优化评论

spider/task.py

@@ -101,7 +101,7 @@ def _edit_privacy(page):
                 continue_btn.click()
                 sleep(1, 2)
                 page.click('//div[@aria-label="Select audience"]//span[text()="Public"]')
-                page.click('//div[@aria-label="Save"]')
+                page.click('//div[@aria-label="Save privacy audience selection and close dialog"]')
                 sleep(1, 2)
                 return
             else:
@@ -111,7 +111,7 @@ def _edit_privacy(page):
         page.click('//div[contains(@aria-label, "Edit privacy")]')
         sleep(1, 2)
         page.click('//div[@aria-label="Select audience"]//span[text()="Public"]')
-        page.click('//div[@aria-label="Done"]')
+        page.click('//div[@aria-label="Done with privacy audience selection and close dialog"]')
         sleep(1, 2)
     except Error as e:
         logger.error(f"Error editing privacy settings: {e}")
@@ -225,6 +225,17 @@ def is_operation_failed(exception):
     return isinstance(exception, OperationFailed) and "更改语言异常" in str(exception)
 
 
+def is_page_crash_error(exception: Exception) -> bool:
+    error_message = str(exception).lower()
+    crash_patterns = (
+        "page crashed",
+        "target crashed",
+        "target page, context or browser has been closed",
+        "browser has been closed",
+    )
+    return any(pattern in error_message for pattern in crash_patterns)
+
+
 @retry(
     stop=stop_after_attempt(3),
     wait=wait_fixed(2),
@@ -259,7 +270,8 @@ def check_account_status(page, cookies):
 
     # 判断是否需要运行cookies页面
     if "flow=user_cookie_choice_v2&source=pft_user_cookie_choice" in page.url:
-        allow_cookies = page.query_selector('//div[@role="dialog"]/div/div/div/div/div[3]/div/div/div[1]/div[1]/div/div')
+        allow_cookies = page.query_selector(
+            '//div[@role="dialog"]/div/div/div/div/div[3]/div/div/div[1]/div[1]/div/div')
         if allow_cookies is None:
             raise OperationFailed("允许cookies设置点击失败")
         allow_cookies.click()
@@ -283,28 +295,58 @@ lock = RLock()
 login_semaphore = threading.Semaphore(2)
 
 
+def is_post_liked_in_dialog(page):
+    """判断弹窗内的帖子是否已点赞"""
+
+    # 检查是否存在已点赞按钮
+    remove_like_button = page.query_selector('[role="dialog"] div[aria-label="Remove Like"]')
+
+    if remove_like_button:
+        # 获取点赞数
+        like_count_elem = remove_like_button.query_selector('..').query_selector('span[dir="auto"]')
+        like_count = like_count_elem.inner_text() if like_count_elem else "未知"
+        print(f"帖子已点赞，点赞数：{like_count}")
+        return True
+    else:
+        # 检查是否存在未点赞按钮
+        like_button = page.query_selector('[role="dialog"] div[aria-label="Like"]')
+        if like_button:
+            like_count_elem = like_button.query_selector('..').query_selector('span[dir="auto"]')
+            like_count = like_count_elem.inner_text() if like_count_elem else "0"
+            print(f"帖子未点赞，点赞数：{like_count}")
+            return False, like_count
+        else:
+            print("未找到点赞按钮")
+            return False
+
 def playwright_like(cookies, target_url):
     path = os.path.join(BASE_PATH, 'chrome', '130-0008', 'chrome.exe')
     with lock:
         with sync_playwright() as playwright:
             update_windows_distinguish()
+            username = 'moremore_51WM1'
+            password = 'TOv5y0nXCZH_JH+5'
+            country = 'US'
 
             browser = playwright.chromium.launch(
-                headless=False, args=['--start-maximized'], executable_path=path
+                headless=False, args=['--start-maximized'], executable_path=path,
+                proxy={
+                    "server": "http://pr.oxylabs.io:7777",  # 必填
+                    "username": f"customer-{username}-cc-{country}",
+                    "password": password
+                }
             )
+
             context = browser.new_context(no_viewport=True)
             context.add_cookies(parse_cookies(cookies))
             page = context.new_page()
             check_account_status(page, parse_cookies(cookies))
 
-            url = 'https://facebook.com'
             try:
-
-                retry_goto(page, url)
                 time.sleep(random.randint(3, 10))
                 if 'permalink.php?story_fbid' in target_url or '/permalink/' in target_url or '/posts/' in target_url:
                     # 文字或图片类型
-                    button_xpath = '//*[@role="dialog"]//span[text()="Like" or @data-ad-rendering-role="like_button"]'
+                    button_xpath = '//*[@role="dialog"]//div[text()="Like" or @data-ad-rendering-role="like_button"]'
                 elif 'watch/?v' in target_url or '/videos/' in target_url:
                     # 视频类型, 视频类型,
                     button_xpath = '//span[@data-ad-rendering-role="like_button"][1]'
@@ -323,9 +365,9 @@ def playwright_like(cookies, target_url):
                     # 滚动到按钮所在处
                     button.scroll_into_view_if_needed()
                     # 判断按钮是否已经点过赞
-                    if button.get_attribute('style') == "" or button.get_attribute('style') is None:
+                    if is_post_liked_in_dialog(page):
                         # 未点过赞进行点赞操作
-                        button.click(force=True)
+                        button.evaluate("element => element.click()")
                 else:
                     raise OperationFailed("未找到点赞按钮")
                 time.sleep(10)
@@ -450,34 +492,37 @@ def retry_get_new_video(page, cookies, post_count):
     raise TimeoutError("未获取到新视频(可能视频上传失败)，已超时")
 
 
-@retry(
-    stop=stop_after_attempt(3),
-    wait=wait_fixed(2),
-    retry=retry_if_exception(is_operation_failed)
-)
 def playwright_post(cookies, content, image_key=None):
     path = os.path.join(BASE_PATH, 'chrome', '130-0008', 'chrome.exe')
     with lock:
         with sync_playwright() as playwright:
             update_windows_distinguish()
+            max_browser_retries = 3
+            last_error = None
+
+            for browser_attempt in range(max_browser_retries):
+                browser = None
+                context = None
+                file_path = None
+                try:
                     browser = playwright.chromium.launch(
                         headless=False, args=['--start-maximized'], executable_path=path
                     )
                     context = browser.new_context(no_viewport=True)
                     context.add_cookies(parse_cookies(cookies))
                     page = context.new_page()
-            page.evaluate(f'document.body.style.zoom = "{const.DISPLAY_SCALE}"')
 
                     check_account_status(page, parse_cookies(cookies))
                     # 声明默认发布视频数量
                     video_count = 0
 
-            url = 'https://facebook.com'
-            try:
+                    url = 'https://www.facebook.com'
                     # 先获取视频数量
                     if image_key is not None and ".mp4" in image_key:
                         video_count = get_post_count(page, cookies)
 
+                    # check_account_status 已经把页面带到 Facebook 首页，避免重复二次跳转导致页面进程崩溃
+                    if "facebook.com" not in page.url:
                         retry_goto(page, url)
                     time.sleep(random.randint(3, 10))
                     time.sleep(5)
@@ -509,19 +554,49 @@ def playwright_post(cookies, content, image_key=None):
                         post_url = retry_get_new_video(page, cookies, video_count)
                         time.sleep(random.randint(3, 10))
 
-            except Error as e:
-                raise OperationFailed(f'操作超时，请重试{e}')
-
                     screenshot_content = _full_screenshot()
-            if image_key:
-                os.remove(file_path)
-            context.close()
-            browser.close()
-
                     key = f'screenshot/{uuid.uuid4()}.png'
                     put_object(key, screenshot_content)
                     return {'response_url': post_url, 'screenshot_key': key}
 
+                except TimeoutError as e:
+                    last_error = e
+                    logger.warning(
+                        f"发布任务超时，尝试重建浏览器重试: attempt {browser_attempt + 1}/{max_browser_retries}, error={e}"
+                    )
+                except Error as e:
+                    last_error = e
+                    if is_page_crash_error(e):
+                        logger.warning(
+                            f"发布任务页面崩溃，尝试重建浏览器重试: attempt {browser_attempt + 1}/{max_browser_retries}, error={e}"
+                        )
+                    else:
+                        logger.warning(
+                            f"发布任务 Playwright 异常，尝试重试: attempt {browser_attempt + 1}/{max_browser_retries}, error={e}"
+                        )
+                finally:
+                    if image_key and file_path and os.path.exists(file_path):
+                        os.remove(file_path)
+                    if context is not None:
+                        try:
+                            context.close()
+                        except Exception:
+                            pass
+                    if browser is not None:
+                        try:
+                            browser.close()
+                        except Exception:
+                            pass
+
+                if browser_attempt < max_browser_retries - 1:
+                    time.sleep(2)
+
+            if isinstance(last_error, TimeoutError):
+                raise OperationFailed(f'操作超时，请重试: {last_error}')
+            if isinstance(last_error, Error) and is_page_crash_error(last_error):
+                raise OperationFailed(f'页面崩溃，请重试: {last_error}')
+            raise OperationFailed(f'操作失败，请重试: {last_error}')
+
 
 def playwright_comment(cookies, target_url, content, image_key=None):
     path = os.path.join(BASE_PATH, 'chrome', '130-0008', 'chrome.exe')
@@ -529,9 +604,19 @@ def playwright_comment(cookies, target_url, content, image_key=None):
         with sync_playwright() as playwright:
             update_windows_distinguish()
 
+            username = 'moremore_51WM1'
+            password = 'TOv5y0nXCZH_JH+5'
+            country = 'US'
+
             browser = playwright.chromium.launch(
-                headless=False, args=['--start-maximized'], executable_path=path
+                headless=False, args=['--start-maximized'], executable_path=path,
+                proxy={
+                    "server": "http://pr.oxylabs.io:7777",  # 必填
+                    "username": f"customer-{username}-cc-{country}",
+                    "password": password
+                }
             )
+
             context = browser.new_context(no_viewport=True)
             context.add_cookies(parse_cookies(cookies))
             page = context.new_page()
@@ -544,9 +629,9 @@ def playwright_comment(cookies, target_url, content, image_key=None):
 
                 if 'permalink.php?story_fbid' in target_url or '/posts/' in target_url or "/permalink/" in target_url:
                     # 文字或图片类型
-                    input_xpath = '//div[@role="dialog"]//span[text()="Comment" or@data-ad-rendering-role="comment_button"]'
+                    input_xpath = '//div[@role="dialog"]//span[text()="Comment" or @data-ad-rendering-role="comment_button"] | //div[@role="dialog"]//div[@aria-label="Write a comment…" and @role="textbox" and @contenteditable="true"]'
                     attach_xpath = '//div[@id="focused-state-actions-list"]//div[@aria-label="Attach a photo or video"]'
-                    comment_xpath = '//div[@aria-label="Comment"]'
+                    comment_xpath = '//div[@aria-label="Comment"] | //div[@id="focused-state-composer-submit"] | //div[@aria-label="Post comment" and @role="button"]'
                     page.click(input_xpath)
                     sleep(1, 2)
                 elif 'watch/?v' in target_url or '/videos/' in target_url:
@@ -600,8 +685,17 @@ def playwright_get_user_profile(cookies, username=None):
     with login_semaphore:
         with sync_playwright() as playwright:
             update_windows_distinguish()
+            username = 'moremore_51WM1'
+            password = 'TOv5y0nXCZH_JH+5'
+            country = 'US'
+
             browser = playwright.chromium.launch(
-                headless=const.HEADLESS, args=['--start-maximized'], executable_path=path
+                headless=const.HEADLESS, args=['--start-maximized'], executable_path=path,
+                proxy={
+                    "server": "http://pr.oxylabs.io:7777",  # 必填
+                    "username": f"customer-{username}-cc-{country}",
+                    "password": password
+                }
             )
             context = browser.new_context(no_viewport=True)
             context.add_cookies(parse_cookies(cookies))
@@ -758,7 +852,7 @@ def retry_goto(page: "Page", url: str, max_retries: int = 3, retry_delay: int =
 
             # Make the synchronous goto call within the try block
             # Use wait_until='load' as in your traceback, or adjust if needed
-            page.goto(url, timeout=30000, wait_until="domcontentloaded")  # Use the synchronous call
+            page.goto(url, timeout=180000)  # Use the synchronous call
 
             print(f"Successfully navigated to {url} on attempt {attempt + 1}")
             break  # Navigation was successful, exit the retry loop
@@ -772,11 +866,15 @@ def retry_goto(page: "Page", url: str, max_retries: int = 3, retry_delay: int =
                 print(f"All {max_retries} attempts failed for {url}.")
                 # If all retries fail, re-raise the exception
                 raise e
+        except Error as e:
+            if is_page_crash_error(e):
+                logger.error(f"Navigation page crashed for {url} on attempt {attempt + 1}: {e}")
+            else:
+                logger.error(f"Navigation playwright error for {url} on attempt {attempt + 1}: {e}")
+            raise e
         except Exception as e:
             # Catch any other unexpected errors during goto
             print(f"An unexpected error occurred during navigation to {url} on attempt {attempt + 1}: {e}")
-            # Decide if other exceptions should also trigger retries
-            # For now, we'll just re-raise other exceptions immediately
             raise e
 
 
@@ -992,14 +1090,21 @@ def playwright_m_login(username, password, code_2fa=None):
             max_proxy_retries = 3
             proxy_retry_count = 0
             while proxy_retry_count < max_proxy_retries:
-                proxy_url = ProxyChecker(timeout=8).get_valid_proxy_url()
-                if proxy_url is None:
-                    raise OperationFailed("获取代理失败")
-                logger.info(f"使用proxy={proxy_url}")
+                # proxy_url = ProxyChecker(timeout=8).get_valid_proxy_url()
+                # if proxy_url is None:
+                #     raise OperationFailed("获取代理失败")
+                # logger.info(f"使用proxy={proxy_url}")
+
+                username = 'moremore_51WM1'
+                password = 'TOv5y0nXCZH_JH+5'
+                country = 'US'
 
                 browser = playwright.chromium.launch(
-                    headless=const.HEADLESS, args=['--start-maximized'], executable_path=path, proxy={
-                        "server": proxy_url,
+                    headless=const.HEADLESS, args=['--start-maximized'], executable_path=path,
+                    proxy={
+                        "server": "http://pr.oxylabs.io:7777",  # 必填
+                        "username": f"customer-{username}-cc-{country}",
+                        "password": password
                     }
                 )
                 # random_user_agent = UserAgent().getBrowser(["Chrome Mobile iOS"]).get("useragent")
@@ -1117,9 +1222,19 @@ def playwright_share(cookies, target_url, content):
         with sync_playwright() as playwright:
             update_windows_distinguish()
 
+            username = 'moremore_51WM1'
+            password = 'TOv5y0nXCZH_JH+5'
+            country = 'US'
+
             browser = playwright.chromium.launch(
-                headless=False, args=['--start-maximized'], executable_path=path
+                headless=False, args=['--start-maximized'], executable_path=path,
+                proxy={
+                    "server": "http://pr.oxylabs.io:7777",  # 必填
+                    "username": f"customer-{username}-cc-{country}",
+                    "password": password
+                }
             )
+
             context = browser.new_context(no_viewport=True)
             context.add_cookies(parse_cookies(cookies))
             page = context.new_page()
@@ -1181,8 +1296,8 @@ def playwright_share(cookies, target_url, content):
 
 
 if __name__ == '__main__':
-    cookies = "c_user=61587708421547; xs=32:70ZigyqtoCfQ0w:2:1770569737:-1:-1; oo=v13:1770569731; datr=A8CIafB8DHVOhB7TfY9GWAIG"
-    print(playwright_check_account_cookies(cookies))
+    cookies = {"c_user":"61585626484995","datr":"NGBPadlTOQH5fwTPcUTGtLI5","dpr":"0.3400000035762787","fr":"0OHgdNICbgL23ZzCq..BptI5c..AAA.0.0.BptI5l.AWf9nkRp50sAbeHCdwhX--SsOVA","m_pixel_ratio":"0.3400000035762787","ps_l":"1","ps_n":"1","sb":"-Y-0adE6TkW7Fd070vzZxn5H","useragent":"TW96aWxsYS81LjAgKExpbnV4OyBBbmRyb2lkIDEwOyBGaW5kIFg1IFBybyBCdWlsZC9UUDFBLjIyMTAwNS4wMDI7IHd2KSBBcHBsZVdlYktpdC81MzcuMzYgKEtIVE1MLCBsaWtlIEdlY2tvKSBWZXJzaW9uLzQuMCBDaHJvbWUvMTI3LjAuMC4wIE1vYmlsZSBTYWZhcmkvNTM3LjM2IFtGQkFOL0VNQTtGQkxDL2VuX1VTO0ZCQVYvNDE5LjAuMC4xMC4xMjA7XQ%3D%3D","wd":"500x402","x-referer":"eyJyIjoiL3NlY3VyaXR5LzJmYWMvc2V0dXAvcXJjb2RlL2dlbmVyYXRlLz9leHQ9MTc3MzcwMDE2NiZoYXNoPUFlY2Vld0J0M0Q3UnhaMDgwTVBYMHR0UDFKUSIsImgiOiIvc2VjdXJpdHkvMmZhYy9zZXR1cC9xcmNvZGUvZ2VuZXJhdGUvP2V4dD0xNzczNzAwMTY2Jmhhc2g9QWVjZWV3QnQzRDdSeFowODBNUFgwdHRQMUpRIiwicyI6Im0ifQ%3D%3D","xs":"42:KzK0Y_fjLkJzUQ:2:1766809815:-1:-1"}
+    # print(playwright_check_account_cookies(cookies))
     # print(playwright_set_user_profile(cookies, "61584735094876", "Inaaya", "Inaaya",
     #                                   "facebook/user_upload/0196f098-851c-7810-b2aa-0833a0a7b09d/8c428558-2d90-4f45-baa3-a25f8a654b5c.png"))
     # cookies = '{"locale": "en_US", "datr": "ZnGnaBBx0yN7pov19-8_A6Gr", "sb": "ZnGnaDQicDSsVuevkudqio1J", "m_pixel_ratio": "1", "wd": "1920x1080", "test_cookie": "CheckForPermission", "c_user": "61579364283503", "xs": "34%3AdWeZoaWzFrtdVQ%3A2%3A1755804022%3A-1%3A-1", "oo": "v1%7C3%3A1755804031"}'
@@ -1194,7 +1309,8 @@ if __name__ == '__main__':
     # post(cookies, 'cs2025')
     # like(cookies, 'ZmVlZGJhY2s6MTIyMTA5NjE0NjU0NzkzNzc5')
     # comment(cookies, 'ZmVlZGJhY2s6MTIyMTA5NjE0NjU0NzkzNzc5', 'game la', 'xzpq.mp4')
-    # playwright_like(cookies, 'https://www.facebook.com/watch/?v=1007800324567828')
+    # playwright_like(cookies,
+    #                 'https://www.facebook.com/permalink.php?story_fbid=pfbid026dQik7NyujSGd2rjWyTKbQVw2CVM5BeJEP9sYxo7K5TBGhWZ4sUfs4gMoSLKq2mJl&id=61575372223339')
     # print(playwright_post(cookies, '1111', image_key="1.mp4"))
     # playwright_post(cookies, '2025-3-26~like', "")
     # playwright_comment(
@@ -1206,7 +1322,7 @@ if __name__ == '__main__':
     #     '2025-3-26~like',
     #     # 'rg.jpg'
     # )
-    # print(playwright_get_user_profile(cookies))
+    print(playwright_get_user_profile(cookies))
     # print(_change_language(cookies))
     # playwright_set_user_profile(
     #     cookies,