@aiready/pattern-detect 0.14.0 → 0.14.3

package/dist/cli.js

@@ -256,7 +256,7 @@ function calculateSimilarity(a, b) {
   return intersection.size / union.size;
 }
 async function detectDuplicatePatterns(fileContents, options) {
-  const { minSimilarity, minLines, streamResults } = options;
+  const { minSimilarity, minLines, streamResults, onProgress } = options;
   const allBlocks = [];
   for (const { file, content } of fileContents) {
     const blocks = extractBlocks(file, content);
@@ -265,12 +265,33 @@ async function detectDuplicatePatterns(fileContents, options) {
     );
   }
   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 norm1 = normalizeCode(b1.code);
     for (let j = i + 1; j < allBlocks.length; j++) {
-      const b1 = allBlocks[i];
+      comparisons++;
       const b2 = allBlocks[j];
       if (b1.file === b2.file) continue;
-      const norm1 = normalizeCode(b1.code);
       const norm2 = normalizeCode(b2.code);
       const sim = calculateSimilarity(norm1, norm2);
       if (sim >= minSimilarity) {
@@ -306,6 +327,13 @@ async function detectDuplicatePatterns(fileContents, options) {
       }
     }
   }
+  if (onProgress) {
+    onProgress(
+      totalComparisons,
+      totalComparisons,
+      `Duplicate detection complete. Found ${duplicates.length} patterns.`
+    );
+  }
   return duplicates.sort((a, b) => b.similarity - a.similarity);
 }
 
@@ -557,7 +585,8 @@ var PatternDetectProvider = {
         ),
         duplicates: results.duplicates,
         groups: results.groups,
-        clusters: results.clusters
+        clusters: results.clusters,
+        config: results.config
       },
       metadata: {
         toolName: import_core5.ToolName.PatternDetect,
@@ -615,29 +644,26 @@ 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(
-    3,
-    Math.min(10, Math.floor(3e4 / estimatedBlocks))
-  );
-  const minSimilarity = Math.min(0.75, 0.5 + estimatedBlocks / 1e4 * 0.25);
+  const fileCount = files.length;
+  const estimatedBlocks = fileCount * 5;
   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))
+    Math.min(20, 6 + Math.floor(estimatedBlocks / 1e3) * 2)
   );
+  const minSimilarity = Math.min(0.85, 0.5 + estimatedBlocks / 5e3 * 0.3);
   const batchSize = estimatedBlocks > 1e3 ? 200 : 100;
-  const severity = estimatedBlocks > 5e3 ? "high" : "all";
+  const severity = estimatedBlocks > 3e3 ? "high" : "all";
+  const maxCandidatesPerBlock = Math.max(
+    5,
+    Math.min(100, Math.floor(1e6 / estimatedBlocks))
+  );
   const defaults = {
     rootDir: directory,
     minSimilarity,
     minLines,
     batchSize,
     approx: true,
-    minSharedTokens,
+    minSharedTokens: 10,
     maxCandidatesPerBlock,
     streamResults: false,
     severity,
@@ -766,7 +792,7 @@ async function analyzePatterns(options) {
       minClusterFiles
     );
   }
-  return { results, duplicates, files, groups, clusters };
+  return { results, duplicates, files, groups, clusters, config: finalOptions };
 }
 function generateSummary(results) {
   const allIssues = results.flatMap((r) => r.issues);

package/dist/cli.mjs

@@ -3,7 +3,7 @@ import {
   analyzePatterns,
   filterBySeverity,
   generateSummary
-} from "./chunk-KDWGWBP5.mjs";
+} from "./chunk-KC2CQMG2.mjs";
 
 // src/cli.ts
 import { Command } from "commander";

package/dist/index.d.mts

@@ -150,6 +150,7 @@ declare function analyzePatterns(options: PatternDetectOptions): Promise<{
     files: string[];
     groups?: DuplicateGroup[];
     clusters?: RefactorCluster[];
+    config: PatternDetectOptions;
 }>;
 /**
  * Generate a summary of pattern analysis

package/dist/index.d.ts

@@ -150,6 +150,7 @@ declare function analyzePatterns(options: PatternDetectOptions): Promise<{
     files: string[];
     groups?: DuplicateGroup[];
     clusters?: RefactorCluster[];
+    config: PatternDetectOptions;
 }>;
 /**
  * Generate a summary of pattern analysis

package/dist/index.js

@@ -280,7 +280,7 @@ function calculateSimilarity(a, b) {
   return intersection.size / union.size;
 }
 async function detectDuplicatePatterns(fileContents, options) {
-  const { minSimilarity, minLines, streamResults } = options;
+  const { minSimilarity, minLines, streamResults, onProgress } = options;
   const allBlocks = [];
   for (const { file, content } of fileContents) {
     const blocks = extractBlocks(file, content);
@@ -289,12 +289,33 @@ async function detectDuplicatePatterns(fileContents, options) {
     );
   }
   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 norm1 = normalizeCode(b1.code);
     for (let j = i + 1; j < allBlocks.length; j++) {
-      const b1 = allBlocks[i];
+      comparisons++;
       const b2 = allBlocks[j];
       if (b1.file === b2.file) continue;
-      const norm1 = normalizeCode(b1.code);
       const norm2 = normalizeCode(b2.code);
       const sim = calculateSimilarity(norm1, norm2);
       if (sim >= minSimilarity) {
@@ -330,6 +351,13 @@ async function detectDuplicatePatterns(fileContents, options) {
       }
     }
   }
+  if (onProgress) {
+    onProgress(
+      totalComparisons,
+      totalComparisons,
+      `Duplicate detection complete. Found ${duplicates.length} patterns.`
+    );
+  }
   return duplicates.sort((a, b) => b.similarity - a.similarity);
 }
 
@@ -581,7 +609,8 @@ var PatternDetectProvider = {
         ),
         duplicates: results.duplicates,
         groups: results.groups,
-        clusters: results.clusters
+        clusters: results.clusters,
+        config: results.config
       },
       metadata: {
         toolName: import_core5.ToolName.PatternDetect,
@@ -639,29 +668,26 @@ 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(
-    3,
-    Math.min(10, Math.floor(3e4 / estimatedBlocks))
-  );
-  const minSimilarity = Math.min(0.75, 0.5 + estimatedBlocks / 1e4 * 0.25);
+  const fileCount = files.length;
+  const estimatedBlocks = fileCount * 5;
   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))
+    Math.min(20, 6 + Math.floor(estimatedBlocks / 1e3) * 2)
   );
+  const minSimilarity = Math.min(0.85, 0.5 + estimatedBlocks / 5e3 * 0.3);
   const batchSize = estimatedBlocks > 1e3 ? 200 : 100;
-  const severity = estimatedBlocks > 5e3 ? "high" : "all";
+  const severity = estimatedBlocks > 3e3 ? "high" : "all";
+  const maxCandidatesPerBlock = Math.max(
+    5,
+    Math.min(100, Math.floor(1e6 / estimatedBlocks))
+  );
   const defaults = {
     rootDir: directory,
     minSimilarity,
     minLines,
     batchSize,
     approx: true,
-    minSharedTokens,
+    minSharedTokens: 10,
     maxCandidatesPerBlock,
     streamResults: false,
     severity,
@@ -790,7 +816,7 @@ async function analyzePatterns(options) {
       minClusterFiles
     );
   }
-  return { results, duplicates, files, groups, clusters };
+  return { results, duplicates, files, groups, clusters, config: finalOptions };
 }
 function generateSummary(results) {
   const allIssues = results.flatMap((r) => r.issues);

package/dist/index.mjs

@@ -9,7 +9,7 @@ import {
   generateSummary,
   getSeverityLabel,
   getSmartDefaults
-} from "./chunk-KDWGWBP5.mjs";
+} from "./chunk-KC2CQMG2.mjs";
 export {
   PatternDetectProvider,
   Severity,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@aiready/pattern-detect",
-  "version": "0.14.0",
+  "version": "0.14.3",
   "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": "0.21.0"
+    "@aiready/core": "0.21.3"
   },
   "devDependencies": {
     "tsup": "^8.3.5",