@skrillex1224/playwright-toolkit 2.0.58 → 2.0.60

package/dist/index.js

@@ -84,7 +84,9 @@ function createLogger(moduleName) {
 // src/errors.js
 var errors_exports = {};
 __export(errors_exports, {
-  CrawlerError: () => CrawlerError
+  CrawlerError: () => CrawlerError,
+  FirstDataTimeoutError: () => FirstDataTimeoutError,
+  OverallTimeoutError: () => OverallTimeoutError
 });
 import { serializeError } from "serialize-error";
 var CrawlerError = class _CrawlerError extends Error {
@@ -137,6 +139,38 @@ var CrawlerError = class _CrawlerError extends Error {
     return crawlerError;
   }
 };
+var FirstDataTimeoutError = class _FirstDataTimeoutError extends Error {
+  /**
+   * @param {number} timeout - 超时时间（毫秒）
+   * @param {Object} [context={}] - 上下文信息
+   */
+  constructor(timeout, context = {}) {
+    super(`\u9996\u6B21\u6570\u636E\u63A5\u6536\u8D85\u65F6 (${timeout}ms)`);
+    this.name = "FirstDataTimeoutError";
+    this.timeout = timeout;
+    this.context = context;
+    this.timestamp = (/* @__PURE__ */ new Date()).toISOString();
+    if (Error.captureStackTrace) {
+      Error.captureStackTrace(this, _FirstDataTimeoutError);
+    }
+  }
+};
+var OverallTimeoutError = class _OverallTimeoutError extends Error {
+  /**
+   * @param {number} timeout - 超时时间（毫秒）
+   * @param {Object} [context={}] - 上下文信息
+   */
+  constructor(timeout, context = {}) {
+    super(`\u6574\u4F53\u8BF7\u6C42\u8D85\u65F6 (${timeout}ms)`);
+    this.name = "OverallTimeoutError";
+    this.timeout = timeout;
+    this.context = context;
+    this.timestamp = (/* @__PURE__ */ new Date()).toISOString();
+    if (Error.captureStackTrace) {
+      Error.captureStackTrace(this, _OverallTimeoutError);
+    }
+  }
+};
 
 // src/apify-kit.js
 import { serializeError as serializeError2 } from "serialize-error";
@@ -243,28 +277,6 @@ var ApifyKit = {
 import delay from "delay";
 var logger2 = createLogger("Utils");
 var Utils = {
-  /**
-  * 解析 SSE 流文本
-  * 支持 `data: {...}` 和 `data:{...}` 两种格式
-  */
-  parseSseStream(sseStreamText) {
-    const events = [];
-    const lines = sseStreamText.split("\n");
-    for (const line of lines) {
-      if (line.startsWith("data:")) {
-        try {
-          const jsonContent = line.substring(5).trim();
-          if (jsonContent && jsonContent !== "[DONE]") {
-            events.push(JSON.parse(jsonContent));
-          }
-        } catch (e) {
-          logger2.debug("parseSseStream", `JSON \u89E3\u6790\u5931\u8D25: ${e.message}, line: ${line.substring(0, 100)}...`);
-        }
-      }
-    }
-    logger2.success("parseSseStream", `\u89E3\u6790\u5B8C\u6210, events \u6570\u91CF: ${events.length}`);
-    return events;
-  },
   /**
   * 解析 Cookie 字符串为 Playwright 格式的 Cookie 数组
   * @param {string} cookieString - Cookie 字符串
@@ -994,6 +1006,175 @@ var Captcha = {
   useCaptchaMonitor
 };
 
+// src/sse.js
+import https from "https";
+import { URL } from "url";
+var logger8 = createLogger("Sse");
+var Sse = {
+  /**
+  * 解析 SSE 流文本
+  * 支持 `data: {...}` 和 `data:{...}` 两种格式
+  * @param {string} sseStreamText
+  * @returns {Array<Object>} events
+  */
+  parseSseStream(sseStreamText) {
+    const events = [];
+    const lines = sseStreamText.split("\n");
+    for (const line of lines) {
+      if (line.startsWith("data:")) {
+        try {
+          const jsonContent = line.substring(5).trim();
+          if (jsonContent) {
+            events.push(JSON.parse(jsonContent));
+          }
+        } catch (e) {
+          logger8.debug("parseSseStream", `JSON \u89E3\u6790\u5931\u8D25: ${e.message}, line: ${line.substring(0, 100)}...`);
+        }
+      }
+    }
+    logger8.success("parseSseStream", `\u89E3\u6790\u5B8C\u6210, events \u6570\u91CF: ${events.length}`);
+    return events;
+  },
+  /**
+   * 拦截网络请求并使用 Node.js 原生 https 模块转发，以解决流式数据捕获问题。
+   * @param {import('playwright').Page} page
+   * @param {string|RegExp} urlPattern - 拦截的 URL 模式
+   * @param {object} options
+   * @param {function(string, function, string): void} [options.onData] - (textChunk, resolve, accumulatedText) => void
+   * @param {function(string, function): void} [options.onEnd] - (fullText, resolve) => void
+   * @param {function(Error): void} [options.onTimeout] - (error) => void，error 为 FirstDataTimeoutError 或 OverallTimeoutError
+   * @param {number} [options.firstDataTimeout=90000] - 首次数据接收超时 (ms)，默认 90s
+   * @param {number} [options.timeout=180000] - 整体请求超时时间 (ms)，默认 180s
+   * @returns {Promise<any>} - 返回 Promise，当流满足条件时 resolve
+   */
+  async intercept(page, urlPattern, options = {}) {
+    const {
+      onData,
+      onEnd,
+      onTimeout,
+      firstDataTimeout = 9e4,
+      timeout = 18e4
+    } = options;
+    let streamResolve, streamReject;
+    const capturePromise = new Promise((resolve, reject) => {
+      streamResolve = resolve;
+      streamReject = reject;
+    });
+    let firstDataTimer = null;
+    let overallTimer = null;
+    let hasReceivedFirstData = false;
+    const clearAllTimers = () => {
+      if (firstDataTimer) {
+        clearTimeout(firstDataTimer);
+        firstDataTimer = null;
+      }
+      if (overallTimer) {
+        clearTimeout(overallTimer);
+        overallTimer = null;
+      }
+    };
+    firstDataTimer = setTimeout(() => {
+      if (!hasReceivedFirstData) {
+        const error = new FirstDataTimeoutError(firstDataTimeout);
+        logger8.error(`[Intercept] \u9996\u6B21\u6570\u636E\u63A5\u6536\u8D85\u65F6`);
+        clearAllTimers();
+        if (onTimeout) {
+          onTimeout(error);
+        }
+        streamReject(error);
+      }
+    }, firstDataTimeout);
+    overallTimer = setTimeout(() => {
+      const error = new OverallTimeoutError(timeout);
+      logger8.error(`[Intercept] \u6574\u4F53\u8BF7\u6C42\u8D85\u65F6`);
+      clearAllTimers();
+      if (onTimeout) {
+        onTimeout(error);
+      }
+      streamReject(error);
+    }, timeout);
+    await page.route(urlPattern, async (route) => {
+      const request = route.request();
+      const requestUrl = request.url();
+      logger8.info(`[MITM] \u5DF2\u62E6\u622A\u8BF7\u6C42: ${requestUrl}`);
+      try {
+        const headers = await request.allHeaders();
+        const postData = request.postData();
+        const urlObj = new URL(requestUrl);
+        delete headers["accept-encoding"];
+        delete headers["content-length"];
+        const reqOptions = {
+          hostname: urlObj.hostname,
+          port: 443,
+          path: urlObj.pathname + urlObj.search,
+          method: request.method(),
+          headers,
+          timeout
+        };
+        const req = https.request(reqOptions, (res) => {
+          const chunks = [];
+          let accumulatedText = "";
+          res.on("data", (chunk) => {
+            if (!hasReceivedFirstData) {
+              hasReceivedFirstData = true;
+              if (firstDataTimer) {
+                clearTimeout(firstDataTimer);
+                firstDataTimer = null;
+              }
+              logger8.debug("[Intercept] \u5DF2\u63A5\u6536\u9996\u6B21\u6570\u636E");
+            }
+            chunks.push(chunk);
+            const textChunk = chunk.toString("utf-8");
+            accumulatedText += textChunk;
+            logger8.debug(`[CHUNK] ${textChunk.length} bytes`);
+            if (onData) {
+              try {
+                onData(textChunk, streamResolve, accumulatedText);
+              } catch (e) {
+                logger8.error(`onData callback error: ${e.message}`);
+              }
+            }
+          });
+          res.on("end", () => {
+            logger8.info("[MITM] \u4E0A\u6E38\u54CD\u5E94\u7ED3\u675F (Stream End)");
+            clearAllTimers();
+            const fullBody = Buffer.concat(chunks);
+            if (onEnd) {
+              try {
+                onEnd(accumulatedText, streamResolve);
+              } catch (e) {
+                logger8.error(`onEnd callback error: ${e.message}`);
+              }
+            } else if (!onData) {
+              streamResolve(accumulatedText);
+            }
+            route.fulfill({
+              status: res.statusCode,
+              headers: res.headers,
+              body: fullBody
+            }).catch((e) => logger8.warning(`Route fulfill failed: ${e.message}`));
+          });
+        });
+        req.on("error", (e) => {
+          logger8.error(`[MITM] Upstream request error: ${e.message}`);
+          clearAllTimers();
+          route.abort();
+          streamReject(e);
+        });
+        if (postData) {
+          req.write(postData);
+        }
+        req.end();
+      } catch (e) {
+        logger8.error(`[MITM] Setup error: ${e.message}`);
+        clearAllTimers();
+        route.continue();
+      }
+    });
+    return capturePromise;
+  }
+};
+
 // index.js
 var usePlaywrightToolKit = () => {
   return {
@@ -1005,6 +1186,7 @@ var usePlaywrightToolKit = () => {
     Constants: constants_exports,
     Utils,
     Captcha,
+    Sse,
     Errors: errors_exports
   };
 };