@raolin2025/claude-code-node 2.0.0 → 2.2.0

package/src/git/pr-reviewer.js

@@ -0,0 +1,533 @@
+/**
+ * PR 自动审查引擎
+ * 结合规则检查 + LLM 智能评估
+ */
+
+import { parseDiff, splitDiffByFile, countFileChanges } from './utils/diff-parser.js'
+import { GitHubAPI } from './github-api.js'
+import { callLLM, buildPRReviewPrompt } from './llm-assistant.js'
+
+/**
+ * PR 审查器
+ */
+export class PRReviewer {
+  /**
+   * @param {GitHubAPI} github - GitHub API 实例
+   * @param {Object} rules - 审查规则配置
+   */
+  constructor(github, rules = {}, options = {}) {
+    this.github = github
+    this.rules = {
+      checks: {
+        codeQuality: true,
+        security: true,
+        tests: true,
+        docs: true,
+        complexity: true,
+        duplication: true
+      },
+      ...rules
+    }
+    this.apiConfig = options.apiConfig || {}
+    this.enableLLM = options.enableLLM !== false
+  }
+
+  /**
+   * 审查 PR
+   * @param {number} prNumber
+   * @param {Object} options - 审查选项
+   * @returns {Promise<ReviewResult>}
+   */
+  async reviewPR(prNumber, options = {}) {
+    const {
+      analyzeLLM = true,  // 是否使用 LLM 智能分析
+      commentThreshold = 'WARNING' // 'INFO' | 'WARNING' | 'ERROR'
+    } = options
+
+    // 1. 获取 PR 数据
+    const pr = await this.github.getPR(prNumber)
+    const diff = await this.github.getPRDiff(prNumber)
+    const files = await this.github.getPRFiles(prNumber)
+
+    // 2. 解析 diff
+    const fileMap = splitDiffByFile(diff)
+    const allHunks = parseDiff(diff)
+
+    // 3. 运行规则检查
+    const findings = await this.runAllChecks(pr, files, diff, fileMap)
+
+    // 4. LLM 智能分析（可选）
+    let llmSummary = null
+    if (analyzeLLM && this.enableLLM && findings.length > 0) {
+      llmSummary = await this.llmAnalysis(pr, findings, diff)
+    } else if (findings.length === 0) {
+      llmSummary = { summary: '✅ No issues found', riskLevel: 'LOW', recommendations: ['PR looks good to merge'] }
+    }
+
+    // 5. 生成审查报告
+    const report = this.generateReviewReport(pr, findings, llmSummary)
+
+    // 6. 准备评论列表（根据 threshold 过滤）
+    const comments = findings
+      .filter(f => this.shouldComment(f, commentThreshold))
+      .map(f => this.buildReviewComment(f))
+
+    return {
+      prNumber,
+      prTitle: pr.title,
+      prBody: pr.body,
+      findings,
+      summary: report,
+      llmSummary,
+      comments,
+      meta: {
+        totalFiles: files.length,
+        totalLinesChanged: files.reduce((sum, f) => sum + f.additions + f.deletions, 0),
+        reviewedAt: new Date().toISOString()
+      }
+    }
+  }
+
+  /**
+   * 执行所有检查
+   */
+  async runAllChecks(pr, files, diff, fileMap) {
+    const findings = []
+
+    // 并行运行所有检查器
+    const checks = []
+
+    if (this.rules.checks.codeQuality) {
+      checks.push(this.checkCodeQuality(files, fileMap))
+    }
+    if (this.rules.checks.security) {
+      checks.push(this.checkSecurity(files, fileMap))
+    }
+    if (this.rules.checks.tests) {
+      checks.push(this.checkTests(files, fileMap))
+    }
+    if (this.rules.checks.docs) {
+      checks.push(this.checkDocs(pr, files, fileMap))
+    }
+    if (this.rules.checks.complexity) {
+      checks.push(this.checkComplexity(files, fileMap))
+    }
+    if (this.rules.checks.duplication) {
+      checks.push(this.checkDuplication(files, fileMap))
+    }
+
+    const results = await Promise.all(checks)
+    findings.push(...results.flat())
+
+    return findings
+  }
+
+  // ============ 检查器实现 ============
+
+  /**
+   * 代码质量检查
+   */
+  async checkCodeQuality(files, fileMap) {
+    const findings = []
+    const patterns = {
+      'console.log': { severity: 'WARNING', message: 'Remove console.log before merging' },
+      'debugger': { severity: 'ERROR', message: 'Remove debugger statements' },
+      'FIXME': { severity: 'INFO', message: 'FIXME comment found' },
+      'TODO': { severity: 'INFO', message: 'TODO comment found' },
+      'XXX': { severity: 'WARNING', message: 'XXX comment indicates incomplete code' },
+      'var ': { severity: 'WARNING', message: 'Consider using const/let instead of var' },
+      '===': { severity: 'WARNING', message: 'Strict equality === is fine, but consider type-safe comparison' } // false positive demo
+    }
+
+    for (const [filePath, hunks] of fileMap.entries()) {
+      const isCodeFile = this.isCodeFile(filePath)
+      if (!isCodeFile) continue
+
+      const changes = countFileChanges(hunks)
+      if (changes.changes === 0) continue
+
+      for (const hunk of hunks) {
+        for (const line of hunk.lines) {
+          if (line.type !== '+') continue // 只检查新增的代码
+
+          const content = line.content
+          for (const [pattern, info] of Object.entries(patterns)) {
+            if (content.includes(pattern)) {
+              findings.push({
+                type: 'code-quality',
+                severity: info.severity,
+                file: filePath,
+                line: line.newLineNum,
+                column: 0,
+                message: `${info.message} (line: ${content.trim()})`,
+                snippet: content.trim(),
+                rule: pattern
+              })
+            }
+          }
+        }
+      }
+    }
+
+    return findings
+  }
+
+  /**
+   * 安全检查（关键漏洞检测）
+   */
+  async checkSecurity(files, fileMap) {
+    const findings = []
+    const securityPatterns = {
+      // 硬编码密钥
+      'password': { severity: 'ERROR', message: 'Potential hardcoded password' },
+      'secret': { severity: 'ERROR', message: 'Potential hardcoded secret' },
+      'api_key': { severity: 'ERROR', message: 'Potential API key' },
+      'token': { severity: 'WARNING', message: 'Potential token hardcoded' },
+      // SQL 注入风险
+      'SELECT .* FROM': { severity: 'ERROR', message: 'Possible SQL injection if string concatenation' },
+      'eval(': { severity: 'CRITICAL', message: 'eval() is dangerous' },
+      // 路径遍历
+      '../': { severity: 'WARNING', message: 'Path traversal risk' },
+      '__dirname': { severity: 'WARNING', message: 'Check path construction safety' },
+      // XSS
+      'innerHTML': { severity: 'WARNING', message: 'Potential XSS when using innerHTML' },
+      'document.write': { severity: 'ERROR', message: 'Avoid document.write' }
+    }
+
+    for (const [filePath, hunks] of fileMap.entries()) {
+      const isCodeFile = this.isCodeFile(filePath)
+      if (!isCodeFile) continue
+
+      for (const hunk of hunks) {
+        for (const line of hunk.lines) {
+          if (line.type !== '+') continue
+
+          const content = line.content.toLowerCase()
+          for (const [pattern, info] of Object.entries(securityPatterns)) {
+            const regex = new RegExp(pattern, 'i')
+            if (regex.test(content)) {
+              findings.push({
+                type: 'security',
+                severity: info.severity,
+                file: filePath,
+                line: line.newLineNum,
+                message: info.message,
+                snippet: line.content.trim(),
+                rule: pattern
+              })
+            }
+          }
+        }
+      }
+    }
+
+    return findings
+  }
+
+  /**
+   * 测试覆盖率检查
+   */
+  async checkTests(files, fileMap) {
+    const findings = []
+
+    // 检查是否有新增代码但缺少对应测试
+    const modifiedFiles = new Set()
+    const testFiles = new Set()
+
+    for (const [filePath] of fileMap.entries()) {
+      modifiedFiles.add(filePath)
+      if (this.isTestFile(filePath)) {
+        testFiles.add(filePath)
+      }
+    }
+
+    // 统计未覆盖的修改文件
+    for (const file of modifiedFiles) {
+      if (this.isTestFile(file)) continue
+
+      const hasCorrespondingTest = this.findTestFile(file, Array.from(testFiles))
+      if (!hasCorrespondingTest) {
+        const changes = countFileChanges(fileMap.get(file) || [])
+        if (changes.additions > 5) { // 只提示较明显的修改
+          findings.push({
+            type: 'tests',
+            severity: 'WARNING',
+            file,
+            line: null,
+            message: `New/modified code but missing test file for ${this.getFileName(file)}`,
+            rule: 'missing-test'
+          })
+        }
+      }
+    }
+
+    return findings
+  }
+
+  /**
+   * 文档检查
+   */
+  async checkDocs(pr, files, fileMap) {
+    const findings = []
+
+    // 检查 PR 描述是否足够详细
+    if (!pr.body || pr.body.trim().length < 50) {
+      findings.push({
+        type: 'docs',
+        severity: 'INFO',
+        file: null,
+        line: null,
+        message: 'PR description is brief; consider adding more context (purpose, testing steps, breaking changes)',
+        rule: 'pr-description'
+      })
+    }
+
+    // 检查 API 变更是否更新文档
+    const hasAPIFiles = Array.from(fileMap.keys()).some(p => p.includes('src/') || p.includes('lib/'))
+    const hasDocsFiles = Array.from(fileMap.keys()).some(p => p.includes('docs/') || p.includes('README'))
+
+    if (hasAPIFiles && !hasDocsFiles) {
+      findings.push({
+        type: 'docs',
+        severity: 'WARNING',
+        file: null,
+        line: null,
+        message: 'Code changes detected but no documentation updates found',
+        rule: 'missing-docs'
+      })
+    }
+
+    return findings
+  }
+
+  /**
+   * 复杂度检查（简易版：检查大函数）
+   */
+  async checkComplexity(files, fileMap) {
+    const findings = []
+    const complexityThreshold = 50 // 行数阈值
+
+    for (const [filePath, hunks] of fileMap.entries()) {
+      const isCodeFile = this.isCodeFile(filePath)
+      if (!isCodeFile) continue
+
+      // 简单统计：统计新增的连续行
+      for (const hunk of hunks) {
+        let addedBlockStart = null
+        let addedBlockLines = 0
+
+        for (const line of hunk.lines) {
+          if (line.type === '+') {
+            if (addedBlockStart === null) {
+              addedBlockStart = line.newLineNum
+            }
+            addedBlockLines++
+          } else {
+            if (addedBlockLines > 0 && addedBlockLines >= complexityThreshold) {
+              findings.push({
+                type: 'complexity',
+                severity: 'WARNING',
+                file: filePath,
+                line: addedBlockStart,
+                message: `Large addition (${addedBlockLines} lines) may indicate complex logic`,
+                rule: 'large-block'
+              })
+            }
+            addedBlockStart = null
+            addedBlockLines = 0
+          }
+        }
+      }
+    }
+
+    return findings
+  }
+
+  /**
+   * 重复代码检查（简易版：检测重复行）
+   */
+  async checkDuplication(files, fileMap) {
+    const findings = []
+    const addedLines = []
+
+    // 收集所有新增行
+    for (const [filePath, hunks] of fileMap.entries()) {
+      const isCodeFile = this.isCodeFile(filePath)
+      if (!isCodeFile) continue
+
+      for (const hunk of hunks) {
+        for (const line of hunk.lines) {
+          if (line.type === '+') {
+            addedLines.push({
+              file: filePath,
+              line: line.newLineNum,
+              content: line.content.trim()
+            })
+          }
+        }
+      }
+    }
+
+    // 检查重复行（简单 O(n^2) 检查，适合增量数量）
+    const duplicates = new Set()
+    for (let i = 0; i < addedLines.length; i++) {
+      for (let j = i + 1; j < addedLines.length; j++) {
+        if (addedLines[i].content === addedLines[j].content &&
+            addedLines[i].content.length > 20 && // 忽略短行
+            !duplicates.has(i) && !duplicates.has(j)) {
+          findings.push({
+            type: 'duplication',
+            severity: 'WARNING',
+            file: addedLines[i].file,
+            line: addedLines[i].line,
+            message: 'Potential duplicate code found in another location',
+            rule: 'duplicate-lines',
+            duplicateLocation: `${addedLines[j].file}:${addedLines[j].line}`
+          })
+          duplicates.add(i)
+          duplicates.add(j)
+        }
+      }
+    }
+
+    return findings
+  }
+
+  // ============ LLM 智能分析 ============
+
+  /**
+   * 使用 LLM 进行整体评估
+   */
+  async llmAnalysis(pr, findings, diff) {
+    try {
+      const prompt = buildPRReviewPrompt(pr, diff, findings)
+      const response = await callLLM(
+        'You are an expert code reviewer. Provide concise, actionable feedback.',
+        prompt,
+        this.apiConfig
+      )
+
+      // 解析 LLM 输出（简化：直接返回文本）
+      return {
+        summary: response,
+        riskLevel: this._inferRiskLevel(response),
+        recommendations: this._extractRecommendations(response)
+      }
+    } catch (error) {
+      console.error('LLM analysis failed:', error)
+      // 降级到规则统计
+      return this._fallbackAnalysis(findings)
+    }
+  }
+
+  _inferRiskLevel(text) {
+    const lower = text.toLowerCase()
+    if (lower.includes('reject') || lower.includes('critical')) return 'HIGH'
+    if (lower.includes('needs work') || lower.includes('issues')) return 'MEDIUM'
+    return 'LOW'
+  }
+
+  _extractRecommendations(text) {
+    const lines = text.split('\n')
+    const recs = []
+    for (const line of lines) {
+      if (line.match(/^\s*[-*•] /i) || line.includes('Recommendation') || line.includes('建议')) {
+        recs.push(line.trim())
+      }
+    }
+    return recs.length > 0 ? recs : ['See LLM analysis for details']
+  }
+
+  _fallbackAnalysis(findings) {
+    const severityCount = { CRITICAL: 0, ERROR: 0, WARNING: 0, INFO: 0 }
+    for (const f of findings) {
+      severityCount[f.severity] = (severityCount[f.severity] || 0) + 1
+    }
+    const total = findings.length
+    const riskLevel = total > 20 ? 'HIGH' : total > 10 ? 'MEDIUM' : 'LOW'
+    return {
+      summary: `PR 审查完成：发现 ${total} 个问题（CRITICAL: ${severityCount.CRITICAL}, ERROR: ${severityCount.ERROR}, WARNING: ${severityCount.WARNING}, INFO: ${severityCount.INFO})`,
+      riskLevel,
+      recommendations: total > 0
+        ? ['建议修复 CRITICAL 和 ERROR 级别问题', '注意安全问题', '补充测试用例']
+        : ['PR 质量良好，可以合并']
+    }
+  }
+
+  // ============ 辅助方法 ============
+
+  shouldComment(finding, threshold) {
+    const severityOrder = { CRITICAL: 4, ERROR: 3, WARNING: 2, INFO: 1 }
+    const thresholdOrder = { ERROR: 3, WARNING: 2, INFO: 1 }
+    return severityOrder[finding.severity] >= thresholdOrder[threshold]
+  }
+
+  buildReviewComment(finding) {
+    const emoji = this.getSeverityEmoji(finding.severity)
+    const body = `${emoji} **${finding.severity}** - ${finding.message}`
+
+    return {
+      body,
+      path: finding.file,
+      line: finding.line, // 保留原始行号，外层调用方决定是否转 position
+    }
+  }
+
+  getSeverityEmoji(severity) {
+    switch (severity) {
+      case 'CRITICAL': return '🔴'
+      case 'ERROR': return '🚫'
+      case 'WARNING': return '⚠️'
+      case 'INFO': return 'ℹ️'
+      default: return '📝'
+    }
+  }
+
+  generateReviewReport(pr, findings, llmSummary) {
+    const severityCount = { CRITICAL: 0, ERROR: 0, WARNING: 0, INFO: 0 }
+    for (const f of findings) {
+      severityCount[f.severity] = (severityCount[f.severity] || 0) + 1
+    }
+
+    const total = findings.length
+    const blockers = severityCount.CRITICAL + severityCount.ERROR
+
+    return {
+      overall: llmSummary?.riskLevel || (blockers > 0 ? 'NEEDS_WORK' : 'CLEAN'),
+      totalFindings: total,
+      severityCount,
+      blockers,
+      summary: llmSummary?.summary || `Review completed: ${total} issues found (${blockers} blocking)`,
+      recommendations: total > 0
+        ? ['Fix CRITICAL/ERROR issues before merging', 'Address WARNING items', 'Consider INFO suggestions']
+        : ['PR looks good to merge']
+    }
+  }
+
+  isCodeFile(filePath) {
+    const codeExt = ['.js', '.ts', '.jsx', '.tsx', '.py', '.java', '.go', '.rs', '.c', '.cpp', '.h', '.hpp']
+    return codeExt.some(ext => filePath.endsWith(ext))
+  }
+
+  isTestFile(filePath) {
+    const testPatterns = ['test', 'spec', '__tests__', '.test.', '.spec.']
+    return testPatterns.some(p => filePath.includes(p))
+  }
+
+  getFileName(filePath) {
+    return filePath.split('/').pop()
+  }
+
+  findTestFile(sourceFile, testFiles) {
+    const name = this.getFileName(sourceFile)
+    // 简单的命名约定：xxx.js → xxx.test.js / xxx.spec.js / test/xxx.test.js
+    const patterns = [
+      `test/${name}.test.js`,
+      `test/${name}.spec.js`,
+      `${name}.test.js`,
+      `${name}.spec.js`,
+      `__tests__/${name}.js`
+    ]
+    return patterns.some(p => testFiles.includes(p))
+  }
+}