAyE;IACzE,6EAA6E;IAC7E,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;QAC5E,OAAO,MAAM,IAAI,EAAE,CAAC;IACrB,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,oBAAoB,CACnC,UAAqC,EACrC,UAA+B,EACZ;IACnB,IAAI,UAAU,CAAC,IAAI,KAAK,CAAC;QAAE,OAAO,CAAC,GAAG,UAAU,CAAC,CAAC;IAClD,MAAM,OAAO,GAAqB,EAAE,CAAC;IACrC,MAAM,IAAI,GAAqB,EAAE,CAAC;IAClC,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACpC,CAAC,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,OAAO,CAAC,MAAM,GAAG,eAAe,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IACvG,CAAC;IACD,OAAO,CAAC,GAAG,OAAO,EAAE,GAAG,IAAI,CAAC,CAAC;AAAA,CAC7B;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,OAAwB,EAA2B;IAC3F,MAAM,EAAE,GAAG,EAAE,KAAK,EAAE,OAAO,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IAChD,MAAM,IAAI,GAAG,mBAAmB,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IAC/C,MAAM,aAAa,GAAG,OAAO,CAAC,IAAI,IAAI,MAAM,CAAC;IAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;IAC/C,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,IAAI,aAAa,CAAC;IAE3C,MAAM,KAAK,GAAG,OAAO,EAAE,QAAQ,EAAE,CAAC;IAClC,MAAM,cAAc,GAAG,OAAO,EAAE,WAAW,EAAE,IAAI,KAAK,CAAC;IACvD,MAAM,sBAAsB,GAC3B,KAAK,KAAK,SAAS;QAClB,CAAC,CAAC,+BAA+B;QACjC,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,aAAa,IAAI,KAAK,CAAC,KAAK,KAAK,SAAS;YAC3D,CAAC,CAAC,KAAK,CAAC,MAAM;YACd,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,MAAM;gBACvB,CAAC,CAAC,gCAAgC;gBAClC,CAAC,CAAC,SAAS,CAAC;IAEhB,MAAM,UAAU,GAAG,iBAAiB,CAAC,KAAK,EAAE,aAAa,EAAE,cAAc,EAAE,sBAAsB,CAAC,CAAC;IACnG,MAAM,IAAI,GAAG,UAAU,CAAC,IAAI,CAAC;IAE7B,0EAA0E;IAC1E,iEAAiE;IACjE,MAAM,WAAW,GAA4B,cAAc;QAC1D,CAAC,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,CAAC,OAAQ,CAAC,kBAAkB,CAAC,GAAG,EAAE,IAAI,CAAC;QACvD,CAAC,CAAC,SAAS,CAAC;IAEb,MAAM,KAAK,GAAG,IAAI,GAAG,EAAyB,CAAC;IAC/C;oEAC8D;IAC9D,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAC;IACnC,MAAM,KAAK,GAAkB,EAAE,CAAC;IAChC,MAAM,cAAc,GAA8B,EAAE,CAAC;IACrD,MAAM,MAAM,GAAY,EAAE,CAAC;IAE3B,MAAM,UAAU,GAAG,KAAK,EAAE,YAAY,GAAG,KAAK,EAAiB,EAAE,CAAC;QACjE,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,sEAAsE;YACtE,oEAAoE;YACpE,sEAAsE;YACtE,uEAAuE;YACvE,MAAM,KAAK,GAAG,YAAY,CAAC,CAAC,CAAC,OAAO,EAAE,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;YACrE,IAAI,KAAK;gBAAE,KAAK,MAAM,GAAG,IAAI,KAAK;oBAAE,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACtD,MAAM,QAAQ,GAAG,MAAM,mBAAmB,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,KAAK,EAAE,mBAAmB,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;YAC5G,MAAM,OAAO,GAAG,aAAa,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;YACrD,mEAAmE;YACnE,oEAAoE;YACpE,MAAM,IAAI,GAAG,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC;YAC1F,KAAK,MAAM,CAAC,EAAE,EAAE,IAAI,CAAC,IAAI,OAAO,CAAC,KAAK;gBAAE,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;oBAAE,KAAK,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;YAChF,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC1E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,QAAQ,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC3C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,kEAAkE;YAClE,mEAAmE;YACnE,yDAAyD;YACzD,4EAAwE;YACxE,sDAAsD;YACtD,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,WAAW,CAAC;YAC/D,MAAM,SAAS,GAAG,CACjB,MAAM,OAAQ,CAAC,YAAY,CAAC,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,CACzF,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;YACjC,MAAM,IAAI,GAAgB,SAAS,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;gBACpD,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC;gBACX,KAAK,EAAE,GAAG,CAAC,KAAK;gBAChB,MAAM,EAAE,OAAO;aACf,CAAC,CAAC,CAAC;YACJ,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;gBAC7B,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;YACvF,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,KAAK,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACrF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,KAAK,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC3E,CAAC;IAAA,CACD,CAAC;IAEF,MAAM,OAAO,GAAG,KAAK,IAAmB,EAAE,CAAC;QAC1C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,IAAI,CAAC;YACJ,iEAAiE;YACjE,qEAAqE;YACrE,+DAA6D;YAC7D,MAAM,SAAS,GAAG,MAAM,OAAQ,CAAC,YAAY,CAAC,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,CAAC,CAAC;YAClF,MAAM,IAAI,GAAgB,SAAS,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;gBACpD,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC;gBACX,KAAK,EAAE,GAAG,CAAC,KAAK;gBAChB,MAAM,EAAE,MAAM;aACd,CAAC,CAAC,CAAC;YACJ,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;gBAC7B,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,CAAC;oBACxB,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;gBACvF,CAAC;YACF,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACjB,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3D,cAAc,CAAC,IAAI,GAAG,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAC1E,CAAC;IAAA,CACD,CAAC;IAEF,2EAA2E;IAC3E,yEAAyE;IACzE,2EAA2E;IAC3E,2EAA2E;IAC3E,8EAA4E;IAC5E,kEAAkE;IAClE,MAAM,gBAAgB,GACrB,CAAC,OAAO,CAAC,OAAO,IAAI,OAAO,EAAE,wBAAwB,EAAE,IAAI,KAAK,CAAC,IAAI,cAAc,IAAI,IAAI,KAAK,SAAS,CAAC;IAE3G,MAAM,IAAI,GAAoB,EAAE,CAAC;IACjC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,QAAQ;QAAE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,gBAAgB,CAAC,CAAC,CAAC;IACrF,IAAI,IAAI,KAAK,UAAU,IAAI,IAAI,KAAK,QAAQ;QAAE,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC;IACpE,wEAAwE;IACxE,sEAAsE;IACtE,IAAI,gBAAgB;QAAE,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC;IAC3C,MAAM,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACxB,IAAI,MAAM,EAAE,OAAO;QAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;IAC1D,wEAAwE;IACxE,iCAAiC;IACjC,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,MAAM,MAAM,CAAC,CAAC,CAAC,IAAI,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC;IAE7F,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,YAAY,CAAC,CAAC;IAC1D,IAAI,UAAU,GAAqB,EAAE,CAAC;IACtC,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAC/B,IAAI,IAAI;YAAE,UAAU,CAAC,IAAI,CAAC,EAAE,GAAG,GAAG,EAAE,GAAG,IAAI,EAAE,CAAC,CAAC;IAChD,CAAC;IAED,IAAI,UAAiC,CAAC;IACtC,IAAI,OAAO,CAAC,YAAY,IAAI,OAAO,EAAE,oBAAoB,EAAE,EAAE,CAAC;QAC7D,MAAM,QAAQ,GAAG,MAAM,kBAAkB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,EAAE,OAAO,CAAC,CAAC;QAC3E,UAAU,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,QAAQ,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,SAAS,EAAE,CAAC;QAC/F,UAAU,GAAG,QAAQ,CAAC,UAAU,CAAC;IAClC,CAAC;SAAM,IAAI,OAAO,CAAC,MAAM,KAAK,KAAK,EAAE,CAAC;QACrC,MAAM,QAAQ,GAAG,gBAAgB,CAAC,KAAK,EAAE,UAAU,EAAE,GAAG,CAAC,CAAC;QAC1D,UAAU,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,UAAU,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,CAAC,SAAS,EAAE,CAAC;QACjG,UAAU,GAAG,QAAQ,CAAC,UAAU,CAAC;IAClC,CAAC;IACD,4EAA4E;IAC5E,yEAAyE;IACzE,uEAAuE;IACvE,8BAA8B;IAC9B,UAAU,GAAG,oBAAoB,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAExE,OAAO;QACN,UAAU;QACV,YAAY,EAAE,IAAI;QAClB,cAAc,EAAE,UAAU,CAAC,cAAc;QACzC,UAAU,EAAE,KAAK,EAAE,KAAK,KAAK,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,EAAE,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,aAAa;QACzG,UAAU,EAAE,cAAc;QAC1B,QAAQ,EAAE,KAAK,EAAE,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS;QAC5F,IAAI;QACJ,MAAM,EAAE,UAAU;KAClB,CAAC;AAAA,CACF;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,OAAyB,EAA4B;IACpF,MAAM,SAAS,GAAG,MAAM,kBAAkB,CAAC,OAAO,CAAC,CAAC;IAEpD,MAAM,SAAS,GAAG,eAAe,CAAC,SAAS,CAAC,UAAU,EAAE,EAAE,GAAG,EAAE,OAAO,CAAC,GAAG,EAAE,WAAW,EAAE,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC;IAEhH,gBAAgB,CAAC,OAAO,CAAC,GAAG,EAAE;QAC7B,WAAW,EAAE,IAAI,CAAC,GAAG,EAAE;QACvB,KAAK,EAAE,OAAO,CAAC,KAAK;QACpB,aAAa,EAAE,OAAO,CAAC,IAAI,IAAI,MAAM;QACrC,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;QACxC,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,IAAI,EAAE,SAAS,CAAC,YAAY,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS;QACtE,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,KAAK,EAAE,SAAS,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC,CAAC;QAC7G,MAAM,EAAE,SAAS,CAAC,MAAM;KACxB,CAAC,CAAC;IAEH,OAAO;QACN,IAAI,EAAE,SAAS,CAAC,IAAI;QACpB,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;QACxC,WAAW,EAAE,SAAS,CAAC,UAAU,CAAC,MAAM;QACxC,QAAQ,EAAE,SAAS,CAAC,QAAQ;KAC5B,CAAC;AAAA,CACF","sourcesContent":["/**\n * Hybrid search orchestrator: resolve mode, run retrievers in parallel, fuse\n * by rank, expand within budget, trace everything\n * (docs/hybrid-retrieval-design.md).\n *\n * Single-retriever modes flow through the same pipeline — rrfFuse over one\n * list preserves its order — so lexical, semantic, and hybrid all produce the\n * same result shape and the same trace record.\n *\n * `retrieveCandidates` is the candidate-level core (also used by the eval\n * harness, which needs forced modes, a configurable `k`, and no trace\n * pollution); `runSearch` wraps it with span expansion and tracing for the\n * tool.\n */\n\nimport type { EmbsearchService } from \"../embsearch/embsearch-service.js\";\nimport { adaptGrepHits, type ChunkLookup } from \"./adapter.js\";\nimport { assembleContext } from \"./context-assembler.js\";\nimport { crossEncoderRerank } from \"./cross-rerank.js\";\nimport { runLexicalRetriever } from \"./lexical-retriever.js\";\nimport { resolveSearchMode } from \"./mode.js\";\nimport { rerankCandidates } from \"./rerank.js\";\nimport { DEFAULT_RRF_K, rrfFuse } from \"./rrf.js\";\nimport { writeSearchTrace } from \"./trace.js\";\nimport type { CandidateSpan, FusedCandidate, RankedHit, ResolvedSearchMode, SearchMode, SearchTrace } from \"./types.js\";\n\n/** Raw grep line-hits fetched per query (pre-collapse). */\nconst LEXICAL_MATCH_LIMIT = 200;\n/** Adapted lexical candidates entering fusion. The eval gate showed the\n *  uncapped lexical tail diluting hybrid below plain semantic: lexical\n *  precision is front-loaded by the adapter's term-evidence ranking, while\n *  RRF weighs a rank-30 lexical candidate like a rank-30 embedding hit. */\nconst LEXICAL_FUSION_CAP = 20;\n/**\n * Embedding hits fetched per query when fusing with the grep leg.\n *\n * Deliberately shallow. The grep list is capped at {@link LEXICAL_FUSION_CAP}\n * because its tail is unranked noise, so a deep dense pool here simply\n * outnumbers it: raising this to 200 alongside a 20-candidate grep list cost\n * `auto +rr` 0.472 -> 0.450 MRR.\n */\nconst EMBED_TOP_K = 50;\n/**\n * Per-leg depth when both retrievers are ranked ones (dense + BM25).\n *\n * Four times {@link FUSED_WINDOW}, matching what the daemon's `query_hybrid`\n * does internally (`pool = 4·k`). Fusing at the same depth as the window loses\n * any candidate ranked well by one retriever but just outside the other's\n * top-50 — worth 12pp R@10 and 8pp R@50 for `bm25+dense`, which is exactly\n * what closed the gap to daemon-side fusion.\n */\nconst FUSION_POOL_TOP_K = 200;\n/** BM25 depth, matching {@link FUSION_POOL_TOP_K} so neither ranked leg is\n *  structurally advantaged by pool size. The daemon returns only documents\n *  sharing a query term, so this is an upper bound, not a fill. */\nconst BM25_TOP_K = FUSION_POOL_TOP_K;\n/** Candidates from unindexed/stale files moved ahead of the fused window.\n *  Small on purpose: this answers \"what did I just write\", not \"search the\n *  working tree\". */\nconst STALE_HOIST_CAP = 5;\n/** Fused candidates kept for reranking / final slicing. */\nconst FUSED_WINDOW = 50;\n\nexport interface RetrieveOptions {\n\tcwd: string;\n\tquery: string;\n\tmode?: SearchMode;\n\t/** Optional glob filter applied to file paths. */\n\tglob?: string;\n\t/** Maximum fused candidates returned. */\n\tlimit?: number;\n\t/** RRF constant override (eval harness sweeps this). Default: {@link DEFAULT_RRF_K}. */\n\trrfK?: number;\n\t/** Rerank the fused top-50 before slicing to `limit`. Default: true. */\n\trerank?: boolean;\n\t/**\n\t * Ask the daemon to fuse its own BM25 index with the vectors and return one\n\t * already-fused ranking, instead of taking a dense-only list. Needs a store\n\t * built with `--hybrid`. The fused list arrives as a single \"embed\" leg,\n\t * because a pre-fused ranking has no per-retriever structure left to record.\n\t *\n\t * Prefer {@link bm25Leg}: fusing here keeps the legs separable in the trace\n\t * and lets the grep leg participate.\n\t */\n\tdaemonHybrid?: boolean;\n\t/**\n\t * Fetch the daemon's BM25 index as its own ranked list and fuse it here,\n\t * alongside dense and grep.\n\t *\n\t * Defaults to on wherever the daemon can serve it, because BM25 is the\n\t * better lexical leg on the indexed corpus: Recall@50 0.790 -> 0.879, 6 of\n\t * 62 queries better and 0 worse (p <= 0.05). When it is on, the grep leg\n\t * narrows to files the index has not read — see `staleFiles`.\n\t *\n\t * Set `false` to force ripgrep as the only lexical leg; the eval harness\n\t * does this to keep measuring what the old rows measured.\n\t */\n\tbm25Leg?: boolean;\n\t/**\n\t * Reorder the fused shortlist with the daemon's cross-encoder instead of\n\t * the deterministic reranker. Needs embsearch >= 0.3.0; costs one model\n\t * pass per scored candidate.\n\t */\n\tcrossEncoder?: boolean;\n\tservice?: EmbsearchService;\n\tsignal?: AbortSignal;\n}\n\nexport interface RetrieveResult {\n\tcandidates: FusedCandidate[];\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tindexPhase: SearchTrace[\"indexPhase\"];\n\tretrievers: SearchTrace[\"retrievers\"];\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n\trrfK: number;\n\trerank?: SearchTrace[\"rerank\"];\n}\n\nexport interface RunSearchOptions extends RetrieveOptions {\n\t/** Approximate token budget for the result text. */\n\ttokenBudget?: number;\n}\n\nexport interface RunSearchResult {\n\ttext: string;\n\tresolvedMode: ResolvedSearchMode;\n\tdegradedReason?: string;\n\tresultCount: number;\n\t/** Set while the embedding index is still building. */\n\tindexing?: { done: number; total: number };\n}\n\nfunction normalizeSearchGlob(glob: string | undefined): string | undefined {\n\tif (!glob) return undefined;\n\t// Match fd/rg semantics: a slash-containing glob is anchored anywhere in\n\t// the tree, so prepend \"**/\" unless it already starts with a slash or \"**/\".\n\tif (glob.includes(\"/\") && !glob.startsWith(\"/\") && !glob.startsWith(\"**/\")) {\n\t\treturn `**/${glob}`;\n\t}\n\treturn glob;\n}\n\n/**\n * Move candidates from files the index has not read to the front.\n *\n * They are there because grep found them and nothing else could: the index is\n * ranking a stale copy of the file, or has never seen it. Left to fuse, they\n * lose — RRF rewards agreement, and one leg reporting a single document is\n * outvoted by two legs agreeing on hundreds. Measured: scoping grep to stale\n * files without this hoist scored 25% on the live-edit set where unscoped grep\n * scored 100%, because the fused window filled with consensus hits about the\n * indexed copy.\n *\n * Capped, because \"stale\" scales with how far behind the index is. A few\n * edited files is the case this exists for; a fresh checkout makes everything\n * stale, and hoisting all of it would quietly turn hybrid search back into\n * grep. Past the cap the remainder keeps its fused position.\n */\nexport function hoistStaleCandidates(\n\tcandidates: readonly FusedCandidate[],\n\tstaleFiles: ReadonlySet<string>,\n): FusedCandidate[] {\n\tif (staleFiles.size === 0) return [...candidates];\n\tconst hoisted: FusedCandidate[] = [];\n\tconst rest: FusedCandidate[] = [];\n\tfor (const candidate of candidates) {\n\t\t(staleFiles.has(candidate.path) && hoisted.length < STALE_HOIST_CAP ? hoisted : rest).push(candidate);\n\t}\n\treturn [...hoisted, ...rest];\n}\n\nexport async function retrieveCandidates(options: RetrieveOptions): Promise<RetrieveResult> {\n\tconst { cwd, query, service, signal } = options;\n\tconst glob = normalizeSearchGlob(options.glob);\n\tconst requestedMode = options.mode ?? \"auto\";\n\tconst limit = Math.max(1, options.limit ?? 10);\n\tconst rrfK = options.rrfK ?? DEFAULT_RRF_K;\n\n\tconst state = service?.getState();\n\tconst embedAvailable = service?.isAvailable() ?? false;\n\tconst embedUnavailableReason =\n\t\tstate === undefined\n\t\t\t? \"semantic index is not enabled\"\n\t\t\t: state.phase === \"unavailable\" || state.phase === \"skipped\"\n\t\t\t\t? state.reason\n\t\t\t\t: state.phase === \"idle\"\n\t\t\t\t\t? \"semantic index has not started\"\n\t\t\t\t\t: undefined;\n\n\tconst resolution = resolveSearchMode(query, requestedMode, embedAvailable, embedUnavailableReason);\n\tconst mode = resolution.mode;\n\n\t// Map lexical hits onto indexed chunk ids whenever the sidecar is usable,\n\t// even in lexical-only mode, so identities line up across modes.\n\tconst lookupChunk: ChunkLookup | undefined = embedAvailable\n\t\t? (rel, line) => service!.findEnclosingChunk(rel, line)\n\t\t: undefined;\n\n\tconst spans = new Map<string, CandidateSpan>();\n\t/** Files the index does not have current content for, when grep is scoped\n\t *  to them. Read again after fusion — see the hoist below. */\n\tconst staleSet = new Set<string>();\n\tconst lists: RankedHit[][] = [];\n\tconst retrieverStats: SearchTrace[\"retrievers\"] = {};\n\tconst errors: Error[] = [];\n\n\tconst runLexical = async (scopeToStale = false): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Scoped run: grep covers only what the index has not read yet, so it\n\t\t\t// adds the one thing BM25 cannot see without re-voting on documents\n\t\t\t// BM25 already ranked. Fusing two lexical views of the same corpus is\n\t\t\t// what made the three-leg configuration lose (18 of 62 queries worse).\n\t\t\tconst paths = scopeToStale ? service?.staleFiles(signal) : undefined;\n\t\t\tif (paths) for (const rel of paths) staleSet.add(rel);\n\t\t\tconst lineHits = await runLexicalRetriever({ cwd, query, limit: LEXICAL_MATCH_LIMIT, glob, signal, paths });\n\t\t\tconst adapted = adaptGrepHits(lineHits, lookupChunk);\n\t\t\t// In single-retriever lexical mode the full list is the result; in\n\t\t\t// hybrid, only the front-loaded head is trustworthy enough to vote.\n\t\t\tconst hits = mode === \"hybrid\" ? adapted.hits.slice(0, LEXICAL_FUSION_CAP) : adapted.hits;\n\t\t\tfor (const [id, span] of adapted.spans) if (!spans.has(id)) spans.set(id, span);\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.grep = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runEmbed = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// The flat index pads top-k with whatever exists; with the cosine\n\t\t\t// metric the store uses, score <= 0 means \"no relation at all\", so\n\t\t\t// those padding hits would cast RRF votes on pure noise.\n\t\t\t// A BM25 leg is itself ranked, so the pair can afford — and needs — the\n\t\t\t// deeper pool; the grep leg cannot (see EMBED_TOP_K).\n\t\t\tconst topK = options.bm25Leg ? FUSION_POOL_TOP_K : EMBED_TOP_K;\n\t\t\tconst chunkHits = (\n\t\t\t\tawait service!.searchChunks(query, topK, glob, options.daemonHybrid ? \"hybrid\" : \"dense\")\n\t\t\t).filter((hit) => hit.score > 0);\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"embed\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.embed = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\tconst runBm25 = async (): Promise<void> => {\n\t\tconst startedMs = Date.now();\n\t\ttry {\n\t\t\t// Raw BM25 sums; only the ordering enters fusion, the score is a\n\t\t\t// diagnostic. Unlike the dense leg there is no zero-score padding to\n\t\t\t// filter — the daemon omits documents sharing no query term.\n\t\t\tconst chunkHits = await service!.searchChunks(query, BM25_TOP_K, glob, \"lexical\");\n\t\t\tconst hits: RankedHit[] = chunkHits.map((hit, i) => ({\n\t\t\t\tid: hit.id,\n\t\t\t\trank: i + 1,\n\t\t\t\tscore: hit.score,\n\t\t\t\tsource: \"bm25\",\n\t\t\t}));\n\t\t\tfor (const hit of chunkHits) {\n\t\t\t\tif (!spans.has(hit.id)) {\n\t\t\t\t\tspans.set(hit.id, { path: hit.path, startLine: hit.startLine, endLine: hit.endLine });\n\t\t\t\t}\n\t\t\t}\n\t\t\tlists.push(hits);\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: hits.length };\n\t\t} catch (e) {\n\t\t\terrors.push(e instanceof Error ? e : new Error(String(e)));\n\t\t\tretrieverStats.bm25 = { latencyMs: Date.now() - startedMs, hitCount: 0 };\n\t\t}\n\t};\n\n\t// BM25 is the better lexical leg where the index is current: it beats grep\n\t// on Recall@50 (0.790 -> 0.879, 6 queries better and 0 worse, p <= 0.05)\n\t// because it ranks the whole corpus rather than truncating a match stream.\n\t// It is also blind to anything indexed later than it was written, which is\n\t// precisely where grep still wins — so grep runs scoped to that set instead\n\t// of being dropped or left to duplicate BM25 over the whole tree.\n\tconst bm25AsLexicalLeg =\n\t\t(options.bm25Leg ?? service?.supportsLexicalRetriever() ?? false) && embedAvailable && mode !== \"lexical\";\n\n\tconst runs: Promise<void>[] = [];\n\tif (mode === \"lexical\" || mode === \"hybrid\") runs.push(runLexical(bm25AsLexicalLeg));\n\tif (mode === \"semantic\" || mode === \"hybrid\") runs.push(runEmbed());\n\t// `semantic` gets the BM25 leg too: the caller asked for the index, and\n\t// BM25 is part of it. Only an explicit `lexical` request excludes it.\n\tif (bm25AsLexicalLeg) runs.push(runBm25());\n\tawait Promise.all(runs);\n\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t// A partial failure in hybrid degrades to whichever retriever survived;\n\t// only a total loss is an error.\n\tif (lists.length === 0) throw errors[0] ?? new Error(\"search produced no retriever results\");\n\n\tconst fused = rrfFuse(lists, rrfK).slice(0, FUSED_WINDOW);\n\tlet candidates: FusedCandidate[] = [];\n\tfor (const hit of fused) {\n\t\tconst span = spans.get(hit.id);\n\t\tif (span) candidates.push({ ...hit, ...span });\n\t}\n\n\tlet rerankInfo: SearchTrace[\"rerank\"];\n\tif (options.crossEncoder && service?.supportsCrossEncoder()) {\n\t\tconst reranked = await crossEncoderRerank(query, candidates, cwd, service);\n\t\trerankInfo = { applied: true, candidateCount: reranked.scored, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t} else if (options.rerank !== false) {\n\t\tconst reranked = rerankCandidates(query, candidates, cwd);\n\t\trerankInfo = { applied: true, candidateCount: candidates.length, latencyMs: reranked.latencyMs };\n\t\tcandidates = reranked.candidates;\n\t}\n\t// After reranking, not before: both rerankers weight the fused prior, and a\n\t// lone grep hit from an unindexed file has the lowest prior there is, so\n\t// hoisting first would simply be undone. The reranker still orders the\n\t// hoisted set against itself.\n\tcandidates = hoistStaleCandidates(candidates, staleSet).slice(0, limit);\n\n\treturn {\n\t\tcandidates,\n\t\tresolvedMode: mode,\n\t\tdegradedReason: resolution.degradedReason,\n\t\tindexPhase: state?.phase === \"ready\" ? \"ready\" : state?.phase === \"indexing\" ? \"indexing\" : \"unavailable\",\n\t\tretrievers: retrieverStats,\n\t\tindexing: state?.phase === \"indexing\" ? { done: state.done, total: state.total } : undefined,\n\t\trrfK,\n\t\trerank: rerankInfo,\n\t};\n}\n\nexport async function runSearch(options: RunSearchOptions): Promise<RunSearchResult> {\n\tconst retrieved = await retrieveCandidates(options);\n\n\tconst assembled = assembleContext(retrieved.candidates, { cwd: options.cwd, tokenBudget: options.tokenBudget });\n\n\twriteSearchTrace(options.cwd, {\n\t\ttimestampMs: Date.now(),\n\t\tquery: options.query,\n\t\trequestedMode: options.mode ?? \"auto\",\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tindexPhase: retrieved.indexPhase,\n\t\trrfK: retrieved.resolvedMode === \"hybrid\" ? retrieved.rrfK : undefined,\n\t\tretrievers: retrieved.retrievers,\n\t\tfused: retrieved.candidates.map(({ id, rrfScore, ranks, rawScores }) => ({ id, rrfScore, ranks, rawScores })),\n\t\trerank: retrieved.rerank,\n\t});\n\n\treturn {\n\t\ttext: assembled.text,\n\t\tresolvedMode: retrieved.resolvedMode,\n\t\tdegradedReason: retrieved.degradedReason,\n\t\tresultCount: retrieved.candidates.length,\n\t\tindexing: retrieved.indexing,\n\t};\n}\n"]}

package/examples/extensions/custom-provider-anthropic/package.json

@@ -1,7 +1,7 @@
 {
 	"name": "@kolisachint/hoocode-extension-custom-provider-anthropic",
 	"private": true,
-	"version": "0.2.165",
+	"version": "0.3.0",
 	"type": "module",
 	"engines": {
 		"bun": ">=1.0.0"

package/examples/extensions/custom-provider-gitlab-duo/package.json

@@ -1,7 +1,7 @@
 {
 	"name": "@kolisachint/hoocode-extension-custom-provider-gitlab-duo",
 	"private": true,
-	"version": "0.2.165",
+	"version": "0.3.0",
 	"type": "module",
 	"engines": {
 		"bun": ">=1.0.0"

package/examples/extensions/sandbox/package.json

@@ -1,7 +1,7 @@
 {
 	"name": "@kolisachint/hoocode-extension-sandbox",
 	"private": true,
-	"version": "0.2.165",
+	"version": "0.3.0",
 	"type": "module",
 	"engines": {
 		"bun": ">=1.0.0"

package/examples/extensions/with-deps/package.json

@@ -1,7 +1,7 @@
 {
 	"name": "@kolisachint/hoocode-extension-with-deps",
 	"private": true,
-	"version": "0.2.165",
+	"version": "0.3.0",
 	"type": "module",
 	"engines": {
 		"bun": ">=1.0.0"

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@kolisachint/hoocode-agent",
-	"version": "0.4.168",
+	"version": "0.5.0",
 	"description": "Coding agent CLI with read, bash, edit, write tools and session management",
 	"type": "module",
 	"hoocodeConfig": {
@@ -48,9 +48,9 @@
 		"prepublishOnly": "npm run clean && npm run build"
 	},
 	"dependencies": {
-		"@kolisachint/hoocode-agent-core": "^0.4.168",
-		"@kolisachint/hoocode-ai": "^0.4.168",
-		"@kolisachint/hoocode-tui": "^0.4.168",
+		"@kolisachint/hoocode-agent-core": "^0.5.0",
+		"@kolisachint/hoocode-ai": "^0.5.0",
+		"@kolisachint/hoocode-tui": "^0.5.0",
 		"@silvia-odwyer/photon-node": "^0.3.4",
 		"chalk": "^5.5.0",
 		"cli-highlight": "^2.1.11",