[SMO-522] Fix heap overflow in large monorepo scans

src/utils/glob.mts

@@ -164,6 +164,14 @@ export function filterBySupportedScanFiles(
   return filepaths.filter(p => micromatch.some(p, patterns, { dot: true }))
 }
 
+export function createSupportedFilesFilter(
+  supportedFiles: SocketSdkSuccessResult<'getReportSupportedFiles'>['data'],
+): (filepath: string) => boolean {
+  const patterns = getSupportedFilePatterns(supportedFiles)
+  return (filepath: string) =>
+    micromatch.some(filepath, patterns, { dot: true })
+}
+
 export function getSupportedFilePatterns(
   supportedFiles: SocketSdkSuccessResult<'getReportSupportedFiles'>['data'],
 ): string[] {
@@ -178,6 +186,10 @@ export function getSupportedFilePatterns(
 }
 
 type GlobWithGitIgnoreOptions = GlobOptions & {
+  // Optional filter function to apply during streaming.
+  // When provided, only files passing this filter are accumulated.
+  // This is critical for memory efficiency when scanning large monorepos.
+  filter?: ((filepath: string) => boolean) | undefined
   socketConfig?: SocketYml | undefined
 }
 
@@ -187,6 +199,7 @@ export async function globWithGitIgnore(
 ): Promise<string[]> {
   const {
     cwd = process.cwd(),
+    filter,
     socketConfig,
     ...additionalOptions
   } = { __proto__: null, ...options } as GlobWithGitIgnoreOptions
@@ -243,27 +256,39 @@ export async function globWithGitIgnore(
     ...additionalOptions,
   } as GlobOptions
 
-  if (!hasNegatedPattern) {
+  // When no filter is provided and no negated patterns exist, use the fast path.
+  if (!hasNegatedPattern && !filter) {
     return await fastGlob.glob(patterns as string[], globOptions)
   }
-
   // Add support for negated "ignore" patterns which many globbing libraries,
   // including 'fast-glob', 'globby', and 'tinyglobby', lack support for.
-  const filtered: string[] = []
-  const ig = ignore().add([...ignores])
+  // Use streaming to avoid unbounded memory accumulation.
+  // This is critical for large monorepos with 100k+ files.
+  const results: string[] = []
+  const ig = hasNegatedPattern ? ignore().add([...ignores]) : null
   const stream = fastGlob.globStream(
     patterns as string[],
     globOptions,
   ) as AsyncIterable<string>
   for await (const p of stream) {
-    // Note: the input files must be INSIDE the cwd. If you get strange looking
-    // relative path errors here, most likely your path is outside the given cwd.
-    const relPath = globOptions.absolute ? path.relative(cwd, p) : p
-    if (!ig.ignores(relPath)) {
-      filtered.push(p)
+    // Check gitignore patterns with negation support.
+    if (ig) {
+      // Note: the input files must be INSIDE the cwd. If you get strange looking
+      // relative path errors here, most likely your path is outside the given cwd.
+      const relPath = globOptions.absolute ? path.relative(cwd, p) : p
+      if (ig.ignores(relPath)) {
+        continue
+      }
+    }
+    // Apply the optional filter to reduce memory usage.
+    // When scanning large monorepos, this filters early (e.g., to manifest files only)
+    // instead of accumulating all 100k+ files and filtering later.
+    if (filter && !filter(p)) {
+      continue
     }
+    results.push(p)
   }
-  return filtered
+  return results
 }
 
 export async function globWorkspace(

src/utils/glob.test.mts

@@ -0,0 +1,252 @@
+import { existsSync, readdirSync, rmSync } from 'node:fs'
+import path from 'node:path'
+import { fileURLToPath } from 'node:url'
+
+import mockFs from 'mock-fs'
+import { afterEach, describe, expect, it } from 'vitest'
+
+import { normalizePath } from '@socketsecurity/registry/lib/path'
+
+import { NODE_MODULES } from '../constants.mjs'
+import {
+  createSupportedFilesFilter,
+  globWithGitIgnore,
+  pathsToGlobPatterns,
+} from './glob.mts'
+
+import type FileSystem from 'mock-fs/lib/filesystem'
+
+// Filter functions defined at module scope to satisfy linting rules.
+function filterJsonFiles(filepath: string): boolean {
+  return filepath.endsWith('.json')
+}
+
+function filterTsFiles(filepath: string): boolean {
+  return filepath.endsWith('.ts')
+}
+
+const __filename = fileURLToPath(import.meta.url)
+const __dirname = path.dirname(__filename)
+
+const rootNmPath = path.join(__dirname, '../..', NODE_MODULES)
+const mockFixturePath = normalizePath(path.join(__dirname, 'glob-mock'))
+const mockNmPath = normalizePath(rootNmPath)
+
+// Remove broken symlinks in node_modules before loading to prevent mock-fs errors.
+function cleanupBrokenSymlinks(dirPath: string): void {
+  try {
+    if (!existsSync(dirPath)) {
+      return
+    }
+    const entries = readdirSync(dirPath, { withFileTypes: true })
+    for (const entry of entries) {
+      const fullPath = path.join(dirPath, entry.name)
+      try {
+        if (entry.isSymbolicLink() && !existsSync(fullPath)) {
+          // Symlink exists but target does not, remove it.
+          rmSync(fullPath, { force: true })
+        } else if (entry.isDirectory()) {
+          // Recursively check subdirectories.
+          cleanupBrokenSymlinks(fullPath)
+        }
+      } catch {
+        // Ignore errors for individual entries.
+      }
+    }
+  } catch {
+    // If we cannot read the directory, skip cleanup.
+  }
+}
+
+// Clean up broken symlinks before loading node_modules.
+cleanupBrokenSymlinks(rootNmPath)
+
+// Load node_modules with error handling for any remaining issues.
+const mockedNmCallback = (() => {
+  try {
+    return mockFs.load(rootNmPath)
+  } catch (e) {
+    // If loading fails due to broken symlinks or missing files, return empty mock.
+    console.warn(
+      `Warning: Failed to load node_modules for mock-fs: ${e instanceof Error ? e.message : String(e)}`,
+    )
+    return {}
+  }
+})()
+
+function mockTestFs(config: FileSystem.DirectoryItems) {
+  return mockFs({
+    ...config,
+    [mockNmPath]: mockedNmCallback,
+  })
+}
+
+describe('glob utilities', () => {
+  afterEach(() => {
+    mockFs.restore()
+  })
+
+  describe('globWithGitIgnore()', () => {
+    it('should find files matching glob patterns', async () => {
+      mockTestFs({
+        [`${mockFixturePath}/package.json`]: '{}',
+        [`${mockFixturePath}/src/index.ts`]: '',
+      })
+
+      const results = await globWithGitIgnore(['**/*.json'], {
+        cwd: mockFixturePath,
+      })
+
+      expect(results.map(normalizePath)).toEqual([
+        `${mockFixturePath}/package.json`,
+      ])
+    })
+
+    it('should respect .gitignore files', async () => {
+      mockTestFs({
+        [`${mockFixturePath}/.gitignore`]: 'ignored/**',
+        [`${mockFixturePath}/package.json`]: '{}',
+        [`${mockFixturePath}/ignored/package.json`]: '{}',
+        [`${mockFixturePath}/included/package.json`]: '{}',
+      })
+
+      const results = await globWithGitIgnore(['**/*.json'], {
+        cwd: mockFixturePath,
+      })
+
+      expect(results.map(normalizePath).sort()).toEqual([
+        `${mockFixturePath}/included/package.json`,
+        `${mockFixturePath}/package.json`,
+      ])
+    })
+
+    it('should handle negated patterns in .gitignore', async () => {
+      mockTestFs({
+        [`${mockFixturePath}/.gitignore`]: 'ignored/**\n!ignored/keep.json',
+        [`${mockFixturePath}/package.json`]: '{}',
+        [`${mockFixturePath}/ignored/excluded.json`]: '{}',
+        [`${mockFixturePath}/ignored/keep.json`]: '{}',
+      })
+
+      const results = await globWithGitIgnore(['**/*.json'], {
+        cwd: mockFixturePath,
+      })
+
+      // The negated pattern should allow keep.json to be included.
+      expect(results.map(normalizePath).sort()).toEqual([
+        `${mockFixturePath}/ignored/keep.json`,
+        `${mockFixturePath}/package.json`,
+      ])
+    })
+
+    it('should apply filter function during streaming to reduce memory', async () => {
+      // Create a mock filesystem with many files.
+      const files: FileSystem.DirectoryItems = {}
+      const fileCount = 100
+      for (let i = 0; i < fileCount; i += 1) {
+        files[`${mockFixturePath}/file${i}.txt`] = 'content'
+        files[`${mockFixturePath}/file${i}.json`] = '{}'
+      }
+      // Add a gitignore with negated pattern to trigger the streaming path.
+      files[`${mockFixturePath}/.gitignore`] = 'temp/\n!temp/keep.json'
+      mockTestFs(files)
+
+      const results = await globWithGitIgnore(['**/*'], {
+        cwd: mockFixturePath,
+        filter: filterJsonFiles,
+      })
+
+      // Should only include .json files (100 files).
+      expect(results).toHaveLength(fileCount)
+      for (const result of results) {
+        expect(result.endsWith('.json')).toBe(true)
+      }
+    })
+
+    it('should apply filter without negated patterns', async () => {
+      mockTestFs({
+        [`${mockFixturePath}/package.json`]: '{}',
+        [`${mockFixturePath}/src/index.ts`]: '',
+        [`${mockFixturePath}/src/utils.ts`]: '',
+        [`${mockFixturePath}/readme.md`]: '',
+      })
+
+      const results = await globWithGitIgnore(['**/*'], {
+        cwd: mockFixturePath,
+        filter: filterTsFiles,
+      })
+
+      expect(results.map(normalizePath).sort()).toEqual([
+        `${mockFixturePath}/src/index.ts`,
+        `${mockFixturePath}/src/utils.ts`,
+      ])
+    })
+
+    it('should combine filter with negated gitignore patterns', async () => {
+      mockTestFs({
+        [`${mockFixturePath}/.gitignore`]: 'build/**\n!build/manifest.json',
+        [`${mockFixturePath}/package.json`]: '{}',
+        [`${mockFixturePath}/src/index.ts`]: '',
+        [`${mockFixturePath}/build/output.js`]: '',
+        [`${mockFixturePath}/build/manifest.json`]: '{}',
+      })
+
+      const results = await globWithGitIgnore(['**/*'], {
+        cwd: mockFixturePath,
+        filter: filterJsonFiles,
+      })
+
+      // Should include package.json and the negated build/manifest.json, but not build/output.js.
+      expect(results.map(normalizePath).sort()).toEqual([
+        `${mockFixturePath}/build/manifest.json`,
+        `${mockFixturePath}/package.json`,
+      ])
+    })
+  })
+
+  describe('createSupportedFilesFilter()', () => {
+    it('should create a filter function matching supported file patterns', () => {
+      const supportedFiles = {
+        npm: {
+          packagejson: { pattern: 'package.json' },
+          packagelockjson: { pattern: 'package-lock.json' },
+        },
+      }
+
+      const filter = createSupportedFilesFilter(supportedFiles)
+
+      expect(filter('/path/to/package.json')).toBe(true)
+      expect(filter('/path/to/package-lock.json')).toBe(true)
+      expect(filter('/path/to/random.txt')).toBe(false)
+      expect(filter('/path/to/nested/package.json')).toBe(true)
+    })
+  })
+
+  describe('pathsToGlobPatterns()', () => {
+    it('should convert "." to "**/*"', () => {
+      expect(pathsToGlobPatterns(['.'])).toEqual(['**/*'])
+      expect(pathsToGlobPatterns(['./'])).toEqual(['**/*'])
+    })
+
+    it('should append "/**/*" to directory paths', () => {
+      mockTestFs({
+        [`${mockFixturePath}/subdir`]: {
+          'file.txt': '',
+        },
+      })
+
+      // The function checks if path is a directory using isDirSync.
+      const result = pathsToGlobPatterns(['subdir'], mockFixturePath)
+      expect(result).toEqual(['subdir/**/*'])
+    })
+
+    it('should keep file paths unchanged', () => {
+      mockTestFs({
+        [`${mockFixturePath}/file.txt`]: '',
+      })
+
+      const result = pathsToGlobPatterns(['file.txt'], mockFixturePath)
+      expect(result).toEqual(['file.txt'])
+    })
+  })
+})

src/utils/path-resolve.mts

@@ -9,7 +9,7 @@ import { isDirSync } from '@socketsecurity/registry/lib/fs'
 
 import constants, { NODE_MODULES, NPM } from '../constants.mts'
 import {
-  filterBySupportedScanFiles,
+  createSupportedFilesFilter,
   globWithGitIgnore,
   pathsToGlobPatterns,
 } from './glob.mts'
@@ -114,13 +114,17 @@ export async function getPackageFilesForScan(
     ...options,
   } as PackageFilesForScanOptions
 
-  const filepaths = await globWithGitIgnore(
+  // Apply the supported files filter during streaming to avoid accumulating
+  // all files in memory. This is critical for large monorepos with 100k+ files
+  // where accumulating all paths before filtering causes OOM errors.
+  const filter = createSupportedFilesFilter(supportedFiles)
+
+  return await globWithGitIgnore(
     pathsToGlobPatterns(inputPaths, options?.cwd),
     {
       cwd,
+      filter,
       socketConfig,
     },
   )
-
-  return filterBySupportedScanFiles(filepaths!, supportedFiles)
 }