fix(codemods): SIGSEV segfault problem. reduce amount of parse() calls. Preserve ast root refe…

packages/codemods/package.json

@@ -37,7 +37,7 @@
   "author": "",
   "license": "MIT",
   "dependencies": {
-    "@ast-grep/napi": "^0.28.0",
+    "@ast-grep/napi": "^0.42.0",
     "chalk": "^5.3.0",
     "commander": "^14.0.0",
     "glob": "^11.0.0",
@@ -60,4 +60,4 @@
   "volta": {
     "extends": "../../package.json"
   }
-}
+}

packages/codemods/src/cli/index.ts

@@ -19,3 +19,4 @@ createListCommand(program);
 program.showHelpAfterError();
 
 await program.parseAsync(process.argv);
+await new Promise((resolve) => setTimeout(resolve, 5000));

packages/codemods/src/schema-migration/processors/model.ts

@@ -1,4 +1,4 @@
-import { parse, type SgNode } from '@ast-grep/napi';
+import type { SgNode } from '@ast-grep/napi';
 import { existsSync, mkdirSync, readFileSync, writeFileSync } from 'fs';
 import { dirname, join, resolve } from 'path';
 
@@ -20,7 +20,6 @@ import {
   FRAGMENT_BASE_SOURCE,
   generateMergedSchemaCode,
   getEmberDataImports,
-  getLanguageFromPath,
   getMixinImports,
   getModelImportSources,
   isModelFile,
@@ -55,10 +54,7 @@ import { normalizePath, removeFileExtension, toKebabCase } from '../utils/string
  * for each identifier in `identifierRefs`. These are included verbatim in the schema file
  * so that schema field options can reference them by name.
  */
-function collectConstantDecls(filePath: string, source: string, identifierRefs: Set<string>): string {
-  const lang = getLanguageFromPath(filePath);
-  const ast = parse(lang, source);
-  const root = ast.root();
+function collectConstantDecls(root: SgNode, identifierRefs: Set<string>): string {
   const declarations: string[] = [];
 
   for (const exportStmt of root.findAll({ rule: { kind: 'export_statement' } })) {
@@ -140,14 +136,10 @@ interface ModelASTValidation {
 }
 
 /**
- * Validate model AST: parse source, find imports, validate default export, check model class and fragment status.
+ * Validate model AST: find imports, validate default export, check model class and fragment status.
  * Returns null if the file is not a valid model.
  */
-function validateModelAST(filePath: string, source: string, options: TransformOptions): ModelASTValidation | null {
-  const lang = getLanguageFromPath(filePath);
-  const ast = parse(lang, source);
-  const root = ast.root();
-
+function validateModelAST(root: SgNode, filePath: string, options: TransformOptions): ModelASTValidation | null {
   const expectedSources = getModelImportSources(options);
   const modelImportLocal = findEmberImportLocalName(root, expectedSources, options, filePath, process.cwd());
   log.debug(`DEBUG: Model import local: ${modelImportLocal}`);
@@ -537,7 +529,7 @@ function generateRegularModelArtifacts(
     }
   }
   const constantDeclarations =
-    identifierRefs.size > 0 ? collectConstantDecls(filePath, source, identifierRefs) : undefined;
+    identifierRefs.size > 0 ? collectConstantDecls(entity.parsedFile.root, identifierRefs) : undefined;
 
   // Generate merged schema code (schema + types in one file)
   const mergedSchemaCode = generateMergedSchemaCode({
@@ -749,7 +741,7 @@ function analyzeModelFromParsed(
   const baseName = parsedFile.baseName;
 
   try {
-    const validation = validateModelAST(filePath, parsedFile.source, options);
+    const validation = validateModelAST(parsedFile.root, filePath, options);
     if (!validation) {
       return createInvalidResult(modelName, baseName);
     }

packages/codemods/src/schema-migration/utils/extension-generation.ts

@@ -40,49 +40,29 @@ function getImportLocalNames(importNode: SgNode): string[] {
   return names.filter(Boolean);
 }
 
+const NOT_INSIDE_IMPORT = { not: { inside: { kind: 'import_statement', stopBy: 'end' } } } as const;
+
 /**
- * Check whether any of the given names appear as identifiers in the non-import
- * portion of the source. Checks both value identifiers and type identifiers
+ * Check whether any of the given names appear as identifiers outside of
+ * import statements. Checks both value identifiers and type identifiers
  * (e.g., in `extends Foo` within interface declarations) to avoid false removals.
  */
-function areAnyNamesUsed(names: string[], nonImportRoot: SgNode, lang: Lang): boolean {
+function areAnyNamesUsed(names: string[], root: SgNode, lang: Lang): boolean {
   const nameSet = new Set(names);
 
   // In TypeScript, interface extends clauses produce `type_identifier` nodes
   // rather than `identifier`, so we must check both kinds.
   // JavaScript grammars don't have `type_identifier`.
-  const identifiers = nonImportRoot.findAll({ rule: { kind: 'identifier' } });
-  if (identifiers.some((id) => nameSet.has(id.text()))) return true;
-
-  if (lang === AstLang.TypeScript) {
-    const typeIdentifiers = nonImportRoot.findAll({ rule: { kind: 'type_identifier' } });
-    if (typeIdentifiers.some((id) => nameSet.has(id.text()))) return true;
-  }
-
-  return false;
+  const kinds: string[] = lang === AstLang.TypeScript ? ['identifier', 'type_identifier'] : ['identifier'];
+  return kinds.some((kind) =>
+    root.findAll({ rule: { kind, ...NOT_INSIDE_IMPORT } }).some((id) => nameSet.has(id.text()))
+  );
 }
 
-/**
- * Shared setup for unused import removal: parse the source, split into
- * import nodes and a non-import AST root for usage checking.
- */
-function prepareUnusedImportRemoval(source: string, lang: Lang) {
+function removeUnusedImports(source: string, lang: Lang): string {
   const ast = parse(lang, source);
   const root = ast.root();
   const importNodes = root.findAll({ rule: { kind: 'import_statement' } });
-
-  // Build source without imports so identifier lookups only check usage sites
-  let nonImportSource = source;
-  for (const imp of importNodes) {
-    nonImportSource = nonImportSource.replaceAll(imp.text(), '');
-  }
-  const nonImportRoot = parse(lang, nonImportSource).root();
-
-  return { root, importNodes, nonImportRoot };
-}
-
-function removeUnusedImports(source: string, lang: Lang): string {
-  const { root, importNodes, nonImportRoot } = prepareUnusedImportRemoval(source, lang);
   if (importNodes.length === 0) return source;
 
   type Edit = ReturnType<SgNode['replace']>;
@@ -92,7 +72,7 @@ function removeUnusedImports(source: string, lang: Lang): string {
     const localNames = getImportLocalNames(imp);
     if (localNames.length === 0) continue;
 
-    if (!areAnyNamesUsed(localNames, nonImportRoot, lang)) {
+    if (!areAnyNamesUsed(localNames, root, lang)) {
       edits.push(imp.replace(''));
     }
   }
@@ -101,7 +81,9 @@ function removeUnusedImports(source: string, lang: Lang): string {
 }
 
 function removeUnusedTypeImports(source: string, lang: Lang): string {
-  const { root, importNodes, nonImportRoot } = prepareUnusedImportRemoval(source, lang);
+  const ast = parse(lang, source);
+  const root = ast.root();
+  const importNodes = root.findAll({ rule: { kind: 'import_statement' } });
   if (importNodes.length === 0) return source;
 
   type Edit = ReturnType<SgNode['replace']>;
@@ -115,7 +97,7 @@ function removeUnusedTypeImports(source: string, lang: Lang): string {
       const localNames = getImportLocalNames(imp);
       if (localNames.length === 0) continue;
 
-      if (!areAnyNamesUsed(localNames, nonImportRoot, lang)) {
+      if (!areAnyNamesUsed(localNames, root, lang)) {
         edits.push(imp.replace(''));
       }
       continue;
@@ -138,7 +120,7 @@ function removeUnusedTypeImports(source: string, lang: Lang): string {
 
       const localName = (specifier.field('alias') ?? specifier.field('name'))?.text();
 
-      if (localName && !areAnyNamesUsed([localName], nonImportRoot, lang)) {
+      if (localName && !areAnyNamesUsed([localName], root, lang)) {
         hasChanges = true;
       } else {
         keptSpecifiers.push(specText);

packages/codemods/src/schema-migration/utils/file-parser.ts

@@ -130,6 +130,9 @@ export interface ParsedFile {
   extension: '.ts' | '.js';
   /** Whether the file should be treated as TypeScript (extension is .ts or forceTypeScript is enabled) */
   readonly isTypeScript: boolean;
+  /** Parsed AST — kept alive so downstream consumers can reuse it instead of re-parsing */
+  ast: ReturnType<typeof parse>;
+  root: SgNode;
   /** All imports in the file */
   imports: ParsedFileImport[];
   /** EmberData schema fields (@attr, @hasMany, @belongsTo, fragment, etc.) */
@@ -878,6 +881,8 @@ export function parseFile(filePath: string, code: string, options: TransformOpti
     get isTypeScript() {
       return this.extension === '.ts' || forceTypeScript;
     },
+    ast,
+    root,
     imports,
     fields,
     behaviors,