瀏覽代碼

药房网修改

zhuoyuncheng 1 周之前
父節點
當前提交
c0272b25de
共有 2 個文件被更改,包括 5 次插入286 次删除
  1. 5 4
      spiders/yaofangwang/yaofangwang_crawl.py
  2. 0 282
      spiders/yaojiujiu/yjj_cookies.json

+ 5 - 4
spiders/yaofangwang/yaofangwang_crawl.py

@@ -310,13 +310,14 @@ class YaofangwangCrawl:
                 "product_brand": self.brand or "",
                 "collect_task_id": self.collect_task_id,
                 "search_name": self.product,
+                "company_name": shop,
                 "collect_config_info": json.dumps(
                 {"sampling_cycle": self.sampling_cycle, "sampling_start_time": self.sampling_start_time,
                  "sampling_end_time": self.sampling_end_time}),
                 "account_id": self.account_id,
                 "collect_region_id": self.collect_region_id,
                 "collect_round": self.collect_round,
-                "is_sold_out": 1
+                "is_sold_out": 0
 
             }
 
@@ -356,9 +357,9 @@ class YaofangwangCrawl:
         li_list = tree.xpath("//div[@id='wrap']//ul[contains(@class,'goodlist_search')]/li")
         id_list = []
         for li in li_list:
-            href_raw = self._x1(li, ".//a/@href")
-            title = self._x1(li, ".//a/text()")
-            print(title)
+            href_raw = self._x1(li, ".//a[@class='txt sc_medicine']/@href")
+            titles = li.xpath(".//a[@class='txt sc_medicine']/@title")
+            title = "".join(titles).replace("\n", "").strip()
             if self.product not in title:
                 break
             if not href_raw:

+ 0 - 282
spiders/yaojiujiu/yjj_cookies.json

@@ -1,282 +0,0 @@
-[
-  {
-    "name": "acw_tc",
-    "value": "1f2a9ed9-c7fc-45eb-961c-4f2432a2b281cad13421bc4c5d3660a38ca83d4cb3a7",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1778479846.390636,
-    "httpOnly": true,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "_abfpc",
-    "value": "c0dde0d7006a107ae3a6a7ed87ac5e6ef35f63aa_2.0",
-    "domain": ".yyjzt.com",
-    "path": "/",
-    "expires": 1813038047.618307,
-    "httpOnly": false,
-    "secure": true,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "cna",
-    "value": "d1c3fe028780e6e4efdd405ff96dc9c3",
-    "domain": "quickaplus-he-api-cn-shanghai.aliyuncs.com",
-    "path": "/",
-    "expires": 1813038070.185668,
-    "httpOnly": false,
-    "secure": true,
-    "sameSite": "None"
-  },
-  {
-    "name": "cna",
-    "value": "d1c3fe028780e6e4efdd405ff96dc9c3",
-    "domain": ".yyjzt.com",
-    "path": "/",
-    "expires": 1813038068.929424,
-    "httpOnly": false,
-    "secure": true,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "acw_tc",
-    "value": "0a15e18817784780675096155e499c8471a8efd8c11ebe237d198ae2cff60a",
-    "domain": "s.yyjzt.com",
-    "path": "/",
-    "expires": 1778479848.116366,
-    "httpOnly": true,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "jzt_browser_id",
-    "value": "d1c3fe028780e6e4efdd405ff96dc9c3",
-    "domain": ".yyjzt.com",
-    "path": "/",
-    "expires": 1813038048.861434,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "yjj-token",
-    "value": "ssr_2053712066124701783",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070068,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "userName",
-    "value": "18971731507",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070067,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "nickName",
-    "value": "%E6%A2%85%E6%9E%97%E5%85%AD%E8%B7%AF",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070067,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "userMobile",
-    "value": "18971731507",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070067,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "userBasicId",
-    "value": "1570949407765906055",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070068,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "loginName",
-    "value": "18971731507",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070067,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "companyId",
-    "value": "1570719328938967459",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070068,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "companyName",
-    "value": "%E6%B9%96%E5%8C%97%E5%92%8C%E5%90%AF%E5%85%B4%E9%A6%99%E9%A2%82%E5%A4%A7%E8%8D%AF%E6%88%BF%E8%BF%9E%E9%94%81%E6%9C%89%E9%99%90%E5%85%AC%E5%8F%B8",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070069,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "userId",
-    "value": "1570715138825075429",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070068,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "existBindCustomer",
-    "value": "1",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070068,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "areaCode",
-    "value": "42010500",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070068,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "creditCode",
-    "value": "91420104591067935X",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070069,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "isClear",
-    "value": "0",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070068,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "cartNum",
-    "value": "cartNum",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1778499668,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "category",
-    "value": "category",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1778499669,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "sloganAndBottom",
-    "value": "sloganAndBottom",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1778478128,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "companyFreeze",
-    "value": "0",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070069,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "compFreezeReason",
-    "value": "",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070069,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "firmList",
-    "value": "firmList",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1778478128,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "dowmnAppCodeUrl",
-    "value": "https://s.yyjzt.com/jzt-cms-web/20221013/1665666007570af1.png",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070069,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "WeChartQRCode",
-    "value": "https://s.yyjzt.com/jzt-cms-web/20231222/17032279632862dc.png",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070069,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  },
-  {
-    "name": "companyQRCode",
-    "value": "https://s.yyjzt.com/jzt-cms-web/20260420/1776653614972f3a0.png",
-    "domain": "www.yyjzt.com",
-    "path": "/",
-    "expires": 1781070069,
-    "httpOnly": false,
-    "secure": false,
-    "sameSite": "Lax"
-  }
-]