@skrillex1224/playwright-toolkit 2.1.9 → 2.1.10

package/dist/index.js

@@ -1252,6 +1252,14 @@ var DEFAULT_BLOCKING_CONFIG = {
   /** 额外自定义扩展名列表 */
   customExtensions: []
 };
+var SHARED_GOT_OPTIONS = {
+  http2: false,
+  // 禁用 HTTP2 避免在拦截场景下的握手兼容性问题
+  retry: { limit: 0 },
+  // 让 Playwright 或外层逻辑处理重试
+  throwHttpErrors: false
+  // 404/500 等错误不抛出异常，直接透传给浏览器
+};
 var Interception = {
   /**
    * 根据配置生成需要屏蔽的扩展名列表
@@ -1295,16 +1303,6 @@ var Interception = {
   /**
    * 设置网络拦截规则（资源屏蔽 + CDN 直连）
    * 
-   * 工作流程：
-   * 1. 检查请求是否在屏蔽列表中 → 如果是，直接 abort
-   * 2. 检查是否匹配直连域名 → 如果是，使用 Node.js fetch 直连
-   * 3. 其他请求正常走代理
-   * 
-   * 适用场景：
-   * - 代理 IP 无法访问某些 CDN 域名
-   * - 需要加速静态资源加载
-   * - 屏蔽不必要的资源请求
-   * 
    * @param {import('playwright').Page} page - Playwright Page 对象
    * @param {Object} [options] - 配置选项
    * @param {string[]} [options.directDomains] - 需要直连的域名列表
@@ -1317,10 +1315,10 @@ var Interception = {
       directDomains = [],
       blockingConfig = {},
       fallbackToProxy = true
-      // 默认回退到代理，保证可用性
     } = options;
     const mergedBlockingConfig = { ...DEFAULT_BLOCKING_CONFIG, ...blockingConfig };
     const blockedExtensions = this.getBlockedExtensions(mergedBlockingConfig);
+    const hasDirectDomains = directDomains.length > 0;
     const enabledCategories = [];
     if (mergedBlockingConfig.blockArchive) enabledCategories.push("\u538B\u7F29\u5305");
     if (mergedBlockingConfig.blockExecutable) enabledCategories.push("\u53EF\u6267\u884C\u6587\u4EF6");
@@ -1330,21 +1328,19 @@ var Interception = {
     if (mergedBlockingConfig.blockFont) enabledCategories.push("\u5B57\u4F53");
     if (mergedBlockingConfig.blockCss) enabledCategories.push("CSS");
     if (mergedBlockingConfig.blockOther) enabledCategories.push("\u5176\u4ED6");
-    const hasDirectDomains = directDomains.length > 0;
-    logger9.start("setup", hasDirectDomains ? `\u76F4\u8FDE\u57DF\u540D: [${directDomains.join(", ")}]` : "\u4EC5\u8D44\u6E90\u5C4F\u853D\u6A21\u5F0F");
+    logger9.start("setup", hasDirectDomains ? `\u76F4\u8FDE\u57DF\u540D: [${directDomains.length} \u4E2A] | \u5C4F\u853D: [${enabledCategories.join(", ")}]` : `\u4EC5\u8D44\u6E90\u5C4F\u853D\u6A21\u5F0F | \u5C4F\u853D: [${enabledCategories.join(", ")}]`);
     await page.route("**/*", async (route) => {
-      const request = route.request();
-      const url = request.url();
-      const urlLower = url.toLowerCase();
-      const urlPath = urlLower.split("?")[0];
-      const shouldBlock = blockedExtensions.some((ext) => urlPath.endsWith(ext));
-      if (shouldBlock) {
-        logger9.debug(`\u5DF2\u5C4F\u853D: ${url.substring(0, 100)}`);
-        return route.abort();
-      }
-      if (hasDirectDomains) {
-        const matchesDomain = directDomains.some((domain) => url.includes(domain));
-        if (matchesDomain) {
+      try {
+        const request = route.request();
+        const url = request.url();
+        const urlLower = url.toLowerCase();
+        const urlPath = urlLower.split("?")[0];
+        const shouldBlock = blockedExtensions.some((ext) => urlPath.endsWith(ext));
+        if (shouldBlock) {
+          await route.abort();
+          return;
+        }
+        if (hasDirectDomains && directDomains.some((domain) => url.includes(domain))) {
           try {
             const reqHeaders = await request.allHeaders();
             delete reqHeaders["host"];
@@ -1352,30 +1348,29 @@ var Interception = {
             const method = request.method();
             const postData = method !== "GET" && method !== "HEAD" ? request.postDataBuffer() : void 0;
             const response = await gotScraping({
+              ...SHARED_GOT_OPTIONS,
+              // 应用通用配置
               url,
               method,
               headers: reqHeaders,
               body: postData,
               responseType: "buffer",
-              // 直接获取 Buffer
-              // 【修复 1】禁用 HTTP/2，防止握手挂起
-              http2: false,
-              // 模拟浏览器指纹 (Header 部分)
+              // 强制获取 Buffer
+              // 模拟浏览器 TLS 指纹
               headerGeneratorOptions: Stealth.getTlsFingerprintOptions(userAgent),
-              retry: { limit: 0 },
-              // 【修复 2】显式使用新的 Agent 实例，彻底无视 Apify 的环境变量代理
-              // 注意：这会牺牲一部分 TLS 指纹模拟能力，但能保证网络通畅
+              // 【核心修复 1】: keepAlive 设置为 false
+              // 每一个拦截请求都是独立的，使用 keepAlive 会导致 Agent 池耗尽，
+              // 从而导致后续请求一直 Pending。
               agent: {
-                http: new HttpAgent({ keepAlive: true }),
-                https: new HttpsAgent({ keepAlive: true, rejectUnauthorized: false })
+                http: new HttpAgent({ keepAlive: false }),
+                https: new HttpsAgent({ keepAlive: false, rejectUnauthorized: false })
               },
-              // 设置超时，防止永久 Pending
+              // 设置较短的超时时间，给回退代理留出机会
               timeout: {
-                request: 15 * 1e3
-                // 建议缩短到 15s
+                request: 12 * 1e3
+                // 12秒超时
               }
             });
-            const body = response.body;
             const resHeaders = {};
             for (const [key, value] of Object.entries(response.headers)) {
               if (Array.isArray(value)) {
@@ -1386,29 +1381,61 @@ var Interception = {
             }
             delete resHeaders["content-encoding"];
             delete resHeaders["content-length"];
-            logger9.debug(`\u76F4\u8FDE\u6210\u529F: ${url.substring(0, 100)}`);
-            await route.fulfill({
+            delete resHeaders["transfer-encoding"];
+            delete resHeaders["connection"];
+            delete resHeaders["keep-alive"];
+            logger9.info(`\u76F4\u8FDE\u6210\u529F: ${url}`);
+            await safeFulfill(route, {
               status: response.statusCode,
               headers: resHeaders,
-              body
+              body: response.body
             });
             return;
           } catch (e) {
             if (fallbackToProxy) {
-              logger9.warn(`\u76F4\u8FDE\u5931\u8D25\uFF0C\u56DE\u9000\u4EE3\u7406: ${url.substring(0, 80)} | \u539F\u56E0: ${e.message}`);
-              return route.continue();
+              logger9.warn(`\u76F4\u8FDE\u5F02\u5E38\uFF0C\u56DE\u9000\u4EE3\u7406: ${url} | Err: ${e.message}`);
+              await safeContinue(route);
+              return;
             } else {
-              logger9.warn(`\u76F4\u8FDE\u5931\u8D25: ${url.substring(0, 80)} | \u539F\u56E0: ${e.message}`);
-              return route.abort();
+              logger9.warn(`\u76F4\u8FDE\u5931\u8D25: ${url} | Err: ${e.message}`);
+              await route.abort();
+              return;
             }
           }
         }
+        await safeContinue(route);
+      } catch (err) {
+        try {
+          if (!route.request().failure()) {
+            await route.continue();
+          }
+        } catch (_) {
+        }
       }
-      return route.continue();
     });
-    logger9.success("setup", `\u5C4F\u853D\u5206\u7C7B: [${enabledCategories.join(", ")}]`);
   }
 };
+async function safeFulfill(route, options) {
+  try {
+    await route.fulfill(options);
+  } catch (error) {
+    if (!isIgnorableError(error)) {
+      console.error(`[Interception] Fulfill Error: ${error.message}`);
+    }
+  }
+}
+async function safeContinue(route) {
+  try {
+    await route.continue();
+  } catch (error) {
+    if (!isIgnorableError(error)) {
+    }
+  }
+}
+function isIgnorableError(error) {
+  const msg = error.message;
+  return msg.includes("already handled") || msg.includes("Target closed") || msg.includes("closed");
+}
 
 // index.js
 var usePlaywrightToolKit = () => {