graph-indexer 1.0.1 → 1.0.3

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "graph-indexer",
-    "version": "1.0.1",
+    "version": "1.0.3",
     "description": "Zero-DB, Air-gapped, In-Memory Graph Indexer MCP Server with Hybrid RRF Search and Cross-Language Topology",
     "type": "module",
     "main": "core-engine.mjs",
@@ -42,17 +42,6 @@
         "chokidar": "5.0.0",
         "ignore": "5.3.2",
         "tree-sitter": "0.21.1",
-        "tree-sitter-c-sharp": "^0.21.3",
-        "tree-sitter-css": "0.21.1",
-        "tree-sitter-go": "0.21.0",
-        "tree-sitter-java": "^0.21.0",
-        "tree-sitter-javascript": "0.21.4",
-        "tree-sitter-kotlin": "^0.3.8",
-        "tree-sitter-php": "0.23.1",
-        "tree-sitter-python": "0.21.0",
-        "tree-sitter-ruby": "^0.23.1",
-        "tree-sitter-rust": "0.21.0",
-        "tree-sitter-typescript": "0.21.2",
         "zod": "3.23.8"
     },
     "keywords": [
@@ -88,6 +77,17 @@
         "access": "public"
     },
     "optionalDependencies": {
-        "hnswlib-node": "^3.0.0"
+        "hnswlib-node": "^3.0.0",
+        "tree-sitter-c-sharp": "^0.21.3",
+        "tree-sitter-css": "0.21.1",
+        "tree-sitter-go": "0.21.0",
+        "tree-sitter-java": "^0.21.0",
+        "tree-sitter-javascript": "0.21.4",
+        "tree-sitter-kotlin": "^0.3.8",
+        "tree-sitter-php": "0.23.1",
+        "tree-sitter-python": "0.21.0",
+        "tree-sitter-ruby": "^0.23.1",
+        "tree-sitter-rust": "0.21.0",
+        "tree-sitter-typescript": "0.21.2"
     }
 }

package/parser-utils.mjs

@@ -19,34 +19,65 @@ import path from 'path';
 import { createHash } from 'crypto';
 import ignore from 'ignore';
 import Parser from 'tree-sitter';
-const { Query } = Parser; // Importación robusta nativa
-
-import TypeScript from 'tree-sitter-typescript';
-import JavaScript from 'tree-sitter-javascript';
-import CSS from 'tree-sitter-css';
-import Python from 'tree-sitter-python';
-import Rust from 'tree-sitter-rust';
-import Go from 'tree-sitter-go';
-import PHP from 'tree-sitter-php';
-import Java from 'tree-sitter-java';
-import Kotlin from 'tree-sitter-kotlin';
-import CSharp from 'tree-sitter-c-sharp';
-import Ruby from 'tree-sitter-ruby';
+const { Query } = Parser; // Native parser query helper
+
 import { truncateForEmbedding } from './core-engine.mjs';
 
 export const MAX_FILE_SIZE_BYTES = 500000;
 export const OLLAMA_HOST = process.env.OLLAMA_HOST || "http://localhost:11434";
 
+// ─── Dynamic Language Loading ─────────────────────────────────────────────────
+
+function _loadProjectConfig() {
+    const configPath = path.join(process.env.MCP_PROJECT_ROOT || process.cwd(), '.graph-indexer.json');
+    try {
+        if (fs.existsSync(configPath)) return JSON.parse(fs.readFileSync(configPath, 'utf-8'));
+    } catch { /* ignore */ }
+    return null;
+}
+
+async function _tryLang(pkg, enabledLangs, key) {
+    if (enabledLangs && !enabledLangs.includes(key)) return null;
+    try {
+        return (await import(pkg)).default;
+    } catch {
+        process.stderr.write(`[graph-indexer] WARNING: ${pkg} not installed — ${key} files will be skipped\n`);
+        return null;
+    }
+}
+
+const _cfg = _loadProjectConfig();
+const _enabled = _cfg?.languages ?? null; // null = all languages
+
+const [
+    TypeScript, JavaScript, CSS, Python, Rust,
+    Go, PHP, Java, Kotlin, CSharp, Ruby
+] = await Promise.all([
+    _tryLang('tree-sitter-typescript', _enabled, 'typescript'),
+    _tryLang('tree-sitter-javascript', _enabled, 'javascript'),
+    _tryLang('tree-sitter-css',        _enabled, 'css'),
+    _tryLang('tree-sitter-python',     _enabled, 'python'),
+    _tryLang('tree-sitter-rust',       _enabled, 'rust'),
+    _tryLang('tree-sitter-go',         _enabled, 'go'),
+    _tryLang('tree-sitter-php',        _enabled, 'php'),
+    _tryLang('tree-sitter-java',       _enabled, 'java'),
+    _tryLang('tree-sitter-kotlin',     _enabled, 'kotlin'),
+    _tryLang('tree-sitter-c-sharp',    _enabled, 'csharp'),
+    _tryLang('tree-sitter-ruby',       _enabled, 'ruby'),
+]);
+
 const LANGUAGE_MAP = {
-    '.ts': TypeScript.typescript, '.tsx': TypeScript.tsx,
-    '.js': JavaScript, '.jsx': JavaScript, '.mjs': JavaScript, '.cjs': JavaScript,
-    '.css': CSS, '.scss': CSS,
-    '.py': Python, '.rs': Rust, '.go': Go,
-    '.php': PHP.php,
-    '.java': Java,
-    '.kt': Kotlin, '.kts': Kotlin,
-    '.cs': CSharp,
-    '.rb': Ruby,
+    ...(TypeScript ? { '.ts': TypeScript.typescript, '.tsx': TypeScript.tsx } : {}),
+    ...(JavaScript ? { '.js': JavaScript, '.jsx': JavaScript, '.mjs': JavaScript, '.cjs': JavaScript } : {}),
+    ...(CSS        ? { '.css': CSS, '.scss': CSS } : {}),
+    ...(Python     ? { '.py': Python }  : {}),
+    ...(Rust       ? { '.rs': Rust }    : {}),
+    ...(Go         ? { '.go': Go }      : {}),
+    ...(PHP        ? { '.php': PHP.php } : {}),
+    ...(Java       ? { '.java': Java }  : {}),
+    ...(Kotlin     ? { '.kt': Kotlin, '.kts': Kotlin } : {}),
+    ...(CSharp     ? { '.cs': CSharp }  : {}),
+    ...(Ruby       ? { '.rb': Ruby }    : {}),
 };
 
 // 🥇 ULTRA-GENERIC QUERIES: Immune to grammar changes between TS/JS/TSX
@@ -398,6 +429,7 @@ export function extractSemanticChunks(rootNode, relPath, sourceCode, ext) {
             const params = extractParams(chunkNode, ext);
             const returnType = extractReturnType(chunkNode, ext);
             const classContext = extractClassContext(chunkNode);
+            const typeRefs = extractTypeAnnotations(chunkNode, ext);
 
             const id = createHash('sha256')
                 .update(`${relPath}::${chunkNode.startPosition.row}::${chunkNode.startPosition.column}`)
@@ -408,7 +440,8 @@ export function extractSemanticChunks(rootNode, relPath, sourceCode, ext) {
                 docstring: docstring, code_snippet: snippet, content_hash: hash,
                 start_line: chunkNode.startPosition.row + 1, end_line: chunkNode.endPosition.row + 1,
                 calls: outgoingCalls,
-                params, return_type: returnType, class_context: classContext
+                params, return_type: returnType, class_context: classContext,
+                type_refs: typeRefs,
             });
         }
     } catch (e) {
@@ -709,6 +742,44 @@ export function extractClassContext(chunkNode) {
     return '';
 }
 
+/**
+ * Frontier 2: Extract TypeScript/Python type annotation names from a chunk node.
+ * Returns simple type names (e.g. ['User', 'AuthToken', 'PaymentService'])
+ * used to enrich the inverted index and the type_refs chunk field.
+ */
+export function extractTypeAnnotations(chunkNode, ext) {
+    const types = new Set();
+    const JS_LIKE = ['.ts', '.tsx', '.js', '.jsx', '.mjs', '.cjs'];
+    if (!JS_LIKE.includes(ext) && ext !== '.py') return [];
+
+    function walk(node) {
+        // TypeScript: type_annotation nodes contain the type text
+        if (node.type === 'type_annotation') {
+            const typeText = node.text.replace(/^:\s*/, '').trim();
+            // Extract simple identifiers from the type (skip primitives)
+            const PRIMITIVES = new Set(['string','number','boolean','void','any','unknown','never','null','undefined','object','symbol','bigint']);
+            for (const match of typeText.matchAll(/\b([A-Z][A-Za-z0-9]*)\b/g)) {
+                if (!PRIMITIVES.has(match[1].toLowerCase())) types.add(match[1]);
+            }
+        }
+        // TypeScript: generic_type, predefined_type (string, number…) skip, named types keep
+        else if (node.type === 'type_identifier' || node.type === 'generic_type') {
+            const name = node.children[0]?.text || node.text;
+            if (name && /^[A-Z]/.test(name)) types.add(name);
+        }
+        // Python: type comments or annotations (annotation nodes)
+        else if (node.type === 'annotation' && ext === '.py') {
+            const typeText = node.text.replace(/^->\s*|^:\s*/, '').trim();
+            for (const match of typeText.matchAll(/\b([A-Z][A-Za-z0-9]*)\b/g)) {
+                types.add(match[1]);
+            }
+        }
+        for (const child of node.children) walk(child);
+    }
+    walk(chunkNode);
+    return Array.from(types).slice(0, 20);
+}
+
 export function extractCalls(rootNode) {
     const calls = new Set();
     function walk(node) {

package/watch-daemon.mjs

@@ -132,11 +132,11 @@ async function processFileChange(absolutePath) {
         db.saveDebounced();
         process.stderr.write(`[daemon] 🔄 Synced: ${filename} (Cache Hit: ${newChunks.length - chunksToEmbed.length}/${newChunks.length})\n`);
     } catch (err) {
-        process.stderr.write(`[daemon] ❌ Error en ${filename}: ${err.message}\n`);
+        process.stderr.write(`[daemon] ❌ Error in ${filename}: ${err.message}\n`);
     }
 }
 
-process.stderr.write(`🚀 Chokidar Watcher Daemon iniciado en: ${PROJECT_ROOT}\n`);
+process.stderr.write(`🚀 Chokidar Watcher Daemon started in: ${PROJECT_ROOT}\n`);
 const watcher = chokidar.watch(PROJECT_ROOT, {
     ignored: /(^|[\/\\])\../, persistent: true,
     awaitWriteFinish: { stabilityThreshold: 300, pollInterval: 100 }