perf: fix v3.0.1 build regression (14.1 → ~5.8 ms/file)

src/builder.js

@@ -1272,7 +1272,7 @@ export async function buildGraph(rootDir, opts = {}) {
   }
   _t.rolesMs = performance.now() - _t.roles0;
 
-  // For incremental builds, filter out reverse-dep-only files from AST/complexity
+  // For incremental builds, filter out reverse-dep-only files from AST/complexity/CFG/dataflow
   // — their content didn't change, so existing ast_nodes/function_complexity rows are valid.
   let astComplexitySymbols = allSymbols;
   if (!isFullBuild) {
@@ -1287,13 +1287,12 @@ export async function buildGraph(rootDir, opts = {}) {
         }
       }
       debug(
-        `AST/complexity: processing ${astComplexitySymbols.size} changed files (skipping ${reverseDepFiles.size} reverse-deps)`,
+        `AST/complexity/CFG/dataflow: processing ${astComplexitySymbols.size} changed files (skipping ${reverseDepFiles.size} reverse-deps)`,
       );
     }
   }
 
   // AST node extraction (calls, new, string, regex, throw, await)
-  // Must run before complexity which releases _tree references
   _t.ast0 = performance.now();
   if (opts.ast !== false) {
     try {
@@ -1317,12 +1316,25 @@ export async function buildGraph(rootDir, opts = {}) {
   }
   _t.complexityMs = performance.now() - _t.complexity0;
 
+  // Pre-parse files missing WASM trees (native builds) so CFG + dataflow
+  // share a single parse pass instead of each creating parsers independently
+  if (opts.cfg !== false || opts.dataflow !== false) {
+    _t.wasmPre0 = performance.now();
+    try {
+      const { ensureWasmTrees } = await import('./parser.js');
+      await ensureWasmTrees(astComplexitySymbols, rootDir);
+    } catch (err) {
+      debug(`WASM pre-parse failed: ${err.message}`);
+    }
+    _t.wasmPreMs = performance.now() - _t.wasmPre0;
+  }
+
   // CFG analysis (skip with --no-cfg)
   if (opts.cfg !== false) {
     _t.cfg0 = performance.now();
     try {
       const { buildCFGData } = await import('./cfg.js');
-      await buildCFGData(db, allSymbols, rootDir, engineOpts);
+      await buildCFGData(db, astComplexitySymbols, rootDir, engineOpts);
     } catch (err) {
       debug(`CFG analysis failed: ${err.message}`);
     }
@@ -1334,7 +1346,7 @@ export async function buildGraph(rootDir, opts = {}) {
     _t.dataflow0 = performance.now();
     try {
       const { buildDataflowEdges } = await import('./dataflow.js');
-      await buildDataflowEdges(db, allSymbols, rootDir, engineOpts);
+      await buildDataflowEdges(db, astComplexitySymbols, rootDir, engineOpts);
     } catch (err) {
       debug(`Dataflow analysis failed: ${err.message}`);
     }
@@ -1434,6 +1446,7 @@ export async function buildGraph(rootDir, opts = {}) {
       rolesMs: +_t.rolesMs.toFixed(1),
       astMs: +_t.astMs.toFixed(1),
       complexityMs: +_t.complexityMs.toFixed(1),
+      ...(_t.wasmPreMs != null && { wasmPreMs: +_t.wasmPreMs.toFixed(1) }),
       ...(_t.cfgMs != null && { cfgMs: +_t.cfgMs.toFixed(1) }),
       ...(_t.dataflowMs != null && { dataflowMs: +_t.dataflowMs.toFixed(1) }),
     },

src/complexity.js

@@ -1769,9 +1769,6 @@ export async function buildComplexityMetrics(db, fileSymbols, rootDir, _engineOp
         );
         analyzed++;
       }
-
-      // Release cached tree for GC
-      symbols._tree = null;
     }
   });
 

src/parser.js

@@ -38,6 +38,9 @@ function grammarPath(name) {
 
 let _initialized = false;
 
+// Memoized parsers — avoids reloading WASM grammars on every createParsers() call
+let _cachedParsers = null;
+
 // Query cache for JS/TS/TSX extractors (populated during createParsers)
 const _queryCache = new Map();
 
@@ -66,6 +69,8 @@ const TS_EXTRA_PATTERNS = [
 ];
 
 export async function createParsers() {
+  if (_cachedParsers) return _cachedParsers;
+
   if (!_initialized) {
     await Parser.init();
     _initialized = true;
@@ -94,6 +99,7 @@ export async function createParsers() {
       parsers.set(entry.id, null);
     }
   }
+  _cachedParsers = parsers;
   return parsers;
 }
 
@@ -104,6 +110,54 @@ export function getParser(parsers, filePath) {
   return parsers.get(entry.id) || null;
 }
 
+/**
+ * Pre-parse files missing `_tree` via WASM so downstream phases (CFG, dataflow)
+ * don't each need to create parsers and re-parse independently.
+ * Only parses files whose extension is in SUPPORTED_EXTENSIONS.
+ *
+ * @param {Map<string, object>} fileSymbols - Map<relPath, { definitions, _tree, _langId, ... }>
+ * @param {string} rootDir - absolute project root
+ */
+export async function ensureWasmTrees(fileSymbols, rootDir) {
+  // Check if any file needs a tree
+  let needsParse = false;
+  for (const [relPath, symbols] of fileSymbols) {
+    if (!symbols._tree) {
+      const ext = path.extname(relPath).toLowerCase();
+      if (_extToLang.has(ext)) {
+        needsParse = true;
+        break;
+      }
+    }
+  }
+  if (!needsParse) return;
+
+  const parsers = await createParsers();
+
+  for (const [relPath, symbols] of fileSymbols) {
+    if (symbols._tree) continue;
+    const ext = path.extname(relPath).toLowerCase();
+    const entry = _extToLang.get(ext);
+    if (!entry) continue;
+    const parser = parsers.get(entry.id);
+    if (!parser) continue;
+
+    const absPath = path.join(rootDir, relPath);
+    let code;
+    try {
+      code = fs.readFileSync(absPath, 'utf-8');
+    } catch {
+      continue;
+    }
+    try {
+      symbols._tree = parser.parse(code);
+      symbols._langId = entry.id;
+    } catch {
+      // skip files that fail to parse
+    }
+  }
+}
+
 /**
  * Check whether the required WASM grammar files exist on disk.
  */