cc-devflow 4.2.0 → 4.4.1

package/.claude/hooks/pre-tool-use-guardrail.ts

@@ -1,342 +0,0 @@
-#!/usr/bin/env node
-/**
- * ==============================================
- * PreToolUse Hook: Guardrail Enforcer
- * ==============================================
- *
- * 功能: 在文件编辑前检查 Guardrail Skills 的触发条件
- *      如果匹配到 guardrail 且 enforcement="block"，返回 exit code 2 阻止操作
- *
- * 集成: devflow-tdd-enforcer, constitution-guardian
- *
- * Exit Codes:
- *   0 - Allow (no violations)
- *   2 - Block (guardrail violation detected)
- *   1 - Error (unexpected failure)
- */
-
-import { readFileSync, existsSync } from 'fs';
-import { join } from 'path';
-
-// ============================================================================
-// 📋 Type Definitions
-// ============================================================================
-
-interface HookInput {
-    session_id: string;
-    file_path: string;
-    content: string;
-    tool_name: string;
-}
-
-interface FileTriggers {
-    pathPatterns?: string[];
-    contentPatterns?: string[];
-}
-
-interface SkipConditions {
-    sessionSkillUsed?: boolean;
-    fileMarkers?: string[];
-    envOverride?: string;
-}
-
-interface GuardrailSkill {
-    type: 'guardrail';
-    enforcement: 'block' | 'suggest' | 'warn';
-    priority: 'critical' | 'high' | 'medium' | 'low';
-    fileTriggers?: FileTriggers;
-    blockMessage?: string;
-    skipConditions?: SkipConditions;
-}
-
-interface SkillRules {
-    version: string;
-    skills: Record<string, GuardrailSkill>;
-}
-
-interface Violation {
-    skillName: string;
-    matchType: 'pathPattern' | 'contentPattern';
-    matchedPattern: string;
-    line?: number;
-}
-
-// ============================================================================
-// 📦 Helper Functions
-// ============================================================================
-
-/**
- * 归一化文件路径为项目相对路径
- *
- * 【问题】
- * hook payload 传递的是绝对路径，但 skill-rules.json 中的 glob 是相对路径
- * 直接比较会导致永不匹配
- *
- * 【解决】
- * 将绝对路径转为相对于项目根目录的路径
- *
- * @example
- * normalizeFilePath('/Users/dimon/cc-devflow/devflow/requirements/REQ-001/TASKS.md', '/Users/dimon/cc-devflow')
- * // → 'devflow/requirements/REQ-001/TASKS.md'
- */
-function normalizeFilePath(filePath: string, projectRoot: string): string {
-    // 如果是绝对路径且在项目内，转为相对路径
-    if (filePath.startsWith(projectRoot)) {
-        const relative = filePath.slice(projectRoot.length);
-        // 去除前导斜杠
-        return relative.startsWith('/') ? relative.slice(1) : relative;
-    }
-    // 已经是相对路径或不在项目内
-    return filePath;
-}
-
-/**
- * 检查文件路径是否匹配 glob 模式
- *
- * 【修复】
- * 1. 先转义正则元字符（防止 . + ^ $ 等被误解释）
- * 2. 再替换 glob 通配符（两个星号、一个星号、问号）
- *
- * 【为什么顺序重要】
- * 错误顺序: "foo.md".replace(星号→点星).escape() → "foo.md" (点匹配任意字符 ❌)
- * 正确顺序: "foo.md".escape().replace(星号→点星) → "foo\.md" (点匹配字面点号 ✓)
- *
- * @example
- * matchesPath('devflow/requirements/REQ-001/TASKS.md', 'devflow/requirements/双星/TASKS.md')
- * // → true
- */
-function matchesPath(filePath: string, pattern: string): boolean {
-    // Step 1: 转义所有正则特殊字符（保留 glob 通配符）
-    const escapeRegex = (str: string) => {
-        // 转义: . + ^ $ { } ( ) | [ ] \
-        // 不转义: * ? (因为这些是 glob 语法)
-        return str.replace(/[.+^${}()|[\]\\]/g, '\\$&');
-    };
-
-    // Step 2: 先转义，再替换 glob 通配符
-    // 注意：escapeRegex 不转义 *，所以字符串中的 * 仍然是 *，不是 \*
-    const regexPattern = escapeRegex(pattern)
-        .replace(/\*\*/g, '.*')          // ** → .* (任意字符，包括 /)
-        .replace(/\*/g, '[^/]*')         // * → [^/]* (任意字符，不包括 /)
-        .replace(/\?/g, '.');            // ? → . (单个字符)
-
-    const regex = new RegExp(`^${regexPattern}$`);
-    return regex.test(filePath);
-}
-
-/**
- * 检查内容是否匹配正则模式,并返回匹配的行号
- */
-function matchesContent(content: string, pattern: string): { matched: boolean; line?: number } {
-    const regex = new RegExp(pattern, 'im'); // i = case-insensitive, m = multiline
-    const match = regex.exec(content);
-
-    if (!match) {
-        return { matched: false };
-    }
-
-    // 计算匹配位置所在的行号
-    const textBeforeMatch = content.substring(0, match.index);
-    const lineNumber = textBeforeMatch.split('\n').length;
-
-    return { matched: true, line: lineNumber };
-}
-
-/**
- * 检查是否应该跳过此 guardrail
- */
-function shouldSkip(
-    skillName: string,
-    content: string,
-    skipConditions?: SkipConditions,
-    sessionId?: string
-): boolean {
-    if (!skipConditions) {
-        return false;
-    }
-
-    // Check 1: Environment variable override
-    if (skipConditions.envOverride && process.env[skipConditions.envOverride] === '1') {
-        console.error(`ℹ️  Skipping ${skillName}: ${skipConditions.envOverride}=1`);
-        return true;
-    }
-
-    // Check 2: File markers (e.g., @skip-tdd-check)
-    if (skipConditions.fileMarkers) {
-        for (const marker of skipConditions.fileMarkers) {
-            if (content.includes(marker)) {
-                console.error(`ℹ️  Skipping ${skillName}: Found marker '${marker}'`);
-                return true;
-            }
-        }
-    }
-
-    // Check 3: Session skill used (已实现完整持久化)
-    if (skipConditions.sessionSkillUsed && sessionId) {
-        const projectDir = process.env.CLAUDE_PROJECT_DIR || process.cwd();
-        const stateDir = join(projectDir, '.claude', 'hooks', 'state');
-        const stateFile = join(stateDir, `skills-used-${sessionId}.json`);
-
-        try {
-            if (existsSync(stateFile)) {
-                const usedSkills: string[] = JSON.parse(readFileSync(stateFile, 'utf-8'));
-                if (usedSkills.includes(skillName)) {
-                    console.error(`ℹ️  Skipping ${skillName}: Already used in this session`);
-                    return true;
-                }
-            }
-        } catch (error) {
-            // 状态文件损坏或读取失败，忽略错误（fail open）
-            console.error(`⚠️  Warning: Failed to read session state for ${skillName}:`, error);
-        }
-    }
-
-    return false;
-}
-
-// ============================================================================
-// 🎯 Main Logic
-// ============================================================================
-
-async function main() {
-    try {
-        // ===== 1. 读取输入 =====
-        const input = readFileSync(0, 'utf-8');
-        const data: HookInput = JSON.parse(input);
-
-        const { session_id, file_path, content, tool_name } = data;
-
-        // 只在文件编辑工具时触发 (Edit, Write)
-        if (!['Edit', 'Write'].includes(tool_name)) {
-            process.exit(0); // Allow non-edit operations
-        }
-
-        // ===== 2. 加载 Guardrail Rules =====
-        const projectDir = process.env.CLAUDE_PROJECT_DIR || process.cwd();
-
-        // ===== 归一化文件路径 =====
-        // hook payload 是绝对路径，需转为相对路径才能与 skill-rules.json 中的 glob 匹配
-        const normalizedPath = normalizeFilePath(file_path, projectDir);
-        const rulesPath = join(projectDir, '.claude', 'skills', 'skill-rules.json');
-
-        if (!existsSync(rulesPath)) {
-            // No skill rules configured, allow all
-            process.exit(0);
-        }
-
-        const rules: SkillRules = JSON.parse(readFileSync(rulesPath, 'utf-8'));
-
-        // ===== 3. 检查所有 Guardrail Skills =====
-        const violations: Violation[] = [];
-
-        for (const [skillName, config] of Object.entries(rules.skills)) {
-            // 只处理 type="guardrail" 且 enforcement="block"
-            if (config.type !== 'guardrail' || config.enforcement !== 'block') {
-                continue;
-            }
-
-            // 检查是否应该跳过
-            if (shouldSkip(skillName, content, config.skipConditions, session_id)) {
-                continue;
-            }
-
-            const triggers = config.fileTriggers;
-            if (!triggers) {
-                continue;
-            }
-
-            // Check 1: Path patterns
-            if (triggers.pathPatterns) {
-                for (const pattern of triggers.pathPatterns) {
-                    // 使用归一化后的相对路径进行匹配
-                    if (matchesPath(normalizedPath, pattern)) {
-                        // Path matched, now check content patterns
-                        if (triggers.contentPatterns) {
-                            for (const contentPattern of triggers.contentPatterns) {
-                                const result = matchesContent(content, contentPattern);
-                                if (result.matched) {
-                                    violations.push({
-                                        skillName,
-                                        matchType: 'contentPattern',
-                                        matchedPattern: contentPattern,
-                                        line: result.line
-                                    });
-                                }
-                            }
-                        }
-                        break; // Path matched, no need to check other path patterns
-                    }
-                }
-            }
-        }
-
-        // ===== 4. 处理违规 =====
-        if (violations.length > 0) {
-            // 按 priority 排序 (critical > high > medium > low)
-            const priorityOrder = { critical: 0, high: 1, medium: 2, low: 3 };
-            violations.sort((a, b) => {
-                const priorityA = priorityOrder[rules.skills[a.skillName].priority] || 999;
-                const priorityB = priorityOrder[rules.skills[b.skillName].priority] || 999;
-                return priorityA - priorityB;
-            });
-
-            // 显示第一个违规的 blockMessage
-            const firstViolation = violations[0];
-            const skill = rules.skills[firstViolation.skillName];
-
-            let blockMessage = skill.blockMessage || `⚠️ BLOCKED by ${firstViolation.skillName}`;
-
-            // 替换占位符
-            blockMessage = blockMessage.replace('{file_path}', file_path);
-
-            // 如果有行号,添加到消息中
-            if (firstViolation.line) {
-                blockMessage += `\n\n📍 Violation at Line ${firstViolation.line}`;
-            }
-
-            // 输出到 stderr (用户可见)
-            console.error('\n' + blockMessage + '\n');
-
-            // ===== 记录 Session 状态（用于 sessionSkillUsed 跳过条件）=====
-            // 在阻塞前记录，以便下次同一 session 中跳过
-            if (skill.skipConditions?.sessionSkillUsed) {
-                try {
-                    const stateDir = join(projectDir, '.claude', 'hooks', 'state');
-                    const stateFile = join(stateDir, `skills-used-${session_id}.json`);
-
-                    // 确保目录存在
-                    if (!existsSync(stateDir)) {
-                        require('fs').mkdirSync(stateDir, { recursive: true });
-                    }
-
-                    // 读取现有状态
-                    let usedSkills: string[] = [];
-                    if (existsSync(stateFile)) {
-                        usedSkills = JSON.parse(readFileSync(stateFile, 'utf-8'));
-                    }
-
-                    // 添加当前 skill（去重）
-                    if (!usedSkills.includes(firstViolation.skillName)) {
-                        usedSkills.push(firstViolation.skillName);
-                        require('fs').writeFileSync(stateFile, JSON.stringify(usedSkills, null, 2), 'utf-8');
-                    }
-                } catch (error) {
-                    // 状态写入失败不应影响阻塞（fail closed）
-                    console.error(`⚠️  Warning: Failed to write session state:`, error);
-                }
-            }
-
-            // 返回 exit code 2 (阻塞)
-            process.exit(2);
-        }
-
-        // ===== 5. 无违规,允许操作 =====
-        process.exit(0);
-
-    } catch (error) {
-        console.error('❌ PreToolUse Hook Error:', error);
-        process.exit(1); // Error, but don't block (fail open)
-    }
-}
-
-main();