@aiready/pattern-detect 0.17.8 → 0.17.12

package/dist/chunk-AQIP4JGM.mjs

@@ -0,0 +1,283 @@
+// src/context-rules.ts
+import {
+  IssueType,
+  getSeverityLabel,
+  filterBySeverity,
+  Severity
+} from "@aiready/core";
+var CONTEXT_RULES = [
+  // Test Fixtures - Intentional duplication for test isolation
+  {
+    name: "test-fixtures",
+    detect: (file, code) => {
+      const isTestFile = file.includes(".test.") || file.includes(".spec.") || file.includes("__tests__") || file.includes("/test/") || file.includes("/tests/");
+      const hasTestFixtures = code.includes("beforeAll") || code.includes("afterAll") || code.includes("beforeEach") || code.includes("afterEach") || code.includes("setUp") || code.includes("tearDown");
+      return isTestFile && hasTestFixtures;
+    },
+    severity: Severity.Info,
+    reason: "Test fixture duplication is intentional for test isolation",
+    suggestion: "Consider if shared test setup would improve maintainability without coupling tests"
+  },
+  // Email/Document Templates - Often intentionally similar for consistency
+  {
+    name: "templates",
+    detect: (file, code) => {
+      const isTemplate = file.includes("/templates/") || file.includes("-template") || file.includes("/email-templates/") || file.includes("/emails/");
+      const hasTemplateContent = (code.includes("return") || code.includes("export")) && (code.includes("html") || code.includes("subject") || code.includes("body"));
+      return isTemplate && hasTemplateContent;
+    },
+    severity: Severity.Info,
+    reason: "Template duplication may be intentional for maintainability and branding consistency",
+    suggestion: "Extract shared structure only if templates become hard to maintain"
+  },
+  // E2E/Integration Test Page Objects - Test independence
+  {
+    name: "e2e-page-objects",
+    detect: (file, code) => {
+      const isE2ETest = file.includes("e2e/") || file.includes("/e2e/") || file.includes(".e2e.") || file.includes("/playwright/") || file.includes("playwright/") || file.includes("/cypress/") || file.includes("cypress/") || file.includes("/integration/") || file.includes("integration/");
+      const hasPageObjectPatterns = code.includes("page.") || code.includes("await page") || code.includes("locator") || code.includes("getBy") || code.includes("selector") || code.includes("click(") || code.includes("fill(");
+      return isE2ETest && hasPageObjectPatterns;
+    },
+    severity: Severity.Info,
+    reason: "E2E test duplication ensures test independence and reduces coupling",
+    suggestion: "Consider page object pattern only if duplication causes maintenance issues"
+  },
+  // Configuration Files - Often necessarily similar by design
+  {
+    name: "config-files",
+    detect: (file) => {
+      return file.endsWith(".config.ts") || file.endsWith(".config.js") || file.includes("jest.config") || file.includes("vite.config") || file.includes("webpack.config") || file.includes("rollup.config") || file.includes("tsconfig");
+    },
+    severity: Severity.Info,
+    reason: "Configuration files often have similar structure by design",
+    suggestion: "Consider shared config base only if configurations become hard to maintain"
+  },
+  // Type Definitions - Duplication for type safety and module independence
+  // Matches type-only files OR interface/type definitions in component files
+  {
+    name: "type-definitions",
+    detect: (file, code) => {
+      const isTypeFile = file.endsWith(".d.ts") || file.includes("/types/");
+      const hasOnlyTypeDefinitions = (code.includes("interface ") || code.includes("type ") || code.includes("enum ")) && !code.includes("function ") && !code.includes("class ") && !code.includes("const ") && !code.includes("let ") && !code.includes("export default");
+      const isInterfaceOnlySnippet = code.trim().startsWith("interface ") && code.includes("{") && code.includes("}") && !code.includes("function ") && !code.includes("const ") && !code.includes("return ");
+      return isTypeFile && hasOnlyTypeDefinitions || isInterfaceOnlySnippet;
+    },
+    severity: Severity.Info,
+    reason: "Type/interface definitions are intentionally duplicated for module independence",
+    suggestion: "Extract to shared types package only if causing maintenance burden"
+  },
+  // Migration Scripts - One-off scripts that are similar by nature
+  {
+    name: "migration-scripts",
+    detect: (file) => {
+      return file.includes("/migrations/") || file.includes("/migrate/") || file.includes(".migration.");
+    },
+    severity: Severity.Info,
+    reason: "Migration scripts are typically one-off and intentionally similar",
+    suggestion: "Duplication is acceptable for migration scripts"
+  },
+  // Mock Data - Test data intentionally duplicated
+  {
+    name: "mock-data",
+    detect: (file, code) => {
+      const isMockFile = file.includes("/mocks/") || file.includes("/__mocks__/") || file.includes("/fixtures/") || file.includes(".mock.") || file.includes(".fixture.");
+      const hasMockData = code.includes("mock") || code.includes("Mock") || code.includes("fixture") || code.includes("stub") || code.includes("export const");
+      return isMockFile && hasMockData;
+    },
+    severity: Severity.Info,
+    reason: "Mock data duplication is expected for comprehensive test coverage",
+    suggestion: "Consider shared factories only for complex mock generation"
+  },
+  // Tool Implementations - Structural Boilerplate
+  {
+    name: "tool-implementations",
+    detect: (file, code) => {
+      const isToolFile = file.includes("/tools/") || file.endsWith(".tool.ts") || code.includes("toolDefinitions");
+      const hasToolStructure = code.includes("execute") && (code.includes("try") || code.includes("catch"));
+      return isToolFile && hasToolStructure;
+    },
+    severity: Severity.Info,
+    reason: "Tool implementations share structural boilerplate but have distinct business logic",
+    suggestion: "Tool duplication is acceptable for boilerplate interface wrappers"
+  },
+  // Common UI Event Handlers - Very specific patterns only
+  {
+    name: "common-ui-handlers",
+    detect: (file, code) => {
+      const isUIFile = file.includes("/components/") || file.includes(".tsx") || file.includes(".jsx") || file.includes("/hooks/");
+      const hasCommonHandler = code.includes("handleClickOutside") && code.includes("dropdownRef.current") && code.includes("!dropdownRef.current.contains") || code.includes("handleEscape") && code.includes("event.key") && code.includes("=== 'Escape'") || code.includes("handleClickInside") && code.includes("event.stopPropagation");
+      return isUIFile && hasCommonHandler;
+    },
+    severity: Severity.Info,
+    reason: "Common UI event handlers are boilerplate patterns that repeat across components",
+    suggestion: "Consider extracting to shared hooks (useClickOutside, useEscapeKey) only if causing maintenance issues"
+  },
+  // Utility Functions - Small helpers in dedicated utility files
+  {
+    name: "utility-functions",
+    detect: (file, code) => {
+      const isUtilFile = file.endsWith(".util.ts") || file.endsWith(".helper.ts") || file.endsWith(".utils.ts");
+      const hasUtilPattern = code.includes("function format") || code.includes("function parse") || code.includes("function sanitize") || code.includes("function normalize") || code.includes("function convert");
+      return isUtilFile && hasUtilPattern;
+    },
+    severity: Severity.Info,
+    reason: "Utility functions in dedicated utility files may be intentionally similar",
+    suggestion: "Consider extracting to shared utilities only if causing significant duplication"
+  },
+  // React/Vue Hooks - Standard patterns
+  {
+    name: "shared-hooks",
+    detect: (file, code) => {
+      const isHookFile = file.includes("/hooks/") || file.endsWith(".hook.ts") || file.endsWith(".hook.tsx");
+      const hasHookPattern = code.includes("function use") || code.includes("export function use") || code.includes("const use") || code.includes("export const use");
+      return isHookFile && hasHookPattern;
+    },
+    severity: Severity.Info,
+    reason: "Hooks follow standard patterns and are often intentionally similar across components",
+    suggestion: "Consider extracting common hook logic only if hooks become complex"
+  },
+  // CLI Command Definitions - Commander.js boilerplate patterns
+  {
+    name: "cli-command-definitions",
+    detect: (file, code) => {
+      const isCliFile = file.includes("/commands/") || file.includes("/cli/") || file.endsWith(".command.ts");
+      const hasCommandPattern = (code.includes(".command(") || code.includes("defineCommand")) && (code.includes(".description(") || code.includes(".option(")) && (code.includes(".action(") || code.includes("async"));
+      return isCliFile && hasCommandPattern;
+    },
+    severity: Severity.Info,
+    reason: "CLI command definitions follow standard Commander.js patterns and are intentionally similar",
+    suggestion: "Command boilerplate duplication is acceptable for CLI interfaces"
+  },
+  // Score/Rating Helper Functions - Common threshold patterns
+  {
+    name: "score-helpers",
+    detect: (file, code) => {
+      const isHelperFile = file.includes("/utils/") || file.includes("/helpers/") || file.endsWith(".util.ts");
+      const hasScorePattern = (code.includes("if (score >=") || code.includes("if (score >")) && (code.includes("return") && code.includes("'")) && code.split("if (score").length >= 3;
+      return isHelperFile && hasScorePattern;
+    },
+    severity: Severity.Info,
+    reason: "Score/rating helper functions use common threshold patterns that are intentionally similar",
+    suggestion: "Score formatting duplication is acceptable for consistent UI display"
+  },
+  // D3/Canvas Event Handlers - Standard visualization patterns
+  {
+    name: "visualization-handlers",
+    detect: (file, code) => {
+      const isVizFile = file.includes("/visualizer/") || file.includes("/charts/") || file.includes("GraphCanvas") || file.includes("ForceDirected");
+      const hasVizPattern = (code.includes("dragstarted") || code.includes("dragged") || code.includes("dragended")) && (code.includes("simulation") || code.includes("d3.") || code.includes("alphaTarget"));
+      return isVizFile && hasVizPattern;
+    },
+    severity: Severity.Info,
+    reason: "D3/visualization event handlers follow standard patterns and are intentionally similar",
+    suggestion: "Visualization boilerplate duplication is acceptable for interactive charts"
+  },
+  // Icon/Switch Statement Helpers - Common enum-to-value patterns
+  {
+    name: "switch-helpers",
+    detect: (file, code) => {
+      const hasSwitchPattern = code.includes("switch (") && code.includes("case '") && code.includes("return") && code.split("case ").length >= 4;
+      const hasIconPattern = code.includes("getIcon") || code.includes("getColor") || code.includes("getLabel") || code.includes("getRating");
+      return hasSwitchPattern && hasIconPattern;
+    },
+    severity: Severity.Info,
+    reason: "Switch statement helpers for enum-to-value mapping are inherently similar",
+    suggestion: "Switch duplication is acceptable for mapping enums to display values"
+  },
+  // Common API/Utility Functions - Legitimate duplication across modules
+  {
+    name: "common-api-functions",
+    detect: (file, code) => {
+      const isApiFile = file.includes("/api/") || file.includes("/lib/") || file.includes("/utils/") || file.endsWith(".ts");
+      const hasCommonApiPattern = code.includes("getStripe") && code.includes("process.env.STRIPE_SECRET_KEY") || code.includes("getUserByEmail") && code.includes("queryItems") || code.includes("updateUser") && code.includes("buildUpdateExpression") || code.includes("listUserRepositories") && code.includes("queryItems") || code.includes("listTeamRepositories") && code.includes("queryItems") || code.includes("getRemediation") && code.includes("queryItems") || code.includes("formatBreakdownKey") && code.includes(".replace(/([A-Z])/g") || code.includes("queryItems") && code.includes("KeyConditionExpression") || code.includes("putItem") && code.includes("createdAt") || code.includes("updateItem") && code.includes("buildUpdateExpression");
+      return isApiFile && hasCommonApiPattern;
+    },
+    severity: Severity.Info,
+    reason: "Common API/utility functions are legitimately duplicated across modules for clarity and independence",
+    suggestion: "Consider extracting to shared utilities only if causing significant duplication"
+  },
+  // Next.js Route Handler Patterns - Boilerplate API route patterns
+  {
+    name: "nextjs-route-handlers",
+    detect: (file, code) => {
+      const isRouteFile = file.includes("/api/") && (file.endsWith("/route.ts") || file.endsWith("/route.js"));
+      const hasRoutePattern = code.includes("export async function POST") || code.includes("export async function GET") || code.includes("export async function PUT") || code.includes("export async function DELETE") || code.includes("NextResponse.json") || code.includes("NextRequest");
+      return isRouteFile && hasRoutePattern;
+    },
+    severity: Severity.Info,
+    reason: "Next.js route handlers follow standard patterns and are intentionally similar across endpoints",
+    suggestion: "Route handler duplication is acceptable for API endpoint boilerplate"
+  },
+  // Validation Functions - Inherently similar patterns
+  {
+    name: "validation-functions",
+    detect: (file, code) => {
+      const hasValidationPattern = code.includes("isValid") || code.includes("validate") || code.includes("checkValid") || code.includes("isEmail") || code.includes("isPhone") || code.includes("isUrl") || code.includes("isNumeric") || code.includes("isAlpha") || code.includes("isAlphanumeric") || code.includes("isEmpty") || code.includes("isNotEmpty") || code.includes("isRequired") || code.includes("isOptional");
+      return hasValidationPattern;
+    },
+    severity: Severity.Info,
+    reason: "Validation functions are inherently similar and often intentionally duplicated for domain clarity",
+    suggestion: "Consider extracting to shared validators only if validation logic becomes complex"
+  }
+];
+function calculateSeverity(file1, file2, code, similarity, linesOfCode) {
+  for (const rule of CONTEXT_RULES) {
+    if (rule.detect(file1, code) || rule.detect(file2, code)) {
+      return {
+        severity: rule.severity,
+        reason: rule.reason,
+        suggestion: rule.suggestion,
+        matchedRule: rule.name
+      };
+    }
+  }
+  if (similarity >= 0.95 && linesOfCode >= 30) {
+    return {
+      severity: Severity.Critical,
+      reason: "Large nearly-identical code blocks waste tokens and create maintenance burden",
+      suggestion: "Extract to shared utility module immediately"
+    };
+  } else if (similarity >= 0.95 && linesOfCode >= 15) {
+    return {
+      severity: Severity.Major,
+      reason: "Nearly identical code should be consolidated",
+      suggestion: "Move to shared utility file"
+    };
+  } else if (similarity >= 0.85) {
+    return {
+      severity: Severity.Major,
+      reason: "High similarity indicates significant duplication",
+      suggestion: "Extract common logic to shared function"
+    };
+  } else if (similarity >= 0.7) {
+    return {
+      severity: Severity.Minor,
+      reason: "Moderate similarity detected",
+      suggestion: "Consider extracting shared patterns if code evolves together"
+    };
+  } else {
+    return {
+      severity: Severity.Minor,
+      reason: "Minor similarity detected",
+      suggestion: "Monitor but refactoring may not be worthwhile"
+    };
+  }
+}
+function getSeverityThreshold(severity) {
+  const thresholds = {
+    [Severity.Critical]: 0.95,
+    [Severity.Major]: 0.85,
+    [Severity.Minor]: 0.5,
+    [Severity.Info]: 0
+  };
+  return thresholds[severity] || 0;
+}
+
+export {
+  IssueType,
+  getSeverityLabel,
+  filterBySeverity,
+  CONTEXT_RULES,
+  calculateSeverity,
+  getSeverityThreshold
+};

package/dist/{chunk-JBUZ6YHE.mjs → chunk-B4NLWKPZ.mjs}

@@ -1,12 +1,12 @@
 import {
   detectDuplicatePatterns
-} from "./chunk-RS73WLNI.mjs";
+} from "./chunk-SXVLRPMF.mjs";
 import {
   calculateSeverity
-} from "./chunk-I6ETJC7L.mjs";
+} from "./chunk-LUUJOUK5.mjs";
 
 // src/grouping.ts
-import { getSeverityLevel } from "@aiready/core";
+import { Severity, getSeverityLevel } from "@aiready/core";
 import path from "path";
 function groupDuplicatesByFilePair(duplicates) {
   const groups = /* @__PURE__ */ new Map();
@@ -124,9 +124,77 @@ function filterClustersByImpact(clusters, minTokenCost = 1e3, minFiles = 3) {
     (c) => c.totalTokenCost >= minTokenCost && c.files.length >= minFiles
   );
 }
+function isPureInterfaceDefinition(code) {
+  const trimmed = code.trim();
+  if (!trimmed.startsWith("interface ") && !trimmed.startsWith("type ") && !trimmed.startsWith("export interface ") && !trimmed.startsWith("export type ") && !trimmed.startsWith("enum ") && !trimmed.startsWith("export enum ")) {
+    return false;
+  }
+  if (trimmed.includes("={") || trimmed.includes("=> {") || trimmed.includes("function ") || trimmed.includes("() {") || trimmed.includes(" implements ")) {
+    return false;
+  }
+  if (trimmed.length > 200) return false;
+  return true;
+}
+var BRAND_INDICATORS = [
+  "cyberpunk",
+  "cyber-blue",
+  "cyber-purple",
+  "slate-900",
+  "slate-400",
+  "zinc-",
+  "indigo-",
+  "neon-",
+  "glassmorphism",
+  "backdrop-blur"
+];
+function isBrandSpecificComponent(filePath) {
+  const lower = filePath.toLowerCase();
+  const brandingTerms = ["landing", "clawmore", "platform", "apps/"];
+  for (const term of brandingTerms) {
+    if (lower.includes(term)) return true;
+  }
+  return false;
+}
+function areBrandSpecificVariants(file1, file2, code1, code2) {
+  const f1IsBrand = isBrandSpecificComponent(file1);
+  const f2IsBrand = isBrandSpecificComponent(file2);
+  if (f1IsBrand && f2IsBrand && file1 !== file2) {
+    const hasBrandKeyword = (code) => {
+      const lowerCode = code.toLowerCase();
+      return BRAND_INDICATORS.some((ind) => lowerCode.includes(ind));
+    };
+    const code1Brand = hasBrandKeyword(code1);
+    const code2Brand = hasBrandKeyword(code2);
+    if (code1Brand && code2Brand) {
+      return true;
+    }
+  }
+  return false;
+}
+function filterBrandSpecificVariants(duplicates) {
+  return duplicates.filter((dup) => {
+    if (dup.file1 === dup.file2) return true;
+    const isBrandVariant = areBrandSpecificVariants(
+      dup.file1,
+      dup.file2,
+      dup.code1,
+      dup.code2
+    );
+    if (isBrandVariant) {
+      dup.severity = Severity.Info;
+      dup.suggestion = "Brand-specific themed component variant (intentional)";
+    }
+    const isInterfaceDef = isPureInterfaceDefinition(dup.code1) && isPureInterfaceDefinition(dup.code2);
+    if (isInterfaceDef) {
+      dup.severity = Severity.Info;
+      dup.suggestion = "Pure interface/type definition - intentional for module independence";
+    }
+    return true;
+  });
+}
 
 // src/analyzer.ts
-import { scanFiles, readFileContent, Severity as Severity2, IssueType } from "@aiready/core";
+import { scanFiles, readFileContent, Severity as Severity2, IssueType as IssueType2 } from "@aiready/core";
 function getRefactoringSuggestion(patternType, similarity) {
   const baseMessages = {
     "api-handler": "Extract common middleware or create a base handler class",
@@ -167,7 +235,7 @@ async function getSmartDefaults(directory, userOptions) {
     6,
     Math.min(20, 6 + Math.floor(estimatedBlocks / 1e3) * 2)
   );
-  const minSimilarity = Math.min(0.85, 0.5 + estimatedBlocks / 5e3 * 0.3);
+  const minSimilarity = Math.min(0.75, 0.45 + estimatedBlocks / 1e4 * 0.3);
   const batchSize = estimatedBlocks > 1e3 ? 200 : 100;
   const severity = estimatedBlocks > 3e3 ? "high" : "all";
   const maxCandidatesPerBlock = Math.max(
@@ -268,15 +336,21 @@ async function analyzePatterns(options) {
     ignoreWhitelist,
     onProgress: options.onProgress
   });
+  filterBrandSpecificVariants(duplicates);
   for (const file of files) {
     const fileDuplicates = duplicates.filter(
       (dup) => dup.file1 === file || dup.file2 === file
     );
     const issues = fileDuplicates.map((dup) => {
       const otherFile = dup.file1 === file ? dup.file2 : dup.file1;
-      const severity2 = dup.similarity > 0.95 ? Severity2.Critical : dup.similarity > 0.9 ? Severity2.Major : Severity2.Minor;
+      let severity2;
+      if (dup.severity === "info" || dup.severity === "Info") {
+        severity2 = Severity2.Info;
+      } else {
+        severity2 = dup.similarity > 0.95 ? Severity2.Critical : dup.similarity > 0.9 ? Severity2.Major : Severity2.Minor;
+      }
       return {
-        type: IssueType.DuplicatePattern,
+        type: IssueType2.DuplicatePattern,
         severity: severity2,
         message: `${dup.patternType} pattern ${Math.round(dup.similarity * 100)}% similar to ${otherFile} (${dup.tokenCost} tokens wasted)`,
         location: {
@@ -327,9 +401,9 @@ async function analyzePatterns(options) {
   return { results, duplicates, files, groups, clusters, config: finalOptions };
 }
 function generateSummary(results) {
-  const allIssues = results.flatMap((r) => r.issues);
+  const allIssues = results.flatMap((r) => r.issues || []);
   const totalTokenCost = results.reduce(
-    (sum, r) => sum + (r.metrics.tokenCost || 0),
+    (sum, r) => sum + (r.metrics?.tokenCost || 0),
     0
   );
   const patternsByType = {
@@ -385,6 +459,8 @@ export {
   groupDuplicatesByFilePair,
   createRefactorClusters,
   filterClustersByImpact,
+  areBrandSpecificVariants,
+  filterBrandSpecificVariants,
   getSmartDefaults,
   analyzePatterns,
   generateSummary

package/dist/chunk-IPBGVPUX.mjs

@@ -0,0 +1,143 @@
+import {
+  calculateSeverity
+} from "./chunk-2P7BQHGR.mjs";
+
+// src/detector.ts
+import {
+  calculateStringSimilarity,
+  calculateHeuristicConfidence,
+  extractCodeBlocks
+} from "@aiready/core";
+
+// src/core/normalizer.ts
+function normalizeCode(code, isPython = false) {
+  if (!code) return "";
+  let normalized = code;
+  if (isPython) {
+    normalized = normalized.replace(/#.*/g, "");
+  } else {
+    normalized = normalized.replace(/\/\/.*$/gm, "").replace(/\/\*[\s\S]*?\*\//g, "");
+  }
+  return normalized.replace(/"[^"]*"/g, '"STR"').replace(/'[^']*'/g, "'STR'").replace(/`[^`]*`/g, "`STR`").replace(/\b\d+\b/g, "NUM").replace(/\s+/g, " ").trim().toLowerCase();
+}
+
+// src/detector.ts
+function extractBlocks(file, content) {
+  return extractCodeBlocks(file, content);
+}
+function calculateSimilarity(a, b) {
+  return calculateStringSimilarity(a, b);
+}
+function calculateConfidence(similarity, tokens, lines) {
+  return calculateHeuristicConfidence(similarity, tokens, lines);
+}
+async function detectDuplicatePatterns(fileContents, options) {
+  const {
+    minSimilarity,
+    minLines,
+    streamResults,
+    onProgress,
+    excludePatterns = [],
+    confidenceThreshold = 0,
+    ignoreWhitelist = []
+  } = options;
+  const allBlocks = [];
+  const excludeRegexes = excludePatterns.map((p) => new RegExp(p, "i"));
+  for (const { file, content } of fileContents) {
+    const blocks = extractBlocks(file, content);
+    for (const b of blocks) {
+      if (b.endLine - b.startLine + 1 < minLines) continue;
+      const isExcluded = excludeRegexes.some((regex) => regex.test(b.code));
+      if (isExcluded) continue;
+      allBlocks.push(b);
+    }
+  }
+  const duplicates = [];
+  const totalBlocks = allBlocks.length;
+  let comparisons = 0;
+  const totalComparisons = totalBlocks * (totalBlocks - 1) / 2;
+  if (onProgress) {
+    onProgress(
+      0,
+      totalComparisons,
+      `Starting duplicate detection on ${totalBlocks} blocks...`
+    );
+  }
+  for (let i = 0; i < allBlocks.length; i++) {
+    if (i % 50 === 0 && i > 0) {
+      await new Promise((resolve) => setImmediate(resolve));
+      if (onProgress) {
+        onProgress(
+          comparisons,
+          totalComparisons,
+          `Analyzing blocks (${i}/${totalBlocks})...`
+        );
+      }
+    }
+    const b1 = allBlocks[i];
+    const isPython1 = b1.file.toLowerCase().endsWith(".py");
+    const norm1 = normalizeCode(b1.code, isPython1);
+    for (let j = i + 1; j < allBlocks.length; j++) {
+      comparisons++;
+      const b2 = allBlocks[j];
+      if (b1.file === b2.file) continue;
+      const isWhitelisted = ignoreWhitelist.some((pattern) => {
+        return b1.file.includes(pattern) && b2.file.includes(pattern) || pattern === `${b1.file}::${b2.file}` || pattern === `${b2.file}::${b1.file}`;
+      });
+      if (isWhitelisted) continue;
+      const isPython2 = b2.file.toLowerCase().endsWith(".py");
+      const norm2 = normalizeCode(b2.code, isPython2);
+      const sim = calculateSimilarity(norm1, norm2);
+      if (sim >= minSimilarity) {
+        const confidence = calculateConfidence(
+          sim,
+          b1.tokens,
+          b1.endLine - b1.startLine + 1
+        );
+        if (confidence < confidenceThreshold) continue;
+        const { severity, reason, suggestion, matchedRule } = calculateSeverity(
+          b1.file,
+          b2.file,
+          b1.code,
+          sim,
+          b1.endLine - b1.startLine + 1
+        );
+        const dup = {
+          file1: b1.file,
+          line1: b1.startLine,
+          endLine1: b1.endLine,
+          file2: b2.file,
+          line2: b2.startLine,
+          endLine2: b2.endLine,
+          code1: b1.code,
+          code2: b2.code,
+          similarity: sim,
+          confidence,
+          patternType: b1.patternType,
+          tokenCost: b1.tokens + b2.tokens,
+          severity,
+          reason,
+          suggestion,
+          matchedRule
+        };
+        duplicates.push(dup);
+        if (streamResults)
+          console.log(
+            `[DUPLICATE] ${dup.file1}:${dup.line1} <-> ${dup.file2}:${dup.line2} (${Math.round(sim * 100)}%, conf: ${Math.round(confidence * 100)}%)`
+          );
+      }
+    }
+  }
+  if (onProgress) {
+    onProgress(
+      totalComparisons,
+      totalComparisons,
+      `Duplicate detection complete. Found ${duplicates.length} patterns.`
+    );
+  }
+  return duplicates.sort((a, b) => b.similarity - a.similarity);
+}
+
+export {
+  detectDuplicatePatterns
+};