ai-unit-test-generator 1.3.0

package/lib/core/scorer.mjs

@@ -0,0 +1,633 @@
+#!/usr/bin/env node
+
+import { readFileSync, writeFileSync, existsSync } from 'node:fs'
+
+async function req(mod, hint) { try { return await import(mod) } catch { throw new Error(`${mod} not installed. Run: npm i -D ${hint || mod}`) } }
+
+function parseArgs(argv) {
+  const args = {}
+  for (let i = 2; i < argv.length; i++) {
+    const a = argv[i]
+    if (a.startsWith('--')) {
+      const [k, v] = a.includes('=') ? a.split('=') : [a, argv[i + 1]]
+      args[k.replace(/^--/, '')] = v === undefined || v.startsWith('--') ? true : v
+      if (v !== undefined && !v.startsWith('--') && !a.includes('=')) i++
+    }
+  }
+  return args
+}
+
+function loadJson(p) {
+  if (!p || !existsSync(p)) return null
+  try { return JSON.parse(readFileSync(p, 'utf8')) } catch { return null }
+}
+
+function toFixedDown(num, digits = 2) { const m = Math.pow(10, digits); return Math.floor(num * m) / m }
+function clamp(n, min, max) { return Math.max(min, Math.min(max, n)) }
+
+function loadConfig(pathFromArg) { return loadJson(pathFromArg) || loadJson('ut_scoring_config.json') || {} }
+function pickWeight(cfg, key, def) { return cfg?.weights?.[key] ?? def }
+function pickThreshold(cfg, key, def) { return cfg?.thresholds?.[key] ?? def }
+
+function isMainChain(path, cfg) { const arr = (cfg?.mainChainPaths || []).map(s => String(s).toLowerCase()); const lower = (path || '').toLowerCase(); return arr.some(s => lower.includes(s)) }
+
+function mapBCByConfig({ name, path, impactHint }, cfg, overrides) {
+  const key = `${path}#${name}`
+  if (overrides?.BC?.[key] !== undefined) return overrides.BC[key]
+  const lower = `${name} ${path} ${impactHint || ''}`.toLowerCase()
+  let bc = 3
+  const kw = cfg?.bcKeywords || {}
+  const ordered = Object.keys(kw).map(Number).sort((a,b)=>b-a)
+  for (const score of ordered) { const list = kw[String(score)] || []; if (list.some(token => lower.includes(String(token).toLowerCase()))) { bc = score; break } }
+  if (!isMainChain(path, cfg)) { const cap = cfg?.bcCapForNonMainChain ?? 8; if (bc > cap) bc = cap }
+  return bc
+}
+
+function mapCCFromMetrics(metrics, cfg, eslintCognitive, target, overrides) {
+  const key = `${target.path}#${target.name}`
+  if (overrides?.CC?.[key] !== undefined) return overrides.CC[key]
+  if (!metrics) throw new Error('Function-level metrics missing')
+  
+  const cyclo = metrics?.cyclomatic ?? 0
+  const fusion = cfg?.ccFusion || {}
+  let cc = 3
+  
+  // P0-1: CC融合认知复杂度（默认策略）
+  const cognitive = eslintCognitive?.[key]
+  if (cognitive !== undefined) {
+    // 融合 cyclomatic + cognitive
+    const wC = fusion.wC ?? 0.7
+    const wK = fusion.wK ?? 0.3
+    const fused = wC * cyclo + wK * cognitive
+    cc = clamp(Math.floor(fused / 5) + 3, 2, fusion.cap ?? 10)
+  } else {
+    // 该函数未被 ESLint 分析（可能太简单），仅用 cyclomatic
+    let base = 3
+    const ranges = cfg?.ccMapping?.cyclomatic || []
+    for (const r of ranges) {
+      const gte = r.gte ?? -Infinity, lte = r.lte ?? Infinity, gt = r.gt, eq = r.eq
+      let hit = false
+      if (typeof eq === 'number') hit = cyclo === eq
+      else if (typeof gt === 'number') hit = cyclo > gt
+      else hit = cyclo >= gte && cyclo <= lte
+      if (hit) { base = r.score; break }
+    }
+    let adj = 0
+    for (const r of (cfg?.ccMapping?.adjustments || [])) { 
+      const val = metrics?.[r.field]
+      if (val === undefined || val === null) continue
+      const op = r.op || '>='
+      if ((op === '>=') && val >= r.value) adj += r.delta || 1
+      else if ((op === '>') && val > r.value) adj += r.delta || 1
+      else if ((op === '==') && val === r.value) adj += r.delta || 1
+    }
+    cc = clamp(base + Math.min(adj, 3), 2, cfg?.ccMapping?.cap ?? 10)
+  }
+  
+  // P1-2: 修复内部函数 LOC 加成读取配置路径
+  const internal = target.internal === true
+  const loc = target.loc || 0
+  const ccAdjust = cfg?.ccAdjust || {}
+  const locBonusThreshold = ccAdjust.locBonusThreshold ?? 50
+  const locBonus = ccAdjust.locBonus ?? 1
+  if (internal && loc >= locBonusThreshold) {
+    cc = clamp(cc + locBonus, 2, cfg?.ccMapping?.cap ?? 10)
+  }
+  
+  return cc
+}
+
+function mapLikelihoodFromGitByConfig(git, depGraphData, cfg) {
+  const rules = cfg?.likelihoodRules || []
+  const c30 = git?.commits30d ?? 0, c90 = git?.commits90d ?? 0, c180 = git?.commits180d ?? 0
+  let score = cfg?.fallbacks?.ERLikelihood ?? 3
+  for (const r of rules) {
+    if (r.field === 'commits30d') { if (r.op === '>=') { if (c30 >= r.value) { score = r.score; break } } if (r.op === 'between') { if (c30 >= r.min && c30 <= r.max) { score = r.score; break } } }
+    if (r.field === 'fallback90d' && r.op === 'gt') { if (c30 === 0 && c90 > r.value) { score = r.score; break } }
+    if (r.field === 'fallback180dZero' && r.op === 'eq') { if (c90 === 0 && c180 === 0 && r.value === true) { score = r.score; break } }
+  }
+  
+  // P0-2: depGraph 提升（基于真实依赖图）
+  const depCfg = cfg?.depGraph || {}
+  if (depCfg.enable && depGraphData) {
+    const { crossModuleScore, fanOut, fanIn } = depGraphData
+    if (crossModuleScore >= (depCfg.neighborCategoryBoost ?? 2)) score = clamp(score + 1, 1, cfg?.boostRules?.cap ?? 5)
+    if ((fanOut + fanIn) >= (depCfg.degreeBoost ?? 8)) score = clamp(score + 1, 1, cfg?.boostRules?.cap ?? 5)
+  }
+  
+  // Git 辅助信号提升
+  const boostCfg = cfg?.boostRules || {}
+  const boosted = ((git?.authors30d ?? 0) >= (boostCfg.authors30dGte ?? 999)) || git?.inCategory || git?.multiPlatform
+  if (boosted) score = clamp(score + 1, 1, boostCfg.cap ?? 5)
+  return score
+}
+
+function mapImpactFromHintByConfig(hint, cfg, localMap) {
+  if (localMap?.[hint] !== undefined) return localMap[hint]
+  const lower = (hint || '').toLowerCase()
+  const impactKw = cfg?.impactKeywords || {}
+  const ordered = Object.keys(impactKw).map(Number).sort((a,b)=>b-a)
+  for (const score of ordered) { const list = impactKw[String(score)] || []; if (list.some(token => lower.includes(String(token).toLowerCase()))) return score }
+  return 3
+}
+
+function mapERFromGitAndImpactConfig(git, impactHint, depGraphData, cfg, overrides, localImpact, target) {
+  const key = `${target.path}#${target.name}`
+  if (overrides?.ER?.[key] !== undefined) return overrides.ER[key]
+  const likelihood = mapLikelihoodFromGitByConfig(git, depGraphData, cfg)
+  const impact = mapImpactFromHintByConfig(impactHint, cfg, localImpact)
+  const matrix = cfg?.erMatrix || {}
+  return matrix?.[likelihood]?.[impact] ?? 6
+}
+
+function mapROIByConfig(hint, cfg, localMap, overrides, target) {
+  const key = `${target.path}#${target.name}`
+  if (overrides?.ROI?.[key] !== undefined) return overrides.ROI[key]
+  if (localMap?.[hint] !== undefined) return localMap[hint]
+  const r = cfg?.roiRules || {}
+  if (hint?.isPure) return r.pure ?? 10
+  if (hint?.dependenciesInjectable) return r.injectable ?? 9
+  if (hint?.multiPlatformStrong) return r.nativeOrNetwork ?? 5
+  if (hint?.needsUI) return r.needsUI ?? 3
+  return r.multiContext ?? 7
+}
+
+// 映射 testability（本质上就是 ROI 的重命名）
+function mapTestabilityByConfig(hint, cfg, localMap, overrides, target) {
+  const key = `${target.path}#${target.name}`
+  if (overrides?.Testability?.[key] !== undefined) return overrides.Testability[key]
+  if (localMap?.[hint] !== undefined) return localMap[hint]
+  const r = cfg?.testabilityRules || cfg?.roiRules || {}
+  if (hint?.isPure) return r.pure ?? 10
+  if (hint?.dependenciesInjectable) return r.injectable ?? 9
+  if (hint?.multiPlatformStrong) return r.nativeOrNetwork ?? 5
+  if (hint?.needsUI) return r.needsUI ?? 3
+  return r.multiContext ?? 7
+}
+
+// 计算依赖计数分数（被多少个模块引用）
+function mapDependencyCount(depGraphData, cfg) {
+  if (!depGraphData) return 2 // 默认分数
+  
+  const fanIn = depGraphData.fanIn || 0
+  const mapping = cfg?.dependencyCountMapping || [
+    { "gte": 10, "score": 10 },
+    { "gte": 5, "lt": 10, "score": 8 },
+    { "gte": 3, "lt": 5, "score": 6 },
+    { "gte": 1, "lt": 3, "score": 4 },
+    { "eq": 0, "score": 2 }
+  ]
+  
+  for (const rule of mapping) {
+    if (rule.eq !== undefined && fanIn === rule.eq) return rule.score
+    if (rule.gte !== undefined && rule.lt !== undefined) {
+      if (fanIn >= rule.gte && fanIn < rule.lt) return rule.score
+    }
+    if (rule.gte !== undefined && rule.lt === undefined) {
+      if (fanIn >= rule.gte) return rule.score
+    }
+  }
+  
+  return 2
+}
+// 传统评分（兼容旧配置）
+function computeScoreLegacy({ BC, CC, ER, ROI }, cfg) { 
+  const s = BC * pickWeight(cfg,'BC',0.4) + CC * pickWeight(cfg,'CC',0.3) + ER * pickWeight(cfg,'ER',0.2) + ROI * pickWeight(cfg,'ROI',0.1)
+  const score = toFixedDown(s, (cfg?.round?.digits ?? 2))
+  const P0 = pickThreshold(cfg,'P0',8.5), P1 = pickThreshold(cfg,'P1',6.5), P2 = pickThreshold(cfg,'P2',4.5)
+  let priority = 'P3'
+  if (score >= P0) priority = 'P0'
+  else if (score >= P1) priority = 'P1'
+  else if (score >= P2) priority = 'P2'
+  return { score, priority, layer: 'N/A' }
+}
+
+// 分层评分（新方法）
+function computeScoreLayered({ BC, CC, ER, testability, dependencyCount }, target, cfg) {
+  const layer = target.layer || 'unknown'
+  const layerDef = cfg?.layers?.[layer]
+  
+  if (!layerDef) {
+    // 如果没有层级定义，回退到传统评分
+    return computeScoreLegacy({ BC, CC, ER, ROI: testability }, cfg)
+  }
+  
+  const weights = layerDef.weights || {}
+  let score = 0
+  
+  // 根据层级定义的权重计算分数
+  if (weights.businessCriticality !== undefined) score += BC * weights.businessCriticality
+  if (weights.complexity !== undefined) score += CC * weights.complexity
+  if (weights.errorRisk !== undefined) score += ER * weights.errorRisk
+  if (weights.testability !== undefined) score += testability * weights.testability
+  if (weights.dependencyCount !== undefined) score += dependencyCount * weights.dependencyCount
+  
+  score = toFixedDown(score, cfg?.round?.digits ?? 2)
+  
+  // 使用层级特定的阈值
+  const thresholds = layerDef.thresholds || { P0: 8.0, P1: 6.5, P2: 4.5 }
+  let priority = 'P3'
+  if (score >= thresholds.P0) priority = 'P0'
+  else if (score >= thresholds.P1) priority = 'P1'
+  else if (score >= thresholds.P2) priority = 'P2'
+  
+  return { score, priority, layer, layerName: layerDef.name }
+}
+
+// 统一评分入口
+function computeScore(metrics, target, cfg) {
+  const mode = cfg?.scoringMode || 'legacy'
+  
+  if (mode === 'layered') {
+    return computeScoreLayered(metrics, target, cfg)
+  } else {
+    return computeScoreLegacy(metrics, cfg)
+  }
+}
+
+function defaultMd(rows, statusMap = new Map()) {
+  // 按总分降序排序（高分在前）
+  const sorted = [...rows].sort((a, b) => b.score - a.score)
+  
+  let md = '<!-- UT Priority Scoring Report -->\n'
+  md += '<!-- Format: Status can be "TODO" | "DONE" | "SKIP" -->\n'
+  md += '<!-- You can mark status by replacing TODO with DONE or SKIP -->\n\n'
+  md += '| Status | Score | Priority | Name | Type | Layer | Path | BC | CC | ER | Testability | DepCount | Notes |\n'
+  md += '|--------|-------|----------|------|------|-------|------|----|----|----|-----------|---------| ------- |\n'
+  
+  sorted.forEach(r => {
+    // 从状态映射中查找现有状态，否则默认为 TODO
+    const key = `${r.path}#${r.name}`
+    const status = statusMap.get(key) || 'TODO'
+    md += `| ${status} | ${r.score} | ${r.priority} | ${r.name} | ${r.type} | ${r.layerName || r.layer} | ${r.path} | ${r.BC} | ${r.CC} | ${r.ER} | ${r.testability || r.ROI} | ${r.dependencyCount || 'N/A'} | ${r.notes} |\n` 
+  })
+  
+  // 添加统计信息
+  md += '\n---\n\n'
+  md += '## 📊 Summary\n\n'
+  const p0 = sorted.filter(r => r.priority === 'P0').length
+  const p1 = sorted.filter(r => r.priority === 'P1').length
+  const p2 = sorted.filter(r => r.priority === 'P2').length
+  const p3 = sorted.filter(r => r.priority === 'P3').length
+  md += `- **Total Targets**: ${sorted.length}\n`
+  md += `- **P0 (Must Test)**: ${p0}\n`
+  md += `- **P1 (High Priority)**: ${p1}\n`
+  md += `- **P2 (Medium Priority)**: ${p2}\n`
+  md += `- **P3 (Low Priority)**: ${p3}\n\n`
+  md += '## 🎯 Quick Commands\n\n'
+  md += '```bash\n'
+  md += '# View P0 targets only\n'
+  md += 'grep "| TODO.*P0 |" reports/ut_scores.md\n\n'
+  md += '# Mark a target as DONE (example)\n'
+  md += 'sed -i "" "s/| TODO | 9.3 | P0 | findRecommendRoom/| DONE | 9.3 | P0 | findRecommendRoom/" reports/ut_scores.md\n\n'
+  md += '# Count remaining TODO items\n'
+  md += 'grep -c "| TODO |" reports/ut_scores.md\n'
+  md += '```\n'
+  
+  return md 
+}
+
+function defaultCsv(rows) {
+  // 按总分降序排序（高分在前）
+  const sorted = [...rows].sort((a, b) => b.score - a.score)
+  
+  const head = ['status','score','priority','name','path','type','layer','layerName','BC','CC','ER','testability','dependencyCount','notes'].join(',')
+  const body = sorted.map(r => [
+    'TODO',
+    r.score,
+    r.priority,
+    r.name,
+    r.path,
+    r.type,
+    r.layer || 'N/A',
+    r.layerName || 'N/A',
+    r.BC,
+    r.CC,
+    r.ER,
+    r.testability || r.ROI,
+    r.dependencyCount || 'N/A',
+    `"${r.notes}"`
+  ].join(',')).join('\n')
+  return head + '\n' + body + '\n'
+}
+
+async function buildFuncMetricsProvider(targets) {
+  const tsMorph = await req('ts-morph','ts-morph')
+  const escomplex = await req('escomplex','escomplex')
+  const { Project, SyntaxKind } = tsMorph
+  const project = new Project({ skipAddingFilesFromTsConfig: true })
+  const byPath = new Map(); targets.forEach(t => { if (!byPath.has(t.path)) byPath.set(t.path, true) }); Array.from(byPath.keys()).forEach(p => project.addSourceFileAtPathIfExists(p))
+  const byFunc = {}
+  const skipped = []
+  for (const t of targets) {
+    const sf = project.getSourceFile(t.path); if (!sf) { process.stderr.write(`⚠️  Source not found: ${t.path}\n`); continue }
+    let node = sf.getFunction(t.name)
+    if (!node) { const v = sf.getVariableDeclaration(t.name); const init = v?.getInitializer(); if (init && (init.getKind() === SyntaxKind.FunctionExpression || init.getKind() === SyntaxKind.ArrowFunction)) node = init }
+    if (!node) { skipped.push(`${t.name} in ${t.path}`); continue }
+    const text = node.getText(); let cyclomatic = 0
+    try { const res = escomplex.analyzeModule ? escomplex.analyzeModule(text) : null; cyclomatic = res?.aggregate?.cyclomatic ?? 0 } catch { skipped.push(`${t.name} (escomplex failed)`); continue }
+    byFunc[`${t.path}#${t.name}`] = { cyclomatic }
+  }
+  if (skipped.length > 0) process.stdout.write(`⚠️  Skipped ${skipped.length} targets (not functions or parse failed)\n`)
+  return { project, byFunc }
+}
+
+function pickMetricsForTarget(provider, target) { const key = `${target.path}#${target.name}`; const func = provider?.byFunc?.[key]; if (!func) return { cyclomatic: 1 }; return func }
+
+// P0-2: 构建文件依赖图
+function buildDepGraph(project, cfg) {
+  const categories = cfg?.crossModuleCategories || []
+  const graph = new Map()
+  
+  function topCategory(path) {
+    const parts = path.replace(/\\/g, '/').split('/')
+    const idx = parts.indexOf('src')
+    return idx >= 0 && idx + 1 < parts.length ? parts[idx + 1] : parts[0]
+  }
+  
+  for (const sf of project.getSourceFiles()) {
+    const path = sf.getFilePath().replace(process.cwd() + '/', '')
+    const deps = new Set()
+    const cat = topCategory(path)
+    
+    for (const imp of sf.getImportDeclarations()) {
+      const modSpec = imp.getModuleSpecifierValue()
+      if (!modSpec.startsWith('.')) continue
+      const impSf = imp.getModuleSpecifierSourceFile()
+      if (!impSf) continue
+      const resolved = sf.getDirectory().getRelativePathTo(impSf)
+      if (resolved) deps.add(resolved)
+    }
+    
+    const depCats = new Set()
+    deps.forEach(d => {
+      const dCat = topCategory(d)
+      if (dCat !== cat && categories.includes(dCat)) depCats.add(dCat)
+    })
+    
+    graph.set(path, {
+      category: cat,
+      deps: Array.from(deps),
+      crossModuleScore: depCats.size,
+      fanOut: deps.size,
+      fanIn: 0
+    })
+  }
+  
+  // 计算 fanIn
+  for (const [path, data] of graph.entries()) {
+    data.deps.forEach(dep => {
+      if (graph.has(dep)) graph.get(dep).fanIn++
+    })
+  }
+  
+  return graph
+}
+
+// P0-1: 加载 ESLint cognitive complexity
+function loadESLintCognitive(eslintJsonPath) {
+  const eslint = loadJson(eslintJsonPath)
+  if (!eslint || !Array.isArray(eslint)) return null
+  const cognitive = {}
+  
+  for (const file of eslint) {
+    const filePath = file.filePath?.replace(process.cwd() + '/', '')
+    if (!file.messages || !filePath) continue
+    
+    // 读取源文件以匹配函数名
+    let sourceLines = null
+    try {
+      sourceLines = readFileSync(file.filePath, 'utf8').split('\n')
+    } catch {
+      continue
+    }
+    
+    for (const msg of file.messages) {
+      if (msg.ruleId === 'sonarjs/cognitive-complexity' && msg.message) {
+        const complexityMatch = msg.message.match(/from\s+(\d+)\s+to/)
+        if (!complexityMatch) continue
+        const complexity = Number(complexityMatch[1])
+        
+        // 从消息行号附近查找函数名
+        const line = msg.line - 1 // 0-based
+        const contextLines = sourceLines.slice(Math.max(0, line - 2), line + 3).join('\n')
+        
+        // 匹配各种函数定义
+        const patterns = [
+          /export\s+(?:const|function)\s+(\w+)/,  // export const/function name
+          /const\s+(\w+)\s*=\s*\(/,               // const name = (
+          /function\s+(\w+)\s*\(/,                // function name(
+          /(\w+)\s*:\s*\([^)]*\)\s*=>/,          // name: () =>
+          /export\s+default\s+function\s+(\w+)/  // export default function name
+        ]
+        
+        let fnName = null
+        for (const pattern of patterns) {
+          const match = contextLines.match(pattern)
+          if (match) {
+            fnName = match[1]
+            break
+          }
+        }
+        
+        if (fnName) {
+          cognitive[`${filePath}#${fnName}`] = complexity
+        }
+      }
+    }
+  }
+  return Object.keys(cognitive).length ? cognitive : null
+}
+
+async function main() {
+  const args = parseArgs(process.argv)
+  const gitPath = args.git
+  const targetsPath = args.targets
+  const outMd = args['out-md']
+  const outCsv = args['out-csv']
+  const configPath = args['config']
+  const coverageJsonPath = args['coverage'] || 'coverage/coverage-summary.json'
+
+  // P2-1: 基本输入校验
+  const cfg = loadConfig(configPath)
+  if (!cfg || typeof cfg !== 'object') throw new Error('Invalid config file')
+  
+  // 根据评分模式验证配置
+  const mode = cfg.scoringMode || 'legacy'
+  if (mode === 'layered') {
+    if (!cfg.layers || typeof cfg.layers !== 'object') throw new Error('Layered mode requires cfg.layers')
+  } else {
+    if (!cfg.weights || !cfg.thresholds) throw new Error('Legacy mode requires cfg.weights and cfg.thresholds')
+  }
+  
+  // Git signals 可选（--skip-git 时为空对象）
+  const gitSignals = loadJson(gitPath) || {}
+  
+  const targets = loadJson(targetsPath)
+  if (!Array.isArray(targets) || targets.length === 0) throw new Error('Missing or empty targets array')
+
+  // P0-3: 加载项目内校准数据
+  const hintMaps = cfg?.hintMaps || {}
+  const localImpact = loadJson(hintMaps.impactLocal)
+  const localROI = loadJson(hintMaps.roiLocal)
+  const overrides = loadJson(cfg?.overrides)
+
+  // P0-1: 加载 ESLint cognitive
+  const eslintCognitive = loadESLintCognitive('reports/eslint.json') || null
+  if (!eslintCognitive) {
+    process.stdout.write('⚠️  ESLint cognitive complexity not available, using cyclomatic only\n')
+  }
+
+  const funcProvider = await buildFuncMetricsProvider(targets)
+  
+  // P0-2: 构建依赖图
+  const depGraph = cfg?.depGraph?.enable ? buildDepGraph(funcProvider.project, cfg) : null
+
+  // 可选覆盖率引导：读取 coverage-summary.json
+  let coverageSummary = null
+  if (existsSync(coverageJsonPath)) {
+    try { coverageSummary = JSON.parse(readFileSync(coverageJsonPath, 'utf8')) } catch {}
+  }
+
+  const rows = []
+  for (const t of targets) {
+    const { name, path, type = 'function', impactHint = '', roiHint = {}, internal = false, loc = 0 } = t
+    const metrics = pickMetricsForTarget(funcProvider, t)
+    const git = gitSignals[path] || null
+    const depGraphData = depGraph?.get(path) || null
+
+    const BC = mapBCByConfig({ name, path, impactHint }, cfg, overrides)
+    const CC = mapCCFromMetrics(metrics, cfg, eslintCognitive, t, overrides)
+    
+    // P0-4: 先计算 likelihood，再应用平台调整
+    let likelihood = 0
+    if (git) {
+      const rules = cfg?.likelihoodRules || []
+      const c30 = git?.commits30d ?? 0, c90 = git?.commits90d ?? 0, c180 = git?.commits180d ?? 0
+      likelihood = cfg?.fallbacks?.ERLikelihood ?? 3
+      for (const r of rules) {
+        if (r.field === 'commits30d') { if (r.op === '>=') { if (c30 >= r.value) { likelihood = r.score; break } } if (r.op === 'between') { if (c30 >= r.min && c30 <= r.max) { likelihood = r.score; break } } }
+        if (r.field === 'fallback90d' && r.op === 'gt') { if (c30 === 0 && c90 > r.value) { likelihood = r.score; break } }
+        if (r.field === 'fallback180dZero' && r.op === 'eq') { if (c90 === 0 && c180 === 0 && r.value === true) { likelihood = r.score; break } }
+      }
+    }
+    
+    // 应用平台调整（仅当 likelihood < 4 时）
+    const plat = cfg?.ccMapping?.platformAdjust
+    let CCFinal = CC
+    if (git?.multiPlatform && likelihood < (plat?.skipIfLikelihoodGte ?? 4)) {
+      CCFinal = clamp(CC + (plat?.delta ?? 0), 2, plat?.cap ?? 10)
+    }
+    
+    const ER = mapERFromGitAndImpactConfig(git, impactHint, depGraphData, cfg, overrides, localImpact, t)
+    const ROI = mapROIByConfig(roiHint, cfg, localROI, overrides, t)
+    const testability = mapTestabilityByConfig(roiHint, cfg, localROI, overrides, t)
+    const dependencyCount = mapDependencyCount(depGraphData, cfg)
+
+    let { score, priority, layer, layerName } = computeScore({ BC, CC: CCFinal, ER, ROI, testability, dependencyCount }, t, cfg)
+
+    // 覆盖率加权（可选）：对低覆盖模块提升优先级或分数
+    if (coverageSummary) {
+      const fileKey = Object.keys(coverageSummary).find(k => k.endsWith(path))
+      const fileCov = fileKey ? coverageSummary[fileKey] : null
+      const linesPct = fileCov?.lines?.pct
+      const cfgBoost = cfg?.coverageBoost || { enable: false }
+      if (cfgBoost.enable && typeof linesPct === 'number') {
+        const threshold = cfgBoost.threshold ?? 60
+        const maxBoost = cfgBoost.maxBoost ?? 0.5
+        if (linesPct < threshold) {
+          const ratio = (threshold - linesPct) / Math.max(threshold, 1)
+          const delta = toFixedDown(Math.min(maxBoost, ratio * (cfgBoost.scale ?? 0.5)), 2)
+          score = toFixedDown(score + delta, cfg?.round?.digits ?? 2)
+          // 重新判定优先级
+          const layerDef = cfg?.layers?.[layer]
+          const thresholds = (layerDef?.thresholds) || { P0: pickThreshold(cfg,'P0',8.5), P1: pickThreshold(cfg,'P1',6.5), P2: pickThreshold(cfg,'P2',4.5) }
+          if (score >= thresholds.P0) priority = 'P0'
+          else if (score >= thresholds.P1) priority = 'P1'
+          else if (score >= thresholds.P2) priority = 'P2'
+        }
+      }
+    }
+    
+    // 构建 notes
+    const notesParts = []
+        const cognitive = eslintCognitive?.[`${path}#${name}`]
+        if (cognitive !== undefined) notesParts.push('CC:fused')
+        else notesParts.push('CC:cyclo-only')
+    if (internal && loc >= (cfg?.ccAdjust?.locBonusThreshold ?? 50)) notesParts.push(`CC+${cfg?.ccAdjust?.locBonus ?? 1}loc`)
+    if (CCFinal > CC) notesParts.push('CC+platform')
+    if (depGraphData && depGraphData.crossModuleScore > 0) notesParts.push(`ER+depGraph(${depGraphData.crossModuleScore})`)
+    if (!git) notesParts.push('ER:fallback')
+    else notesParts.push('ER:git')
+    if (overrides) {
+      const key = `${path}#${name}`
+      if (overrides.BC?.[key]) notesParts.push('BC:override')
+      if (overrides.CC?.[key]) notesParts.push('CC:override')
+      if (overrides.ER?.[key]) notesParts.push('ER:override')
+      if (overrides.ROI?.[key]) notesParts.push('ROI:override')
+    }
+    if (localImpact) notesParts.push('Impact:local')
+    if (localROI) notesParts.push('ROI:local')
+    notesParts.push(internal ? 'Internal:Y' : 'Internal:N')
+    if (!isMainChain(path, cfg)) notesParts.push(`BC:cap≤${cfg?.bcCapForNonMainChain ?? 8}`)
+    
+    rows.push({ 
+      name, 
+      path, 
+      type, 
+      layer: layer || 'N/A',
+      layerName: layerName || 'N/A',
+      BC, 
+      CC: CCFinal, 
+      ER, 
+      ROI, 
+      testability, 
+      dependencyCount,
+      score, 
+      priority, 
+      notes: notesParts.join('; ') 
+    })
+  }
+
+  // 保留旧状态（DONE/SKIP）
+  const statusMap = readExistingStatus(outMd)
+  
+  if (outMd) writeFileSync(outMd, defaultMd(rows, statusMap))
+  if (outCsv) writeFileSync(outCsv, defaultCsv(rows))
+}
+
+/**
+ * 读取现有报告的状态信息
+ */
+function readExistingStatus(mdPath) {
+  const statusMap = new Map()
+  
+  if (!mdPath || !existsSync(mdPath)) {
+    return statusMap
+  }
+  
+  try {
+    const content = readFileSync(mdPath, 'utf-8')
+    const lines = content.split('\n')
+    
+    for (const line of lines) {
+      // 匹配表格行: | Status | ... | Name | ... | Path |
+      if (line.includes('| DONE |') || line.includes('| SKIP |')) {
+        const parts = line.split('|').map(p => p.trim()).filter(Boolean)
+        if (parts.length >= 4) {
+          const status = parts[0] // DONE 或 SKIP
+          const name = parts[3]   // Name 列
+          const path = parts[6]   // Path 列
+          const key = `${path}#${name}`
+          statusMap.set(key, status)
+        }
+      }
+    }
+  } catch (err) {
+    // 忽略读取错误，返回空 map
+  }
+  
+  return statusMap
+}
+
+main()

package/lib/index.js

@@ -0,0 +1,18 @@
+/**
+ * @ctrip/ut-priority-scorer
+ * 
+ * AI-friendly Unit Test priority scoring system
+ * Based on layered architecture and testability metrics
+ */
+
+export { default as genTargets } from './gen-targets.mjs'
+export { default as genGitSignals } from './gen-git-signals.mjs'
+export { default as score } from './score-ut.mjs'
+
+// 版本信息
+export const version = '1.0.0'
+
+// 默认配置路径
+export const DEFAULT_CONFIG_PATH = 'ut_scoring_config.json'
+export const DEFAULT_OUTPUT_DIR = 'reports'
+

package/lib/index.mjs

@@ -0,0 +1,25 @@
+/**
+ * ai-test-generator - AI-powered unit test generator with smart priority scoring
+ * 
+ * @example
+ * ```javascript
+ * // 使用 Core 模块进行代码分析
+ * import { scanCode, scoreTargets } from 'ai-test-generator'
+ * 
+ * // 使用 AI 模块生成测试
+ * import { buildPrompt, callAI, extractTests } from 'ai-test-generator/ai'
+ * 
+ * // 使用 Workflows 模块进行批量生成
+ * import { runBatch, runAll } from 'ai-test-generator/workflows'
+ * ```
+ */
+
+// 导出核心模块（最常用）
+export * from './core/index.mjs'
+
+// 导出其他模块（通过子路径导入）
+// import * from 'ai-test-generator/ai'
+// import * from 'ai-test-generator/testing'
+// import * from 'ai-test-generator/workflows'
+// import * from 'ai-test-generator/utils'
+