context-mode 1.0.52 → 1.0.54

package/build/store.js

@@ -212,13 +212,17 @@ export class ContentStore {
     // Search path (hot)
     #stmtSearchPorter;
     #stmtSearchPorterFiltered;
+    #stmtSearchPorterExact;
     #stmtSearchTrigram;
     #stmtSearchTrigramFiltered;
+    #stmtSearchTrigramExact;
     #stmtFuzzyVocab;
     #stmtSearchPorterContentType;
     #stmtSearchPorterFilteredContentType;
+    #stmtSearchPorterExactContentType;
     #stmtSearchTrigramContentType;
     #stmtSearchTrigramFilteredContentType;
+    #stmtSearchTrigramExactContentType;
     // Read path
     #stmtListSources;
     #stmtChunksBySource;
@@ -278,6 +282,8 @@ export class ContentStore {
       CREATE TABLE IF NOT EXISTS vocabulary (
         word TEXT PRIMARY KEY
       );
+
+      CREATE INDEX IF NOT EXISTS idx_sources_label ON sources(label);
     `);
     }
     #prepareStatements() {
@@ -320,6 +326,20 @@ export class ContentStore {
       WHERE chunks MATCH ? AND sources.label LIKE ?
       ORDER BY rank
       LIMIT ?
+    `);
+        this.#stmtSearchPorterExact = this.#db.prepare(`
+      SELECT
+        chunks.title,
+        chunks.content,
+        chunks.content_type,
+        sources.label,
+        bm25(chunks, 5.0, 1.0) AS rank,
+        highlight(chunks, 1, char(2), char(3)) AS highlighted
+      FROM chunks
+      JOIN sources ON sources.id = chunks.source_id
+      WHERE chunks MATCH ? AND sources.label = ?
+      ORDER BY rank
+      LIMIT ?
     `);
         this.#stmtSearchTrigram = this.#db.prepare(`
       SELECT
@@ -348,6 +368,20 @@ export class ContentStore {
       WHERE chunks_trigram MATCH ? AND sources.label LIKE ?
       ORDER BY rank
       LIMIT ?
+    `);
+        this.#stmtSearchTrigramExact = this.#db.prepare(`
+      SELECT
+        chunks_trigram.title,
+        chunks_trigram.content,
+        chunks_trigram.content_type,
+        sources.label,
+        bm25(chunks_trigram, 5.0, 1.0) AS rank,
+        highlight(chunks_trigram, 1, char(2), char(3)) AS highlighted
+      FROM chunks_trigram
+      JOIN sources ON sources.id = chunks_trigram.source_id
+      WHERE chunks_trigram MATCH ? AND sources.label = ?
+      ORDER BY rank
+      LIMIT ?
     `);
         // Content-type filtered variants
         this.#stmtSearchPorterContentType = this.#db.prepare(`
@@ -377,6 +411,20 @@ export class ContentStore {
       WHERE chunks MATCH ? AND sources.label LIKE ? AND chunks.content_type = ?
       ORDER BY rank
       LIMIT ?
+    `);
+        this.#stmtSearchPorterExactContentType = this.#db.prepare(`
+      SELECT
+        chunks.title,
+        chunks.content,
+        chunks.content_type,
+        sources.label,
+        bm25(chunks, 5.0, 1.0) AS rank,
+        highlight(chunks, 1, char(2), char(3)) AS highlighted
+      FROM chunks
+      JOIN sources ON sources.id = chunks.source_id
+      WHERE chunks MATCH ? AND sources.label = ? AND chunks.content_type = ?
+      ORDER BY rank
+      LIMIT ?
     `);
         this.#stmtSearchTrigramContentType = this.#db.prepare(`
       SELECT
@@ -405,6 +453,20 @@ export class ContentStore {
       WHERE chunks_trigram MATCH ? AND sources.label LIKE ? AND chunks_trigram.content_type = ?
       ORDER BY rank
       LIMIT ?
+    `);
+        this.#stmtSearchTrigramExactContentType = this.#db.prepare(`
+      SELECT
+        chunks_trigram.title,
+        chunks_trigram.content,
+        chunks_trigram.content_type,
+        sources.label,
+        bm25(chunks_trigram, 5.0, 1.0) AS rank,
+        highlight(chunks_trigram, 1, char(2), char(3)) AS highlighted
+      FROM chunks_trigram
+      JOIN sources ON sources.id = chunks_trigram.source_id
+      WHERE chunks_trigram MATCH ? AND sources.label = ? AND chunks_trigram.content_type = ?
+      ORDER BY rank
+      LIMIT ?
     `);
         // Fuzzy path
         this.#stmtFuzzyVocab = this.#db.prepare("SELECT word FROM vocabulary WHERE length(word) BETWEEN ? AND ?");
@@ -514,17 +576,34 @@ export class ContentStore {
         };
     }
     // ── Search ──
-    search(query, limit = 3, source, mode = "AND", contentType) {
+    #mapSearchRows(rows) {
+        return rows.map((r) => ({
+            title: r.title,
+            content: r.content,
+            source: r.label,
+            rank: r.rank,
+            contentType: r.content_type,
+            highlighted: r.highlighted,
+        }));
+    }
+    #sourceFilterParam(source, sourceMatchMode) {
+        return sourceMatchMode === "exact" ? source : `%${source}%`;
+    }
+    search(query, limit = 3, source, mode = "AND", contentType, sourceMatchMode = "like") {
         const sanitized = sanitizeQuery(query, mode);
         let stmt;
         let params;
         if (source && contentType) {
-            stmt = this.#stmtSearchPorterFilteredContentType;
-            params = [sanitized, `%${source}%`, contentType, limit];
+            stmt = sourceMatchMode === "exact"
+                ? this.#stmtSearchPorterExactContentType
+                : this.#stmtSearchPorterFilteredContentType;
+            params = [sanitized, this.#sourceFilterParam(source, sourceMatchMode), contentType, limit];
         }
         else if (source) {
-            stmt = this.#stmtSearchPorterFiltered;
-            params = [sanitized, `%${source}%`, limit];
+            stmt = sourceMatchMode === "exact"
+                ? this.#stmtSearchPorterExact
+                : this.#stmtSearchPorterFiltered;
+            params = [sanitized, this.#sourceFilterParam(source, sourceMatchMode), limit];
         }
         else if (contentType) {
             stmt = this.#stmtSearchPorterContentType;
@@ -534,30 +613,26 @@ export class ContentStore {
             stmt = this.#stmtSearchPorter;
             params = [sanitized, limit];
         }
-        const rows = stmt.all(...params);
-        return rows.map((r) => ({
-            title: r.title,
-            content: r.content,
-            source: r.label,
-            rank: r.rank,
-            contentType: r.content_type,
-            highlighted: r.highlighted,
-        }));
+        return this.#mapSearchRows(stmt.all(...params));
     }
     // ── Trigram Search (Layer 2) ──
-    searchTrigram(query, limit = 3, source, mode = "AND", contentType) {
+    searchTrigram(query, limit = 3, source, mode = "AND", contentType, sourceMatchMode = "like") {
         const sanitized = sanitizeTrigramQuery(query, mode);
         if (!sanitized)
             return [];
         let stmt;
         let params;
         if (source && contentType) {
-            stmt = this.#stmtSearchTrigramFilteredContentType;
-            params = [sanitized, `%${source}%`, contentType, limit];
+            stmt = sourceMatchMode === "exact"
+                ? this.#stmtSearchTrigramExactContentType
+                : this.#stmtSearchTrigramFilteredContentType;
+            params = [sanitized, this.#sourceFilterParam(source, sourceMatchMode), contentType, limit];
         }
         else if (source) {
-            stmt = this.#stmtSearchTrigramFiltered;
-            params = [sanitized, `%${source}%`, limit];
+            stmt = sourceMatchMode === "exact"
+                ? this.#stmtSearchTrigramExact
+                : this.#stmtSearchTrigramFiltered;
+            params = [sanitized, this.#sourceFilterParam(source, sourceMatchMode), limit];
         }
         else if (contentType) {
             stmt = this.#stmtSearchTrigramContentType;
@@ -567,15 +642,7 @@ export class ContentStore {
             stmt = this.#stmtSearchTrigram;
             params = [sanitized, limit];
         }
-        const rows = stmt.all(...params);
-        return rows.map((r) => ({
-            title: r.title,
-            content: r.content,
-            source: r.label,
-            rank: r.rank,
-            contentType: r.content_type,
-            highlighted: r.highlighted,
-        }));
+        return this.#mapSearchRows(stmt.all(...params));
     }
     // ── Fuzzy Correction (Layer 3) ──
     fuzzyCorrect(query) {
@@ -598,11 +665,11 @@ export class ContentStore {
         return bestDist <= maxDist ? bestWord : null;
     }
     // ── Reciprocal Rank Fusion (Cormack et al. 2009) ──
-    #rrfSearch(query, limit, source, contentType) {
+    #rrfSearch(query, limit, source, contentType, sourceMatchMode = "like") {
         const K = 60; // Standard RRF constant
         const fetchLimit = Math.max(limit * 2, 10);
-        const porterResults = this.search(query, fetchLimit, source, "OR", contentType);
-        const trigramResults = this.searchTrigram(query, fetchLimit, source, "OR", contentType);
+        const porterResults = this.search(query, fetchLimit, source, "OR", contentType, sourceMatchMode);
+        const trigramResults = this.searchTrigram(query, fetchLimit, source, "OR", contentType, sourceMatchMode);
         const scoreMap = new Map();
         const key = (r) => `${r.source}::${r.title}`;
         for (const [i, r] of porterResults.entries()) {
@@ -655,9 +722,9 @@ export class ContentStore {
             .map(({ result }) => result);
     }
     // ── Unified Fallback Search ──
-    searchWithFallback(query, limit = 3, source, contentType) {
+    searchWithFallback(query, limit = 3, source, contentType, sourceMatchMode = "like") {
         // Step 1: RRF fusion (porter OR + trigram OR → merge)
-        const rrfResults = this.#rrfSearch(query, limit, source, contentType);
+        const rrfResults = this.#rrfSearch(query, limit, source, contentType, sourceMatchMode);
         if (rrfResults.length > 0) {
             const reranked = this.#applyProximityReranking(rrfResults, query);
             return reranked.map((r) => ({ ...r, matchLayer: "rrf" }));
@@ -672,7 +739,7 @@ export class ContentStore {
         const correctedWords = words.map((w) => this.fuzzyCorrect(w) ?? w);
         const correctedQuery = correctedWords.join(" ");
         if (correctedQuery !== original) {
-            const fuzzyResults = this.#rrfSearch(correctedQuery, limit, source, contentType);
+            const fuzzyResults = this.#rrfSearch(correctedQuery, limit, source, contentType, sourceMatchMode);
             if (fuzzyResults.length > 0) {
                 const reranked = this.#applyProximityReranking(fuzzyResults, correctedQuery);
                 return reranked.map((r) => ({ ...r, matchLayer: "rrf-fuzzy" }));