codeimpact 0.4.2 → 0.4.3

package/dist/index.js

@@ -28332,15 +28332,16 @@ var EXCLUDED_PATH_PATTERNS = [
   "knowledge/"
 ];
 function PATH_EXCLUSION_SQL(col = "path") {
-  return `${col} NOT LIKE '%node_modules%'
-        AND ${col} NOT LIKE '%.git%'
+  return `${col} NOT LIKE '%node_modules/%'
+        AND ${col} NOT LIKE '%.git/%'
         AND ${col} NOT LIKE '%/dist/%'
         AND ${col} NOT LIKE '%/build/%'
         AND ${col} NOT LIKE '%/venv/%'
         AND ${col} NOT LIKE '%/.venv/%'
         AND ${col} NOT LIKE '%/env/%'
         AND ${col} NOT LIKE '%__pycache__%'
-        AND ${col} NOT LIKE '%/knowledge/%'`;
+        AND ${col} NOT LIKE '%/knowledge/%'
+        AND ${col} NOT LIKE 'knowledge/%'`;
 }
 var Tier2Storage = class {
   db;
@@ -28392,6 +28393,33 @@ var Tier2Storage = class {
     const stmt = this.db.prepare("DELETE FROM files WHERE path = ?");
     stmt.run(path);
   }
+  /**
+   * Bulk-delete files whose paths match excluded patterns (node_modules, venv, etc.).
+   * Call this during initialization to clean up stale data from previous indexing runs.
+   * Returns the number of rows deleted.
+   */
+  purgeExcludedFiles() {
+    const excludedIds = this.db.prepare(`
+      SELECT id FROM files
+      WHERE NOT (${PATH_EXCLUSION_SQL("path")})
+    `).all();
+    if (excludedIds.length === 0) return 0;
+    const ids = excludedIds.map((r) => r.id);
+    const chunkSize = 500;
+    let deleted = 0;
+    for (let i = 0; i < ids.length; i += chunkSize) {
+      const chunk = ids.slice(i, i + chunkSize);
+      const placeholders = chunk.map(() => "?").join(",");
+      this.db.prepare(`DELETE FROM embeddings WHERE file_id IN (${placeholders})`).run(...chunk);
+      this.db.prepare(`DELETE FROM symbols WHERE file_id IN (${placeholders})`).run(...chunk);
+      this.db.prepare(`DELETE FROM imports WHERE file_id IN (${placeholders})`).run(...chunk);
+      this.db.prepare(`DELETE FROM exports WHERE file_id IN (${placeholders})`).run(...chunk);
+      this.db.prepare(`DELETE FROM dependencies WHERE source_file_id IN (${placeholders}) OR target_file_id IN (${placeholders})`).run(...chunk, ...chunk);
+      const result = this.db.prepare(`DELETE FROM files WHERE id IN (${placeholders})`).run(...chunk);
+      deleted += result.changes;
+    }
+    return deleted;
+  }
   getAllFiles() {
     const stmt = this.db.prepare(`
       SELECT id, path, content_hash as contentHash, preview, language,
@@ -28671,6 +28699,10 @@ var Tier2Storage = class {
     const stmt = this.db.prepare("DELETE FROM dependencies WHERE source_file_id = ?");
     stmt.run(fileId);
   }
+  /** Remove ALL dependency edges. Used before a full rebuild pass. */
+  clearAllDependencies() {
+    this.db.prepare("DELETE FROM dependencies").run();
+  }
   // Project summary
   updateProjectSummary(name, description, languages, keyDirectories, architectureNotes) {
     const stmt = this.db.prepare(`
@@ -30294,6 +30326,28 @@ var ASTParser = class {
 import chokidar from "chokidar";
 import { EventEmitter } from "events";
 import { relative } from "path";
+function buildIgnoreMatcher(patterns) {
+  const segments = /* @__PURE__ */ new Set();
+  for (const p of patterns) {
+    const m = p.match(/^\*\*\/([^/*]+)\/\*\*?$/);
+    if (m && m[1]) {
+      segments.add(m[1]);
+    }
+  }
+  for (const s of ["node_modules", ".git", "venv", ".venv", "env", "__pycache__", "vendor", "knowledge", ".next", ".nuxt"]) {
+    segments.add(s);
+  }
+  const segmentArray = [...segments];
+  return (testPath) => {
+    const normalised = testPath.replace(/\\/g, "/");
+    for (const seg of segmentArray) {
+      if (normalised.includes("/" + seg + "/") || normalised.endsWith("/" + seg) || normalised === seg) {
+        return true;
+      }
+    }
+    return false;
+  };
+}
 var FileWatcher = class extends EventEmitter {
   watcher = null;
   projectPath;
@@ -30307,15 +30361,16 @@ var FileWatcher = class extends EventEmitter {
     if (this.watcher) {
       return;
     }
+    const ignoreFn = buildIgnoreMatcher(this.ignorePatterns);
     this.watcher = chokidar.watch(this.projectPath, {
       ignored: [
         /(^|[\/\\])\../,
         // dotfiles
-        ...this.ignorePatterns
+        ignoreFn
       ],
       persistent: true,
-      ignoreInitial: false,
-      // We want initial add events for indexing
+      ignoreInitial: true,
+      // Don't emit 'add' for existing files — initial index uses glob
       awaitWriteFinish: {
         stabilityThreshold: 300,
         pollInterval: 100
@@ -30334,7 +30389,7 @@ var FileWatcher = class extends EventEmitter {
     });
   }
   handleEvent(type, path) {
-    const relativePath = relative(this.projectPath, path);
+    const relativePath = relative(this.projectPath, path).replace(/\\/g, "/");
     const event = {
       type,
       path,
@@ -30460,6 +30515,31 @@ function countLines(content) {
 }
 
 // src/indexing/indexer.ts
+var EXCLUDED_PATH_SEGMENTS = [
+  "node_modules",
+  ".git",
+  "venv",
+  ".venv",
+  "env",
+  "__pycache__",
+  ".pytest_cache",
+  ".mypy_cache",
+  ".ruff_cache",
+  ".tox",
+  ".nox",
+  "vendor",
+  ".gradle",
+  "Pods",
+  "DerivedData",
+  ".next",
+  ".nuxt",
+  ".svelte-kit",
+  "knowledge"
+];
+function isExcludedPath(relPath) {
+  const segments = relPath.replace(/\\/g, "/").split("/");
+  return segments.some((seg) => EXCLUDED_PATH_SEGMENTS.includes(seg));
+}
 var Indexer = class extends EventEmitter2 {
   config;
   embeddingGenerator;
@@ -30525,9 +30605,12 @@ var Indexer = class extends EventEmitter2 {
   }
   async indexFile(absolutePath) {
     try {
+      const relativePath = relative2(this.config.projectPath, absolutePath).replace(/\\/g, "/");
+      if (isExcludedPath(relativePath)) {
+        return false;
+      }
       const content = readFileSync3(absolutePath, "utf-8");
       const stats = statSync(absolutePath);
-      const relativePath = relative2(this.config.projectPath, absolutePath);
       const contentHash = hashContent(content);
       const existingFile = this.tier2.getFile(relativePath);
       if (existingFile && existingFile.contentHash === contentHash) {
@@ -30604,6 +30687,10 @@ var Indexer = class extends EventEmitter2 {
     this.isIndexing = true;
     this.emit("indexingStarted");
     try {
+      const purged = this.tier2.purgeExcludedFiles();
+      if (purged > 0) {
+        console.error(`[Index] Purged ${purged} excluded files from database`);
+      }
       const patterns = [
         "**/*.ts",
         "**/*.tsx",
@@ -30662,6 +30749,10 @@ var Indexer = class extends EventEmitter2 {
           console.error(`Error indexing ${file2}:`, error2);
         }
       }
+      const depCount = this.rebuildDependencies();
+      if (depCount > 0) {
+        console.error(`[Index] Built ${depCount} dependency edges`);
+      }
       this.emit("indexingComplete", {
         total: checked,
         indexed,
@@ -30671,6 +30762,23 @@ var Indexer = class extends EventEmitter2 {
       this.isIndexing = false;
     }
   }
+  /**
+   * Rebuild all dependency edges from stored imports.
+   * Run after initial indexing so all target files are in the DB.
+   */
+  rebuildDependencies() {
+    this.tier2.clearAllDependencies();
+    const allImports = this.tier2.getAllImports();
+    let count = 0;
+    for (const imp of allImports) {
+      const targetFile = this.tier2.resolveImportToFile(imp.filePath, imp.importedFrom, this.config.projectPath);
+      if (targetFile) {
+        this.tier2.addDependency(imp.fileId, targetFile.id, "imports");
+        count++;
+      }
+    }
+    return count;
+  }
   startWatching() {
     this.watcher.start();
   }
@@ -30885,7 +30993,7 @@ ${result.preview}
       if (currentFile && dirname3(r.file) === dirname3(currentFile)) {
         score *= 1.5;
       }
-      const hoursSinceModified = (Date.now() - r.lastModified) / 36e5;
+      const hoursSinceModified = (Date.now() - r.lastModified * 1e3) / 36e5;
       if (hoursSinceModified < 24) {
         score *= 1 + 0.3 * (24 - hoursSinceModified) / 24;
       }
@@ -35630,7 +35738,7 @@ ${code}` : code;
             file: result.path,
             similarity,
             snippet: result.preview,
-            lastModified: file2 ? new Date(file2.lastModified) : void 0,
+            lastModified: file2 ? new Date(file2.lastModified * 1e3) : void 0,
             usageCount: dependents.length + 1,
             function: symbols.length > 0 ? symbols[0].name : void 0
           });
@@ -45196,7 +45304,10 @@ var CodeImpactEngine = class {
       this.featureContextManager.onFileOpened(path);
       this.summarizer.invalidateSummaryByPath(path);
       this.knowledgeOrchestrator.schedule("file_indexed", [path]);
-      this.pendingComponentDocPaths.add(path);
+      const normalizedPath = path.replace(/\\/g, "/");
+      if (!normalizedPath.includes("knowledge/") && !normalizedPath.includes(".code-impact/")) {
+        this.pendingComponentDocPaths.add(path);
+      }
       if (this.componentDocTimer) clearTimeout(this.componentDocTimer);
       this.componentDocTimer = setTimeout(() => {
         this.batchGenerateComponentDocs(Array.from(this.pendingComponentDocPaths)).catch((err) => {
@@ -45367,6 +45478,11 @@ var CodeImpactEngine = class {
         FROM files f
         LEFT JOIN dependencies d ON d.target_file_id = f.id
         WHERE f.language IS NOT NULL
+          AND f.path NOT LIKE '%node_modules%'
+          AND f.path NOT LIKE '%knowledge/%'
+          AND f.path NOT LIKE '%.code-impact/%'
+          AND f.path NOT LIKE '%/venv/%'
+          AND f.path NOT LIKE '%/.venv/%'
         GROUP BY f.id
         ORDER BY dep_count DESC
         LIMIT 20