@oculum/scanner 1.0.11 → 1.0.13

package/src/{layer2/ai-package-hallucination.ts → detect/ai-code/package-hallucination.ts}

@@ -12,14 +12,17 @@
  * 3. Generic unscoped names that are too vague to be real
  */
 
-import type { Vulnerability, VulnerabilitySeverity } from '../types'
+import type { Vulnerability, VulnerabilitySeverity } from '../../shared/types'
+import type { ParsedFile } from '../../shared/parsed-file'
 import {
   isComment,
   isTestOrMockFile,
   isDocumentationFile,
   isScannerOrFixtureFile,
   isExampleDirectory,
-} from '../utils/context-helpers'
+} from '../../parse/file-classifier'
+
+const BASE_CONFIDENCE = 0.55
 
 // ============================================================================
 // Known Hallucinated Package Database
@@ -143,6 +146,107 @@ const KNOWN_LEGITIMATE_PACKAGES: Set<string> = new Set([
   'beautifulsoup4', 'lxml', 'scrapy',
 ])
 
+/**
+ * Packages that should NEVER be flagged as typosquats despite similarity to popular packages.
+ * These are commonly used legitimate packages that happen to be lexically similar to others.
+ *
+ * Examples of FPs this prevents:
+ * - 'test' flagged as typosquat of 'nest' (59 FPs in cal-com)
+ * - 'geist' (Vercel's font) flagged as typosquat of 'nest'
+ * - 'botid' flagged as typosquat of 'solid'
+ * - 'cypress' flagged as typosquat (major testing framework)
+ */
+const TYPOSQUAT_WHITELIST: Set<string> = new Set([
+  // Common test/build packages (easily confused with 'nest', etc.)
+  'test', 'jest', 'mocha', 'chai', 'karma', 'ava', 'tape',
+  'nest', 'next', 'best', 'rest', 'fest', 'west',  // Short common words
+
+  // Major testing frameworks - NEVER flag these
+  'cypress', 'playwright', 'vitest', 'jasmine', 'puppeteer',
+
+  // Vercel ecosystem
+  'geist', 'turbo', 'swr', 'next', 'vercel', 'ai',
+
+  // Common frameworks that might match other frameworks
+  'solid', 'react', 'vue', 'svelte', 'angular', 'ember',
+  'hono', 'koa', 'oak', 'hapi', 'fastify', 'express',
+
+  // Short legitimate package names that could false-positive
+  'botid', 'zod', 'yup', 'joi', 'got', 'ora', 'ink', 'arg',
+  'fs', 'os', 'vm', 'ws', 'ip', 'ms', 'qs', 'p-',
+  'co', 'on', 'yn', 'is', 'it', 'to', 'at', 'no', 'of',
+
+  // Font packages
+  'geist', 'inter', 'roboto', 'lato', 'poppins',
+
+  // Common utility packages with similar names
+  'uuid', 'ulid', 'cuid', 'nanoid',
+  'date', 'data', 'deta', 'meta',
+  'path', 'glob', 'node', 'deno',
+  'form', 'from', 'fork', 'core',
+  'fast', 'last', 'past', 'cast', 'vast',
+  'hash', 'cash', 'dash', 'bash', 'rash',
+  'find', 'bind', 'mind', 'kind', 'wind',
+  'send', 'lend', 'tend', 'mend', 'bend',
+  'read', 'lead', 'head', 'dead', 'bead',
+  'sort', 'port', 'fort', 'tort', 'mort',
+
+  // Database/ORM related
+  'pg', 'mysql', 'mssql', 'sqlite', 'redis', 'mongo',
+  'knex', 'drizzle', 'prisma', 'typeorm',
+
+  // Build tools
+  'vite', 'esbuild', 'rollup', 'parcel', 'turbo',
+
+  // Linting/formatting
+  'eslint', 'biome', 'oxlint', 'rome',
+
+  // Popular packages that sound like typosquats but aren't
+  'lodash', 'axios', 'chalk', 'dotenv', 'bcrypt', 'bcryptjs',
+  'dayjs', 'luxon', 'sharp', 'jimp', 'pino', 'bunyan', 'winston',
+
+  // Math/rendering libraries (katex is NOT a typosquat of knex)
+  'katex', 'latex', 'mathjax',
+
+  // Unified ecosystem (unist is NOT a typosquat of nest - 5M+ weekly downloads)
+  'unist', 'mdast', 'hast', 'xast', 'nlcst', 'esast',
+  'unified', 'remark', 'rehype', 'retext', 'redot',
+  'micromark', 'mdx', 'mdxjs',
+])
+
+/**
+ * Scoped package prefixes that are always legitimate
+ * Packages starting with these scopes should never be flagged
+ */
+const LEGITIMATE_SCOPED_PREFIXES: string[] = [
+  '@vercel/',
+  '@types/',
+  '@testing-library/',
+  '@babel/',
+  '@eslint/',
+  '@typescript-eslint/',
+  '@aws-sdk/',
+  '@azure/',
+  '@google-cloud/',
+  '@octokit/',
+  '@prisma/',
+  '@trpc/',
+  '@tanstack/',
+  '@radix-ui/',
+  '@headlessui/',
+  '@heroicons/',
+  '@tailwindcss/',
+  '@fontsource/',
+  '@next/',
+  '@supabase/',
+  '@clerk/',
+  '@auth/',
+  '@sentry/',
+  '@opentelemetry/',
+  '@anthropic-ai/',
+  '@langchain/',
+]
+
 // ============================================================================
 // Typosquatting Detection
 // ============================================================================
@@ -237,6 +341,20 @@ function hasCharacterSubstitution(packageName: string, popularPackage: string):
 function checkTyposquatting(packageName: string): { isTyposquat: boolean; similarTo?: string; reason?: string } {
   const name = packageName.toLowerCase()
 
+  // Skip packages in the typosquat whitelist - these are known legitimate packages
+  if (TYPOSQUAT_WHITELIST.has(name)) {
+    return { isTyposquat: false }
+  }
+
+  // Skip scoped packages with legitimate prefixes
+  if (packageName.startsWith('@')) {
+    for (const prefix of LEGITIMATE_SCOPED_PREFIXES) {
+      if (packageName.startsWith(prefix)) {
+        return { isTyposquat: false }
+      }
+    }
+  }
+
   for (const popular of POPULAR_PACKAGES_FOR_TYPOSQUAT) {
     const popularLower = popular.toLowerCase()
 
@@ -398,10 +516,42 @@ function isRelativeImport(importPath: string): boolean {
 }
 
 /**
- * Check if this is an alias import (@/, ~/, #)
+ * Check if this is an alias import (@/, ~/, #) or TypeScript path alias
+ * These are local imports configured in tsconfig.json, NOT npm packages
+ *
+ * Common patterns:
+ * - @/lib/auth (Next.js/Vite common pattern)
+ * - ~/components (Nuxt.js pattern)
+ * - #/utils (private imports)
+ * - @components/Button (custom tsconfig paths)
  */
 function isAliasImport(importPath: string): boolean {
-  return /^[@~#]\//.test(importPath)
+  // Standard alias patterns
+  if (/^[@~#]\//.test(importPath)) {
+    return true
+  }
+
+  // Scoped packages are NOT aliases (they have @org/package format)
+  // But @/path and @alias/path ARE aliases when:
+  // 1. They start with @ followed by / (e.g., @/lib/auth)
+  // 2. They have a short "scope" that looks like an alias (e.g., @components/Button)
+  if (importPath.startsWith('@')) {
+    // Check if it looks like a scoped npm package (@org/package-name)
+    // vs a path alias (@/path or @alias/path)
+    const match = importPath.match(/^@([^/]+)\//)
+    if (match) {
+      const scope = match[1]
+      // Empty scope means @/ which is always an alias
+      if (!scope) return true
+      // Single char scopes are aliases (@a/foo, @x/bar)
+      if (scope.length === 1) return true
+      // Common alias names
+      const commonAliases = ['app', 'src', 'lib', 'utils', 'components', 'hooks', 'services', 'config', 'types', 'styles', 'assets', 'public', 'pages', 'layouts', 'features', 'modules', 'core', 'shared']
+      if (commonAliases.includes(scope.toLowerCase())) return true
+    }
+  }
+
+  return false
 }
 
 /**
@@ -511,6 +661,15 @@ function extractPackageName(importPath: string): string | null {
     return null
   }
 
+  // Skip bare path aliases — unscoped generic names with sub-paths
+  // e.g., 'models/challenge', 'data/types', 'lib/auth' are tsconfig path aliases
+  if (!importPath.startsWith('@') && importPath.includes('/')) {
+    const baseName = importPath.split('/')[0]
+    if (GENERIC_UNSCOPED_NAMES.has(baseName)) {
+      return null  // Local path alias, not an npm package
+    }
+  }
+
   // Handle scoped packages (@org/package)
   if (importPath.startsWith('@')) {
     const parts = importPath.split('/')
@@ -658,7 +817,8 @@ function extractRequirementsDeps(_content: string, lines: string[]): ImportMatch
  */
 export function detectAIPackageHallucination(
   content: string,
-  filePath: string
+  filePath: string,
+  options?: { parsed?: ParsedFile }
 ): Vulnerability[] {
   const vulnerabilities: Vulnerability[] = []
 
@@ -666,7 +826,7 @@ export function detectAIPackageHallucination(
   if (isScannerOrFixtureFile(filePath)) return vulnerabilities
   if (isDocumentationFile(filePath)) return vulnerabilities
 
-  const lines = content.split('\n')
+  const lines = options?.parsed?.lines ?? content.split('\n')
   const isTestFile = isTestOrMockFile(filePath)
   const isExample = isExampleDirectory(filePath)
   const isManifest = isPackageManifest(filePath)
@@ -733,7 +893,9 @@ export function detectAIPackageHallucination(
         suggestedFix,
         confidence: 'high',
         layer: 2,
+        source: 'ai_code' as const,
         requiresAIValidation: false, // Typosquats don't need AI validation - pattern is clear
+        baseConfidence: BASE_CONFIDENCE,
       })
       continue // Don't also flag as hallucination
     }
@@ -782,7 +944,9 @@ export function detectAIPackageHallucination(
         suggestedFix,
         confidence: KNOWN_HALLUCINATED_PACKAGES.has(imp.packageName) ? 'high' : 'medium',
         layer: 2,
+        source: 'ai_code' as const,
         requiresAIValidation: severity !== 'info' && !KNOWN_HALLUCINATED_PACKAGES.has(imp.packageName),
+        baseConfidence: BASE_CONFIDENCE,
       })
     }
   }