@skrillex1224/playwright-toolkit 2.1.35 → 2.1.37

package/README.md

@@ -38,9 +38,6 @@ const crawler = new PlaywrightCrawler({
     },
     preNavigationHooks: [
         async ({ page }) => {
-            // 统一反爬：时区/语言/权限/视口
-            await AntiCheat.applyPage(page);
-            
             // 验证码监控
             Captcha.useCaptchaMonitor(page, {
                 domSelector: '#captcha_container',

package/dist/index.cjs

@@ -183,37 +183,86 @@ async function createApifyKit() {
   const { Actor: Actor2 } = apify;
   return {
     /**
-     * 核心封装：执行步骤，带自动日志确认和失败截图处理
+     * 核心封装：执行步骤，带自动日志确认、失败截图处理和重试机制
+     * 
+     * @param {string} step - 步骤名称
+     * @param {import('playwright').Page} page - Playwright page 对象
+     * @param {Function} actionFn - 执行的异步操作
+     * @param {Object} [options] - 配置选项
+     * @param {boolean} [options.failActor=true] - 失败时是否调用 Actor.fail
+     * @param {Object} [options.retry] - 重试配置
+     * @param {number} [options.retry.times=0] - 重试次数
+     * @param {'direct'|'refresh'} [options.retry.mode='direct'] - 重试模式
+     * @param {Function} [options.retry.before] - 重试前钩子，可覆盖默认等待行为
      */
     async runStep(step, page, actionFn, options = {}) {
-      const { failActor = true } = options;
-      logger.start(`[Step] ${step}`);
-      try {
-        const result = await actionFn();
-        logger.success(`[Step] ${step}`);
-        return result;
-      } catch (error) {
-        logger.fail(`[Step] ${step}`, error);
-        if (failActor) {
-          let base64 = "\u622A\u56FE\u5931\u8D25";
-          try {
-            if (page) {
-              const buffer = await page.screenshot({ fullPage: true, type: "jpeg", quality: 60 });
-              base64 = `data:image/jpeg;base64,${buffer.toString("base64")}`;
-            }
-          } catch (snapErr) {
-            logger.warn(`\u622A\u56FE\u751F\u6210\u5931\u8D25: ${snapErr.message}`);
-          }
-          await this.pushFailed(error, {
-            step,
-            page,
-            options,
-            base64
-          });
-          await Actor2.fail(`Run Step ${step} \u5931\u8D25: ${error.message}`);
+      const { failActor = true, retry = {} } = options;
+      const { times: retryTimes = 0, mode: retryMode = "direct", before: beforeRetry } = retry;
+      const executeAction = async (attemptNumber) => {
+        const attemptLabel = attemptNumber > 0 ? ` (\u91CD\u8BD5 #${attemptNumber})` : "";
+        logger.start(`[Step] ${step}${attemptLabel}`);
+        try {
+          const result = await actionFn();
+          logger.success(`[Step] ${step}${attemptLabel}`);
+          return { success: true, result };
+        } catch (error) {
+          logger.fail(`[Step] ${step}${attemptLabel}`, error);
+          return { success: false, error };
+        }
+      };
+      const prepareForRetry = async (attemptNumber) => {
+        if (typeof beforeRetry === "function") {
+          logger.start(`[RetryStep] \u6267\u884C\u81EA\u5B9A\u4E49 before \u94A9\u5B50 (\u7B2C ${attemptNumber} \u6B21\u91CD\u8BD5)`);
+          await beforeRetry(page, attemptNumber);
+          logger.success(`[RetryStep] before \u94A9\u5B50\u5B8C\u6210`);
+        } else if (retryMode === "refresh") {
+          logger.start(`[RetryStep] \u5237\u65B0\u9875\u9762 (\u7B2C ${attemptNumber} \u6B21\u91CD\u8BD5)`);
+          await page.reload({ waitUntil: "domcontentloaded" });
+          logger.success(`[RetryStep] \u9875\u9762\u5237\u65B0\u5B8C\u6210`);
         } else {
-          throw error;
+          logger.start(`[RetryStep] \u7B49\u5F85 3 \u79D2 (\u7B2C ${attemptNumber} \u6B21\u91CD\u8BD5)`);
+          await new Promise((resolve) => setTimeout(resolve, 3e3));
+          logger.success(`[RetryStep] \u7B49\u5F85\u5B8C\u6210`);
         }
+      };
+      let lastResult = await executeAction(0);
+      if (lastResult.success) {
+        return lastResult.result;
+      }
+      for (let attempt = 1; attempt <= retryTimes; attempt++) {
+        logger.start(`[RetryStep] \u51C6\u5907\u7B2C ${attempt}/${retryTimes} \u6B21\u91CD\u8BD5: ${step}`);
+        try {
+          await prepareForRetry(attempt);
+        } catch (prepareError) {
+          logger.warn(`[RetryStep] \u91CD\u8BD5\u51C6\u5907\u5931\u8D25: ${prepareError.message}`);
+          continue;
+        }
+        lastResult = await executeAction(attempt);
+        if (lastResult.success) {
+          return lastResult.result;
+        }
+      }
+      const finalError = lastResult.error;
+      if (failActor) {
+        let base64 = "\u622A\u56FE\u5931\u8D25";
+        try {
+          if (page) {
+            const buffer = await page.screenshot({ fullPage: true, type: "jpeg", quality: 60 });
+            base64 = `data:image/jpeg;base64,${buffer.toString("base64")}`;
+          }
+        } catch (snapErr) {
+          logger.warn(`\u622A\u56FE\u751F\u6210\u5931\u8D25: ${snapErr.message}`);
+        }
+        await this.pushFailed(finalError, {
+          step,
+          page,
+          options,
+          base64,
+          retryAttempts: retryTimes
+        });
+        await Actor2.fail(`Run Step ${step} \u5931\u8D25 (\u5DF2\u91CD\u8BD5 ${retryTimes} \u6B21): ${finalError.message}`);
+      } else {
+        throw finalError;
       }
     },
     /**
@@ -376,25 +425,12 @@ var BASE_CONFIG = Object.freeze({
   geolocation: null
 });
 var DEFAULT_LAUNCH_ARGS = [
-  "--disable-blink-features=AutomationControlled",
+  // '--disable-blink-features=AutomationControlled', // Crawlee 可能会自动处理，过多干预反而会被识别
   "--no-sandbox",
   "--disable-setuid-sandbox",
   "--window-position=0,0",
   `--lang=${BASE_CONFIG.locale}`
 ];
-var ADVANCED_LAUNCH_ARGS = [
-  ...DEFAULT_LAUNCH_ARGS,
-  "--disable-dev-shm-usage",
-  "--disable-background-networking",
-  "--disable-default-apps",
-  "--disable-extensions",
-  "--disable-sync",
-  "--disable-translate",
-  "--metrics-recording-only",
-  "--mute-audio",
-  "--no-first-run"
-];
-var CONTEXT_CONFIG_CACHE = /* @__PURE__ */ new WeakMap();
 function buildFingerprintOptions(locale) {
   return {
     browsers: [{ name: "chrome", minVersion: 110 }],
@@ -403,95 +439,9 @@ function buildFingerprintOptions(locale) {
     locales: [locale]
   };
 }
-function parseAcceptLanguage(acceptLanguage) {
-  if (!acceptLanguage) return [];
-  return acceptLanguage.split(",").map((part) => part.trim().split(";")[0]).filter(Boolean);
-}
-function normalizeLanguages(acceptLanguage, fallbackLocale) {
-  const languages = parseAcceptLanguage(acceptLanguage);
-  if (languages.length === 0) return [fallbackLocale];
-  if (!languages.includes(fallbackLocale)) {
-    return [fallbackLocale, ...languages];
-  }
-  return languages;
-}
-function getOperatingSystemsFromUserAgent(userAgent) {
-  const lowerUA = userAgent.toLowerCase();
-  if (lowerUA.includes("windows")) return ["windows"];
-  if (lowerUA.includes("mac os") || lowerUA.includes("macintosh")) return ["macos"];
-  if (lowerUA.includes("linux")) return ["linux"];
-  return [];
-}
-function buildContextConfigKey(config) {
-  return JSON.stringify({
-    locale: config.locale,
-    acceptLanguage: config.acceptLanguage,
-    timezoneId: config.timezoneId,
-    timezoneOffset: config.timezoneOffset
-  });
-}
-async function applyContextSettings(context, config, languages, permissions, injectLocaleTimezone) {
-  const contextKey = buildContextConfigKey(config);
-  const cached = CONTEXT_CONFIG_CACHE.get(context);
-  const isFirstInit = !cached;
-  const effectiveConfig = cached?.config || config;
-  const effectiveLanguages = cached?.languages || languages;
-  if (isFirstInit) {
-    CONTEXT_CONFIG_CACHE.set(context, {
-      key: contextKey,
-      config,
-      languages
-    });
-  } else if (cached.key !== contextKey) {
-    logger3.warn("applyContext", "Context already initialized; ignore conflicting locale/timezone.");
-  }
-  await context.setExtraHTTPHeaders({
-    "accept-language": effectiveConfig.acceptLanguage
-  });
-  if (isFirstInit) {
-    await context.addInitScript(({ locale, timezoneId, timezoneOffset, languages: languages2, applyLocaleTimezone }) => {
-      const originalDateTimeFormat = Intl.DateTimeFormat;
-      if (applyLocaleTimezone) {
-        Intl.DateTimeFormat = function(locales, initOptions) {
-          const nextLocales = locales || locale;
-          const nextOptions = initOptions ? { ...initOptions } : {};
-          nextOptions.timeZone = nextOptions.timeZone || timezoneId;
-          return new originalDateTimeFormat(nextLocales, nextOptions);
-        };
-        Intl.DateTimeFormat.prototype = originalDateTimeFormat.prototype;
-        Date.prototype.getTimezoneOffset = function() {
-          return timezoneOffset;
-        };
-        Object.defineProperty(navigator, "language", { get: () => languages2[0] });
-        Object.defineProperty(navigator, "languages", { get: () => languages2 });
-      }
-      Object.defineProperty(navigator, "webdriver", { get: () => void 0 });
-    }, {
-      locale: effectiveConfig.locale,
-      timezoneId: effectiveConfig.timezoneId,
-      timezoneOffset: effectiveConfig.timezoneOffset,
-      languages: effectiveLanguages,
-      applyLocaleTimezone: injectLocaleTimezone
-    });
-  }
-  if (effectiveConfig.geolocation) {
-    await context.setGeolocation(effectiveConfig.geolocation);
-    await context.grantPermissions(["geolocation"]);
-  }
-  if (permissions?.length) {
-    await context.grantPermissions(permissions);
-  }
-}
-function resolveConfig(overrides = {}) {
-  return {
-    ...BASE_CONFIG,
-    ...overrides,
-    geolocation: overrides.geolocation === null ? null : overrides.geolocation || BASE_CONFIG.geolocation
-  };
-}
 var AntiCheat = {
   /**
-   * 获取统一的基础配置（中国、桌面端、中文语言）。
+   * 获取统一的基础配置
    */
   getBaseConfig() {
     return { ...BASE_CONFIG };
@@ -508,100 +458,18 @@ var AntiCheat = {
   getLaunchArgs() {
     return [...DEFAULT_LAUNCH_ARGS];
   },
-  /**
-   * 获取增强启动参数（高风险场景）。
-   */
-  getAdvancedLaunchArgs() {
-    return [...ADVANCED_LAUNCH_ARGS];
-  },
-  /**
-   * 统一应用到 BrowserContext（时区/语言/权限/地理位置）。
-   *
-   * @param {import('playwright').BrowserContext} context
-   * @param {Object} [options]
-   * @param {string} [options.locale]
-   * @param {string} [options.acceptLanguage]
-   * @param {string} [options.timezoneId]
-   * @param {number} [options.timezoneOffset]
-   * @param {import('playwright').Geolocation|null} [options.geolocation]
-   * @param {string[]} [options.permissions]
-   */
-  async applyContext(context, options = {}) {
-    const config = resolveConfig(options);
-    const languages = normalizeLanguages(config.acceptLanguage, config.locale);
-    const permissions = Array.isArray(options.permissions) ? options.permissions : [];
-    await applyContextSettings(context, config, languages, permissions, true);
-    logger3.success("applyContext", `${config.locale} | ${config.timezoneId}`);
-  },
-  /**
-   * 统一应用到 Page（Context + 视口同步）。
-   *
-   * @param {import('playwright').Page} page
-   * @param {Object} [options] - 传递给 applyContext 的选项
-   */
-  async applyPage(page, options = {}) {
-    const config = resolveConfig(options);
-    const languages = normalizeLanguages(config.acceptLanguage, config.locale);
-    const permissions = Array.isArray(options.permissions) ? options.permissions : [];
-    let injectLocaleTimezone = true;
-    try {
-      const env = await page.evaluate(() => ({
-        language: navigator.language,
-        languages: Array.isArray(navigator.languages) ? navigator.languages : [],
-        timeZone: Intl.DateTimeFormat().resolvedOptions().timeZone,
-        tzOffset: (/* @__PURE__ */ new Date()).getTimezoneOffset()
-      }));
-      const languageMatch = env.language === languages[0];
-      const timeZoneMatch = env.timeZone === config.timezoneId && env.tzOffset === config.timezoneOffset;
-      injectLocaleTimezone = !(languageMatch && timeZoneMatch);
-    } catch (e) {
-      injectLocaleTimezone = true;
-    }
-    await applyContextSettings(page.context(), config, languages, permissions, injectLocaleTimezone);
-    await this.syncViewportWithScreen(page);
-  },
-  /**
-   * 同步 Page 视口到 window.screen，避免视口/屏幕不一致检测。
-   */
-  async syncViewportWithScreen(page) {
-    try {
-      const screen = await page.evaluate(() => ({
-        width: window.screen.width,
-        height: window.screen.height
-      }));
-      await page.setViewportSize({
-        width: screen.width,
-        height: screen.height
-      });
-      logger3.success("syncViewport", `size=${screen.width}x${screen.height}`);
-    } catch (e) {
-      logger3.warn(`syncViewport \u5931\u8D25: ${e.message}\uFF0C\u56DE\u9000\u5230 1920x1080`);
-      await page.setViewportSize({ width: 1920, height: 1080 });
-    }
-  },
   /**
    * 为 got-scraping 生成与浏览器一致的 TLS 指纹配置（桌面端）。
-   *
-   * @param {string} [userAgent]
    */
   getTlsFingerprintOptions(userAgent = "", acceptLanguage = "") {
-    const primaryLocale = parseAcceptLanguage(acceptLanguage || BASE_CONFIG.acceptLanguage)[0] || BASE_CONFIG.locale;
-    const fingerprint = buildFingerprintOptions(primaryLocale);
-    const os = getOperatingSystemsFromUserAgent(userAgent);
-    if (os.length > 0) fingerprint.operatingSystems = os;
-    return fingerprint;
+    return buildFingerprintOptions(BASE_CONFIG.locale);
   },
   /**
-   * 规范化请求头，确保语言与浏览器一致。
-   *
-   * @param {Record<string, string>} headers
-   * @returns {Record<string, string>}
+   * 规范化请求头
    */
   applyLocaleHeaders(headers, acceptLanguage = "") {
-    if (acceptLanguage) {
-      headers["accept-language"] = acceptLanguage;
-    } else if (!headers["accept-language"]) {
-      headers["accept-language"] = BASE_CONFIG.acceptLanguage;
+    if (!headers["accept-language"]) {
+      headers["accept-language"] = acceptLanguage || BASE_CONFIG.acceptLanguage;
     }
     return headers;
   }
@@ -1037,18 +905,6 @@ var Launch = {
       ignoreDefaultArgs: ["--enable-automation"]
     };
   },
-  /**
-   * 获取增强版启动选项（用于高风险反爬场景）
-   */
-  getAdvancedLaunchOptions(customArgs = []) {
-    return {
-      args: [
-        ...AntiCheat.getAdvancedLaunchArgs(),
-        ...customArgs
-      ],
-      ignoreDefaultArgs: ["--enable-automation"]
-    };
-  },
   /**
    * 推荐的 Fingerprint Generator 选项
    * 确保生成的是桌面端、较新的 Chrome，以匹配我们的脚本逻辑
@@ -1582,8 +1438,6 @@ var Interception = {
           try {
             const reqHeaders = await request.allHeaders();
             delete reqHeaders["host"];
-            const currentAcceptLanguage = reqHeaders["accept-language"] || "";
-            AntiCheat.applyLocaleHeaders(reqHeaders, currentAcceptLanguage);
             const resolvedAcceptLanguage = reqHeaders["accept-language"] || "";
             const userAgent = reqHeaders["user-agent"] || "";
             const method = request.method();
@@ -1597,8 +1451,8 @@ var Interception = {
               body: postData,
               responseType: "buffer",
               // 强制获取 Buffer
-              // 模拟浏览器 TLS 指纹
-              headerGeneratorOptions: AntiCheat.getTlsFingerprintOptions(userAgent, resolvedAcceptLanguage),
+              // 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,