@aiready/pattern-detect 0.7.7 → 0.7.11

package/dist/cli.mjs

@@ -2,7 +2,7 @@
 import {
   analyzePatterns,
   generateSummary
-} from "./chunk-YCGV65F5.mjs";
+} from "./chunk-GSJFORRO.mjs";
 
 // src/cli.ts
 import { Command } from "commander";
@@ -11,7 +11,7 @@ import { writeFileSync } from "fs";
 import { join } from "path";
 import { loadConfig, mergeConfigWithDefaults } from "@aiready/core";
 var program = new Command();
-program.name("aiready-patterns").description("Detect duplicate patterns in your codebase").version("0.1.0").addHelpText("after", "\nCONFIGURATION:\n  Supports config files: aiready.json, aiready.config.json, .aiready.json, .aireadyrc.json, aiready.config.js, .aireadyrc.js\n  CLI options override config file settings").argument("<directory>", "Directory to analyze").option("-s, --similarity <number>", "Minimum similarity score (0-1)").option("-l, --min-lines <number>", "Minimum lines to consider").option("--batch-size <number>", "Batch size for comparisons").option("--no-approx", "Disable approximate candidate selection (faster on small repos, slower on large)").option("--min-shared-tokens <number>", "Minimum shared tokens to consider a candidate").option("--max-candidates <number>", "Maximum candidates per block").option("--no-stream-results", "Disable incremental output (default: enabled)").option("--include <patterns>", "File patterns to include (comma-separated)").option("--exclude <patterns>", "File patterns to exclude (comma-separated)").option("--severity <level>", "Filter by severity: critical|high|medium|all").option("--include-tests", "Include test files in analysis (excluded by default)").option("--max-results <number>", "Maximum number of results to show in console output").option(
+program.name("aiready-patterns").description("Detect duplicate patterns in your codebase").version("0.1.0").addHelpText("after", "\nCONFIGURATION:\n  Supports config files: aiready.json, aiready.config.json, .aiready.json, .aireadyrc.json, aiready.config.js, .aireadyrc.js\n  CLI options override config file settings\n\nPARAMETER TUNING:\n  If you get too few results: decrease --similarity, --min-lines, or --min-shared-tokens\n  If analysis is too slow: increase --min-lines, --min-shared-tokens, or decrease --max-candidates\n  If you get too many false positives: increase --similarity or --min-lines\n\nEXAMPLES:\n  aiready-patterns .                                    # Basic analysis with smart defaults\n  aiready-patterns . --similarity 0.3 --min-lines 3     # More sensitive detection\n  aiready-patterns . --max-candidates 50 --no-approx    # Slower but more thorough\n  aiready-patterns . --output json > report.json       # JSON export").argument("<directory>", "Directory to analyze").option("-s, --similarity <number>", "Minimum similarity score (0-1). Lower = more results, higher = fewer but more accurate. Default: 0.4").option("-l, --min-lines <number>", "Minimum lines to consider. Lower = more results, higher = faster analysis. Default: 5").option("--batch-size <number>", "Batch size for comparisons. Higher = faster but more memory. Default: 100").option("--no-approx", "Disable approximate candidate selection. Slower but more thorough on small repos").option("--min-shared-tokens <number>", "Minimum shared tokens to consider a candidate. Higher = faster, fewer results. Default: 8").option("--max-candidates <number>", "Maximum candidates per block. Higher = more thorough but slower. Default: 100").option("--no-stream-results", "Disable incremental output (default: enabled)").option("--include <patterns>", "File patterns to include (comma-separated)").option("--exclude <patterns>", "File patterns to exclude (comma-separated)").option("--severity <level>", "Filter by severity: critical|high|medium|all. Default: all").option("--include-tests", "Include test files in analysis (excluded by default)").option("--max-results <number>", "Maximum number of results to show in console output. Default: 10").option(
   "-o, --output <format>",
   "Output format: console, json, html",
   "console"
@@ -163,6 +163,20 @@ program.name("aiready-patterns").description("Detect duplicate patterns in your
   }
   if (totalIssues === 0) {
     console.log(chalk.green("\n\u2728 Great! No duplicate patterns detected.\n"));
+    console.log(chalk.yellow("\u{1F4A1} If you expected to find duplicates, try adjusting parameters:"));
+    console.log(chalk.dim("   \u2022 Lower similarity threshold: --similarity 0.3"));
+    console.log(chalk.dim("   \u2022 Reduce minimum lines: --min-lines 3"));
+    console.log(chalk.dim("   \u2022 Include test files: --include-tests"));
+    console.log(chalk.dim("   \u2022 Lower shared tokens threshold: --min-shared-tokens 5"));
+    console.log("");
+  }
+  if (totalIssues > 0 && totalIssues < 5) {
+    console.log(chalk.yellow("\n\u{1F4A1} Few results found. To find more duplicates, try:"));
+    console.log(chalk.dim("   \u2022 Lower similarity threshold: --similarity 0.3"));
+    console.log(chalk.dim("   \u2022 Reduce minimum lines: --min-lines 3"));
+    console.log(chalk.dim("   \u2022 Include test files: --include-tests"));
+    console.log(chalk.dim("   \u2022 Lower shared tokens threshold: --min-shared-tokens 5"));
+    console.log("");
   }
   console.log(chalk.cyan(divider));
   if (totalIssues > 0) {

package/dist/index.d.mts

@@ -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.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
@@ -348,48 +361,31 @@ async function getSmartDefaults(directory, userOptions) {
   const { scanFiles: scanFiles2 } = await import("@aiready/core");
   const files = await scanFiles2(scanOptions);
   const estimatedBlocks = files.length * 3;
-  let defaults;
-  if (estimatedBlocks < 1e3) {
-    defaults = {
-      rootDir: directory,
-      minSimilarity: 0.4,
-      minLines: 5,
-      batchSize: 100,
-      approx: true,
-      minSharedTokens: 8,
-      maxCandidatesPerBlock: 100,
-      streamResults: false,
-      severity: "all",
-      includeTests: false
-    };
-  } else if (estimatedBlocks < 5e3) {
-    defaults = {
-      rootDir: directory,
-      minSimilarity: 0.5,
-      minLines: 6,
-      batchSize: 100,
-      approx: true,
-      minSharedTokens: 10,
-      maxCandidatesPerBlock: 50,
-      streamResults: false,
-      severity: "all",
-      includeTests: false
-    };
-  } else {
-    defaults = {
-      rootDir: directory,
-      minSimilarity: 0.6,
-      minLines: 8,
-      batchSize: 200,
-      approx: true,
-      minSharedTokens: 12,
-      maxCandidatesPerBlock: 25,
-      streamResults: false,
-      severity: "high",
-      includeTests: false
-    };
+  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,
+    minLines,
+    batchSize,
+    approx: true,
+    minSharedTokens,
+    maxCandidatesPerBlock,
+    streamResults: false,
+    severity,
+    includeTests: false
+  };
+  const result = { ...defaults };
+  for (const [key, value] of Object.entries(defaults)) {
+    if (key in userOptions && userOptions[key] !== void 0) {
+      result[key] = userOptions[key];
+    }
   }
-  return { ...defaults, ...scanOptions };
+  return result;
 }
 function logConfiguration(config, estimatedBlocks) {
   console.log("\u{1F4CB} Configuration:");
@@ -540,5 +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-YCGV65F5.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": "0.7.7",
+  "version": "0.7.11",
   "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": "0.3.3"
+    "@aiready/core": "0.3.6"
   },
   "devDependencies": {
     "tsup": "^8.3.5",