@softerist/heuristic-mcp 3.0.12 → 3.0.14

package/lib/cache.js

@@ -356,13 +356,18 @@ export class EmbeddingsCache {
     }
   }
 
-  async ensureLoaded() {
+  async ensureLoaded({ preferDisk = false } = {}) {
     if (!this.config.enableCache) return;
     if (!this._clearedAfterIndex) return;
     if (this._loadPromise) return this._loadPromise;
 
     this._loadPromise = (async () => {
-      await this.load();
+      if (preferDisk && this.config.verbose) {
+        console.info('[Cache] ensureLoaded: forcing disk vector mode for incremental low-RAM reload');
+      }
+      await this.load({
+        forceVectorLoadMode: preferDisk ? 'disk' : undefined,
+      });
       this._clearedAfterIndex = false;
     })().finally(() => {
       this._loadPromise = null;
@@ -507,7 +512,7 @@ export class EmbeddingsCache {
 
   // -------------------- Load --------------------
 
-  async load() {
+  async load({ forceVectorLoadMode } = {}) {
     if (!this.config.enableCache) return;
 
     try {
@@ -637,16 +642,25 @@ export class EmbeddingsCache {
         }
       }
 
+      const configuredVectorLoadMode =
+        typeof this.config.vectorStoreLoadMode === 'string'
+          ? this.config.vectorStoreLoadMode.toLowerCase()
+          : 'memory';
+      const effectiveVectorLoadMode =
+        forceVectorLoadMode === 'disk' || forceVectorLoadMode === 'memory'
+          ? forceVectorLoadMode
+          : configuredVectorLoadMode;
+
       if (useBinary) {
         try {
           this.binaryStore = await BinaryVectorStore.load(this.config.cacheDirectory, {
             contentCacheEntries: this.config.contentCacheEntries,
             vectorCacheEntries: this.config.vectorCacheEntries,
-            vectorLoadMode: this.config.vectorStoreLoadMode,
+            vectorLoadMode: effectiveVectorLoadMode,
           });
           cacheData = await this.binaryStore.toChunkViews({
             includeContent: this.config.vectorStoreContentMode === 'inline',
-            includeVector: this.config.vectorStoreLoadMode !== 'disk',
+            includeVector: effectiveVectorLoadMode !== 'disk',
           });
         } catch (err) {
           this.binaryStore = null;
@@ -661,7 +675,7 @@ export class EmbeddingsCache {
           if (this.sqliteStore) {
             cacheData = this.sqliteStore.toChunkViews({
               includeContent: this.config.vectorStoreContentMode === 'inline',
-              includeVector: this.config.vectorStoreLoadMode !== 'disk',
+              includeVector: effectiveVectorLoadMode !== 'disk',
             });
           } else {
             // SQLite file missing, need reindex
@@ -684,7 +698,13 @@ export class EmbeddingsCache {
         const allowedExtensions = new Set(
           (this.config.fileExtensions || []).map((ext) => `.${ext}`)
         );
+        const allowedFileNames = new Set(this.config.fileNames || []);
         const applyExtensionFilter = !this.binaryStore;
+        const shouldKeepFile = (filePath) => {
+          const ext = path.extname(filePath);
+          if (allowedExtensions.has(ext)) return true;
+          return allowedFileNames.has(path.basename(filePath));
+        };
 
         const rawHashes = hasHashData ? new Map(Object.entries(hashData)) : new Map();
         this.vectorStore = [];
@@ -693,8 +713,7 @@ export class EmbeddingsCache {
         // Single-pass filter + normalization
         for (const chunk of cacheData) {
           if (applyExtensionFilter) {
-            const ext = path.extname(chunk.file);
-            if (!allowedExtensions.has(ext)) continue;
+            if (!shouldKeepFile(chunk.file)) continue;
           }
           normalizeChunkVector(chunk);
           this.vectorStore.push(chunk);
@@ -707,7 +726,7 @@ export class EmbeddingsCache {
         if (hasHashData) {
           // Only keep hashes for allowed extensions
           for (const [file, entry] of rawHashes) {
-            if (!applyExtensionFilter || allowedExtensions.has(path.extname(file))) {
+            if (!applyExtensionFilter || shouldKeepFile(file)) {
               const normalized = normalizeFileHashEntry(entry);
               if (normalized) {
                 this.fileHashes.set(file, normalized);
@@ -814,24 +833,22 @@ export class EmbeddingsCache {
       const hashFile = path.join(this.config.cacheDirectory, 'file-hashes.json');
       const metaFile = path.join(this.config.cacheDirectory, CACHE_META_FILE);
 
-      // Snapshot to avoid race conditions during async write
+      // Snapshot to avoid race conditions during async write.
+      // Keep this shallow for binary/sqlite to prevent multi-GB vector materialization.
+      const snapshotStore = Array.isArray(this.vectorStore) ? [...this.vectorStore] : [];
+      const supportsBackendVectorResolve =
+        this.config.vectorStoreFormat === 'binary' || this.config.vectorStoreFormat === 'sqlite';
+      const hasMissingVectors = snapshotStore.some(
+        (chunk) => chunk && (chunk.vector === undefined || chunk.vector === null)
+      );
       const useDiskVectors =
-        this.config.vectorStoreLoadMode === 'disk' &&
-        (this.config.vectorStoreFormat === 'binary' || this.config.vectorStoreFormat === 'sqlite');
-      const snapshotStore = Array.isArray(this.vectorStore)
-        ? useDiskVectors
-          ? this.vectorStore.map((chunk, index) => {
-              if (!chunk) return null;
-              const vector = this.getChunkVector(chunk, index);
-              if (!vector) {
-                throw new Error(
-                  `Missing vector data for cache write in disk mode at index ${index}`
-                );
-              }
-              return { ...chunk, vector };
-            })
-          : [...this.vectorStore]
-        : [];
+        supportsBackendVectorResolve &&
+        (this.config.vectorStoreLoadMode === 'disk' || hasMissingVectors);
+      if (hasMissingVectors && !useDiskVectors) {
+        throw new Error(
+          'Missing vector data for cache write and backend vector resolution is unavailable'
+        );
+      }
 
       this.cacheMeta = {
         version: CACHE_META_VERSION,
@@ -856,7 +873,10 @@ export class EmbeddingsCache {
           snapshotStore,
           {
             contentCacheEntries: this.config.contentCacheEntries,
+            vectorCacheEntries: this.config.vectorCacheEntries,
+            vectorLoadMode: useDiskVectors ? 'disk' : this.config.vectorStoreLoadMode,
             getContent: (chunk, index) => this.getChunkContent(chunk, index),
+            getVector: useDiskVectors ? (chunk, index) => this.getChunkVector(chunk, index) : null,
             preRename: async () => {
               if (this.activeReads > 0) {
                 await this.waitForReadersWithTimeout(
@@ -890,6 +910,7 @@ export class EmbeddingsCache {
           snapshotStore,
           {
             getContent: (chunk, index) => this.getChunkContent(chunk, index),
+            getVector: useDiskVectors ? (chunk, index) => this.getChunkVector(chunk, index) : null,
             preRename: async () => {
               if (this.activeReads > 0) {
                 await this.waitForReadersWithTimeout(
@@ -1582,6 +1603,14 @@ export class EmbeddingsCache {
         }
       }
       this.binaryStore = null;
+      if (this.sqliteStore) {
+        try {
+          this.sqliteStore.close();
+        } catch {
+          // ignore close errors
+        }
+      }
+      this.sqliteStore = null;
       this.fileHashes = new Map();
       this.invalidateAnnIndex();
       await this.clearCallGraphData();