@aiready/pattern-detect 0.7.9 → 0.7.12

package/dist/index.d.ts

@@ -61,6 +61,10 @@ interface PatternSummary {
         tokenCost: number;
     }>;
 }
+/**
+ * Determine smart defaults based on repository size estimation
+ */
+declare function getSmartDefaults(directory: string, userOptions: Partial<PatternDetectOptions>): Promise<PatternDetectOptions>;
 declare function analyzePatterns(options: PatternDetectOptions): Promise<{
     results: AnalysisResult[];
     duplicates: DuplicatePattern[];
@@ -71,4 +75,4 @@ declare function analyzePatterns(options: PatternDetectOptions): Promise<{
  */
 declare function generateSummary(results: AnalysisResult[]): PatternSummary;
 
-export { type DuplicatePattern, type PatternDetectOptions, type PatternSummary, type PatternType, analyzePatterns, detectDuplicatePatterns, generateSummary };
+export { type DuplicatePattern, type PatternDetectOptions, type PatternSummary, type PatternType, analyzePatterns, detectDuplicatePatterns, generateSummary, getSmartDefaults };

package/dist/index.js

@@ -32,7 +32,8 @@ var index_exports = {};
 __export(index_exports, {
   analyzePatterns: () => analyzePatterns,
   detectDuplicatePatterns: () => detectDuplicatePatterns,
-  generateSummary: () => generateSummary
+  generateSummary: () => generateSummary,
+  getSmartDefaults: () => getSmartDefaults
 });
 module.exports = __toCommonJS(index_exports);
 var import_core2 = require("@aiready/core");
@@ -220,7 +221,13 @@ async function detectDuplicatePatterns(files, options) {
     let candidates = null;
     if (approx) {
       const counts = /* @__PURE__ */ new Map();
-      for (const tok of blockTokens[i]) {
+      const block1Tokens = new Set(blockTokens[i]);
+      const block1Size = block1Tokens.size;
+      const rareTokens = blockTokens[i].filter((tok) => {
+        const blocksWithToken = invertedIndex.get(tok)?.length || 0;
+        return blocksWithToken < allBlocks.length * 0.1;
+      });
+      for (const tok of rareTokens) {
         const ids = invertedIndex.get(tok);
         if (!ids) continue;
         for (const j of ids) {
@@ -229,7 +236,13 @@ async function detectDuplicatePatterns(files, options) {
           counts.set(j, (counts.get(j) || 0) + 1);
         }
       }
-      candidates = Array.from(counts.entries()).filter(([, shared]) => shared >= minSharedTokens).sort((a, b) => b[1] - a[1]).slice(0, maxCandidatesPerBlock).map(([j, shared]) => ({ j, shared }));
+      candidates = Array.from(counts.entries()).filter(([j, shared]) => {
+        const block2Tokens = blockTokens[j];
+        const block2Size = block2Tokens.length;
+        const minSize = Math.min(block1Size, block2Size);
+        const sharedPercentage = shared / minSize;
+        return shared >= minSharedTokens && sharedPercentage >= 0.3;
+      }).sort((a, b) => b[1] - a[1]).slice(0, Math.min(maxCandidatesPerBlock, 5)).map(([j, shared]) => ({ j, shared }));
     }
     if (approx && candidates) {
       for (const { j } of candidates) {
@@ -322,12 +335,12 @@ async function getSmartDefaults(directory, userOptions) {
   if (userOptions.useSmartDefaults === false) {
     return {
       rootDir: directory,
-      minSimilarity: 0.4,
-      minLines: 5,
+      minSimilarity: 0.6,
+      minLines: 8,
       batchSize: 100,
       approx: true,
-      minSharedTokens: 8,
-      maxCandidatesPerBlock: 100,
+      minSharedTokens: 12,
+      maxCandidatesPerBlock: 5,
       streamResults: false,
       severity: "all",
       includeTests: false
@@ -348,12 +361,12 @@ async function getSmartDefaults(directory, userOptions) {
   const { scanFiles: scanFiles2 } = await import("@aiready/core");
   const files = await scanFiles2(scanOptions);
   const estimatedBlocks = files.length * 3;
-  const maxCandidatesPerBlock = Math.max(10, Math.min(100, Math.floor(8e4 / estimatedBlocks)));
-  const minSimilarity = Math.min(0.65, 0.4 + estimatedBlocks / 15e3 * 0.25);
-  const minLines = Math.max(5, Math.min(10, 5 + Math.floor(estimatedBlocks / 3e3)));
-  const minSharedTokens = Math.max(8, Math.min(15, 8 + Math.floor(estimatedBlocks / 4e3)));
-  const batchSize = estimatedBlocks > 2e3 ? 300 : 150;
-  const severity = estimatedBlocks > 8e3 ? "high" : "all";
+  const maxCandidatesPerBlock = Math.max(3, Math.min(10, Math.floor(3e4 / estimatedBlocks)));
+  const minSimilarity = Math.min(0.75, 0.5 + estimatedBlocks / 1e4 * 0.25);
+  const minLines = Math.max(6, Math.min(12, 6 + Math.floor(estimatedBlocks / 2e3)));
+  const minSharedTokens = Math.max(10, Math.min(20, 10 + Math.floor(estimatedBlocks / 2e3)));
+  const batchSize = estimatedBlocks > 1e3 ? 200 : 100;
+  const severity = estimatedBlocks > 5e3 ? "high" : "all";
   let defaults = {
     rootDir: directory,
     minSimilarity,
@@ -523,5 +536,6 @@ function generateSummary(results) {
 0 && (module.exports = {
   analyzePatterns,
   detectDuplicatePatterns,
-  generateSummary
+  generateSummary,
+  getSmartDefaults
 });

package/dist/index.mjs

@@ -1,10 +1,12 @@
 import {
   analyzePatterns,
   detectDuplicatePatterns,
-  generateSummary
-} from "./chunk-S2KQFII2.mjs";
+  generateSummary,
+  getSmartDefaults
+} from "./chunk-GSJFORRO.mjs";
 export {
   analyzePatterns,
   detectDuplicatePatterns,
-  generateSummary
+  generateSummary,
+  getSmartDefaults
 };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@aiready/pattern-detect",
-  "version": "0.7.9",
+  "version": "0.7.12",
   "description": "Semantic duplicate pattern detection for AI-generated code - finds similar implementations that waste AI context tokens",
   "main": "./dist/index.js",
   "module": "./dist/index.mjs",
@@ -45,7 +45,7 @@
   "dependencies": {
     "commander": "^14.0.0",
     "chalk": "^5.3.0",
-    "@aiready/core": "0.3.6"
+    "@aiready/core": "0.3.7"
   },
   "devDependencies": {
     "tsup": "^8.3.5",