@aiready/context-analyzer 0.5.1 → 0.6.0

package/dist/chunk-EX7HCWAO.mjs

@@ -0,0 +1,625 @@
+// src/index.ts
+import { scanFiles, readFileContent } from "@aiready/core";
+
+// src/analyzer.ts
+import { estimateTokens } from "@aiready/core";
+function buildDependencyGraph(files) {
+  const nodes = /* @__PURE__ */ new Map();
+  const edges = /* @__PURE__ */ new Map();
+  for (const { file, content } of files) {
+    const imports = extractImportsFromContent(content);
+    const exports = extractExports(content);
+    const tokenCost = estimateTokens(content);
+    const linesOfCode = content.split("\n").length;
+    nodes.set(file, {
+      file,
+      imports,
+      exports,
+      tokenCost,
+      linesOfCode
+    });
+    edges.set(file, new Set(imports));
+  }
+  return { nodes, edges };
+}
+function extractImportsFromContent(content) {
+  const imports = [];
+  const patterns = [
+    /import\s+.*?\s+from\s+['"](.+?)['"]/g,
+    // import ... from '...'
+    /import\s+['"](.+?)['"]/g,
+    // import '...'
+    /require\(['"](.+?)['"]\)/g
+    // require('...')
+  ];
+  for (const pattern of patterns) {
+    let match;
+    while ((match = pattern.exec(content)) !== null) {
+      const importPath = match[1];
+      if (importPath && !importPath.startsWith("@") && !importPath.startsWith("node:")) {
+        imports.push(importPath);
+      }
+    }
+  }
+  return [...new Set(imports)];
+}
+function calculateImportDepth(file, graph, visited = /* @__PURE__ */ new Set(), depth = 0) {
+  if (visited.has(file)) {
+    return depth;
+  }
+  const dependencies = graph.edges.get(file);
+  if (!dependencies || dependencies.size === 0) {
+    return depth;
+  }
+  visited.add(file);
+  let maxDepth = depth;
+  for (const dep of dependencies) {
+    const depDepth = calculateImportDepth(dep, graph, visited, depth + 1);
+    maxDepth = Math.max(maxDepth, depDepth);
+  }
+  visited.delete(file);
+  return maxDepth;
+}
+function getTransitiveDependencies(file, graph, visited = /* @__PURE__ */ new Set()) {
+  if (visited.has(file)) {
+    return [];
+  }
+  visited.add(file);
+  const dependencies = graph.edges.get(file);
+  if (!dependencies || dependencies.size === 0) {
+    return [];
+  }
+  const allDeps = [];
+  for (const dep of dependencies) {
+    allDeps.push(dep);
+    allDeps.push(...getTransitiveDependencies(dep, graph, visited));
+  }
+  return [...new Set(allDeps)];
+}
+function calculateContextBudget(file, graph) {
+  const node = graph.nodes.get(file);
+  if (!node) return 0;
+  let totalTokens = node.tokenCost;
+  const deps = getTransitiveDependencies(file, graph);
+  for (const dep of deps) {
+    const depNode = graph.nodes.get(dep);
+    if (depNode) {
+      totalTokens += depNode.tokenCost;
+    }
+  }
+  return totalTokens;
+}
+function detectCircularDependencies(graph) {
+  const cycles = [];
+  const visited = /* @__PURE__ */ new Set();
+  const recursionStack = /* @__PURE__ */ new Set();
+  function dfs(file, path) {
+    if (recursionStack.has(file)) {
+      const cycleStart = path.indexOf(file);
+      if (cycleStart !== -1) {
+        cycles.push([...path.slice(cycleStart), file]);
+      }
+      return;
+    }
+    if (visited.has(file)) {
+      return;
+    }
+    visited.add(file);
+    recursionStack.add(file);
+    path.push(file);
+    const dependencies = graph.edges.get(file);
+    if (dependencies) {
+      for (const dep of dependencies) {
+        dfs(dep, [...path]);
+      }
+    }
+    recursionStack.delete(file);
+  }
+  for (const file of graph.nodes.keys()) {
+    if (!visited.has(file)) {
+      dfs(file, []);
+    }
+  }
+  return cycles;
+}
+function calculateCohesion(exports, filePath) {
+  if (exports.length === 0) return 1;
+  if (exports.length === 1) return 1;
+  if (filePath && isTestFile(filePath)) {
+    return 1;
+  }
+  const domains = exports.map((e) => e.inferredDomain || "unknown");
+  const domainCounts = /* @__PURE__ */ new Map();
+  for (const domain of domains) {
+    domainCounts.set(domain, (domainCounts.get(domain) || 0) + 1);
+  }
+  const total = domains.length;
+  let entropy = 0;
+  for (const count of domainCounts.values()) {
+    const p = count / total;
+    if (p > 0) {
+      entropy -= p * Math.log2(p);
+    }
+  }
+  const maxEntropy = Math.log2(total);
+  return maxEntropy > 0 ? 1 - entropy / maxEntropy : 1;
+}
+function isTestFile(filePath) {
+  const lower = filePath.toLowerCase();
+  return lower.includes("test") || lower.includes("spec") || lower.includes("mock") || lower.includes("fixture") || lower.includes("__tests__") || lower.includes(".test.") || lower.includes(".spec.");
+}
+function calculateFragmentation(files, domain) {
+  if (files.length <= 1) return 0;
+  const directories = new Set(files.map((f) => f.split("/").slice(0, -1).join("/")));
+  return (directories.size - 1) / (files.length - 1);
+}
+function detectModuleClusters(graph) {
+  const domainMap = /* @__PURE__ */ new Map();
+  for (const [file, node] of graph.nodes.entries()) {
+    const domains = node.exports.map((e) => e.inferredDomain || "unknown");
+    const primaryDomain = domains[0] || "unknown";
+    if (!domainMap.has(primaryDomain)) {
+      domainMap.set(primaryDomain, []);
+    }
+    domainMap.get(primaryDomain).push(file);
+  }
+  const clusters = [];
+  for (const [domain, files] of domainMap.entries()) {
+    if (files.length < 2) continue;
+    const totalTokens = files.reduce((sum, file) => {
+      const node = graph.nodes.get(file);
+      return sum + (node?.tokenCost || 0);
+    }, 0);
+    const fragmentationScore = calculateFragmentation(files, domain);
+    const avgCohesion = files.reduce((sum, file) => {
+      const node = graph.nodes.get(file);
+      return sum + (node ? calculateCohesion(node.exports, file) : 0);
+    }, 0) / files.length;
+    const targetFiles = Math.max(1, Math.ceil(files.length / 3));
+    const consolidationPlan = generateConsolidationPlan(
+      domain,
+      files,
+      targetFiles
+    );
+    clusters.push({
+      domain,
+      files,
+      totalTokens,
+      fragmentationScore,
+      avgCohesion,
+      suggestedStructure: {
+        targetFiles,
+        consolidationPlan
+      }
+    });
+  }
+  return clusters.sort((a, b) => b.fragmentationScore - a.fragmentationScore);
+}
+function extractExports(content) {
+  const exports = [];
+  const patterns = [
+    /export\s+function\s+(\w+)/g,
+    /export\s+class\s+(\w+)/g,
+    /export\s+const\s+(\w+)/g,
+    /export\s+type\s+(\w+)/g,
+    /export\s+interface\s+(\w+)/g,
+    /export\s+default/g
+  ];
+  const types = [
+    "function",
+    "class",
+    "const",
+    "type",
+    "interface",
+    "default"
+  ];
+  patterns.forEach((pattern, index) => {
+    let match;
+    while ((match = pattern.exec(content)) !== null) {
+      const name = match[1] || "default";
+      const type = types[index];
+      const inferredDomain = inferDomain(name);
+      exports.push({ name, type, inferredDomain });
+    }
+  });
+  return exports;
+}
+function inferDomain(name) {
+  const lower = name.toLowerCase();
+  const domainKeywords = [
+    "authentication",
+    "authorization",
+    "payment",
+    "invoice",
+    "customer",
+    "product",
+    "order",
+    "cart",
+    "user",
+    "admin",
+    "repository",
+    "controller",
+    "service",
+    "config",
+    "model",
+    "view",
+    "auth",
+    "api",
+    "helper",
+    "util"
+  ];
+  for (const keyword of domainKeywords) {
+    const wordBoundaryPattern = new RegExp(`\\b${keyword}\\b`, "i");
+    if (wordBoundaryPattern.test(name)) {
+      return keyword;
+    }
+  }
+  for (const keyword of domainKeywords) {
+    if (lower.includes(keyword)) {
+      return keyword;
+    }
+  }
+  return "unknown";
+}
+function generateConsolidationPlan(domain, files, targetFiles) {
+  const plan = [];
+  if (files.length <= targetFiles) {
+    return [`No consolidation needed for ${domain}`];
+  }
+  plan.push(
+    `Consolidate ${files.length} ${domain} files into ${targetFiles} cohesive file(s):`
+  );
+  const dirGroups = /* @__PURE__ */ new Map();
+  for (const file of files) {
+    const dir = file.split("/").slice(0, -1).join("/");
+    if (!dirGroups.has(dir)) {
+      dirGroups.set(dir, []);
+    }
+    dirGroups.get(dir).push(file);
+  }
+  plan.push(`1. Create unified ${domain} module file`);
+  plan.push(
+    `2. Move related functionality from ${files.length} scattered files`
+  );
+  plan.push(`3. Update imports in dependent files`);
+  plan.push(
+    `4. Remove old files after consolidation (verify with tests first)`
+  );
+  return plan;
+}
+
+// src/index.ts
+async function getSmartDefaults(directory, userOptions) {
+  const files = await scanFiles({
+    rootDir: directory,
+    include: userOptions.include,
+    exclude: userOptions.exclude
+  });
+  const estimatedBlocks = files.length;
+  let maxDepth;
+  let maxContextBudget;
+  let minCohesion;
+  let maxFragmentation;
+  if (estimatedBlocks < 100) {
+    maxDepth = 4;
+    maxContextBudget = 8e3;
+    minCohesion = 0.5;
+    maxFragmentation = 0.5;
+  } else if (estimatedBlocks < 500) {
+    maxDepth = 5;
+    maxContextBudget = 15e3;
+    minCohesion = 0.45;
+    maxFragmentation = 0.6;
+  } else if (estimatedBlocks < 2e3) {
+    maxDepth = 7;
+    maxContextBudget = 25e3;
+    minCohesion = 0.4;
+    maxFragmentation = 0.7;
+  } else {
+    maxDepth = 10;
+    maxContextBudget = 4e4;
+    minCohesion = 0.35;
+    maxFragmentation = 0.8;
+  }
+  return {
+    maxDepth,
+    maxContextBudget,
+    minCohesion,
+    maxFragmentation,
+    focus: "all",
+    includeNodeModules: false,
+    rootDir: userOptions.rootDir || directory,
+    include: userOptions.include,
+    exclude: userOptions.exclude
+  };
+}
+async function analyzeContext(options) {
+  const {
+    maxDepth = 5,
+    maxContextBudget = 1e4,
+    minCohesion = 0.6,
+    maxFragmentation = 0.5,
+    focus = "all",
+    includeNodeModules = false,
+    ...scanOptions
+  } = options;
+  const files = await scanFiles({
+    ...scanOptions,
+    // Only add node_modules to exclude if includeNodeModules is false
+    // The DEFAULT_EXCLUDE already includes node_modules, so this is only needed
+    // if user overrides the default exclude list
+    exclude: includeNodeModules && scanOptions.exclude ? scanOptions.exclude.filter((pattern) => pattern !== "**/node_modules/**") : scanOptions.exclude
+  });
+  const fileContents = await Promise.all(
+    files.map(async (file) => ({
+      file,
+      content: await readFileContent(file)
+    }))
+  );
+  const graph = buildDependencyGraph(fileContents);
+  const circularDeps = detectCircularDependencies(graph);
+  const clusters = detectModuleClusters(graph);
+  const fragmentationMap = /* @__PURE__ */ new Map();
+  for (const cluster of clusters) {
+    for (const file of cluster.files) {
+      fragmentationMap.set(file, cluster.fragmentationScore);
+    }
+  }
+  const results = [];
+  for (const { file } of fileContents) {
+    const node = graph.nodes.get(file);
+    if (!node) continue;
+    const importDepth = focus === "depth" || focus === "all" ? calculateImportDepth(file, graph) : 0;
+    const dependencyList = focus === "depth" || focus === "all" ? getTransitiveDependencies(file, graph) : [];
+    const contextBudget = focus === "all" ? calculateContextBudget(file, graph) : node.tokenCost;
+    const cohesionScore = focus === "cohesion" || focus === "all" ? calculateCohesion(node.exports, file) : 1;
+    const fragmentationScore = fragmentationMap.get(file) || 0;
+    const relatedFiles = [];
+    for (const cluster of clusters) {
+      if (cluster.files.includes(file)) {
+        relatedFiles.push(...cluster.files.filter((f) => f !== file));
+        break;
+      }
+    }
+    const { severity, issues, recommendations, potentialSavings } = analyzeIssues({
+      file,
+      importDepth,
+      contextBudget,
+      cohesionScore,
+      fragmentationScore,
+      maxDepth,
+      maxContextBudget,
+      minCohesion,
+      maxFragmentation,
+      circularDeps
+    });
+    const domains = [
+      ...new Set(node.exports.map((e) => e.inferredDomain || "unknown"))
+    ];
+    results.push({
+      file,
+      tokenCost: node.tokenCost,
+      linesOfCode: node.linesOfCode,
+      importDepth,
+      dependencyCount: dependencyList.length,
+      dependencyList,
+      circularDeps: circularDeps.filter((cycle) => cycle.includes(file)),
+      cohesionScore,
+      domains,
+      exportCount: node.exports.length,
+      contextBudget,
+      fragmentationScore,
+      relatedFiles,
+      severity,
+      issues,
+      recommendations,
+      potentialSavings
+    });
+  }
+  const issuesOnly = results.filter((r) => r.severity !== "info");
+  const sorted = issuesOnly.sort((a, b) => {
+    const severityOrder = { critical: 0, major: 1, minor: 2, info: 3 };
+    const severityDiff = severityOrder[a.severity] - severityOrder[b.severity];
+    if (severityDiff !== 0) return severityDiff;
+    return b.contextBudget - a.contextBudget;
+  });
+  return sorted.length > 0 ? sorted : results;
+}
+function generateSummary(results) {
+  if (results.length === 0) {
+    return {
+      totalFiles: 0,
+      totalTokens: 0,
+      avgContextBudget: 0,
+      maxContextBudget: 0,
+      avgImportDepth: 0,
+      maxImportDepth: 0,
+      deepFiles: [],
+      avgFragmentation: 0,
+      fragmentedModules: [],
+      avgCohesion: 0,
+      lowCohesionFiles: [],
+      criticalIssues: 0,
+      majorIssues: 0,
+      minorIssues: 0,
+      totalPotentialSavings: 0,
+      topExpensiveFiles: []
+    };
+  }
+  const totalFiles = results.length;
+  const totalTokens = results.reduce((sum, r) => sum + r.tokenCost, 0);
+  const totalContextBudget = results.reduce(
+    (sum, r) => sum + r.contextBudget,
+    0
+  );
+  const avgContextBudget = totalContextBudget / totalFiles;
+  const maxContextBudget = Math.max(...results.map((r) => r.contextBudget));
+  const avgImportDepth = results.reduce((sum, r) => sum + r.importDepth, 0) / totalFiles;
+  const maxImportDepth = Math.max(...results.map((r) => r.importDepth));
+  const deepFiles = results.filter((r) => r.importDepth >= 5).map((r) => ({ file: r.file, depth: r.importDepth })).sort((a, b) => b.depth - a.depth).slice(0, 10);
+  const avgFragmentation = results.reduce((sum, r) => sum + r.fragmentationScore, 0) / totalFiles;
+  const moduleMap = /* @__PURE__ */ new Map();
+  for (const result of results) {
+    for (const domain of result.domains) {
+      if (!moduleMap.has(domain)) {
+        moduleMap.set(domain, []);
+      }
+      moduleMap.get(domain).push(result);
+    }
+  }
+  const fragmentedModules = [];
+  for (const [domain, files] of moduleMap.entries()) {
+    if (files.length < 2) continue;
+    const fragmentationScore = files.reduce((sum, f) => sum + f.fragmentationScore, 0) / files.length;
+    if (fragmentationScore < 0.3) continue;
+    const totalTokens2 = files.reduce((sum, f) => sum + f.tokenCost, 0);
+    const avgCohesion2 = files.reduce((sum, f) => sum + f.cohesionScore, 0) / files.length;
+    const targetFiles = Math.max(1, Math.ceil(files.length / 3));
+    fragmentedModules.push({
+      domain,
+      files: files.map((f) => f.file),
+      totalTokens: totalTokens2,
+      fragmentationScore,
+      avgCohesion: avgCohesion2,
+      suggestedStructure: {
+        targetFiles,
+        consolidationPlan: [
+          `Consolidate ${files.length} ${domain} files into ${targetFiles} cohesive file(s)`,
+          `Current token cost: ${totalTokens2.toLocaleString()}`,
+          `Estimated savings: ${Math.floor(totalTokens2 * 0.3).toLocaleString()} tokens (30%)`
+        ]
+      }
+    });
+  }
+  fragmentedModules.sort((a, b) => b.fragmentationScore - a.fragmentationScore);
+  const avgCohesion = results.reduce((sum, r) => sum + r.cohesionScore, 0) / totalFiles;
+  const lowCohesionFiles = results.filter((r) => r.cohesionScore < 0.6).map((r) => ({ file: r.file, score: r.cohesionScore })).sort((a, b) => a.score - b.score).slice(0, 10);
+  const criticalIssues = results.filter((r) => r.severity === "critical").length;
+  const majorIssues = results.filter((r) => r.severity === "major").length;
+  const minorIssues = results.filter((r) => r.severity === "minor").length;
+  const totalPotentialSavings = results.reduce(
+    (sum, r) => sum + r.potentialSavings,
+    0
+  );
+  const topExpensiveFiles = results.sort((a, b) => b.contextBudget - a.contextBudget).slice(0, 10).map((r) => ({
+    file: r.file,
+    contextBudget: r.contextBudget,
+    severity: r.severity
+  }));
+  return {
+    totalFiles,
+    totalTokens,
+    avgContextBudget,
+    maxContextBudget,
+    avgImportDepth,
+    maxImportDepth,
+    deepFiles,
+    avgFragmentation,
+    fragmentedModules: fragmentedModules.slice(0, 10),
+    avgCohesion,
+    lowCohesionFiles,
+    criticalIssues,
+    majorIssues,
+    minorIssues,
+    totalPotentialSavings,
+    topExpensiveFiles
+  };
+}
+function analyzeIssues(params) {
+  const {
+    file,
+    importDepth,
+    contextBudget,
+    cohesionScore,
+    fragmentationScore,
+    maxDepth,
+    maxContextBudget,
+    minCohesion,
+    maxFragmentation,
+    circularDeps
+  } = params;
+  const issues = [];
+  const recommendations = [];
+  let severity = "info";
+  let potentialSavings = 0;
+  if (circularDeps.length > 0) {
+    severity = "critical";
+    issues.push(
+      `Part of ${circularDeps.length} circular dependency chain(s)`
+    );
+    recommendations.push("Break circular dependencies by extracting interfaces or using dependency injection");
+    potentialSavings += contextBudget * 0.2;
+  }
+  if (importDepth > maxDepth * 1.5) {
+    severity = severity === "critical" ? "critical" : "critical";
+    issues.push(`Import depth ${importDepth} exceeds limit by 50%`);
+    recommendations.push("Flatten dependency tree or use facade pattern");
+    potentialSavings += contextBudget * 0.3;
+  } else if (importDepth > maxDepth) {
+    severity = severity === "critical" ? "critical" : "major";
+    issues.push(`Import depth ${importDepth} exceeds recommended maximum ${maxDepth}`);
+    recommendations.push("Consider reducing dependency depth");
+    potentialSavings += contextBudget * 0.15;
+  }
+  if (contextBudget > maxContextBudget * 1.5) {
+    severity = severity === "critical" ? "critical" : "critical";
+    issues.push(`Context budget ${contextBudget.toLocaleString()} tokens is 50% over limit`);
+    recommendations.push("Split into smaller modules or reduce dependency tree");
+    potentialSavings += contextBudget * 0.4;
+  } else if (contextBudget > maxContextBudget) {
+    severity = severity === "critical" || severity === "major" ? severity : "major";
+    issues.push(`Context budget ${contextBudget.toLocaleString()} exceeds ${maxContextBudget.toLocaleString()}`);
+    recommendations.push("Reduce file size or dependencies");
+    potentialSavings += contextBudget * 0.2;
+  }
+  if (cohesionScore < minCohesion * 0.5) {
+    severity = severity === "critical" ? "critical" : "major";
+    issues.push(`Very low cohesion (${(cohesionScore * 100).toFixed(0)}%) - mixed concerns`);
+    recommendations.push("Split file by domain - separate unrelated functionality");
+    potentialSavings += contextBudget * 0.25;
+  } else if (cohesionScore < minCohesion) {
+    severity = severity === "critical" || severity === "major" ? severity : "minor";
+    issues.push(`Low cohesion (${(cohesionScore * 100).toFixed(0)}%)`);
+    recommendations.push("Consider grouping related exports together");
+    potentialSavings += contextBudget * 0.1;
+  }
+  if (fragmentationScore > maxFragmentation) {
+    severity = severity === "critical" || severity === "major" ? severity : "minor";
+    issues.push(`High fragmentation (${(fragmentationScore * 100).toFixed(0)}%) - scattered implementation`);
+    recommendations.push("Consolidate with related files in same domain");
+    potentialSavings += contextBudget * 0.3;
+  }
+  if (issues.length === 0) {
+    issues.push("No significant issues detected");
+    recommendations.push("File is well-structured for AI context usage");
+  }
+  if (isBuildArtifact(file)) {
+    issues.push("Detected build artifact (bundled/output file)");
+    recommendations.push("Exclude build outputs (e.g., cdk.out, dist, build, .next) from analysis");
+    severity = downgradeSeverity(severity);
+    potentialSavings = 0;
+  }
+  return { severity, issues, recommendations, potentialSavings: Math.floor(potentialSavings) };
+}
+function isBuildArtifact(filePath) {
+  const lower = filePath.toLowerCase();
+  return lower.includes("/node_modules/") || lower.includes("/dist/") || lower.includes("/build/") || lower.includes("/out/") || lower.includes("/output/") || lower.includes("/cdk.out/") || lower.includes("/.next/") || /\/asset\.[^/]+\//.test(lower);
+}
+function downgradeSeverity(s) {
+  switch (s) {
+    case "critical":
+      return "minor";
+    case "major":
+      return "minor";
+    case "minor":
+      return "info";
+    default:
+      return "info";
+  }
+}
+
+export {
+  getSmartDefaults,
+  analyzeContext,
+  generateSummary
+};