@aiready/pattern-detect 0.16.18 → 0.16.20

package/dist/chunk-I6ETJC7L.mjs

@@ -0,0 +1,179 @@
+// src/context-rules.ts
+import { Severity } from "@aiready/core";
+var CONTEXT_RULES = [
+  // Test Fixtures - Intentional duplication for test isolation
+  {
+    name: "test-fixtures",
+    detect: (file, code) => {
+      const isTestFile = file.includes(".test.") || file.includes(".spec.") || file.includes("__tests__") || file.includes("/test/") || file.includes("/tests/");
+      const hasTestFixtures = code.includes("beforeAll") || code.includes("afterAll") || code.includes("beforeEach") || code.includes("afterEach") || code.includes("setUp") || code.includes("tearDown");
+      return isTestFile && hasTestFixtures;
+    },
+    severity: Severity.Info,
+    reason: "Test fixture duplication is intentional for test isolation",
+    suggestion: "Consider if shared test setup would improve maintainability without coupling tests"
+  },
+  // Email/Document Templates - Often intentionally similar for consistency
+  {
+    name: "templates",
+    detect: (file, code) => {
+      const isTemplate = file.includes("/templates/") || file.includes("-template") || file.includes("/email-templates/") || file.includes("/emails/");
+      const hasTemplateContent = (code.includes("return") || code.includes("export")) && (code.includes("html") || code.includes("subject") || code.includes("body"));
+      return isTemplate && hasTemplateContent;
+    },
+    severity: Severity.Minor,
+    reason: "Template duplication may be intentional for maintainability and branding consistency",
+    suggestion: "Extract shared structure only if templates become hard to maintain"
+  },
+  // E2E/Integration Test Page Objects - Test independence
+  {
+    name: "e2e-page-objects",
+    detect: (file, code) => {
+      const isE2ETest = file.includes("e2e/") || file.includes("/e2e/") || file.includes(".e2e.") || file.includes("/playwright/") || file.includes("playwright/") || file.includes("/cypress/") || file.includes("cypress/") || file.includes("/integration/") || file.includes("integration/");
+      const hasPageObjectPatterns = code.includes("page.") || code.includes("await page") || code.includes("locator") || code.includes("getBy") || code.includes("selector") || code.includes("click(") || code.includes("fill(");
+      return isE2ETest && hasPageObjectPatterns;
+    },
+    severity: Severity.Minor,
+    reason: "E2E test duplication ensures test independence and reduces coupling",
+    suggestion: "Consider page object pattern only if duplication causes maintenance issues"
+  },
+  // Configuration Files - Often necessarily similar by design
+  {
+    name: "config-files",
+    detect: (file) => {
+      return file.endsWith(".config.ts") || file.endsWith(".config.js") || file.includes("jest.config") || file.includes("vite.config") || file.includes("webpack.config") || file.includes("rollup.config") || file.includes("tsconfig");
+    },
+    severity: Severity.Minor,
+    reason: "Configuration files often have similar structure by design",
+    suggestion: "Consider shared config base only if configurations become hard to maintain"
+  },
+  // Type Definitions - Duplication for type safety and module independence
+  {
+    name: "type-definitions",
+    detect: (file, code) => {
+      const isTypeFile = file.endsWith(".d.ts") || file.includes("/types/");
+      const hasTypeDefinitions = code.includes("interface ") || code.includes("type ") || code.includes("enum ");
+      return isTypeFile && hasTypeDefinitions;
+    },
+    severity: Severity.Info,
+    reason: "Type duplication may be intentional for module independence and type safety",
+    suggestion: "Extract to shared types package only if causing maintenance burden"
+  },
+  // Migration Scripts - One-off scripts that are similar by nature
+  {
+    name: "migration-scripts",
+    detect: (file) => {
+      return file.includes("/migrations/") || file.includes("/migrate/") || file.includes(".migration.");
+    },
+    severity: Severity.Info,
+    reason: "Migration scripts are typically one-off and intentionally similar",
+    suggestion: "Duplication is acceptable for migration scripts"
+  },
+  // Mock Data - Test data intentionally duplicated
+  {
+    name: "mock-data",
+    detect: (file, code) => {
+      const isMockFile = file.includes("/mocks/") || file.includes("/__mocks__/") || file.includes("/fixtures/") || file.includes(".mock.") || file.includes(".fixture.");
+      const hasMockData = code.includes("mock") || code.includes("Mock") || code.includes("fixture") || code.includes("stub") || code.includes("export const");
+      return isMockFile && hasMockData;
+    },
+    severity: Severity.Info,
+    reason: "Mock data duplication is expected for comprehensive test coverage",
+    suggestion: "Consider shared factories only for complex mock generation"
+  },
+  // Tool Implementations - Structural Boilerplate
+  {
+    name: "tool-implementations",
+    detect: (file, code) => {
+      const isToolFile = file.includes("/tools/") || file.endsWith(".tool.ts") || code.includes("toolDefinitions");
+      const hasToolStructure = code.includes("execute") && (code.includes("try") || code.includes("catch"));
+      return isToolFile && hasToolStructure;
+    },
+    severity: Severity.Info,
+    reason: "Tool implementations share structural boilerplate but have distinct business logic",
+    suggestion: "Tool duplication is acceptable for boilerplate interface wrappers"
+  }
+];
+function calculateSeverity(file1, file2, code, similarity, linesOfCode) {
+  for (const rule of CONTEXT_RULES) {
+    if (rule.detect(file1, code) || rule.detect(file2, code)) {
+      return {
+        severity: rule.severity,
+        reason: rule.reason,
+        suggestion: rule.suggestion,
+        matchedRule: rule.name
+      };
+    }
+  }
+  if (similarity >= 0.95 && linesOfCode >= 30) {
+    return {
+      severity: Severity.Critical,
+      reason: "Large nearly-identical code blocks waste tokens and create maintenance burden",
+      suggestion: "Extract to shared utility module immediately"
+    };
+  } else if (similarity >= 0.95 && linesOfCode >= 15) {
+    return {
+      severity: Severity.Major,
+      reason: "Nearly identical code should be consolidated",
+      suggestion: "Move to shared utility file"
+    };
+  } else if (similarity >= 0.85) {
+    return {
+      severity: Severity.Major,
+      reason: "High similarity indicates significant duplication",
+      suggestion: "Extract common logic to shared function"
+    };
+  } else if (similarity >= 0.7) {
+    return {
+      severity: Severity.Minor,
+      reason: "Moderate similarity detected",
+      suggestion: "Consider extracting shared patterns if code evolves together"
+    };
+  } else {
+    return {
+      severity: Severity.Minor,
+      reason: "Minor similarity detected",
+      suggestion: "Monitor but refactoring may not be worthwhile"
+    };
+  }
+}
+function getSeverityLabel(severity) {
+  const labels = {
+    [Severity.Critical]: "\u{1F534} CRITICAL",
+    [Severity.Major]: "\u{1F7E1} MAJOR",
+    [Severity.Minor]: "\u{1F535} MINOR",
+    [Severity.Info]: "\u2139\uFE0F  INFO"
+  };
+  return labels[severity];
+}
+function filterBySeverity(duplicates, minSeverity) {
+  const severityOrder = [
+    Severity.Info,
+    Severity.Minor,
+    Severity.Major,
+    Severity.Critical
+  ];
+  const minIndex = severityOrder.indexOf(minSeverity);
+  if (minIndex === -1) return duplicates;
+  return duplicates.filter((dup) => {
+    const dupIndex = severityOrder.indexOf(dup.severity);
+    return dupIndex >= minIndex;
+  });
+}
+function getSeverityThreshold(severity) {
+  const thresholds = {
+    [Severity.Critical]: 0.95,
+    [Severity.Major]: 0.85,
+    [Severity.Minor]: 0.5,
+    [Severity.Info]: 0
+  };
+  return thresholds[severity] || 0;
+}
+
+export {
+  CONTEXT_RULES,
+  calculateSeverity,
+  getSeverityLabel,
+  filterBySeverity,
+  getSeverityThreshold
+};

package/dist/chunk-JBUZ6YHE.mjs

@@ -0,0 +1,391 @@
+import {
+  detectDuplicatePatterns
+} from "./chunk-RS73WLNI.mjs";
+import {
+  calculateSeverity
+} from "./chunk-I6ETJC7L.mjs";
+
+// src/grouping.ts
+import { getSeverityLevel } from "@aiready/core";
+import path from "path";
+function groupDuplicatesByFilePair(duplicates) {
+  const groups = /* @__PURE__ */ new Map();
+  for (const dup of duplicates) {
+    const files = [dup.file1, dup.file2].sort();
+    const key = files.join("::");
+    if (!groups.has(key)) {
+      groups.set(key, {
+        filePair: key,
+        severity: dup.severity,
+        occurrences: 0,
+        totalTokenCost: 0,
+        averageSimilarity: 0,
+        patternTypes: /* @__PURE__ */ new Set(),
+        lineRanges: []
+      });
+    }
+    const group = groups.get(key);
+    group.occurrences++;
+    group.totalTokenCost += dup.tokenCost;
+    group.averageSimilarity += dup.similarity;
+    group.patternTypes.add(dup.patternType);
+    group.lineRanges.push({
+      file1: { start: dup.line1, end: dup.endLine1 },
+      file2: { start: dup.line2, end: dup.endLine2 }
+    });
+    const currentSev = dup.severity;
+    if (getSeverityLevel(currentSev) > getSeverityLevel(group.severity)) {
+      group.severity = currentSev;
+    }
+  }
+  return Array.from(groups.values()).map((g) => ({
+    ...g,
+    averageSimilarity: g.averageSimilarity / g.occurrences
+  }));
+}
+function createRefactorClusters(duplicates) {
+  const adjacency = /* @__PURE__ */ new Map();
+  const visited = /* @__PURE__ */ new Set();
+  const components = [];
+  for (const dup of duplicates) {
+    if (!adjacency.has(dup.file1)) adjacency.set(dup.file1, /* @__PURE__ */ new Set());
+    if (!adjacency.has(dup.file2)) adjacency.set(dup.file2, /* @__PURE__ */ new Set());
+    adjacency.get(dup.file1).add(dup.file2);
+    adjacency.get(dup.file2).add(dup.file1);
+  }
+  for (const file of adjacency.keys()) {
+    if (visited.has(file)) continue;
+    const component = [];
+    const queue = [file];
+    visited.add(file);
+    while (queue.length > 0) {
+      const curr = queue.shift();
+      component.push(curr);
+      for (const neighbor of adjacency.get(curr) || []) {
+        if (!visited.has(neighbor)) {
+          visited.add(neighbor);
+          queue.push(neighbor);
+        }
+      }
+    }
+    components.push(component);
+  }
+  const clusters = [];
+  for (const component of components) {
+    if (component.length < 2) continue;
+    const componentDups = duplicates.filter(
+      (d) => component.includes(d.file1) && component.includes(d.file2)
+    );
+    const totalTokenCost = componentDups.reduce(
+      (sum, d) => sum + d.tokenCost,
+      0
+    );
+    const avgSimilarity = componentDups.reduce((sum, d) => sum + d.similarity, 0) / Math.max(1, componentDups.length);
+    const name = determineClusterName(component);
+    const { severity, reason, suggestion } = calculateSeverity(
+      component[0],
+      component[1],
+      "",
+      // Code not available here
+      avgSimilarity,
+      30
+      // Assume substantial if clustered
+    );
+    clusters.push({
+      id: `cluster-${clusters.length}`,
+      name,
+      files: component,
+      severity,
+      duplicateCount: componentDups.length,
+      totalTokenCost,
+      averageSimilarity: avgSimilarity,
+      reason,
+      suggestion
+    });
+  }
+  return clusters;
+}
+function determineClusterName(files) {
+  if (files.length === 0) return "Unknown Cluster";
+  if (files.some((f) => f.includes("blog"))) return "Blog SEO Boilerplate";
+  if (files.some((f) => f.includes("buttons")))
+    return "Button Component Variants";
+  if (files.some((f) => f.includes("cards"))) return "Card Component Variants";
+  if (files.some((f) => f.includes("login.test"))) return "E2E Test Patterns";
+  const first = files[0];
+  const dirName = path.dirname(first).split(path.sep).pop();
+  if (dirName && dirName !== "." && dirName !== "..") {
+    return `${dirName.charAt(0).toUpperCase() + dirName.slice(1)} Domain Group`;
+  }
+  return "Shared Pattern Group";
+}
+function filterClustersByImpact(clusters, minTokenCost = 1e3, minFiles = 3) {
+  return clusters.filter(
+    (c) => c.totalTokenCost >= minTokenCost && c.files.length >= minFiles
+  );
+}
+
+// src/analyzer.ts
+import { scanFiles, readFileContent, Severity as Severity2, IssueType } from "@aiready/core";
+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 getSmartDefaults(directory, userOptions) {
+  if (userOptions.useSmartDefaults === false) {
+    return {
+      rootDir: directory,
+      minSimilarity: 0.6,
+      minLines: 8,
+      batchSize: 100,
+      approx: true,
+      minSharedTokens: 12,
+      maxCandidatesPerBlock: 5,
+      streamResults: false,
+      severity: "all",
+      includeTests: false
+    };
+  }
+  const scanOptions = {
+    rootDir: directory,
+    include: userOptions.include || ["**/*.{ts,tsx,js,jsx,py,java}"],
+    exclude: userOptions.exclude
+  };
+  const files = await scanFiles(scanOptions);
+  const fileCount = files.length;
+  const estimatedBlocks = fileCount * 5;
+  const minLines = Math.max(
+    6,
+    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 > 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: 10,
+    maxCandidatesPerBlock,
+    streamResults: false,
+    severity,
+    includeTests: false
+  };
+  const result = { ...defaults };
+  for (const key of Object.keys(defaults)) {
+    if (key in userOptions && userOptions[key] !== void 0) {
+      result[key] = userOptions[key];
+    }
+  }
+  return result;
+}
+function logConfiguration(config, estimatedBlocks) {
+  if (config.suppressToolConfig) return;
+  console.log("\u{1F4CB} Configuration:");
+  console.log(`   Repository size: ~${estimatedBlocks} code blocks`);
+  console.log(`   Similarity threshold: ${config.minSimilarity}`);
+  console.log(`   Minimum lines: ${config.minLines}`);
+  console.log(`   Approximate mode: ${config.approx ? "enabled" : "disabled"}`);
+  console.log(`   Max candidates per block: ${config.maxCandidatesPerBlock}`);
+  console.log(`   Min shared tokens: ${config.minSharedTokens}`);
+  console.log(`   Severity filter: ${config.severity}`);
+  console.log(`   Include tests: ${config.includeTests}`);
+  if (config.excludePatterns && config.excludePatterns.length > 0) {
+    console.log(`   Exclude patterns: ${config.excludePatterns.length} active`);
+  }
+  if (config.confidenceThreshold && config.confidenceThreshold > 0) {
+    console.log(`   Confidence threshold: ${config.confidenceThreshold}`);
+  }
+  if (config.ignoreWhitelist && config.ignoreWhitelist.length > 0) {
+    console.log(
+      `   Ignore whitelist: ${config.ignoreWhitelist.length} entries`
+    );
+  }
+  console.log("");
+}
+async function analyzePatterns(options) {
+  const smartDefaults = await getSmartDefaults(options.rootDir || ".", options);
+  const finalOptions = { ...smartDefaults, ...options };
+  const {
+    minSimilarity = 0.4,
+    minLines = 5,
+    batchSize = 100,
+    approx = true,
+    minSharedTokens = 8,
+    maxCandidatesPerBlock = 100,
+    streamResults = false,
+    severity = "all",
+    groupByFilePair = true,
+    createClusters = true,
+    minClusterTokenCost = 1e3,
+    minClusterFiles = 3,
+    excludePatterns = [],
+    confidenceThreshold = 0,
+    ignoreWhitelist = [],
+    ...scanOptions
+  } = finalOptions;
+  const files = await scanFiles(scanOptions);
+  const estimatedBlocks = files.length * 3;
+  logConfiguration(finalOptions, estimatedBlocks);
+  const results = [];
+  const READ_BATCH_SIZE = 50;
+  const fileContents = [];
+  for (let i = 0; i < files.length; i += READ_BATCH_SIZE) {
+    const batch = files.slice(i, i + READ_BATCH_SIZE);
+    const batchContents = await Promise.all(
+      batch.map(async (file) => ({
+        file,
+        content: await readFileContent(file)
+      }))
+    );
+    fileContents.push(...batchContents);
+  }
+  const duplicates = await detectDuplicatePatterns(fileContents, {
+    minSimilarity,
+    minLines,
+    batchSize,
+    approx,
+    minSharedTokens,
+    maxCandidatesPerBlock,
+    streamResults,
+    excludePatterns,
+    confidenceThreshold,
+    ignoreWhitelist,
+    onProgress: options.onProgress
+  });
+  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 severity2 = dup.similarity > 0.95 ? Severity2.Critical : dup.similarity > 0.9 ? Severity2.Major : Severity2.Minor;
+      return {
+        type: IssueType.DuplicatePattern,
+        severity: severity2,
+        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)
+      };
+    });
+    let filteredIssues = issues;
+    if (severity !== "all") {
+      const severityMap = {
+        critical: [Severity2.Critical],
+        high: [Severity2.Critical, Severity2.Major],
+        medium: [Severity2.Critical, Severity2.Major, Severity2.Minor]
+      };
+      const allowedSeverities = severityMap[severity] || [Severity2.Critical, Severity2.Major, Severity2.Minor];
+      filteredIssues = issues.filter(
+        (issue) => allowedSeverities.includes(issue.severity)
+      );
+    }
+    const totalTokenCost = fileDuplicates.reduce(
+      (sum, dup) => sum + dup.tokenCost,
+      0
+    );
+    results.push({
+      fileName: file,
+      issues: filteredIssues,
+      metrics: {
+        tokenCost: totalTokenCost,
+        consistencyScore: Math.max(0, 1 - fileDuplicates.length * 0.1)
+      }
+    });
+  }
+  let groups;
+  let clusters;
+  if (groupByFilePair) {
+    groups = groupDuplicatesByFilePair(duplicates);
+  }
+  if (createClusters) {
+    const allClusters = createRefactorClusters(duplicates);
+    clusters = filterClustersByImpact(
+      allClusters,
+      minClusterTokenCost,
+      minClusterFiles
+    );
+  }
+  return { results, duplicates, files, groups, clusters, config: finalOptions };
+}
+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 {
+      files: [
+        {
+          path: issue.location.file,
+          startLine: issue.location.line,
+          endLine: 0
+        },
+        {
+          path: fileMatch?.[1] || "unknown",
+          startLine: 0,
+          endLine: 0
+        }
+      ],
+      similarity: similarityMatch ? parseInt(similarityMatch[1]) / 100 : 0,
+      confidence: similarityMatch ? parseInt(similarityMatch[1]) / 100 : 0,
+      // Fallback for summary
+      patternType: typeMatch?.[1] || "unknown",
+      tokenCost: tokenMatch ? parseInt(tokenMatch[1]) : 0
+    };
+  });
+  return {
+    totalPatterns: allIssues.length,
+    totalTokenCost,
+    patternsByType,
+    topDuplicates
+  };
+}
+
+export {
+  groupDuplicatesByFilePair,
+  createRefactorClusters,
+  filterClustersByImpact,
+  getSmartDefaults,
+  analyzePatterns,
+  generateSummary
+};