提交货源图片和相似ASIN优化

app/amazon/detail_spider.py

@@ -79,7 +79,8 @@ class ChromeAmzone(ChromeAmzoneBase):
         except Exception as e:
             error_msg = f"运行出错: {traceback.format_exc()}"
             self.log(error_msg)
-            show_notification(f"采集失败: {str(e)}", "error")
+            # show_notification(f"采集失败: {str(e)}", "error")
+            raise RuntimeError(error_msg)
             return {}
 
     def _scrape_data(self):
@@ -100,8 +101,27 @@ class ChromeAmzone(ChromeAmzoneBase):
             title_ele = self.tab.ele('xpath://h1[@id="title"]',timeout=30)
             title = title_ele.text
             data["title"] = title
+            sku_ele_ls = self.tab.eles('xpath://ul[@class="a-unordered-list a-vertical a-spacing-mini"]', timeout=20)
+            if len(sku_ele_ls) > 0:
+                data["title"] += sku_ele_ls[0].text
             imge_ele = self.tab.ele('xpath://div[@id="imgTagWrapperId"]//img',timeout=20)
-            image_url = imge_ele.attr("src")
+
+            image_url = ""
+            data_a_dynamic_image = imge_ele.attr("data-a-dynamic-image")
+            if data_a_dynamic_image:
+                dynamic_image_json = json.loads(data_a_dynamic_image)
+                self.log(f"图片信息:{dynamic_image_json}")
+                max_area = 0
+
+                for url, (width, height) in dynamic_image_json.items():
+                    area = width * height
+                    if area > max_area:
+                        max_area = area
+                        image_url = url
+
+            if not image_url:
+                image_url = imge_ele.attr("src")
+
             data["image_url"] = image_url
             return data
 
@@ -263,7 +283,7 @@ class SpiderTask(TaskBase):
                     result.append(res)
                     print("================")
                     is_done = gp_index == len(groups)-1
-                    if len(result) > 20 or is_done:
+                    if len(result) > 10 or is_done:
                         self.post_result(task_id=task_id,chunkIndex=gp_index+1,chunkTotal=len(groups),
                                          asin=asin,item_data=result,is_done=is_done)
                         result = []