@codeyam/codeyam-cli 0.1.0-staging.6e699e5 → 0.1.0-staging.7481014

package/analyzer-template/packages/ai/src/lib/worker/analyzeScopeWorker.ts

@@ -9,13 +9,14 @@ import { Entity } from '~codeyam/types';
 import { FileAnalyzer, ProjectAnalyzer } from '~codeyam/analyze';
 import { analyzeScopeLocal } from '../analyzeScope';
 import {
-  ScopeInfo,
   resetScopeDataStructureMetrics,
+  ScopeInfo,
 } from '../dataStructure/ScopeDataStructure';
 import { SerializableDataStructure } from './SerializableDataStructure';
 import { AI } from '~codeyam/ai';
 import { setThreadContext } from '~codeyam/utils';
 import { threadId } from 'worker_threads';
+import v8 from 'v8';
 
 // Set thread context for logging
 setThreadContext(`<worker-${threadId}>`);
@@ -45,6 +46,80 @@ let projectAnalyzerKey: string | null = null;
 let fileAnalyzer: FileAnalyzer | null = null;
 let fileAnalyzerPath: string | null = null;
 
+// Memory tracking
+let entitiesProcessed = 0;
+let lastMemoryLogAt = 0;
+const MEMORY_LOG_INTERVAL = 100; // Log every N entities
+const MEMORY_LOG_TIME_INTERVAL = 60_000; // Or every 60 seconds
+
+/**
+ * Get memory stats for logging
+ */
+function getWorkerMemoryStats() {
+  const mem = process.memoryUsage();
+  const heapStats = v8.getHeapStatistics();
+  return {
+    rss: `${Math.round(mem.rss / 1024 / 1024)}MB`,
+    heapUsed: `${Math.round(mem.heapUsed / 1024 / 1024)}MB`,
+    heapTotal: `${Math.round(mem.heapTotal / 1024 / 1024)}MB`,
+    external: `${Math.round(mem.external / 1024 / 1024)}MB`,
+    // V8 heap stats give more detail
+    v8TotalHeapSize: `${Math.round(heapStats.total_heap_size / 1024 / 1024)}MB`,
+    v8UsedHeapSize: `${Math.round(heapStats.used_heap_size / 1024 / 1024)}MB`,
+    v8HeapSizeLimit: `${Math.round(heapStats.heap_size_limit / 1024 / 1024)}MB`,
+    v8MallocedMemory: `${Math.round(heapStats.malloced_memory / 1024 / 1024)}MB`,
+  };
+}
+
+/**
+ * Get ts.Program stats for understanding TypeScript memory usage
+ */
+function getTsProgramStats(pa: ProjectAnalyzer) {
+  const sourceFiles = pa.program.getSourceFiles();
+  const projectSourceFiles = sourceFiles.filter(
+    (sf) => !sf.fileName.includes('node_modules'),
+  );
+  const nodeModulesFiles = sourceFiles.filter((sf) =>
+    sf.fileName.includes('node_modules'),
+  );
+
+  // Estimate AST size by sampling text length (rough proxy)
+  let totalProjectTextLength = 0;
+  for (const sf of projectSourceFiles) {
+    totalProjectTextLength += sf.getFullText().length;
+  }
+
+  return {
+    totalSourceFiles: sourceFiles.length,
+    projectSourceFiles: projectSourceFiles.length,
+    nodeModulesFiles: nodeModulesFiles.length,
+    projectTextSizeMB: `${(totalProjectTextLength / 1024 / 1024).toFixed(1)}MB`,
+    // Note: Actual AST memory is typically 10-20x the source text size
+    estimatedAstSizeMB: `${((totalProjectTextLength * 15) / 1024 / 1024).toFixed(0)}MB`,
+  };
+}
+
+/**
+ * Write a heap snapshot for detailed profiling.
+ * The snapshot can be loaded in Chrome DevTools for analysis.
+ *
+ * TODO: Once we understand memory usage, change this back to env-var controlled
+ */
+const HEAP_SNAPSHOT_MODE: 'off' | 'once' | 'periodic' = 'off';
+
+function maybeWriteHeapSnapshot(label: string) {
+  if (HEAP_SNAPSHOT_MODE === 'off') return;
+
+  try {
+    const filename = v8.writeHeapSnapshot();
+    console.log(
+      `CodeYam Worker: Heap snapshot written (${label}): ${filename}`,
+    );
+  } catch (error) {
+    console.error('CodeYam Worker: Failed to write heap snapshot:', error);
+  }
+}
+
 /**
  * Main worker function called by Piscina
  */
@@ -61,8 +136,25 @@ export default async function runAnalyzeScope(
     // Create or reuse ProjectAnalyzer
     const paKey = `${dirPath}`;
     if (!projectAnalyzer || projectAnalyzerKey !== paKey) {
-      projectAnalyzer = await ProjectAnalyzer.from({ project, dirPath });
+      const memBefore = process.memoryUsage();
+      console.log('CodeYam Worker: Creating ProjectAnalyzer (ts.Program)...', {
+        memBefore: getWorkerMemoryStats(),
+      });
+
+      projectAnalyzer = ProjectAnalyzer.from({ project, dirPath });
       projectAnalyzerKey = paKey;
+
+      const memAfter = process.memoryUsage();
+      const tsProgramStats = getTsProgramStats(projectAnalyzer);
+      console.log('CodeYam Worker: ProjectAnalyzer created', {
+        memAfter: getWorkerMemoryStats(),
+        rssDelta: `+${Math.round((memAfter.rss - memBefore.rss) / 1024 / 1024)}MB`,
+        heapDelta: `+${Math.round((memAfter.heapUsed - memBefore.heapUsed) / 1024 / 1024)}MB`,
+        tsProgramStats,
+      });
+
+      // Optionally write heap snapshot after ts.Program creation
+      maybeWriteHeapSnapshot('after-ts-program');
     }
 
     // Create or reuse FileAnalyzer
@@ -95,6 +187,26 @@ export default async function runAnalyzeScope(
       );
     }
 
+    // Periodic memory logging
+    entitiesProcessed++;
+    const now = Date.now();
+    const shouldLogMemory =
+      entitiesProcessed % MEMORY_LOG_INTERVAL === 0 ||
+      now - lastMemoryLogAt > MEMORY_LOG_TIME_INTERVAL;
+
+    if (shouldLogMemory) {
+      lastMemoryLogAt = now;
+      console.log('CodeYam Worker: Memory stats', {
+        entitiesProcessed,
+        ...getWorkerMemoryStats(),
+      });
+
+      // Optionally write periodic heap snapshots
+      if (HEAP_SNAPSHOT_MODE === 'periodic' && entitiesProcessed % 500 === 0) {
+        maybeWriteHeapSnapshot(`after-${entitiesProcessed}-entities`);
+      }
+    }
+
     return { dataStructure: dataStructure.toSerializable() };
   } catch (error) {
     console.error(

package/analyzer-template/packages/analyze/index.ts

@@ -36,11 +36,16 @@ export { getAllEntityNodes } from './src/lib/asts/sourceFiles/getAllEntityNodes'
 export { default as fileAnalyzerFromCode } from './src/lib/files/fileAnalyzerFromCode';
 
 export { default as isolateDataStructure } from './src/lib/files/scenarios/isolateDataStructure';
-export { default as mergeInDependentDataStructure } from './src/lib/files/scenarios/mergeInDependentDataStructure';
+export {
+  default as mergeInDependentDataStructure,
+  DataStructureTimeoutError,
+} from './src/lib/files/scenarios/mergeInDependentDataStructure';
 export { default as mergeValidatedDataStructures } from './src/lib/files/scenarios/mergeValidatedDataStructures';
 
 export { discoverDirectDependencies } from './src/lib/files/analyze/dependencyResolver';
 
+export { transformationTracer } from './src/lib/files/scenarios/TransformationTracer';
+
 export { default as getAnalysisError } from './src/lib/utils/getAnalysisError';
 export { default as measureAndReportExecutionTime } from './src/lib/utils/measureAndReportExecutionTime';
 

package/analyzer-template/packages/analyze/src/lib/FileAnalyzer.ts

@@ -9,6 +9,7 @@ import {
 import { FunctionNode } from './types';
 import getEntityType from './files/getEntityType';
 import getEntityCode from './files/getEntityCode';
+import { getFileByPathSafe } from './utils/getFileByPath';
 
 export class FileAnalyzer {
   public projectAnalyzer: lib.ProjectAnalyzer;
@@ -19,6 +20,7 @@ export class FileAnalyzer {
   // Current per-file caching helps when multiple entities in the same file call getResolvedImportsRelevantToFunctionName,
   // but a project-level cache would also benefit the recursive import chain traversals that cross file boundaries.
   private resolvedImportsCache: Map<string, FunctionDependenciesMap>;
+  private allEntityNodesOnlyAliasesCache?: Record<string, ts.Node>;
 
   constructor({
     projectAnalyzer,
@@ -62,18 +64,23 @@ export class FileAnalyzer {
 
   getAllEntityNodesOnlyAliases() {
     if (!this.sourceFile) return {};
+    if (this.allEntityNodesOnlyAliasesCache) {
+      return this.allEntityNodesOnlyAliasesCache;
+    }
     try {
       const allEntityNodes = lib.asts.sourceFiles.getAllEntityNodes({
         sourceFile: this.sourceFile,
       });
-      if (allEntityNodes.default) {
-        const namedDefault = Object.entries(allEntityNodes).find(
+      const aliasesOnly = { ...allEntityNodes };
+      if (aliasesOnly.default) {
+        const namedDefault = Object.entries(aliasesOnly).find(
           ([otherEntityName, node]) =>
-            otherEntityName !== 'default' && node === allEntityNodes.default,
+            otherEntityName !== 'default' && node === aliasesOnly.default,
         );
-        if (namedDefault) delete allEntityNodes.default;
+        if (namedDefault) delete aliasesOnly.default;
       }
-      return allEntityNodes;
+      this.allEntityNodesOnlyAliasesCache = aliasesOnly;
+      return aliasesOnly;
     } catch (error) {
       console.log('CodeYam ERROR: getAllEntityNodesOnlyAliases failed:', {
         filePath: this.file?.path,
@@ -223,15 +230,11 @@ export class FileAnalyzer {
 
       // Handle dist -> src mapping for workspace packages
       let targetPath = resolvedPath;
-      let targetFile = this.projectAnalyzer.project.files.find(
-        (f) => f.path === targetPath,
-      );
+      let targetFile = getFileByPathSafe(this.projectAnalyzer, targetPath);
 
       if (!targetFile && targetPath.includes('/dist/')) {
         const srcPath = targetPath.replace('/dist/', '/src/');
-        targetFile = this.projectAnalyzer.project.files.find(
-          (f) => f.path === srcPath,
-        );
+        targetFile = getFileByPathSafe(this.projectAnalyzer, srcPath);
         if (targetFile) {
           targetPath = srcPath;
         }
@@ -239,9 +242,9 @@ export class FileAnalyzer {
         // Try with different extensions if still not found
         if (!targetFile) {
           const basePath = srcPath.replace(/\.d\.ts$/, '').replace(/\.js$/, '');
-          targetFile = this.projectAnalyzer.project.files.find(
-            (f) => f.path === basePath + '.ts' || f.path === basePath + '.tsx',
-          );
+          targetFile =
+            getFileByPathSafe(this.projectAnalyzer, basePath + '.ts') ??
+            getFileByPathSafe(this.projectAnalyzer, basePath + '.tsx');
           if (targetFile) {
             targetPath = targetFile.path;
           }
@@ -301,24 +304,25 @@ export class FileAnalyzer {
 
   getLocalEntitiesUsedInFunction(functionName: string) {
     if (!this.sourceFile) return [];
-    const entityNodes = lib.asts.sourceFiles.getAllEntityNodes({
-      sourceFile: this.sourceFile,
-    });
+    const entityNodes = this.getAllEntityNodes();
 
     // Defensive check: ensure the function exists in entityNodes
-    if (!entityNodes[functionName]) {
+    const functionNode = entityNodes[functionName];
+    if (!functionNode) {
       return [];
     }
 
     try {
-      const identifiers = this.getIdentifiersInNode(entityNodes[functionName]);
+      const identifiers = this.getIdentifiersInNode(functionNode);
       return identifiers
-        .filter(
-          (identifier, index, self) =>
-            entityNodes[identifier] !== entityNodes[functionName] &&
-            Object.keys(entityNodes).includes(identifier) &&
-            self.indexOf(identifier) === index,
-        )
+        .filter((identifier) => {
+          const node = entityNodes[identifier];
+          return (
+            !!node &&
+            node !== functionNode &&
+            typeof (node as ts.Node).kind === 'number'
+          );
+        })
         .map((identifier) => ({
           name: identifier,
           node: entityNodes[identifier],
@@ -348,14 +352,13 @@ export class FileAnalyzer {
       sourceFile: this.sourceFile,
     });
     const identifiersInRelevantNode = this.getIdentifiersInNode(relevantNode);
+    const identifiersSet = new Set(identifiersInRelevantNode);
 
     const entries = Object.entries(importsAnalysis)
       .map(([importedFilePath, importedExpressions]) => {
         const relevantExpressions = importedExpressions.filter(
           (importedExpression) => {
-            return identifiersInRelevantNode.includes(
-              importedExpression.importAlias,
-            );
+            return identifiersSet.has(importedExpression.importAlias);
           },
         );
 
@@ -439,8 +442,9 @@ export class FileAnalyzer {
             resolvedImport.resolvedFileName,
           );
 
-          let importedFile = this.projectAnalyzer.project.files.find(
-            (f) => f.path === projectPath,
+          let importedFile = getFileByPathSafe(
+            this.projectAnalyzer,
+            projectPath,
           );
 
           // Workspace packages often resolve to dist/ paths (from package.json "types" or "main"),
@@ -448,9 +452,7 @@ export class FileAnalyzer {
           if (!importedFile && projectPath.includes('/dist/')) {
             const srcPath = projectPath.replace('/dist/', '/src/');
             // Try exact match first (e.g., dist/index.d.ts -> src/index.ts)
-            importedFile = this.projectAnalyzer.project.files.find(
-              (f) => f.path === srcPath,
-            );
+            importedFile = getFileByPathSafe(this.projectAnalyzer, srcPath);
             if (importedFile) {
               projectPath = srcPath;
             }
@@ -460,10 +462,9 @@ export class FileAnalyzer {
               const basePath = srcPath
                 .replace(/\.d\.ts$/, '')
                 .replace(/\.js$/, '');
-              importedFile = this.projectAnalyzer.project.files.find(
-                (f) =>
-                  f.path === basePath + '.ts' || f.path === basePath + '.tsx',
-              );
+              importedFile =
+                getFileByPathSafe(this.projectAnalyzer, basePath + '.ts') ??
+                getFileByPathSafe(this.projectAnalyzer, basePath + '.tsx');
               if (importedFile) {
                 projectPath = importedFile.path;
               }
@@ -591,24 +592,35 @@ export class FileAnalyzer {
 
   getNamespaceIdentifiersInNode(node: ts.Node, namespaceIdentifier: string) {
     const identifiers: string[] = [];
+    const seen = new Set<string>();
 
     const identifierVisitor = (child: ts.Node) => {
       if (ts.isPropertyAccessExpression(child)) {
         if (child.expression.getText() === namespaceIdentifier) {
-          identifiers.push(child.name.getText());
+          const name = child.name.getText();
+          if (!seen.has(name)) {
+            seen.add(name);
+            identifiers.push(name);
+          }
         }
       }
       ts.forEachChild(child, identifierVisitor);
     };
 
     ts.forEachChild(node, identifierVisitor);
-    return identifiers.filter(
-      (identifier, index, self) => self.indexOf(identifier) === index,
-    );
+    return identifiers;
   }
 
   getIdentifiersInNode(node: ts.Node): string[] {
     const identifiers: string[] = [];
+    const seen = new Set<string>();
+
+    const addIdentifier = (text: string, allowDefault = false) => {
+      if (!allowDefault && text === 'default') return;
+      if (seen.has(text)) return;
+      seen.add(text);
+      identifiers.push(text);
+    };
 
     // Helper to safely get text from an identifier or string literal
     const getNodeText = (node: ts.Identifier | ts.StringLiteral): string => {
@@ -637,63 +649,57 @@ export class FileAnalyzer {
     };
 
     if (ts.isIdentifier(node)) {
-      identifiers.push(getNodeText(node));
+      addIdentifier(getNodeText(node));
     }
 
     const identifierVisitor = (child: ts.Node) => {
       if (ts.isIdentifier(child)) {
-        const text = getNodeText(child);
-        if (text === 'default') return;
-        identifiers.push(text);
+        addIdentifier(getNodeText(child));
       } else if (ts.isExportDeclaration(child)) {
         if (child.exportClause && ts.isNamedExports(child.exportClause)) {
           child.exportClause.elements.forEach((element) => {
-            identifiers.push(getNodeText(element.name));
+            addIdentifier(getNodeText(element.name), true);
           });
         } else if (
           child.exportClause &&
           ts.isNamespaceExport(child.exportClause)
         ) {
-          identifiers.push(getNodeText(child.exportClause.name));
+          addIdentifier(getNodeText(child.exportClause.name));
         }
       } else if (ts.isExportAssignment(child)) {
         if (ts.isIdentifier(child.expression)) {
-          identifiers.push(getNodeText(child.expression));
+          addIdentifier(getNodeText(child.expression));
         }
       } else if (ts.isExportSpecifier(child)) {
-        identifiers.push(getNodeText(child.name));
+        addIdentifier(getNodeText(child.name), true);
         return;
       } else if (ts.isFunctionDeclaration(child) && child.name) {
-        identifiers.push(getNodeText(child.name));
+        addIdentifier(getNodeText(child.name));
       } else if (ts.isVariableStatement(child)) {
         child.declarationList.declarations.forEach((declaration) => {
           if (ts.isIdentifier(declaration.name)) {
-            identifiers.push(getNodeText(declaration.name));
+            addIdentifier(getNodeText(declaration.name));
           }
         });
       } else if (ts.isImportDeclaration(child)) {
         if (child.importClause && child.importClause.namedBindings) {
           if (ts.isNamespaceImport(child.importClause.namedBindings)) {
-            identifiers.push(
-              getNodeText(child.importClause.namedBindings.name),
-            );
+            addIdentifier(getNodeText(child.importClause.namedBindings.name));
           } else if (ts.isNamedImports(child.importClause.namedBindings)) {
             child.importClause.namedBindings.elements.forEach((element) => {
-              identifiers.push(getNodeText(element.name));
+              addIdentifier(getNodeText(element.name));
             });
           }
         }
       } else if (ts.isImportSpecifier(child)) {
-        identifiers.push(getNodeText(child.name));
+        addIdentifier(getNodeText(child.name), true);
         return;
       }
       ts.forEachChild(child, identifierVisitor);
     };
 
     ts.forEachChild(node, identifierVisitor);
-    return identifiers.filter(
-      (identifier, index, self) => self.indexOf(identifier) === index,
-    );
+    return identifiers;
   }
 
   resolveNamespaceImport(

package/analyzer-template/packages/analyze/src/lib/ProjectAnalyzer.ts

@@ -13,6 +13,8 @@ export class ProjectAnalyzer {
   programs: ts.Program[]; // All programs from all tsconfigs
   typeChecker: ts.TypeChecker;
   private filesByPath: Map<string, File>; // O(1) lookup cache for files
+  private sourceFileByPath: Map<string, ts.SourceFile>; // O(1) lookup cache for source files
+  private typeCheckerByPath: Map<string, ts.TypeChecker>; // O(1) lookup cache for type checkers by file
 
   private constructor({
     project,
@@ -44,9 +46,20 @@ export class ProjectAnalyzer {
       lib.asts.createProgramFromTsConfigPath(configPath, excludePatterns),
     );
 
+    if (this.programs.length === 0) {
+      throw new Error(
+        'No TypeScript programs could be created — no tsconfig.json found',
+      );
+    }
+
     // Primary program is the first one (usually root) for backward compatibility
     this.program = this.programs[0];
     this.typeChecker = this.program.getTypeChecker();
+
+    // Build source file and type checker indexes for fast lookups
+    this.sourceFileByPath = new Map();
+    this.typeCheckerByPath = new Map();
+    this.buildSourceFileIndex();
   }
 
   static from({
@@ -73,32 +86,7 @@ export class ProjectAnalyzer {
   }
 
   getFileAnalyzerByPath(filePath: string) {
-    if (this.project.files?.length !== this.filesByPath.size) {
-      // Rebuild the cache if project.files has changed
-      this.filesByPath.clear();
-      for (const file of this.project.files) {
-        this.filesByPath.set(file.path, file);
-      }
-    }
-
-    // Convert absolute path to relative if needed
-    let relativePath = filePath;
-    if (filePath.startsWith(this.dirPath + '/')) {
-      relativePath = this.getRelativePath(filePath);
-    } else if (filePath.startsWith('/')) {
-      // Might be an absolute path with a different prefix (e.g., symlink)
-      // Try to normalize it to match our dirPath
-      try {
-        const normalizedPath = realpathSync(filePath);
-        if (normalizedPath.startsWith(this.dirPath + '/')) {
-          relativePath = normalizedPath.replace(this.dirPath + '/', '');
-        }
-      } catch (error) {
-        // If normalization fails, keep the original path
-      }
-    }
-
-    const file = this.filesByPath.get(relativePath);
+    const file = this.getFileByPath(filePath);
     if (!file) {
       // Technically FileAnalyzer was built to accept null file, but most of
       // the code we've written since assumes fileAnalyzer.sourceFile exists,
@@ -109,6 +97,33 @@ export class ProjectAnalyzer {
     return new lib.FileAnalyzer({ projectAnalyzer: this, file });
   }
 
+  getFileByPath(filePath: string): File | undefined {
+    this.ensureFileCache();
+
+    const candidates = new Set<string>();
+    if (filePath) {
+      candidates.add(filePath);
+    }
+
+    if (!filePath.startsWith('/')) {
+      candidates.add(filePath);
+    } else if (filePath.startsWith(this.dirPath + '/')) {
+      candidates.add(filePath.slice(this.dirPath.length + 1));
+    }
+
+    const relativePath = this.toProjectRelativePath(filePath);
+    if (relativePath) {
+      candidates.add(relativePath);
+    }
+
+    for (const candidate of candidates) {
+      const file = this.filesByPath.get(candidate);
+      if (file) return file;
+    }
+
+    return undefined;
+  }
+
   getFileAnalyzer(file: File) {
     this.filesByPath.set(file.path, file);
     return new lib.FileAnalyzer({ projectAnalyzer: this, file });
@@ -142,6 +157,16 @@ export class ProjectAnalyzer {
 
   getSourceFile(filePath: string): ts.SourceFile | undefined {
     const absolutePath = this.getAbsolutePath(filePath);
+    const relativePath = filePath.startsWith('/')
+      ? this.toProjectRelativePath(filePath)
+      : filePath;
+
+    if (relativePath) {
+      const cached = this.sourceFileByPath.get(relativePath);
+      if (cached) {
+        return cached;
+      }
+    }
 
     // Search across all programs
     for (const program of this.programs) {
@@ -241,6 +266,16 @@ export class ProjectAnalyzer {
    */
   getTypeCheckerForFile(filePath: string): ts.TypeChecker {
     const absolutePath = this.getAbsolutePath(filePath);
+    const relativePath = filePath.startsWith('/')
+      ? this.toProjectRelativePath(filePath)
+      : filePath;
+
+    if (relativePath) {
+      const cachedChecker = this.typeCheckerByPath.get(relativePath);
+      if (cachedChecker) {
+        return cachedChecker;
+      }
+    }
 
     // Search for the file in each program and return that program's typeChecker
     for (const program of this.programs) {
@@ -305,20 +340,84 @@ export class ProjectAnalyzer {
    * Refresh the TypeScript program to pick up file changes
    */
   refreshProgram(): void {
-    // Memory logging: Capture state before refresh
-    const memBefore = process.memoryUsage();
-
     // Apply the same unapprovedPaths filtering when refreshing
     const excludePatterns =
       this.project.metadata?.unapprovedPaths?.filter(Boolean) || [];
 
-    // Refresh all programs
-    this.programs = this.tsConfigPaths.map((configPath) =>
-      lib.asts.createProgramFromTsConfigPath(configPath, excludePatterns),
-    );
+    // Release old programs and caches BEFORE creating new ones to reduce peak memory.
+    // Without this, both old (~1.7GB) and new (~1.7GB) programs coexist during creation,
+    // causing OOM on monorepos with multiple tsconfigs.
+    this.programs = [];
+    this.sourceFileByPath.clear();
+    this.typeCheckerByPath.clear();
+
+    // Create programs one at a time so each old program can be GC'd before the next
+    for (const configPath of this.tsConfigPaths) {
+      this.programs.push(
+        lib.asts.createProgramFromTsConfigPath(configPath, excludePatterns),
+      );
+    }
 
     // Update primary program for backward compatibility
     this.program = this.programs[0];
     this.typeChecker = this.program.getTypeChecker();
+
+    this.buildSourceFileIndex();
+  }
+
+  private ensureFileCache(): void {
+    if (this.project.files?.length !== this.filesByPath.size) {
+      // Rebuild the cache if project.files has changed
+      this.filesByPath.clear();
+      for (const file of this.project.files) {
+        this.filesByPath.set(file.path, file);
+      }
+    }
+  }
+
+  private toProjectRelativePath(filePath: string): string | null {
+    if (!filePath) return null;
+
+    if (filePath.startsWith(this.dirPath + '/')) {
+      return filePath.replace(this.dirPath + '/', '');
+    }
+
+    if (!filePath.startsWith('/')) {
+      return null;
+    }
+
+    try {
+      const normalizedPath = realpathSync(filePath);
+      if (normalizedPath.startsWith(this.dirPath + '/')) {
+        return normalizedPath.replace(this.dirPath + '/', '');
+      }
+    } catch (error) {
+      // If normalization fails, return null
+    }
+
+    return null;
+  }
+
+  private buildSourceFileIndex(): void {
+    this.sourceFileByPath.clear();
+    this.typeCheckerByPath.clear();
+
+    for (const program of this.programs) {
+      const checker = program.getTypeChecker();
+      for (const sf of program.getSourceFiles()) {
+        // Skip node_modules files to keep the index small
+        if (sf.fileName.includes('node_modules')) {
+          continue;
+        }
+
+        const relativePath = this.toProjectRelativePath(sf.fileName);
+        if (!relativePath) continue;
+
+        if (!this.sourceFileByPath.has(relativePath)) {
+          this.sourceFileByPath.set(relativePath, sf);
+          this.typeCheckerByPath.set(relativePath, checker);
+        }
+      }
+    }
   }
 }

package/analyzer-template/packages/analyze/src/lib/asts/index.ts

@@ -95,8 +95,13 @@ export function createProgramFromTsConfigPath(
     throw new Error('Could not parse tsconfig.json');
   }
 
-  // Filter file names based on exclude patterns to reduce memory usage
-  let fileNames = parsedCommandLine.fileNames;
+  // Always filter out node_modules from root files to prevent OOM on large projects.
+  // This only removes them as root compilation targets — TypeScript's module resolution
+  // still follows imports into node_modules for type info when processing source files.
+  let fileNames = parsedCommandLine.fileNames.filter(
+    (f) => !f.includes('/node_modules/'),
+  );
+
   if (excludePatterns && excludePatterns.length > 0) {
     const regexPatterns = excludePatterns
       .map((pattern) => {