autosnippet 3.0.10 → 3.0.13

package/lib/shared/LanguageService.js

@@ -11,8 +11,12 @@
  *   const lang = LanguageService.inferLang('App.swift');      // 'swift'
  *   const display = LanguageService.displayName('swift');       // 'Swift'
  *   const primary = LanguageService.detectPrimary(langStats);   // 'typescript'
+ *   const langs  = LanguageService.detectProjectLanguages('/path/to/project');
  */
 
+import { existsSync, readdirSync } from 'node:fs';
+import { join } from 'node:path';
+
 // ═══════════════════════════════════════════════════════════
 // 1) 文件扩展名 → 规范化语言 ID
 // ═══════════════════════════════════════════════════════════
@@ -193,6 +197,105 @@ const SOURCE_CODE_EXTS = Object.freeze(
   ])
 );
 
+// ═══════════════════════════════════════════════════════════
+// 5.5) 语言别名映射 — 将常见缩写/变体归一化为规范 ID
+// ═══════════════════════════════════════════════════════════
+
+/** @type {Readonly<Record<string, string>>} */
+const LANG_ALIASES = Object.freeze({
+  // Objective-C variants
+  objc: 'objectivec',
+  'objective-c': 'objectivec',
+  'obj-c': 'objectivec',
+  // TypeScript
+  ts: 'typescript',
+  tsx: 'typescript',
+  // JavaScript
+  js: 'javascript',
+  jsx: 'javascript',
+  // C++
+  'c++': 'cpp',
+  cxx: 'cpp',
+  // C#
+  'c#': 'csharp',
+  cs: 'csharp',
+  // Python
+  py: 'python',
+  python3: 'python',
+  // Kotlin
+  kt: 'kotlin',
+  // Rust
+  rs: 'rust',
+  // Go
+  golang: 'go',
+  // Ruby
+  rb: 'ruby',
+});
+
+// ═══════════════════════════════════════════════════════════
+// 6) 生态系统/Discoverer ID → 对应编程语言 ID 数组
+// ═══════════════════════════════════════════════════════════
+
+/** @type {Readonly<Record<string, readonly string[]>>} */
+const ECO_TO_LANGS = Object.freeze({
+  spm:     Object.freeze(['swift', 'objectivec']),
+  node:    Object.freeze(['javascript', 'typescript']),
+  go:      Object.freeze(['go']),
+  jvm:     Object.freeze(['java', 'kotlin']),
+  python:  Object.freeze(['python']),
+  dart:    Object.freeze(['dart']),
+  rust:    Object.freeze(['rust']),
+  dotnet:  Object.freeze(['csharp']),
+  ruby:    Object.freeze(['ruby']),
+  generic: Object.freeze([]),
+});
+
+// ═══════════════════════════════════════════════════════════
+// 7) 构建系统标志文件 → 生态系统映射（项目级语言检测的核心数据）
+// ═══════════════════════════════════════════════════════════
+
+/**
+ * @typedef {{ file: string, eco: string, buildTool: string }} BuildSystemMarker
+ * @type {ReadonlyArray<BuildSystemMarker>}
+ */
+const BUILD_SYSTEM_MARKERS = Object.freeze([
+  // Apple / iOS
+  { file: 'Package.swift',    eco: 'spm',     buildTool: 'SPM' },
+  { file: 'Podfile',          eco: 'spm',     buildTool: 'CocoaPods' },
+  // JS / TS (lock files before package.json to detect specific tool)
+  { file: 'yarn.lock',        eco: 'node',    buildTool: 'Yarn' },
+  { file: 'pnpm-lock.yaml',   eco: 'node',    buildTool: 'pnpm' },
+  { file: 'package.json',     eco: 'node',    buildTool: 'npm' },
+  // Python
+  { file: 'Pipfile',          eco: 'python',  buildTool: 'Pipenv' },
+  { file: 'pyproject.toml',   eco: 'python',  buildTool: 'Poetry' },
+  { file: 'setup.py',         eco: 'python',  buildTool: 'setuptools' },
+  { file: 'requirements.txt', eco: 'python',  buildTool: 'pip' },
+  // Go
+  { file: 'go.mod',           eco: 'go',      buildTool: 'Go Modules' },
+  // Rust
+  { file: 'Cargo.toml',       eco: 'rust',    buildTool: 'Cargo' },
+  // JVM
+  { file: 'pom.xml',          eco: 'jvm',     buildTool: 'Maven' },
+  { file: 'build.gradle',     eco: 'jvm',     buildTool: 'Gradle' },
+  { file: 'build.gradle.kts', eco: 'jvm',     buildTool: 'Gradle (Kotlin)' },
+  // Dart / Flutter
+  { file: 'pubspec.yaml',     eco: 'dart',    buildTool: 'Flutter' },
+  { file: 'melos.yaml',       eco: 'dart',    buildTool: 'Melos' },
+  // C# / .NET
+  { file: '*.csproj',         eco: 'dotnet',  buildTool: '.NET' },
+  { file: '*.sln',            eco: 'dotnet',  buildTool: '.NET' },
+  // Ruby
+  { file: 'Gemfile',          eco: 'ruby',    buildTool: 'Bundler' },
+]);
+
+/** 扫描目录时跳过的标准目录（性能优化） */
+const SCAN_SKIP_DIRS = Object.freeze(new Set([
+  '.git', 'node_modules', '.build', 'build', 'dist', 'target', 'out',
+  'vendor', '.cache', 'Pods', 'DerivedData', '__pycache__', '.venv', 'venv',
+  '.gradle', 'Carthage', '.fvm', '.dart_tool', '.cargo',
+]));
+
 // ═══════════════════════════════════════════════════════════
 // LanguageService — 静态单例
 // ═══════════════════════════════════════════════════════════
@@ -206,6 +309,7 @@ export class LanguageService {
    * @returns {string} 语言 ID，如 'swift', 'typescript', 'python', 'unknown'
    */
   static inferLang(filename) {
+    if (!filename || typeof filename !== 'string') return 'unknown';
     const dot = filename.lastIndexOf('.');
     if (dot === -1) {
       return 'unknown';
@@ -220,9 +324,45 @@ export class LanguageService {
    * @returns {string}
    */
   static langFromExt(ext) {
+    if (!ext || typeof ext !== 'string') return 'unknown';
     return EXT_TO_LANG[ext.toLowerCase()] || 'unknown';
   }
 
+  // ─── 别名归一化 ───────────────────────────────
+
+  /**
+   * 将语言 ID 别名/缩写归一化为规范 ID
+   *
+   * 示例:
+   *   normalize('objc')     → 'objectivec'
+   *   normalize('ts')       → 'typescript'
+   *   normalize('golang')   → 'go'
+   *   normalize('swift')    → 'swift' (已是规范 ID)
+   *   normalize('unknown')  → 'unknown'
+   *
+   * @param {string} langId - 语言 ID（可能是别名）
+   * @returns {string} 规范化语言 ID
+   */
+  static normalize(langId) {
+    if (!langId || typeof langId !== 'string') return 'unknown';
+    const lower = langId.toLowerCase().trim();
+    if (KNOWN_PROGRAMMING_LANGS.has(lower)) return lower;
+    return LANG_ALIASES[lower] || lower;
+  }
+
+  /**
+   * 将规范语言 ID 转为 Guard 兼容 ID
+   *
+   * Guard 内置规则使用 'objc' 而非 'objectivec'。
+   * 其他语言 ID 不变。
+   *
+   * @param {string} langId
+   * @returns {string}
+   */
+  static toGuardLangId(langId) {
+    return langId === 'objectivec' ? 'objc' : langId;
+  }
+
   // ─── 显示名 ────────────────────────────────────
 
   /**
@@ -394,6 +534,165 @@ export class LanguageService {
     }
     return null;
   }
+
+  // ─── 生态系统 / 项目级语言检测 ────────────────
+
+  /**
+   * 获取语言别名映射表（不可变）
+   * @returns {Readonly<Record<string, string>>}
+   */
+  static get langAliases() {
+    return LANG_ALIASES;
+  }
+
+  /**
+   * 获取 ECO_TO_LANGS 映射（不可变）
+   * @returns {Readonly<Record<string, readonly string[]>>}
+   */
+  static get ecoToLangs() {
+    return ECO_TO_LANGS;
+  }
+
+  /**
+   * 获取 BUILD_SYSTEM_MARKERS（不可变）
+   * @returns {ReadonlyArray<BuildSystemMarker>}
+   */
+  static get buildSystemMarkers() {
+    return BUILD_SYSTEM_MARKERS;
+  }
+
+  /**
+   * 获取 SCAN_SKIP_DIRS（不可变）
+   * @returns {ReadonlySet<string>}
+   */
+  static get scanSkipDirs() {
+    return SCAN_SKIP_DIRS;
+  }
+
+  /**
+   * 根据生态系统/Discoverer ID 获取对应的语言 ID 数组
+   * @param {string} ecoId - 如 'spm', 'node', 'rust', 'dart'
+   * @returns {readonly string[]}
+   */
+  static langsForEco(ecoId) {
+    return ECO_TO_LANGS[ecoId] || [];
+  }
+
+  /**
+   * 检测构建系统标志文件 — 纯数据匹配，不访问文件系统
+   *
+   * @param {string[]} entryNames - 目录内文件/目录名列表
+   * @returns {Array<{ eco: string, buildTool: string }>}
+   */
+  static matchBuildMarkers(entryNames) {
+    if (!Array.isArray(entryNames) || entryNames.length === 0) return [];
+    const nameSet = new Set(entryNames);
+    const results = [];
+    const seenEco = new Set();
+
+    for (const marker of BUILD_SYSTEM_MARKERS) {
+      if (seenEco.has(marker.eco)) continue;
+      const isGlob = marker.file.startsWith('*');
+      const matched = isGlob
+        ? entryNames.some((n) => n.endsWith(marker.file.slice(1)))
+        : nameSet.has(marker.file);
+      if (matched) {
+        results.push({ eco: marker.eco, buildTool: marker.buildTool });
+        seenEco.add(marker.eco);
+      }
+    }
+    return results;
+  }
+
+  /**
+   * 检测项目使用的编程语言 — 统一入口
+   *
+   * 策略（按优先级）：
+   *   1. 若传入 discovererIds（来自 ModuleService），直接映射为语言
+   *   2. 否则扫描项目目录的构建系统标记文件（支持 monorepo 多层扫描）
+   *
+   * @param {string} projectRoot - 项目根目录绝对路径
+   * @param {object} [opts]
+   * @param {string[]} [opts.discovererIds] - ModuleService 检测到的生态 ID
+   * @param {number}   [opts.maxDepth=2]    - 最大扫描深度：0=仅根目录，1=+子目录，2=+孙目录
+   * @returns {string[]} 规范化语言 ID 数组（如 ['rust', 'dart']）
+   */
+  static detectProjectLanguages(projectRoot, opts = {}) {
+    if (!projectRoot || typeof projectRoot !== 'string') return [];
+    const { discovererIds, maxDepth = 2 } = opts;
+
+    // ── Path 1: 从 Discoverer ID 映射 ──
+    if (discovererIds && discovererIds.length > 0) {
+      const nonGeneric = discovererIds.filter((id) => id !== 'generic');
+      if (nonGeneric.length > 0) {
+        const langSet = new Set();
+        for (const did of nonGeneric) {
+          for (const lang of ECO_TO_LANGS[did] || []) langSet.add(lang);
+        }
+        if (langSet.size > 0) return [...langSet];
+      }
+    }
+
+    // ── Path 2: 扫描构建系统标记文件 ──
+    const seenEco = new Set();
+
+    const scanDir = (dir) => {
+      try {
+        for (const marker of BUILD_SYSTEM_MARKERS) {
+          if (seenEco.has(marker.eco)) continue;
+          const isGlob = marker.file.startsWith('*');
+          let matched = false;
+          if (isGlob) {
+            try {
+              const suffix = marker.file.slice(1);
+              matched = readdirSync(dir).some((n) => n.endsWith(suffix));
+            } catch { /* skip */ }
+          } else {
+            matched = existsSync(join(dir, marker.file));
+          }
+          if (matched) seenEco.add(marker.eco);
+        }
+      } catch { /* skip unreadable dir */ }
+    };
+
+    // Level 0: 项目根目录
+    scanDir(projectRoot);
+
+    // Level 1..maxDepth: 子目录（支持 monorepo）
+    if (seenEco.size === 0 && maxDepth >= 1) {
+      /** @type {Array<[string, number]>} */
+      const queue = [[projectRoot, 0]];
+      while (queue.length > 0) {
+        const [dir, depth] = queue.shift();
+        if (depth >= maxDepth) continue;
+        try {
+          for (const ent of readdirSync(dir, { withFileTypes: true })) {
+            if (!ent.isDirectory() || ent.name.startsWith('.') || SCAN_SKIP_DIRS.has(ent.name)) continue;
+            const sub = join(dir, ent.name);
+            scanDir(sub);
+            if (depth + 1 < maxDepth) queue.push([sub, depth + 1]);
+          }
+        } catch { /* skip */ }
+      }
+    }
+
+    // ── 将生态 ID 转为语言 ID ──
+    const langSet = new Set();
+    for (const eco of seenEco) {
+      for (const lang of ECO_TO_LANGS[eco] || []) langSet.add(lang);
+    }
+
+    // ── 启发式: node 与其他生态共存时，JS/TS 通常只是构建工具，去掉 ──
+    if (seenEco.size > 1 && seenEco.has('node')) {
+      const hasOther = [...seenEco].some((e) => e !== 'node');
+      if (hasOther) {
+        langSet.delete('javascript');
+        langSet.delete('typescript');
+      }
+    }
+
+    return [...langSet];
+  }
 }
 
 export default LanguageService;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "autosnippet",
-  "version": "3.0.10",
+  "version": "3.0.13",
   "description": "Extract code patterns into a knowledge base for AI coding assistants",
   "type": "module",
   "main": "lib/bootstrap.js",