@aiready/pattern-detect 0.1.2 → 0.1.3

package/README.md

@@ -47,6 +47,9 @@ aiready-patterns ./src --similarity 0.9
 # Only look at larger patterns
 aiready-patterns ./src --min-lines 10
 
+# Memory optimization for large codebases
+aiready-patterns ./src --max-blocks 1000 --batch-size 200
+
 # Export to JSON
 aiready-patterns ./src --output json --output-file report.json
 
@@ -165,12 +168,37 @@ router.get('/posts/:id', createResourceHandler('Post', database.posts.findOne));
 
 ## ⚙️ Configuration
 
+### Common Options
+
 | Option | Description | Default |
 |--------|-------------|---------|
-| `minSimilarity` | Similarity threshold (0-1) | `0.85` |
+| `minSimilarity` | Similarity threshold (0-1). Use 0.40 for Jaccard (default), 0.85+ for Levenshtein | `0.40` |
 | `minLines` | Minimum lines to consider a pattern | `5` |
-| `include` | File patterns to include | `['**/*.ts', '**/*.js']` |
-| `exclude` | File patterns to exclude | `['**/node_modules/**', '**/*.test.*']` |
+| `maxBlocks` | Maximum code blocks to analyze (prevents OOM) | `500` |
+| `include` | File patterns to include | `['**/*.{ts,tsx,js,jsx,py,java}']` |
+| `exclude` | File patterns to exclude | See below |
+
+### Exclude Patterns (Default)
+
+By default, these patterns are excluded:
+```bash
+**/node_modules/**
+**/dist/**
+**/build/**
+**/.git/**
+**/coverage/**
+**/*.min.js
+**/*.bundle.js
+```
+
+Override with `--exclude` flag:
+```bash
+# Exclude test files and generated code
+aiready-patterns ./src --exclude "**/test/**,**/generated/**,**/__snapshots__/**"
+
+# Add to defaults (comma-separated)
+aiready-patterns ./src --exclude "**/node_modules/**,**/dist/**,**/build/**,**/*.spec.ts"
+```
 
 ## 📈 Understanding the Output
 
@@ -205,6 +233,78 @@ Estimated tokens wasted when AI tools process duplicate code:
 4. **Use pattern types**: Prioritize refactoring by category (API handlers → validators → utilities)
 5. **Export reports**: Generate HTML reports for team reviews
 
+## ⚠️ Performance & Memory
+
+### Algorithm Complexity
+
+**Fast Mode (default)**: **O(B × C × T)** where:
+- B = number of blocks
+- C = average candidates per block (~100)  
+- T = average tokens per block (~50)
+- **Jaccard similarity** is O(T) instead of O(N²) Levenshtein
+- **Default threshold: 0.40** (comprehensive detection including tests and helpers)
+
+**Exact Mode** (`--no-approx --no-fast-mode`): **O(B² × N²)** where:
+- B = number of blocks
+- N = average characters per block
+- **Levenshtein similarity** - more accurate, much slower
+- **Recommended threshold: 0.85+**
+- **Not recommended for >100 files**
+
+### Performance Benchmarks
+
+| Repo Size | Blocks | Fast Mode | Exact Mode |
+|-----------|--------|-----------|------------|
+| Small (<100 files) | ~50 | <1s | ~10s |
+| Medium (100-500 files) | ~500 | ~2s | ~8 min |
+| Large (500+ files) | ~500 (capped) | ~2s | ~76 min |
+
+**Example:** 828 code blocks → limited to 500 → **2.4s** (fast) vs **76 min** (exact)
+
+### Tuning Options
+
+```bash
+# Default (fast Jaccard mode, 40% threshold - comprehensive detection)
+aiready-patterns ./src
+
+# Higher threshold for only obvious duplicates
+aiready-patterns ./src --similarity 0.65
+
+# Lower threshold for more potential duplicates
+aiready-patterns ./src --similarity 0.55
+
+# Increase quality at cost of speed  
+aiready-patterns ./src --no-fast-mode --max-comparisons 100000
+
+# Exact mode with progress tracking (slowest, shows % and ETA)
+aiready-patterns ./src --no-approx --no-fast-mode --stream-results --max-blocks 100
+
+# Maximum speed (aggressive filtering)
+aiready-patterns ./src --max-blocks 200 --min-shared-tokens 12
+
+# Exact mode (slowest, most accurate)
+aiready-patterns ./src --no-approx --no-fast-mode --max-comparisons 500000
+```
+
+**CLI Options:**
+- `--stream-results` - Output duplicates as found (useful for long analysis)
+- `--no-fast-mode` - Use Levenshtein instead of Jaccard (more accurate, much slower)
+- `--no-approx` - Disable candidate filtering (enables progress % and ETA)
+- `--max-comparisons N` - Cap total comparisons (default 50K)
+- `--max-blocks N` - Limit blocks analyzed (default 500)
+
+**Progress Indicators:**
+- **Approx mode**: Shows blocks processed + duplicates found
+- **Exact mode**: Shows % complete, ETA, and comparisons processed
+- **Stream mode**: Prints each duplicate immediately when found
+
+**Recommendations:**
+- **< 100 files**: Use defaults, or try `--no-fast-mode` for higher accuracy
+- **100-500 files**: Use defaults with fast mode (2-5s typical)
+- **500-1000 files**: Use `--max-blocks 500 --min-lines 10` (~3-10s)
+- **1000+ files**: Use `--max-blocks 300 --min-lines 15` or analyze by module
+- **Slow analysis**: Add `--stream-results` to see progress in real-time
+
 ## 🔧 CI/CD Integration
 
 ### GitHub Actions

package/dist/chunk-4CZGZIDL.mjs

@@ -0,0 +1,409 @@
+// src/index.ts
+import { scanFiles, readFileContent } from "@aiready/core";
+
+// src/detector.ts
+import { similarityScore, estimateTokens } from "@aiready/core";
+function categorizePattern(code) {
+  const lower = code.toLowerCase();
+  if (lower.includes("request") && lower.includes("response") || lower.includes("router.") || lower.includes("app.get") || lower.includes("app.post") || lower.includes("express") || lower.includes("ctx.body")) {
+    return "api-handler";
+  }
+  if (lower.includes("validate") || lower.includes("schema") || lower.includes("zod") || lower.includes("yup") || lower.includes("if") && lower.includes("throw")) {
+    return "validator";
+  }
+  if (lower.includes("return (") || lower.includes("jsx") || lower.includes("component") || lower.includes("props")) {
+    return "component";
+  }
+  if (lower.includes("class ") || lower.includes("this.")) {
+    return "class-method";
+  }
+  if (lower.includes("return ") && !lower.includes("this") && !lower.includes("new ")) {
+    return "utility";
+  }
+  if (lower.includes("function") || lower.includes("=>")) {
+    return "function";
+  }
+  return "unknown";
+}
+function extractCodeBlocks(content, minLines) {
+  const lines = content.split("\n");
+  const blocks = [];
+  let currentBlock = [];
+  let blockStart = 0;
+  let braceDepth = 0;
+  let inFunction = false;
+  for (let i = 0; i < lines.length; i++) {
+    const line = lines[i];
+    const trimmed = line.trim();
+    if (!inFunction && (trimmed.includes("function ") || trimmed.includes("=>") || trimmed.includes("async ") || /^(export\s+)?(async\s+)?function\s+/.test(trimmed) || /^(export\s+)?const\s+\w+\s*=\s*(async\s*)?\(/.test(trimmed))) {
+      inFunction = true;
+      blockStart = i;
+    }
+    for (const char of line) {
+      if (char === "{") braceDepth++;
+      if (char === "}") braceDepth--;
+    }
+    if (inFunction) {
+      currentBlock.push(line);
+    }
+    if (inFunction && braceDepth === 0 && currentBlock.length >= minLines) {
+      const blockContent = currentBlock.join("\n");
+      const linesOfCode = currentBlock.filter(
+        (l) => l.trim() && !l.trim().startsWith("//")
+      ).length;
+      blocks.push({
+        content: blockContent,
+        startLine: blockStart + 1,
+        endLine: i + 1,
+        patternType: categorizePattern(blockContent),
+        linesOfCode
+      });
+      currentBlock = [];
+      inFunction = false;
+    } else if (inFunction && braceDepth === 0) {
+      currentBlock = [];
+      inFunction = false;
+    }
+  }
+  return blocks;
+}
+function normalizeCode(code) {
+  return code.replace(/\/\/.*$/gm, "").replace(/\/\*[\s\S]*?\*\//g, "").replace(/"[^"]*"/g, '"STR"').replace(/'[^']*'/g, "'STR'").replace(/`[^`]*`/g, "`STR`").replace(/\b\d+\b/g, "NUM").replace(/\s+/g, " ").trim();
+}
+function jaccardSimilarity(tokens1, tokens2) {
+  const set1 = new Set(tokens1);
+  const set2 = new Set(tokens2);
+  let intersection = 0;
+  for (const token of set1) {
+    if (set2.has(token)) intersection++;
+  }
+  const union = set1.size + set2.size - intersection;
+  return union === 0 ? 0 : intersection / union;
+}
+function calculateSimilarity(block1, block2) {
+  const norm1 = normalizeCode(block1);
+  const norm2 = normalizeCode(block2);
+  const baseSimilarity = similarityScore(norm1, norm2);
+  const tokens1 = norm1.split(/[\s(){}[\];,]+/).filter(Boolean);
+  const tokens2 = norm2.split(/[\s(){}[\];,]+/).filter(Boolean);
+  const tokenSimilarity = similarityScore(tokens1.join(" "), tokens2.join(" "));
+  return baseSimilarity * 0.4 + tokenSimilarity * 0.6;
+}
+async function detectDuplicatePatterns(files, options) {
+  const {
+    minSimilarity,
+    minLines,
+    maxBlocks = 500,
+    batchSize = 100,
+    approx = true,
+    minSharedTokens = 8,
+    maxCandidatesPerBlock = 100,
+    fastMode = true,
+    maxComparisons = 5e4,
+    // Cap at 50K comparisons by default
+    streamResults = false
+  } = options;
+  const duplicates = [];
+  let allBlocks = files.flatMap(
+    (file) => extractCodeBlocks(file.content, minLines).map((block) => ({
+      content: block.content,
+      startLine: block.startLine,
+      endLine: block.endLine,
+      file: file.file,
+      normalized: normalizeCode(block.content),
+      patternType: block.patternType,
+      tokenCost: estimateTokens(block.content),
+      linesOfCode: block.linesOfCode
+    }))
+  );
+  console.log(`Extracted ${allBlocks.length} code blocks for analysis`);
+  if (allBlocks.length > maxBlocks) {
+    console.log(`\u26A0\uFE0F  Limiting to ${maxBlocks} blocks (sorted by size) to prevent memory issues`);
+    console.log(`   Use --max-blocks to increase limit or --min-lines to filter smaller blocks`);
+    allBlocks = allBlocks.sort((a, b) => b.linesOfCode - a.linesOfCode).slice(0, maxBlocks);
+  }
+  const stopwords = /* @__PURE__ */ new Set([
+    "return",
+    "const",
+    "let",
+    "var",
+    "function",
+    "class",
+    "new",
+    "if",
+    "else",
+    "for",
+    "while",
+    "async",
+    "await",
+    "try",
+    "catch",
+    "switch",
+    "case",
+    "default",
+    "import",
+    "export",
+    "from",
+    "true",
+    "false",
+    "null",
+    "undefined",
+    "this"
+  ]);
+  const tokenize = (norm) => norm.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) {
+    for (let i = 0; i < blockTokens.length; i++) {
+      for (const tok of blockTokens[i]) {
+        let arr = invertedIndex.get(tok);
+        if (!arr) {
+          arr = [];
+          invertedIndex.set(tok, arr);
+        }
+        arr.push(i);
+      }
+    }
+  }
+  const totalComparisons = approx ? void 0 : allBlocks.length * (allBlocks.length - 1) / 2;
+  if (totalComparisons !== void 0) {
+    console.log(`Processing ${totalComparisons.toLocaleString()} comparisons in batches...`);
+  } else {
+    console.log(`Using approximate candidate selection to reduce comparisons...`);
+  }
+  let comparisonsProcessed = 0;
+  let comparisonsBudgetExhausted = false;
+  const startTime = Date.now();
+  for (let i = 0; i < allBlocks.length; i++) {
+    if (maxComparisons && comparisonsProcessed >= maxComparisons) {
+      comparisonsBudgetExhausted = true;
+      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)`);
+      } else {
+        console.log(`   Processed ${i.toLocaleString()}/${allBlocks.length} blocks (${elapsed}s elapsed, ${duplicatesFound} duplicates)`);
+      }
+      await new Promise((resolve) => setImmediate(resolve));
+    }
+    const block1 = allBlocks[i];
+    let candidates = null;
+    if (approx) {
+      const counts = /* @__PURE__ */ new Map();
+      for (const tok of blockTokens[i]) {
+        const ids = invertedIndex.get(tok);
+        if (!ids) continue;
+        for (const j of ids) {
+          if (j <= i) continue;
+          if (allBlocks[j].file === block1.file) continue;
+          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 }));
+    }
+    if (approx && candidates) {
+      for (const { j } of candidates) {
+        if (maxComparisons && comparisonsProcessed >= maxComparisons) break;
+        comparisonsProcessed++;
+        const block2 = allBlocks[j];
+        const similarity = fastMode ? jaccardSimilarity(blockTokens[i], blockTokens[j]) : calculateSimilarity(block1.content, block2.content);
+        if (similarity >= minSimilarity) {
+          const duplicate = {
+            file1: block1.file,
+            file2: block2.file,
+            line1: block1.startLine,
+            line2: block2.startLine,
+            endLine1: block1.endLine,
+            endLine2: block2.endLine,
+            similarity,
+            snippet: block1.content.split("\n").slice(0, 5).join("\n") + "\n...",
+            patternType: block1.patternType,
+            tokenCost: block1.tokenCost + block2.tokenCost,
+            linesOfCode: block1.linesOfCode
+          };
+          duplicates.push(duplicate);
+          if (streamResults) {
+            console.log(`
+   \u2705 Found: ${duplicate.patternType} ${Math.round(similarity * 100)}% similar`);
+            console.log(`      ${duplicate.file1}:${duplicate.line1}-${duplicate.endLine1} \u21D4 ${duplicate.file2}:${duplicate.line2}-${duplicate.endLine2}`);
+            console.log(`      Token cost: ${duplicate.tokenCost.toLocaleString()}`);
+          }
+        }
+      }
+    } else {
+      for (let j = i + 1; j < allBlocks.length; j++) {
+        if (maxComparisons && comparisonsProcessed >= maxComparisons) break;
+        comparisonsProcessed++;
+        const block2 = allBlocks[j];
+        if (block1.file === block2.file) continue;
+        const similarity = fastMode ? jaccardSimilarity(blockTokens[i], blockTokens[j]) : calculateSimilarity(block1.content, block2.content);
+        if (similarity >= minSimilarity) {
+          const duplicate = {
+            file1: block1.file,
+            file2: block2.file,
+            line1: block1.startLine,
+            line2: block2.startLine,
+            endLine1: block1.endLine,
+            endLine2: block2.endLine,
+            similarity,
+            snippet: block1.content.split("\n").slice(0, 5).join("\n") + "\n...",
+            patternType: block1.patternType,
+            tokenCost: block1.tokenCost + block2.tokenCost,
+            linesOfCode: block1.linesOfCode
+          };
+          duplicates.push(duplicate);
+          if (streamResults) {
+            console.log(`
+   \u2705 Found: ${duplicate.patternType} ${Math.round(similarity * 100)}% similar`);
+            console.log(`      ${duplicate.file1}:${duplicate.line1}-${duplicate.endLine1} \u21D4 ${duplicate.file2}:${duplicate.line2}-${duplicate.endLine2}`);
+            console.log(`      Token cost: ${duplicate.tokenCost.toLocaleString()}`);
+          }
+        }
+      }
+    }
+  }
+  if (comparisonsBudgetExhausted) {
+    console.log(`\u26A0\uFE0F  Comparison budget exhausted (${maxComparisons.toLocaleString()} comparisons). Use --max-comparisons to increase.`);
+  }
+  return duplicates.sort(
+    (a, b) => b.similarity - a.similarity || b.tokenCost - a.tokenCost
+  );
+}
+
+// src/index.ts
+function getRefactoringSuggestion(patternType, similarity) {
+  const baseMessages = {
+    "api-handler": "Extract common middleware or create a base handler class",
+    validator: "Consolidate validation logic into shared schema validators (Zod/Yup)",
+    utility: "Move to a shared utilities file and reuse across modules",
+    "class-method": "Consider inheritance or composition to share behavior",
+    component: "Extract shared logic into a custom hook or HOC",
+    function: "Extract into a shared helper function",
+    unknown: "Extract common logic into a reusable module"
+  };
+  const urgency = similarity > 0.95 ? " (CRITICAL: Nearly identical code)" : similarity > 0.9 ? " (HIGH: Very similar, refactor soon)" : "";
+  return baseMessages[patternType] + urgency;
+}
+async function analyzePatterns(options) {
+  const {
+    minSimilarity = 0.65,
+    // Lower default for fast Jaccard mode (Levenshtein would be 0.85+)
+    minLines = 5,
+    maxBlocks = 500,
+    batchSize = 100,
+    approx = true,
+    minSharedTokens = 8,
+    maxCandidatesPerBlock = 100,
+    fastMode = true,
+    maxComparisons = 5e4,
+    streamResults = false,
+    ...scanOptions
+  } = options;
+  const files = await scanFiles(scanOptions);
+  const results = [];
+  const fileContents = await Promise.all(
+    files.map(async (file) => ({
+      file,
+      content: await readFileContent(file)
+    }))
+  );
+  const duplicates = await detectDuplicatePatterns(fileContents, {
+    minSimilarity,
+    minLines,
+    maxBlocks,
+    batchSize,
+    approx,
+    minSharedTokens,
+    maxCandidatesPerBlock,
+    fastMode,
+    maxComparisons,
+    streamResults
+  });
+  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 severity = dup.similarity > 0.95 ? "critical" : dup.similarity > 0.9 ? "major" : "minor";
+      return {
+        type: "duplicate-pattern",
+        severity,
+        message: `${dup.patternType} pattern ${Math.round(dup.similarity * 100)}% similar to ${otherFile} (${dup.tokenCost} tokens wasted)`,
+        location: {
+          file,
+          line: dup.file1 === file ? dup.line1 : dup.line2
+        },
+        suggestion: getRefactoringSuggestion(dup.patternType, dup.similarity)
+      };
+    });
+    const totalTokenCost = fileDuplicates.reduce(
+      (sum, dup) => sum + dup.tokenCost,
+      0
+    );
+    results.push({
+      fileName: file,
+      issues,
+      metrics: {
+        tokenCost: totalTokenCost,
+        consistencyScore: Math.max(0, 1 - fileDuplicates.length * 0.1)
+      }
+    });
+  }
+  return results;
+}
+function generateSummary(results) {
+  const allIssues = results.flatMap((r) => r.issues);
+  const totalTokenCost = results.reduce(
+    (sum, r) => sum + (r.metrics.tokenCost || 0),
+    0
+  );
+  const patternsByType = {
+    "api-handler": 0,
+    validator: 0,
+    utility: 0,
+    "class-method": 0,
+    component: 0,
+    function: 0,
+    unknown: 0
+  };
+  allIssues.forEach((issue) => {
+    const match = issue.message.match(/^(\S+(?:-\S+)*) pattern/);
+    if (match) {
+      const type = match[1];
+      patternsByType[type] = (patternsByType[type] || 0) + 1;
+    }
+  });
+  const topDuplicates = allIssues.slice(0, 10).map((issue) => {
+    const similarityMatch = issue.message.match(/(\d+)% similar/);
+    const tokenMatch = issue.message.match(/\((\d+) tokens/);
+    const typeMatch = issue.message.match(/^(\S+(?:-\S+)*) pattern/);
+    const fileMatch = issue.message.match(/similar to (.+?) \(/);
+    return {
+      file1: issue.location.file,
+      file2: fileMatch?.[1] || "unknown",
+      similarity: similarityMatch ? parseInt(similarityMatch[1]) / 100 : 0,
+      patternType: typeMatch?.[1] || "unknown",
+      tokenCost: tokenMatch ? parseInt(tokenMatch[1]) : 0
+    };
+  });
+  return {
+    totalPatterns: allIssues.length,
+    totalTokenCost,
+    patternsByType,
+    topDuplicates
+  };
+}
+
+export {
+  detectDuplicatePatterns,
+  analyzePatterns,
+  generateSummary
+};