autosnippet 3.0.1 → 3.0.2

package/lib/service/search/SearchEngine.js

@@ -1,6 +1,6 @@
 /**
  * SearchEngine - 统一搜索引擎
- * 
+ *
  * 三级搜索策略: keyword → BM25 ranking → semantic(可选)
  * 从 V1 SearchServiceV2 迁移，适配 V2 架构
  */
@@ -21,13 +21,15 @@ const BM25_B = 0.75;
  * 中文: 单字 + 二元组（bigram）— 无需分词词典即可支持子串匹配
  */
 export function tokenize(text) {
-  if (!text) return [];
+  if (!text) {
+    return [];
+  }
   // 先拆 camelCase/PascalCase（必须在 toLowerCase 之前，否则大小写边界丢失）
   let expanded = text.replace(/([a-z])([A-Z])/g, '$1 $2');
   // 拆全大写前缀：URLSession → URL Session, UITableView → UI Table View
   expanded = expanded.replace(/([A-Z]+)([A-Z][a-z])/g, '$1 $2');
   const normalized = expanded.toLowerCase().replace(/[^\p{L}\p{N}\s_-]/gu, ' ');
-  const rawTokens = normalized.split(/[\s_-]+/).filter(t => t.length >= 1);
+  const rawTokens = normalized.split(/[\s_-]+/).filter((t) => t.length >= 1);
 
   const tokens = [];
   // CJK 正则（中日韩统一表意文字 + 扩展区）
@@ -39,17 +41,25 @@ export function tokenize(text) {
       const cjkChars = raw.match(/[\u4e00-\u9fff\u3400-\u4dbf\uf900-\ufaff]+/g) || [];
       for (const seg of cjkChars) {
         // 单字
-        for (const ch of seg) tokens.push(ch);
+        for (const ch of seg) {
+          tokens.push(ch);
+        }
         // bigram
-        for (let i = 0; i < seg.length - 1; i++) tokens.push(seg[i] + seg[i + 1]);
+        for (let i = 0; i < seg.length - 1; i++) {
+          tokens.push(seg[i] + seg[i + 1]);
+        }
         // 完整片段（≥3 字时额外保留，提升精确匹配权重）
-        if (seg.length >= 3) tokens.push(seg);
+        if (seg.length >= 3) {
+          tokens.push(seg);
+        }
       }
       // 非 CJK 部分（英文/数字）也保留
       const nonCjk = raw.replace(/[\u4e00-\u9fff\u3400-\u4dbf\uf900-\ufaff]+/g, ' ').trim();
       if (nonCjk) {
         for (const t of nonCjk.split(/\s+/)) {
-          if (t.length >= 2) tokens.push(t);
+          if (t.length >= 2) {
+            tokens.push(t);
+          }
         }
       }
     } else if (raw.length >= 2) {
@@ -64,9 +74,9 @@ export function tokenize(text) {
  */
 export class BM25Scorer {
   constructor() {
-    this.documents = [];   // [{id, tokens, tokenFreq, length, meta}]
+    this.documents = []; // [{id, tokens, tokenFreq, length, meta}]
     this.avgLength = 0;
-    this.docFreq = {};     // token → 出现在多少文档中
+    this.docFreq = {}; // token → 出现在多少文档中
     this.totalDocs = 0;
     this._totalLength = 0; // 累计文档长度，避免 O(N) 重算
   }
@@ -95,7 +105,9 @@ export class BM25Scorer {
    */
   search(query, limit = 20) {
     const queryTokens = tokenize(query);
-    if (queryTokens.length === 0) return [];
+    if (queryTokens.length === 0) {
+      return [];
+    }
 
     const scores = [];
 
@@ -104,12 +116,15 @@ export class BM25Scorer {
       const dl = doc.length;
 
       for (const qt of queryTokens) {
-        const tf = doc.tokenFreq[qt] || 0;   // O(1) 查找，替代 O(T) filter
-        if (tf === 0) continue;
+        const tf = doc.tokenFreq[qt] || 0; // O(1) 查找，替代 O(T) filter
+        if (tf === 0) {
+          continue;
+        }
 
         const df = this.docFreq[qt] || 0;
         const idf = Math.log((this.totalDocs - df + 0.5) / (df + 0.5) + 1);
-        const tfNorm = (tf * (BM25_K1 + 1)) / (tf + BM25_K1 * (1 - BM25_B + BM25_B * (dl / this.avgLength)));
+        const tfNorm =
+          (tf * (BM25_K1 + 1)) / (tf + BM25_K1 * (1 - BM25_B + BM25_B * (dl / this.avgLength)));
         score += idf * tfNorm;
       }
 
@@ -163,28 +178,41 @@ export class SearchEngine {
       let entries = [];
 
       try {
-        entries = this.db.prepare(
-          `SELECT id, title, description, language, category, knowledgeType, kind,
+        entries = this.db
+          .prepare(
+            `SELECT id, title, description, language, category, knowledgeType, kind,
                   content, lifecycle, tags, trigger, difficulty, quality, stats,
                   updatedAt, createdAt
            FROM knowledge_entries WHERE lifecycle != 'deprecated'`
-        ).all();
-        entries = entries.map(e => ({
+          )
+          .all();
+        entries = entries.map((e) => ({
           ...e,
           status: e.lifecycle,
         }));
-      } catch { /* table may not exist */ }
+      } catch {
+        /* table may not exist */
+      }
 
       for (const r of entries) {
         let contentText = '';
         try {
           const content = JSON.parse(r.content || '{}');
-          contentText = [content.pattern, content.rationale, content.markdown].filter(Boolean).join(' ');
-        } catch { /* ignore parse error */ }
+          contentText = [content.pattern, content.rationale, content.markdown]
+            .filter(Boolean)
+            .join(' ');
+        } catch {
+          /* ignore parse error */
+        }
         // 包含 tags + trigger 提升召回率
         let tagText = '';
         let parsedTags = [];
-        try { parsedTags = JSON.parse(r.tags || '[]'); tagText = parsedTags.join(' '); } catch { /* ignore */ }
+        try {
+          parsedTags = JSON.parse(r.tags || '[]');
+          tagText = parsedTags.join(' ');
+        } catch {
+          /* ignore */
+        }
         // 解析 stats / quality JSON — 供排序信号使用
         let usageCount = 0;
         let authorityScore = 0;
@@ -192,18 +220,43 @@ export class SearchEngine {
           const stats = JSON.parse(r.stats || '{}');
           usageCount = (stats.adoptions || 0) + (stats.applications || 0) + (stats.searchHits || 0);
           authorityScore = stats.authority || 0;
-        } catch { /* ignore */ }
+        } catch {
+          /* ignore */
+        }
         let qualityOverall = 0;
-        try { qualityOverall = JSON.parse(r.quality || '{}').overall || 0; } catch { /* ignore */ }
-        const text = [r.title, r.description, r.trigger, r.language, r.category, r.knowledgeType, tagText, contentText]
-          .filter(Boolean).join(' ');
+        try {
+          qualityOverall = JSON.parse(r.quality || '{}').overall || 0;
+        } catch {
+          /* ignore */
+        }
+        const text = [
+          r.title,
+          r.description,
+          r.trigger,
+          r.language,
+          r.category,
+          r.knowledgeType,
+          tagText,
+          contentText,
+        ]
+          .filter(Boolean)
+          .join(' ');
         this.scorer.addDocument(r.id, text, {
-          type: 'knowledge', title: r.title, trigger: r.trigger || '', status: r.status,
-          knowledgeType: r.knowledgeType, kind: r.kind || 'pattern',
-          language: r.language || '', category: r.category || '',
-          updatedAt: r.updatedAt || null, createdAt: r.createdAt || null,
-          difficulty: r.difficulty || 'intermediate', tags: parsedTags,
-          usageCount, authorityScore, qualityScore: qualityOverall,
+          type: 'knowledge',
+          title: r.title,
+          trigger: r.trigger || '',
+          status: r.status,
+          knowledgeType: r.knowledgeType,
+          kind: r.kind || 'pattern',
+          language: r.language || '',
+          category: r.category || '',
+          updatedAt: r.updatedAt || null,
+          createdAt: r.createdAt || null,
+          difficulty: r.difficulty || 'intermediate',
+          tags: parsedTags,
+          usageCount,
+          authorityScore,
+          qualityScore: qualityOverall,
         });
       }
 
@@ -233,7 +286,7 @@ export class SearchEngine {
    */
   async search(query, options = {}) {
     const { type = 'all', limit = 20, mode = 'keyword', context } = options;
-    const shouldRank = options.rank ?? (mode !== 'keyword');
+    const shouldRank = options.rank ?? mode !== 'keyword';
 
     if (!query || !query.trim()) {
       return { items: [], total: 0, query };
@@ -246,7 +299,9 @@ export class SearchEngine {
       : `${query}:${type}:${limit}:${mode}:${shouldRank ? 'r' : ''}:${options.groupByKind ? 'g' : ''}`;
     if (cacheKey) {
       const cached = this._getCache(cacheKey);
-      if (cached) return cached;
+      if (cached) {
+        return cached;
+      }
     }
 
     // 确保索引已构建
@@ -262,7 +317,10 @@ export class SearchEngine {
         // 同时做 BM25 + semantic，融合去重取最优分数
         const [bm25Items, semResult] = await Promise.all([
           Promise.resolve(this._bm25Search(query, type, recallLimit)),
-          this._semanticSearch(query, type, recallLimit).catch(() => ({ items: [], actualMode: 'bm25' })),
+          this._semanticSearch(query, type, recallLimit).catch(() => ({
+            items: [],
+            actualMode: 'bm25',
+          })),
         ]);
         const semItems = semResult.items || [];
         const merged = new Map();
@@ -279,7 +337,10 @@ export class SearchEngine {
           }
         }
         results = [...merged.values()].sort((a, b) => b.score - a.score);
-        for (const it of results) { delete it._bm25; delete it._sem; }
+        for (const it of results) {
+          delete it._bm25;
+          delete it._sem;
+        }
         const semActuallyUsed = semResult.actualMode === 'semantic';
         actualMode = semActuallyUsed ? 'auto(bm25+semantic)' : 'auto(bm25-only)';
         break;
@@ -294,7 +355,6 @@ export class SearchEngine {
         actualMode = semResult.actualMode || 'semantic';
         break;
       }
-      case 'keyword':
       default:
         results = this._keywordSearch(query, type, limit);
         break;
@@ -323,7 +383,9 @@ export class SearchEngine {
       }
     }
 
-    if (cacheKey) this._setCache(cacheKey, response);
+    if (cacheKey) {
+      this._setCache(cacheKey, response);
+    }
     return response;
   }
 
@@ -336,13 +398,14 @@ export class SearchEngine {
     const normalized = this._normalizeForRanking(items);
     let ranked = this._coarseRanker.rank(normalized);
     ranked = this._multiSignalRanker.rank(ranked, {
-      ...context, query,
+      ...context,
+      query,
       scenario: context?.intent || 'search',
     });
     if (context?.sessionHistory?.length > 0) {
       ranked = this._contextBoost(ranked, context);
     }
-    return ranked.map(r => ({
+    return ranked.map((r) => ({
       ...r,
       recallScore: r.bm25Score || 0,
       score: r.contextScore || r.rankerScore || r.coarseScore || r.bm25Score || 0,
@@ -354,17 +417,23 @@ export class SearchEngine {
    * 保留原始 content 供下游消费者使用
    */
   _normalizeForRanking(items) {
-    return items.map(item => {
+    return items.map((item) => {
       let codeText = '';
       if (item.content) {
         try {
           const parsed = typeof item.content === 'string' ? JSON.parse(item.content) : item.content;
           codeText = parsed.pattern || parsed.code || '';
-        } catch { /* ignore */ }
+        } catch {
+          /* ignore */
+        }
       }
       let tags = item.tags || [];
       if (typeof tags === 'string') {
-        try { tags = JSON.parse(tags); } catch { tags = []; }
+        try {
+          tags = JSON.parse(tags);
+        } catch {
+          tags = [];
+        }
       }
       return {
         ...item,
@@ -384,24 +453,34 @@ export class SearchEngine {
    */
   _contextBoost(items, context) {
     const { sessionHistory = [], language } = context || {};
-    if (!sessionHistory.length) return items;
+    if (!sessionHistory.length) {
+      return items;
+    }
     const sessionKeywords = new Set();
     for (const turn of sessionHistory) {
       const tokens = tokenize(turn.content || turn.rawInput || '');
-      for (const t of tokens) sessionKeywords.add(t);
+      for (const t of tokens) {
+        sessionKeywords.add(t);
+      }
     }
-    return items.map(item => {
-      let boost = 0;
-      const textTokens = tokenize(
-        [item.title, item.trigger, item.content].filter(Boolean).join(' ')
-      );
-      const overlap = textTokens.filter(t => sessionKeywords.has(t)).length;
-      if (overlap > 0) boost += 0.2 * Math.min(overlap / 5, 1);
-      if (language && item.language === language) boost += 0.1;
-      const baseScore = item.rankerScore || item.coarseScore || item.score || 0;
-      const contextScore = baseScore * (1 + boost);
-      return { ...item, contextScore, contextBoost: boost };
-    }).sort((a, b) => b.contextScore - a.contextScore);
+    return items
+      .map((item) => {
+        let boost = 0;
+        const textTokens = tokenize(
+          [item.title, item.trigger, item.content].filter(Boolean).join(' ')
+        );
+        const overlap = textTokens.filter((t) => sessionKeywords.has(t)).length;
+        if (overlap > 0) {
+          boost += 0.2 * Math.min(overlap / 5, 1);
+        }
+        if (language && item.language === language) {
+          boost += 0.1;
+        }
+        const baseScore = item.rankerScore || item.coarseScore || item.score || 0;
+        const contextScore = baseScore * (1 + boost);
+        return { ...item, contextScore, contextBoost: boost };
+      })
+      .sort((a, b) => b.contextScore - a.contextScore);
   }
 
   /**
@@ -411,31 +490,54 @@ export class SearchEngine {
   _keywordSearch(query, type, limit) {
     const results = [];
     // 转义 LIKE 通配符 (% → \%, _ → \_)
-    const escaped = query.replace(/[%_\\]/g, ch => `\\${ch}`);
+    const escaped = query.replace(/[%_\\]/g, (ch) => `\\${ch}`);
     const pattern = `%${escaped}%`;
 
-    if (type === 'all' || type === 'recipe' || type === 'knowledge' || type === 'rule' || type === 'solution') {
+    if (
+      type === 'all' ||
+      type === 'recipe' ||
+      type === 'knowledge' ||
+      type === 'rule' ||
+      type === 'solution'
+    ) {
       try {
         let rows = [];
         try {
-          rows = this.db.prepare(
-            `SELECT id, title, description, language, category, knowledgeType, kind, lifecycle as status, content, trigger, headers, moduleName, 'knowledge' as type
+          rows = this.db
+            .prepare(
+              `SELECT id, title, description, language, category, knowledgeType, kind, lifecycle as status, content, trigger, headers, moduleName, 'knowledge' as type
              FROM knowledge_entries
              WHERE lifecycle != 'deprecated' AND (title LIKE ? ESCAPE '\\' OR description LIKE ? ESCAPE '\\' OR trigger LIKE ? ESCAPE '\\' OR content LIKE ? ESCAPE '\\')
              LIMIT ?`
-          ).all(pattern, pattern, pattern, pattern, limit);
-        } catch { /* table may not exist */ }
+            )
+            .all(pattern, pattern, pattern, pattern, limit);
+        } catch {
+          /* table may not exist */
+        }
         // 基础相关性排序：trigger 精确 > 标题匹配 > 描述匹配 > 内容匹配
         const lowerQ = query.toLowerCase();
-        results.push(...rows.map(r => {
-          let score = 0.5;
-          if (r.trigger && r.trigger.toLowerCase().includes(lowerQ)) score = 1.2;
-          else if (r.title && r.title.toLowerCase().includes(lowerQ)) score = 1.0;
-          else if (r.description && r.description.toLowerCase().includes(lowerQ)) score = 0.8;
-          return { ...r, trigger: r.trigger || '', kind: r.kind || 'pattern', score: Math.round(score * 1000) / 1000 };
-        }));
+        results.push(
+          ...rows.map((r) => {
+            let score = 0.5;
+            if (r.trigger?.toLowerCase().includes(lowerQ)) {
+              score = 1.2;
+            } else if (r.title?.toLowerCase().includes(lowerQ)) {
+              score = 1.0;
+            } else if (r.description?.toLowerCase().includes(lowerQ)) {
+              score = 0.8;
+            }
+            return {
+              ...r,
+              trigger: r.trigger || '',
+              kind: r.kind || 'pattern',
+              score: Math.round(score * 1000) / 1000,
+            };
+          })
+        );
         results.sort((a, b) => b.score - a.score);
-      } catch { /* table may not exist */ }
+      } catch {
+        /* table may not exist */
+      }
     }
 
     return results.slice(0, limit);
@@ -449,13 +551,15 @@ export class SearchEngine {
 
     if (type !== 'all') {
       // All types now map to 'recipe' since everything is unified
-      results = results.filter(r => {
-        if (type === 'rule') return r.meta.knowledgeType === 'boundary-constraint';
+      results = results.filter((r) => {
+        if (type === 'rule') {
+          return r.meta.knowledgeType === 'boundary-constraint';
+        }
         return r.meta.type === 'recipe';
       });
     }
 
-    const items = results.slice(0, limit).map(r => ({
+    const items = results.slice(0, limit).map((r) => ({
       id: r.id,
       title: r.meta.title,
       trigger: r.meta.trigger || '',
@@ -503,16 +607,21 @@ export class SearchEngine {
         try {
           let vectorResults;
           if (typeof this.vectorStore.hybridSearch === 'function') {
-            const hybrid = await this.vectorStore.hybridSearch(queryEmbedding, query, { topK: limit * 2 });
-            vectorResults = hybrid.map(r => ({
-              id: r.item.id, similarity: r.score, score: r.score,
-              content: r.item.content, metadata: r.item.metadata || {},
+            const hybrid = await this.vectorStore.hybridSearch(queryEmbedding, query, {
+              topK: limit * 2,
+            });
+            vectorResults = hybrid.map((r) => ({
+              id: r.item.id,
+              similarity: r.score,
+              score: r.score,
+              content: r.item.content,
+              metadata: r.item.metadata || {},
             }));
           } else {
             vectorResults = await this.vectorStore.query(queryEmbedding, limit * 2);
           }
           if (vectorResults && vectorResults.length > 0) {
-            let results = vectorResults.map(vr => ({
+            let results = vectorResults.map((vr) => ({
               id: vr.id,
               title: vr.metadata?.title || vr.id,
               type: 'recipe',
@@ -521,8 +630,10 @@ export class SearchEngine {
               score: Math.round((vr.similarity || vr.score || 0) * 1000) / 1000,
             }));
             if (type !== 'all') {
-              results = results.filter(r => {
-                if (type === 'rule') return r.kind === 'rule';
+              results = results.filter((r) => {
+                if (type === 'rule') {
+                  return r.kind === 'rule';
+                }
                 return r.type === 'recipe';
               });
             }
@@ -532,7 +643,9 @@ export class SearchEngine {
             return { items: results, actualMode: 'semantic' };
           }
         } catch (vecErr) {
-          this.logger.warn('Vector store query failed, falling back to BM25', { error: vecErr.message });
+          this.logger.warn('Vector store query failed, falling back to BM25', {
+            error: vecErr.message,
+          });
         }
       }
 
@@ -550,52 +663,87 @@ export class SearchEngine {
    * 用于向量搜索结果与 BM25 结果的一致性
    */
   _supplementDetails(items) {
-    if (!items || items.length === 0) return;
+    if (!items || items.length === 0) {
+      return;
+    }
     try {
-      const ids = items.map(it => it.id);
+      const ids = items.map((it) => it.id);
       const placeholders = ids.map(() => '?').join(',');
       let rows = [];
       try {
-        rows = this.db.prepare(
-          `SELECT id, content, description, trigger, headers, moduleName,
+        rows = this.db
+          .prepare(
+            `SELECT id, content, description, trigger, headers, moduleName,
                   tags, language, category, updatedAt, createdAt, quality, stats, difficulty
            FROM knowledge_entries WHERE id IN (${placeholders})`
-        ).all(...ids);
-      } catch { /* table may not exist */ }
-      const rowMap = new Map(rows.map(r => [r.id, r]));
+          )
+          .all(...ids);
+      } catch {
+        /* table may not exist */
+      }
+      const rowMap = new Map(rows.map((r) => [r.id, r]));
       for (const item of items) {
         const row = rowMap.get(item.id);
         if (row) {
           item.content = item.content || row.content || null;
           item.description = item.description || row.description || '';
           item.trigger = item.trigger || row.trigger || '';
-          if (row.headers) item.headers = row.headers;
-          if (row.moduleName) item.moduleName = row.moduleName;
+          if (row.headers) {
+            item.headers = row.headers;
+          }
+          if (row.moduleName) {
+            item.moduleName = row.moduleName;
+          }
           // 排序信号补充 — 确保 Funnel/Ranker 有真实数据
-          if (!item.language && row.language) item.language = row.language;
-          if (!item.category && row.category) item.category = row.category;
-          if (!item.updatedAt && row.updatedAt) item.updatedAt = row.updatedAt;
-          if (!item.createdAt && row.createdAt) item.createdAt = row.createdAt;
-          if (!item.difficulty && row.difficulty) item.difficulty = row.difficulty;
+          if (!item.language && row.language) {
+            item.language = row.language;
+          }
+          if (!item.category && row.category) {
+            item.category = row.category;
+          }
+          if (!item.updatedAt && row.updatedAt) {
+            item.updatedAt = row.updatedAt;
+          }
+          if (!item.createdAt && row.createdAt) {
+            item.createdAt = row.createdAt;
+          }
+          if (!item.difficulty && row.difficulty) {
+            item.difficulty = row.difficulty;
+          }
           // 解析 tags
           if (!item.tags || (Array.isArray(item.tags) && item.tags.length === 0)) {
-            try { item.tags = JSON.parse(row.tags || '[]'); } catch { /* ignore */ }
+            try {
+              item.tags = JSON.parse(row.tags || '[]');
+            } catch {
+              /* ignore */
+            }
           }
           // 解析 quality JSON → qualityScore
           if (!item.qualityScore) {
-            try { item.qualityScore = JSON.parse(row.quality || '{}').overall || 0; } catch { /* ignore */ }
+            try {
+              item.qualityScore = JSON.parse(row.quality || '{}').overall || 0;
+            } catch {
+              /* ignore */
+            }
           }
           // 解析 stats JSON → usageCount + authorityScore
           if (!item.usageCount) {
             try {
               const stats = JSON.parse(row.stats || '{}');
-              item.usageCount = (stats.adoptions || 0) + (stats.applications || 0) + (stats.searchHits || 0);
-              if (!item.authorityScore) item.authorityScore = stats.authority || 0;
-            } catch { /* ignore */ }
+              item.usageCount =
+                (stats.adoptions || 0) + (stats.applications || 0) + (stats.searchHits || 0);
+              if (!item.authorityScore) {
+                item.authorityScore = stats.authority || 0;
+              }
+            } catch {
+              /* ignore */
+            }
           }
         }
       }
-    } catch { /* DB may not be available */ }
+    } catch {
+      /* DB may not be available */
+    }
   }
 
   /**
@@ -623,7 +771,9 @@ export class SearchEngine {
 
   _getCache(key) {
     const entry = this._cache.get(key);
-    if (!entry) return null;
+    if (!entry) {
+      return null;
+    }
     if (Date.now() - entry.time > this._cacheMaxAge) {
       this._cache.delete(key);
       return null;
@@ -641,7 +791,9 @@ export class SearchEngine {
       const keys = this._cache.keys();
       for (let i = 0; i < toDelete; i++) {
         const k = keys.next().value;
-        if (k !== undefined) this._cache.delete(k);
+        if (k !== undefined) {
+          this._cache.delete(k);
+        }
       }
     }
     this._cache.set(key, { data, time: Date.now() });