@aiready/pattern-detect 0.11.32 → 0.11.36

package/dist/cli.js

@@ -428,7 +428,9 @@ async function detectDuplicatePatterns(files, options) {
       linesOfCode: block.linesOfCode
     }))
   );
-  console.log(`Extracted ${allBlocks.length} code blocks for analysis`);
+  if (!options.onProgress) {
+    console.log(`Extracted ${allBlocks.length} code blocks for analysis`);
+  }
   const pythonFiles = files.filter((f) => f.file.toLowerCase().endsWith(".py"));
   if (pythonFiles.length > 0) {
     const { extractPythonPatterns: extractPythonPatterns2 } = await Promise.resolve().then(() => (init_python_extractor(), python_extractor_exports));
@@ -446,7 +448,9 @@ async function detectDuplicatePatterns(files, options) {
       linesOfCode: p.endLine - p.startLine + 1
     }));
     allBlocks.push(...pythonBlocks);
-    console.log(`Added ${pythonBlocks.length} Python patterns`);
+    if (!options.onProgress) {
+      console.log(`Added ${pythonBlocks.length} Python patterns`);
+    }
   }
   if (!approx && allBlocks.length > 500) {
     console.log(
@@ -484,7 +488,11 @@ async function detectDuplicatePatterns(files, options) {
     "undefined",
     "this"
   ]);
-  const tokenize = (norm) => norm.split(/[\s(){}\[\];,\.]+/).filter((t) => t && t.length >= 3 && !stopwords.has(t.toLowerCase()));
+  const tokenize = (norm) => {
+    const punctuation = "(){}[];.,";
+    const cleaned = norm.split("").map((ch) => punctuation.includes(ch) ? " " : ch).join("");
+    return cleaned.split(/\s+/).filter((t) => t && t.length >= 3 && !stopwords.has(t.toLowerCase()));
+  };
   const blockTokens = allBlocks.map((b) => tokenize(b.normalized));
   const invertedIndex = /* @__PURE__ */ new Map();
   if (approx) {
@@ -518,20 +526,24 @@ async function detectDuplicatePatterns(files, options) {
       break;
     }
     if (i % batchSize === 0 && i > 0) {
-      const elapsed = ((Date.now() - startTime) / 1e3).toFixed(1);
-      const duplicatesFound = duplicates.length;
-      if (totalComparisons !== void 0) {
-        const progress = (comparisonsProcessed / totalComparisons * 100).toFixed(1);
-        const remaining = totalComparisons - comparisonsProcessed;
-        const rate = comparisonsProcessed / parseFloat(elapsed);
-        const eta = remaining > 0 ? (remaining / rate).toFixed(0) : 0;
-        console.log(
-          `   ${progress}% (${comparisonsProcessed.toLocaleString()}/${totalComparisons.toLocaleString()} comparisons, ${elapsed}s elapsed, ~${eta}s remaining, ${duplicatesFound} duplicates)`
-        );
+      if (options.onProgress) {
+        options.onProgress(i, allBlocks.length, `pattern-detect: analyzing blocks`);
       } else {
-        console.log(
-          `   Processed ${i.toLocaleString()}/${allBlocks.length} blocks (${elapsed}s elapsed, ${duplicatesFound} duplicates)`
-        );
+        const elapsed = ((Date.now() - startTime) / 1e3).toFixed(1);
+        const duplicatesFound = duplicates.length;
+        if (totalComparisons !== void 0) {
+          const progress = (comparisonsProcessed / totalComparisons * 100).toFixed(1);
+          const remaining = totalComparisons - comparisonsProcessed;
+          const rate = comparisonsProcessed / parseFloat(elapsed);
+          const eta = remaining > 0 ? (remaining / rate).toFixed(0) : 0;
+          console.log(
+            `   ${progress}% (${comparisonsProcessed.toLocaleString()}/${totalComparisons.toLocaleString()} comparisons, ${elapsed}s elapsed, ~${eta}s remaining, ${duplicatesFound} duplicates)`
+          );
+        } else {
+          console.log(
+            `   Processed ${i.toLocaleString()}/${allBlocks.length} blocks (${elapsed}s elapsed, ${duplicatesFound} duplicates)`
+          );
+        }
       }
       await new Promise((resolve) => setImmediate(resolve));
     }
@@ -1000,7 +1012,7 @@ async function getSmartDefaults(directory, userOptions) {
   );
   const batchSize = estimatedBlocks > 1e3 ? 200 : 100;
   const severity = estimatedBlocks > 5e3 ? "high" : "all";
-  let defaults = {
+  const defaults = {
     rootDir: directory,
     minSimilarity,
     minLines,
@@ -1070,7 +1082,8 @@ async function analyzePatterns(options) {
     approx,
     minSharedTokens,
     maxCandidatesPerBlock,
-    streamResults
+    streamResults,
+    onProgress: options.onProgress
   });
   for (const file of files) {
     const fileDuplicates = duplicates.filter(

package/dist/cli.mjs

@@ -3,7 +3,7 @@ import {
   analyzePatterns,
   filterBySeverity,
   generateSummary
-} from "./chunk-SLDK5PQK.mjs";
+} from "./chunk-YSDOUNJJ.mjs";
 
 // src/cli.ts
 import { Command } from "commander";

package/dist/index.d.mts

@@ -57,6 +57,7 @@ interface DetectionOptions {
     maxCandidatesPerBlock?: number;
     maxComparisons?: number;
     streamResults?: boolean;
+    onProgress?: (processed: number, total: number, message: string) => void;
 }
 /**
  * Detect duplicate patterns across files with enhanced analysis
@@ -130,6 +131,7 @@ interface PatternDetectOptions extends ScanOptions {
     createClusters?: boolean;
     minClusterTokenCost?: number;
     minClusterFiles?: number;
+    onProgress?: (processed: number, total: number, message: string) => void;
 }
 interface PatternSummary {
     totalPatterns: number;

package/dist/index.d.ts

@@ -57,6 +57,7 @@ interface DetectionOptions {
     maxCandidatesPerBlock?: number;
     maxComparisons?: number;
     streamResults?: boolean;
+    onProgress?: (processed: number, total: number, message: string) => void;
 }
 /**
  * Detect duplicate patterns across files with enhanced analysis
@@ -130,6 +131,7 @@ interface PatternDetectOptions extends ScanOptions {
     createClusters?: boolean;
     minClusterTokenCost?: number;
     minClusterFiles?: number;
+    onProgress?: (processed: number, total: number, message: string) => void;
 }
 interface PatternSummary {
     totalPatterns: number;

package/dist/index.js

@@ -446,7 +446,9 @@ async function detectDuplicatePatterns(files, options) {
       linesOfCode: block.linesOfCode
     }))
   );
-  console.log(`Extracted ${allBlocks.length} code blocks for analysis`);
+  if (!options.onProgress) {
+    console.log(`Extracted ${allBlocks.length} code blocks for analysis`);
+  }
   const pythonFiles = files.filter((f) => f.file.toLowerCase().endsWith(".py"));
   if (pythonFiles.length > 0) {
     const { extractPythonPatterns: extractPythonPatterns2 } = await Promise.resolve().then(() => (init_python_extractor(), python_extractor_exports));
@@ -464,7 +466,9 @@ async function detectDuplicatePatterns(files, options) {
       linesOfCode: p.endLine - p.startLine + 1
     }));
     allBlocks.push(...pythonBlocks);
-    console.log(`Added ${pythonBlocks.length} Python patterns`);
+    if (!options.onProgress) {
+      console.log(`Added ${pythonBlocks.length} Python patterns`);
+    }
   }
   if (!approx && allBlocks.length > 500) {
     console.log(
@@ -502,7 +506,11 @@ async function detectDuplicatePatterns(files, options) {
     "undefined",
     "this"
   ]);
-  const tokenize = (norm) => norm.split(/[\s(){}\[\];,\.]+/).filter((t) => t && t.length >= 3 && !stopwords.has(t.toLowerCase()));
+  const tokenize = (norm) => {
+    const punctuation = "(){}[];.,";
+    const cleaned = norm.split("").map((ch) => punctuation.includes(ch) ? " " : ch).join("");
+    return cleaned.split(/\s+/).filter((t) => t && t.length >= 3 && !stopwords.has(t.toLowerCase()));
+  };
   const blockTokens = allBlocks.map((b) => tokenize(b.normalized));
   const invertedIndex = /* @__PURE__ */ new Map();
   if (approx) {
@@ -536,20 +544,24 @@ async function detectDuplicatePatterns(files, options) {
       break;
     }
     if (i % batchSize === 0 && i > 0) {
-      const elapsed = ((Date.now() - startTime) / 1e3).toFixed(1);
-      const duplicatesFound = duplicates.length;
-      if (totalComparisons !== void 0) {
-        const progress = (comparisonsProcessed / totalComparisons * 100).toFixed(1);
-        const remaining = totalComparisons - comparisonsProcessed;
-        const rate = comparisonsProcessed / parseFloat(elapsed);
-        const eta = remaining > 0 ? (remaining / rate).toFixed(0) : 0;
-        console.log(
-          `   ${progress}% (${comparisonsProcessed.toLocaleString()}/${totalComparisons.toLocaleString()} comparisons, ${elapsed}s elapsed, ~${eta}s remaining, ${duplicatesFound} duplicates)`
-        );
+      if (options.onProgress) {
+        options.onProgress(i, allBlocks.length, `pattern-detect: analyzing blocks`);
       } else {
-        console.log(
-          `   Processed ${i.toLocaleString()}/${allBlocks.length} blocks (${elapsed}s elapsed, ${duplicatesFound} duplicates)`
-        );
+        const elapsed = ((Date.now() - startTime) / 1e3).toFixed(1);
+        const duplicatesFound = duplicates.length;
+        if (totalComparisons !== void 0) {
+          const progress = (comparisonsProcessed / totalComparisons * 100).toFixed(1);
+          const remaining = totalComparisons - comparisonsProcessed;
+          const rate = comparisonsProcessed / parseFloat(elapsed);
+          const eta = remaining > 0 ? (remaining / rate).toFixed(0) : 0;
+          console.log(
+            `   ${progress}% (${comparisonsProcessed.toLocaleString()}/${totalComparisons.toLocaleString()} comparisons, ${elapsed}s elapsed, ~${eta}s remaining, ${duplicatesFound} duplicates)`
+          );
+        } else {
+          console.log(
+            `   Processed ${i.toLocaleString()}/${allBlocks.length} blocks (${elapsed}s elapsed, ${duplicatesFound} duplicates)`
+          );
+        }
       }
       await new Promise((resolve) => setImmediate(resolve));
     }
@@ -1119,7 +1131,7 @@ async function getSmartDefaults(directory, userOptions) {
   );
   const batchSize = estimatedBlocks > 1e3 ? 200 : 100;
   const severity = estimatedBlocks > 5e3 ? "high" : "all";
-  let defaults = {
+  const defaults = {
     rootDir: directory,
     minSimilarity,
     minLines,
@@ -1189,7 +1201,8 @@ async function analyzePatterns(options) {
     approx,
     minSharedTokens,
     maxCandidatesPerBlock,
-    streamResults
+    streamResults,
+    onProgress: options.onProgress
   });
   for (const file of files) {
     const fileDuplicates = duplicates.filter(

package/dist/index.mjs

@@ -7,7 +7,7 @@ import {
   generateSummary,
   getSeverityLabel,
   getSmartDefaults
-} from "./chunk-SLDK5PQK.mjs";
+} from "./chunk-YSDOUNJJ.mjs";
 export {
   analyzePatterns,
   calculatePatternScore,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@aiready/pattern-detect",
-  "version": "0.11.32",
+  "version": "0.11.36",
   "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": "0.9.33"
+    "@aiready/core": "0.9.37"
   },
   "devDependencies": {
     "tsup": "^8.3.5",