refactor(ai_seo): 重构 DeepSeek 爬虫

- 更新了页面元素定位方式，以适应网站结构变化
- 优化了深度思考和回答获取的逻辑
- 调整了搜索结果处理的方式
- 更新了数据返回结构的解析

spiders/ai_seo/deepseek.py

@@ -37,15 +37,16 @@ class DeepseekSpider(AbstractAiSeoSpider):
         search_btn = self.browser_page.locator("span:text('联网搜索')").locator('..')
         if await search_btn.is_visible():
             await search_btn.click()
+        self.think = True
         if self.think:
         # 开启深度思考
-            think_btn = self.browser_page.locator("span:text('深度思考 (R1)')").locator('..')
+            think_btn = self.browser_page.locator("span:text('深度思考')").locator('..')
             if await think_btn.is_visible():
-                styles = css_to_dict(await think_btn.get_attribute('style'))
-                if styles.get('--ds-button-color') == '#fff':
+                # styles = css_to_dict(await think_btn.get_attribute('style'))
+                # if styles.get('--ds-button-color') == '#fff':
                 await think_btn.click()
                 await asyncio.sleep(1)
-        chat_input_element = self.browser_page.locator("//textarea[@id='chat-input']")
+        chat_input_element = self.browser_page.locator("//textarea[@placeholder='给 DeepSeek 发送消息 ']")
         await chat_input_element.click()
         # 输入提问词
         await self.browser_page.keyboard.type(self.prompt)
@@ -72,7 +73,8 @@ class DeepseekSpider(AbstractAiSeoSpider):
                 await think_element.nth(-1).click()
                 await asyncio.sleep(2)
         # 获取回答元素
-        answer = self.browser_page.locator("//div[@class='ds-markdown ds-markdown--block']").nth(-1)
+        # answer = self.browser_page.locator("//div[@class='ds-markdown ds-markdown--block']").nth(-1)
+        answer = self.browser_page.locator("//div[contains(@class, 'ds-message')]").nth(-1)
         box = await answer.bounding_box()
         # 设置视口大小
         await self.browser_page.set_viewport_size({
@@ -130,9 +132,11 @@ class DeepseekSpider(AbstractAiSeoSpider):
         body = stream.decode('utf-8')
         datas = body.split("\n\n")
         for data_str in datas:
+            # 返回数据为空 跳过
             if not data_str:
                 continue
             data_str = data_str.replace('data: ', '')
+            # 服务器繁忙 跳过
             try:
                 data = json.loads(data_str)
                 if glom(data, 'v.0.v', default='') == 'TIMEOUT':
@@ -145,27 +149,27 @@ class DeepseekSpider(AbstractAiSeoSpider):
                 logger.debug(f"获取到联网搜索结果")
                 search_result_list = data.get('v', [])
                 search_result_lists.extend(search_result_list)
-                # # 保存搜索结果
-                # ai_search_result_list = []
-                # for search_result in search_result_list:
-                #     url = search_result.get('url', '')
-                #     title = search_result.get('title', '')
-                #     body = search_result.get('snippet', '')
-                #     publish_time = search_result.get('published_at', '')
-                #     host_name = search_result.get('site_name', '未知')
-                #     ai_result = AiSearchResult(url=url, title=title, body=body, publish_time=publish_time, host_name=host_name)
-                #     if ai_result.title and ai_result.url:
-                #         ai_search_result_list.append(ai_result)
-                #     logger.debug(f"ai参考资料: [{host_name}]{title}({url})")
-                # if ai_search_result_list:
-                #     self.ai_answer.search_result = ai_search_result_list
-                #     self.search_result_count = len(self.ai_answer.search_result)
+                # 保存搜索结果
+                ai_search_result_list = []
+                for search_result in search_result_list:
+                    url = search_result.get('url', '')
+                    title = search_result.get('title', '')
+                    body = search_result.get('snippet', '')
+                    publish_time = search_result.get('published_at', '')
+                    host_name = search_result.get('site_name', '未知')
+                    ai_result = AiSearchResult(url=url, title=title, body=body, publish_time=publish_time, host_name=host_name)
+                    if ai_result.title and ai_result.url:
+                        ai_search_result_list.append(ai_result)
+                    logger.debug(f"ai参考资料: [{host_name}]{title}({url})")
+                if ai_search_result_list:
+                    self.ai_answer.search_result = ai_search_result_list
+                    self.search_result_count = len(self.ai_answer.search_result)
                 continue
 
             # 是否开始返回深度思考数据
-            if data.get('p', '') == 'response/thinking_content':
+            if data.get('p', '') == 'response/fragments/1/content':
                 start_thinking = True
-            if data.get('p', '') == 'response/thinking_elapsed_secs':
+            if data.get('p', '') == 'response/fragments/1/elapsed_secs':
                 start_thinking = False
             if start_thinking:
                 # 获取深度思考回复
@@ -177,7 +181,7 @@ class DeepseekSpider(AbstractAiSeoSpider):
                     value = glom(data, target, default="")
                 thinking_text = thinking_text + str(value)
             # 是否开始返回回复数据
-            if data.get('p', '') == 'response/content':
+            if data.get('p', '') == 'response/fragments/2/content':
                 start_content = True
             if start_content:
                 # 获取ai回复