smart-image-scraper-mcp 2.11.0 → 2.11.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "smart-image-scraper-mcp",
-  "version": "2.11.0",
+  "version": "2.11.2",
   "description": "全网智能图片抓取 MCP 服务器 - 支持 Bing/Google 图片搜索、验证和下载",
   "main": "src/index.js",
   "type": "module",

package/src/services/linkValidator.js

@@ -44,11 +44,11 @@ export class LinkValidator {
     }
 
     const controller = new AbortController();
-    const timeoutId = setTimeout(() => controller.abort(), 3000); // 超时 3 秒
+    const timeoutId = setTimeout(() => controller.abort(), 2000); // 超时 2 秒
 
     try {
       const response = await httpClient.head(url, {
-        timeout: 2500, // 超时 2.5 秒
+        timeout: 1800, // 超时 1.8 秒
         maxRedirects: 1, // 最多 1 次重定向
         signal: controller.signal,
       });
@@ -64,21 +64,12 @@ export class LinkValidator {
         return { url, valid: true, quality };
       }
 
-      // 某些服务器不支持 HEAD，返回 405/403 时尝试 GET 降级
-      if (response.status === 405 || response.status === 403) {
-        return await this._validateWithGet(url, fetchQuality);
-      }
-
       return { url, valid: false, error: `status=${response.status}` };
     } catch (error) {
       clearTimeout(timeoutId);
       if (!controller.signal.aborted) {
         controller.abort();
       }
-      // 网络错误时也尝试 GET 降级（某些 CDN 完全拒绝 HEAD）
-      if (error.response && (error.response.status === 405 || error.response.status === 403)) {
-        return await this._validateWithGet(url, fetchQuality);
-      }
       return { url, valid: false, error: 'timeout' };
     }
   }
@@ -236,12 +227,18 @@ export class LinkValidator {
    * @returns {Promise<{valid: Array, invalid: Array}>}
    */
   async validateMany(urls, options = {}) {
-    const { fetchQuality = false, sortByQuality = false } = options;
+    const { fetchQuality = false, sortByQuality = false, signal = null } = options;
     
     logger.info(`Validating ${urls.length} URLs with concurrency ${config.MAX_VALIDATE_CONCURRENCY}${fetchQuality ? ' (with quality check)' : ''}`);
 
     const results = await Promise.all(
-      urls.map(url => this.limit(() => this.validateOne(url, fetchQuality)))
+      urls.map(url => this.limit(() => {
+        // 如果已中止，跳过验证直接返回失败
+        if (signal?.aborted) {
+          return { url, valid: false, error: 'aborted' };
+        }
+        return this.validateOne(url, fetchQuality);
+      }))
     );
 
     let valid = [];

package/src/services/orchestrator.js

@@ -117,6 +117,7 @@ export class Orchestrator {
     const fastMode = qualityMode === 'fast';
     const prioritizeQuality = qualityMode === 'high';
     const minFileSize = this._parseMinFileSize(options.minFileSize);
+    const signal = this.abortController?.signal;
     
     try {
       const scraper = getScraper(source);
@@ -132,6 +133,8 @@ export class Orchestrator {
         rawUrls = cachedUrls;
         metrics.recordCacheHit();
       } else {
+        // 检查是否已中止
+        if (signal?.aborted) throw new Error('操作已取消');
         logger.info(`[SEARCH] "${keyword}" (target: ${searchCount})...`);
         rawUrls = await scraper.search(keyword, searchCount, options);
         if (rawUrls.length > 0) {
@@ -153,17 +156,23 @@ export class Orchestrator {
       let resultUrls;
       let qualityModeLabel;
       
+      // 检查是否已中止
+      if (signal?.aborted) throw new Error('操作已取消');
+
       if (fastMode) {
         // fast 模式：直接使用搜索结果，不验证
         resultUrls = rawUrls.slice(0, count);
         qualityModeLabel = '快速模式（跳过验证）';
         logger.info(`[FAST] "${keyword}" - ${resultUrls.length} URLs`);
       } else {
-        // balanced 或 high 模式：验证链接
-        const { valid } = await this.linkValidator.validateMany(rawUrls, {
+        // balanced 或 high 模式：验证链接（限制验证数量避免超时）
+        const maxValidate = Math.min(rawUrls.length, count * 2 + 5);
+        const urlsToValidate = rawUrls.slice(0, maxValidate);
+        const { valid } = await this.linkValidator.validateMany(urlsToValidate, {
           fetchQuality: prioritizeQuality,
           sortByQuality: prioritizeQuality,
           minFileSize: minFileSize,
+          signal,
         });
         
         // 过滤最小文件大小
@@ -246,6 +255,7 @@ export class Orchestrator {
     const qualityMode = options.quality || 'balanced';
     const prioritizeQuality = qualityMode !== 'fast';
     const minFileSize = this._parseMinFileSize(options.minFileSize);
+    const signal = this.abortController?.signal;
     
     try {
       const scraper = getScraper(source);
@@ -261,6 +271,7 @@ export class Orchestrator {
         rawUrls = cachedUrls;
         metrics.recordCacheHit();
       } else {
+        if (signal?.aborted) throw new Error('操作已取消');
         logger.info(`[SEARCH] "${keyword}" (target: ${searchCount})...`);
         rawUrls = await scraper.search(keyword, searchCount, options);
         if (rawUrls.length > 0) {
@@ -278,8 +289,12 @@ export class Orchestrator {
         };
       }
 
-      // 根据 quality 模式处理
-      let urlsToDownload = rawUrls.slice(0, searchCount);
+      // 检查是否已中止
+      if (signal?.aborted) throw new Error('操作已取消');
+
+      // 根据 quality 模式处理（限制验证数量避免超时）
+      const maxValidate = Math.min(rawUrls.length, count * 2 + 5);
+      let urlsToDownload = rawUrls.slice(0, maxValidate);
       if (prioritizeQuality) {
         const sortByQuality = qualityMode === 'high';
         logger.info(`Validating ${urlsToDownload.length} URLs (quality=${qualityMode})...`);
@@ -287,6 +302,7 @@ export class Orchestrator {
           fetchQuality: sortByQuality,
           sortByQuality: sortByQuality,
           minFileSize: minFileSize,
+          signal,
         });
         
         // 过滤最小文件大小
@@ -364,10 +380,10 @@ export class Orchestrator {
     metrics.recordRequest();
     const requestId = `req_${Date.now()}_${Math.random().toString(36).substr(2, 6)}`;
     
-    // 根据关键词数量动态计算超时时间（每个关键词 8 秒，最少 20 秒，最多 50 秒）
+    // 根据关键词数量动态计算超时时间（每个关键词 15 秒，最少 25 秒，最多 50 秒）
     const keywords = this.parseKeywords(params.query);
     const keywordCount = keywords.length;
-    const GLOBAL_TIMEOUT = Math.min(Math.max(keywordCount * 8000, 20000), 50000);
+    const GLOBAL_TIMEOUT = Math.min(Math.max(keywordCount * 15000, 25000), 50000);
     
     logger.info(`[Orchestrator] Starting request: ${requestId}, keywords: ${keywordCount}, timeout: ${GLOBAL_TIMEOUT/1000}s`);