no message

py_spider/amazon_every_day_spider/dow_new_selection_category.py

@@ -54,21 +54,14 @@ class dow_category_Product():
                 print("强制关闭chrome.exe失败：", e)
             port = 9222
             params_ = ""
-            # params_ = "--blink-settings=imagesEnabled=false"
             os.system(f'start Chrome {params_} --remote-debugging-port={port}')
             chrome_options = Options()
             # 禁止加载图片
             chrome_options.add_argument('--blink-settings=imagesEnabled=false')
             chrome_options.add_experimental_option("debuggerAddress", f"127.0.0.1:{port}")
-            driver = webdriver.Chrome(r'chromedriver.exe', options=chrome_options)
-            # 无界面模式
-            # chrome_options.add_argument('--headless')
-            # 禁用 GPU，加快在 headless 下的渲染
             chrome_options.add_argument('--disable-gpu')
-            # 禁用沙箱，规避权限问题
-            chrome_options.add_argument('--no‑sandbox')
-            # 改用 /tmp 而不是 /dev/shm（避免共享内存不足）
             chrome_options.add_argument('--disable-dev-shm-usage')
+            driver = webdriver.Chrome(r'chromedriver.exe', options=chrome_options)
             self.get_category(site, driver)
 
     def get_category(self, site, driver):
@@ -138,21 +131,6 @@ class dow_category_Product():
                         if Product_name[0] in self.click_product_name_list:
                             print(product_nums, "已经抓取::", Product_name[0].upper())
                             continue
-
-                        driver.execute_script("localStorage.clear();")  # 清除本地存储
-                        time.sleep(0.5)
-                        driver.execute_script(
-                            "caches.keys().then(function(names) { for (let name of names) { caches.delete(name); } });")
-                        driver.execute_script("window.performance.clearResourceTimings();")
-                        time.sleep(0.5)
-                        # 假设你已经有了 driver
-                        # 先 enable heap profiler
-                        driver.execute_cdp_cmd('HeapProfiler.enable', {})
-                        # 然后强制 GC
-                        driver.execute_cdp_cmd('HeapProfiler.collectGarbage', {})
-                        # 最后可选地 disable 掉
-                        driver.execute_cdp_cmd('HeapProfiler.disable', {})
-                        time.sleep(0.5)
                         self.click_product_name_list.append(Product_name[0])
                         self.update_cagetory_state = False
                         driver.execute_script(f"document.querySelector('#{Product_name[0]} > kat-radiobutton').click()")
@@ -395,36 +373,6 @@ class dow_category_Product():
                         while True:
                             try:
                                 if save_Category_list:
-                                    # with self.engine_mysql.begin() as conn_mysql:
-                                    #     for i in save_Category_list:
-                                    #         dele_sql = f"DELETE from {site}_aba_profit_category_insights where category='{i[0]}' and product_type='{i[1]}' and item_type_keyword='{i[2]}' and year_week='{self.y_w}'"
-                                    #         print('删除删除mysql：', dele_sql)
-                                    #         conn_mysql.execute(dele_sql)
-                                    # df = pd.DataFrame(data=save_Category_list,
-                                    #                   columns=['category', "product_type", "item_type_keyword",
-                                    #                            "search_ratio", "product_average", "return_ratio",
-                                    #                            "return_product_average", "year_week", 'sellers',
-                                    #                            'new_brands',
-                                    #                            'asin', 'new_asin', 'per_asin', 'advertisement_spend',
-                                    #                            'star_ratings', 'new_brands_int', 'asin_int',
-                                    #                            'new_asin_int', 'per_asin_int', 'five_star',
-                                    #                            'three_star', 'two_star', 'one_star', 'ad_spend',
-                                    #                            'majority_spend', 'most_popular_keywords_item',
-                                    #                            'reasons_returns_json', 'top_data_json',
-                                    #                            'news_data_json',
-                                    #                            'top_sales_amount', 'top_sales_volume',
-                                    #                            'top_search_ratio',
-                                    #                            'top_return_ratio', 'top_adv_spend',
-                                    #                            'top_majority_spend',
-                                    #                            'news_sales_amount',
-                                    #                            'news_sales_volume',
-                                    #                            'news_search_ratio', 'news_return_ratio',
-                                    #                            'news_adv_spend',
-                                    #                            'news_majority_spend'
-                                    #                            ])
-                                    # self.engine_mysql.to_sql(df, f'{site}_aba_profit_category_insights',
-                                    #                          if_exists="append")
-                                    # print('存储成功 mysql')
                                     with self.engine_pg.begin() as conn_pg:
                                         for i in save_Category_list:
                                             dele_sql = f"DELETE from {site}_aba_profit_category_insights where category='{i[0]}' and product_type='{i[1]}' and item_type_keyword='{i[2]}' and year_week='{self.y_w}'"