@skrillex1224/playwright-toolkit 2.1.135 → 2.1.137

package/dist/index.js

@@ -1432,7 +1432,7 @@ var Captcha = {
 
 // src/sse.js
 import https from "https";
-import { URL as URL2 } from "url";
+import { URL } from "url";
 var logger8 = createInternalLogger("Sse");
 var Sse = {
   /**
@@ -1542,7 +1542,7 @@ var Sse = {
         try {
           const headers = await request.allHeaders();
           const postData = request.postData();
-          const urlObj = new URL2(request.url());
+          const urlObj = new URL(request.url());
           delete headers["accept-encoding"];
           delete headers["content-length"];
           const reqOptions = {
@@ -1824,7 +1824,6 @@ var Interception = {
    * @param {import('playwright').Page} page - Playwright Page 对象
    * @param {Object} [options] - 配置选项
    * @param {string[]} [options.directDomains] - [已过时] 需要直连的域名列表，请改用 directConfig.domains
-   * @param {string[]} [options.directExtensions] - 需要直连的扩展名（如 .js/.css）
    * @param {Object} [options.directConfig] - 直连配置（参考 blockingConfig），支持 domains/directAllXxx 系列开关
    * @param {Object} [options.blockingConfig] - 资源屏蔽配置
    * @param {boolean} [options.fallbackToProxy] - 直连失败时是否回退到代理（默认 true）
@@ -1833,7 +1832,6 @@ var Interception = {
   async setup(page, options = {}) {
     const {
       directDomains = [],
-      directExtensions = [],
       directConfig = {},
       blockingConfig = {},
       fallbackToProxy = true
@@ -1850,10 +1848,20 @@ var Interception = {
     const hasDirectDomains = normalizedDirectDomains.length > 0;
     const normalizedDirectExtensions = Array.from(
       new Set(
-        [...directExtensions, ...configuredDirectExtensions].map((ext) => String(ext || "").toLowerCase().trim()).filter(Boolean).map((ext) => ext.startsWith(".") ? ext : `.${ext}`)
+        configuredDirectExtensions.map((ext) => String(ext || "").toLowerCase().trim()).filter(Boolean).map((ext) => ext.startsWith(".") ? ext : `.${ext}`)
       )
     );
     const hasDirectExtensions = normalizedDirectExtensions.length > 0;
+    const blockedExtensionsSet = new Set(blockedExtensions);
+    const directOnlyExtensions = normalizedDirectExtensions.filter((ext) => !blockedExtensionsSet.has(ext));
+    const directRouteMatchers = [
+      createDomainsRegex(normalizedDirectDomains),
+      createExtensionsRegex(directOnlyExtensions)
+    ].filter(Boolean);
+    const blockRouteMatchers = [createExtensionsRegex(blockedExtensions)].filter(Boolean);
+    const hasDirectRules = directRouteMatchers.length > 0;
+    const hasBlockRules = blockRouteMatchers.length > 0;
+    const hasInterceptionRules = hasDirectRules || hasBlockRules;
     const enabledCategories = [];
     if (mergedBlockingConfig.blockArchive) enabledCategories.push("\u538B\u7F29\u5305");
     if (mergedBlockingConfig.blockExecutable) enabledCategories.push("\u53EF\u6267\u884C\u6587\u4EF6");
@@ -1867,104 +1875,83 @@ var Interception = {
     if (hasDirectDomains) directRules.push(`\u76F4\u8FDE\u57DF\u540D: [${normalizedDirectDomains.length} \u4E2A]`);
     if (hasDirectExtensions) directRules.push(`\u76F4\u8FDE\u6269\u5C55\u540D: [${normalizedDirectExtensions.join(", ")}]`);
     logger9.start("setup", directRules.length > 0 ? `${directRules.join(" | ")} | \u5C4F\u853D: [${enabledCategories.join(", ")}]` : `\u4EC5\u8D44\u6E90\u5C4F\u853D\u6A21\u5F0F | \u5C4F\u853D: [${enabledCategories.join(", ")}]`);
-    await page.route("**/*", async (route) => {
-      let handled = false;
+    if (!hasInterceptionRules) {
+      logger9.info("\u65E0\u547D\u4E2D\u89C4\u5219\uFF0C\u8DF3\u8FC7 page.route \u6CE8\u518C");
+      return;
+    }
+    const directRouteHandler = async (route) => {
       try {
         const request = route.request();
         const url = request.url();
         const urlLower = url.toLowerCase();
         const urlPath = urlLower.split("?")[0];
         const isSilent = DirectConfig.silentExtensions.some((ext) => urlPath.endsWith(ext));
-        const shouldBlock = blockedExtensions.some((ext) => urlPath.endsWith(ext));
-        if (shouldBlock) {
-          await route.abort();
-          handled = true;
-          return;
-        }
-        let isDirectByDomain = false;
-        if (hasDirectDomains) {
-          try {
-            const hostname = new URL(url).hostname;
-            isDirectByDomain = normalizedDirectDomains.some((domain) => hostname.startsWith(domain));
-          } catch (e) {
-          }
-        }
-        const isDirectByExtension = hasDirectExtensions ? normalizedDirectExtensions.some((ext) => urlPath.endsWith(ext)) : false;
-        const isDirect = isDirectByDomain || isDirectByExtension;
-        if (isDirect) {
-          try {
-            const reqHeaders = await request.allHeaders();
-            delete reqHeaders["host"];
-            const resolvedAcceptLanguage = reqHeaders["accept-language"] || "";
-            const userAgent = reqHeaders["user-agent"] || "";
-            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",
-              // 强制获取 Buffer
-              // 移除手动 TLS 指纹配置，使用 got-scraping 默认的高质量指纹
-              // headerGeneratorOptions: ... 
-              // 使用共享的 Agent 单例（keepAlive: false，不会池化连接）
-              agent: {
-                http: SHARED_HTTP_AGENT,
-                https: SHARED_HTTPS_AGENT
-              },
-              // 超时时间
-              timeout: { request: DirectConfig.directTimeout * 1e3 }
-            });
-            const resHeaders = {};
-            for (const [key, value] of Object.entries(response.headers)) {
-              if (Array.isArray(value)) {
-                resHeaders[key] = value.join(", ");
-              } else if (value) {
-                resHeaders[key] = String(value);
-              }
-            }
-            delete resHeaders["content-encoding"];
-            delete resHeaders["content-length"];
-            delete resHeaders["transfer-encoding"];
-            delete resHeaders["connection"];
-            delete resHeaders["keep-alive"];
-            isSilent ? logger9.debug(`\u76F4\u8FDE\u6210\u529F: ${urlPath}`) : logger9.info(`\u76F4\u8FDE\u6210\u529F: ${urlPath}`);
-            await safeFulfill(route, {
-              status: response.statusCode,
-              headers: resHeaders,
-              body: response.body
-            });
-            handled = true;
-            return;
-          } catch (e) {
-            const isTimeout = e.code === "ETIMEDOUT" || e.message.toLowerCase().includes("timeout");
-            const action = fallbackToProxy ? "\u56DE\u9000\u4EE3\u7406" : "\u5DF2\u653E\u5F03";
-            const reason = isTimeout ? `\u8D85\u65F6(${DirectConfig.directTimeout}s)` : `\u5F02\u5E38: ${e.message}`;
-            logger9.warn(`\u76F4\u8FDE${reason}\uFF0C${action}: ${urlPath}`);
-            if (fallbackToProxy) {
-              await safeContinue(route);
-            } else {
-              await route.abort();
-            }
-            handled = true;
-            return;
+        const reqHeaders = await request.allHeaders();
+        delete reqHeaders["host"];
+        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",
+          // 强制获取 Buffer
+          // 使用共享的 Agent 单例（keepAlive: false，不会池化连接）
+          agent: {
+            http: SHARED_HTTP_AGENT,
+            https: SHARED_HTTPS_AGENT
+          },
+          // 超时时间
+          timeout: { request: DirectConfig.directTimeout * 1e3 }
+        });
+        const resHeaders = {};
+        for (const [key, value] of Object.entries(response.headers)) {
+          if (Array.isArray(value)) {
+            resHeaders[key] = value.join(", ");
+          } else if (value) {
+            resHeaders[key] = String(value);
           }
         }
-        await safeContinue(route);
-        handled = true;
+        delete resHeaders["content-encoding"];
+        delete resHeaders["content-length"];
+        delete resHeaders["transfer-encoding"];
+        delete resHeaders["connection"];
+        delete resHeaders["keep-alive"];
+        isSilent ? logger9.debug(`\u76F4\u8FDE\u6210\u529F: ${urlPath}`) : logger9.info(`\u76F4\u8FDE\u6210\u529F: ${urlPath}`);
+        await safeFulfill(route, {
+          status: response.statusCode,
+          headers: resHeaders,
+          body: response.body
+        });
       } catch (err) {
-        logger9.warn(`\u8DEF\u7531\u5904\u7406\u5F02\u5E38: ${err.message}`);
-        if (!handled) {
-          try {
-            await route.continue();
-          } catch (_) {
-          }
+        const message = String(err?.message || "");
+        const isTimeout = err?.code === "ETIMEDOUT" || message.toLowerCase().includes("timeout");
+        const action = fallbackToProxy ? "\u56DE\u9000\u4EE3\u7406" : "\u5DF2\u653E\u5F03";
+        const reason = isTimeout ? `\u8D85\u65F6(${DirectConfig.directTimeout}s)` : `\u5F02\u5E38: ${message}`;
+        logger9.warn(`\u76F4\u8FDE${reason}\uFF0C${action}`);
+        if (fallbackToProxy) {
+          await safeContinue(route);
+        } else {
+          await safeAbort(route);
         }
       }
-    });
+    };
+    const blockRouteHandler = async (route) => {
+      await safeAbort(route);
+    };
+    if (hasDirectRules) {
+      for (const matcher of directRouteMatchers) {
+        await page.route(matcher, directRouteHandler);
+      }
+    }
+    if (hasBlockRules) {
+      for (const matcher of blockRouteMatchers) {
+        await page.route(matcher, blockRouteHandler);
+      }
+    }
   }
 };
 async function safeFulfill(route, options) {
@@ -1984,10 +1971,37 @@ async function safeContinue(route) {
     }
   }
 }
+async function safeAbort(route) {
+  try {
+    await route.abort();
+  } catch (error) {
+    if (!isIgnorableError(error)) {
+    }
+  }
+}
 function isIgnorableError(error) {
   const msg = error.message;
   return msg.includes("already handled") || msg.includes("Target closed") || msg.includes("closed");
 }
+function createExtensionsRegex(extensions = []) {
+  const normalized = Array.from(new Set(
+    extensions.map((ext) => String(ext || "").toLowerCase().trim()).filter(Boolean).map((ext) => ext.startsWith(".") ? ext : `.${ext}`)
+  ));
+  if (!normalized.length) return null;
+  const escaped = normalized.map((ext) => escapeRegex(ext));
+  return new RegExp(`(?:${escaped.join("|")})(?:\\?.*)?$`, "i");
+}
+function createDomainsRegex(domains = []) {
+  const normalized = Array.from(new Set(
+    domains.map((domain) => String(domain || "").toLowerCase().trim()).filter(Boolean)
+  ));
+  if (!normalized.length) return null;
+  const domainParts = normalized.map((domain) => `${escapeRegex(domain)}[^/:?#]*`);
+  return new RegExp(`^https?:\\/\\/(?:${domainParts.join("|")})(?::\\d+)?(?:[/?#]|$)`, "i");
+}
+function escapeRegex(value) {
+  return String(value).replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+}
 
 // src/mutation.js
 import { v4 as uuidv42 } from "uuid";