1

JD_Cookies/main.py

+# https://www.tapd.cn/36769433/prong/stories/view/1136769433001003855
\ No newline at end of file

JD_Cookies/putCookies.py

+import pymysql
+
+def returnDB(host,user,password,database):
+
+    Conn = pymysql.connect(
+        host=host,
+        user=user,
+        password=password,
+        database=database,
+        charset="utf8",
+        cursorclass=pymysql.cursors.DictCursor,
+        autocommit=True
+
+    )
+
+    Cur = Conn.cursor()
+
+    return Cur,Conn
+
+def updateCookie(Driver):
+
+    Cur, Conn = returnDB("59.110.219.171", "zgcindex", "zgcprice2019", "zdzs_proxy")
+
+    cookie = ''
+
+    for i in Driver.get_cookies():
+
+        cookie = cookie + i['name'] + '=' + i['value'] + ';'
+
+    sql = f"UPDATE jd_cookies SET cookie = '{cookie}' WHERE user = 'ruoyu'"
+
+    Cur.execute(sql)
+
+    Cur.close()
+
+    Conn.close()
\ No newline at end of file

putCookies.py

+import pymysql
+
+def returnDB(host,user,password,database):
+
+    Conn = pymysql.connect(
+        host=host,
+        user=user,
+        password=password,
+        database=database,
+        charset="utf8",
+        cursorclass=pymysql.cursors.DictCursor,
+        autocommit=True
+
+    )
+
+    Cur = Conn.cursor()
+
+    return Cur,Conn
+
+def updateCookie(Driver):
+
+    Cur, Conn = returnDB("59.110.219.171", "zgcindex", "zgcprice2019", "zdzs_proxy")
+
+    cookie = ''
+
+    for i in Driver.get_cookies():
+
+        cookie = cookie + i['name'] + '=' + i['value'] + ';'
+
+    sql = f"UPDATE jd_cookies SET cookie = '{cookie}' WHERE user = 'ruoyu'"
+
+    Cur.execute(sql)
+
+    Cur.close()
+
+    Conn.close()
\ No newline at end of file

京东搜索_自动版.py

@@ -4,6 +4,7 @@
 import time
 from urllib import request
 import cv2
+from putCookies import updateCookie
 import numpy as np
 import pyautogui
 from selenium import webdriver
@@ -115,12 +116,12 @@ class SlideUtils:
         pyautogui.moveTo(x + int(offset * random.randint(15, 23) / 20), y, duration=0.28)
         y += random.randint(-9, 0)
         pyautogui.moveTo(x + int(offset * random.randint(17, 21) / 20), y, duration=random.randint(20, 31) / 100)
-        y += random.randint(2, 5)
+        y += random.randint(-3, 3)
         pyautogui.moveTo(xx, y, duration=0.8)
         print("睡眠1秒后松开")
-        time.sleep(1)
+        time.sleep(0.5)
         import math
-        pyautogui.moveRel(0, random.randint(2, 5), duration=0.3)
+        pyautogui.moveRel(0, random.randint(-50, 50), duration=0.5)
         print("松开")
         pyautogui.mouseUp()
 
@@ -132,18 +133,18 @@ class SlideUtils:
         # print(f"睡眠2秒后点击移动")
         # time.sleep(2)
         xx = x + offset + offset_increase
-        pyautogui.moveTo(x, y, duration=0.3)
+        pyautogui.moveTo(x, y, duration=0.5)
         pyautogui.mouseDown()
         y += random.randint(9, 19)
-        pyautogui.moveTo(x + int(offset * random.randint(15, 23) / 20), y, duration=0.5)
+        pyautogui.moveTo(x + int(offset * random.randint(15, 23) / 20), y, duration=0.2)
         y += random.randint(-9, 0)
         pyautogui.moveTo(x + int(offset * random.randint(17, 21) / 20), y, duration=random.randint(20, 31) / 100)
-        y += random.randint(0, 2)
-        pyautogui.moveTo(xx, y, duration=0.5)
+        y += random.randint(-3, 3)
+        pyautogui.moveTo(xx, y, duration=0.3)
         print("睡眠1秒后松开")
-        time.sleep(1)
+        time.sleep(0.5)
         import math
-        pyautogui.moveRel(0, y+3, duration=0.3)
+        pyautogui.moveRel(0, random.randint(-300, -100), duration=0.5)
         print("松开")
         pyautogui.mouseUp()
         # print(f"睡眠2秒后点击
@@ -274,7 +275,7 @@ def find_best_match_angle(background_img_path, rotating_img_path):
         distortion_level = average_distance
         if angle == -360:
             best_score = distortion_level
-        print(f'中心区域扭曲程度估计（数值越小表示越接近未扭曲）: {distortion_level}')
+        # print(f'中心区域扭曲程度估计（数值越小表示越接近未扭曲）: {angle} {distortion_level}')
         # res = cv2.matchTemplate(background, rotated_captcha, cv2.TM_CCOEFF_NORMED)
         # _, max_val, _, max_loc = cv2.minMaxLoc(res)
         #
@@ -360,13 +361,20 @@ def get_cookie():
         print("元素的X位置：", x_position)
         print("元素的Y位置：", y_position)
         # 其中x为屏幕左上角至滑块中心的横向像素值，y为屏幕左上角至滑块中心纵向像素值, 可根据自己屏幕配置
+
         SlideUtils.slide_by_pyautogui(x_position, y_position, x, offset_increase)
         time.sleep(1)
     # query_spider_sql = f"SELECT id,url,mallId FROM DMP_SEARCH_SPIDER WHERE ID NOT IN (SELECT ID FROM DMP_SEARCH_DATA) and mallId = 'DS-JD' and project_id = '110' order by docID desc limit 10000"
     # query_spider_sql = f"SELECT id,url,mallId FROM DMP_SEARCH_SPIDER WHERE  data_batch ='202401251550270001' and ID NOT IN (SELECT ID FROM DMP_SEARCH_DATA) and mallId = 'DS-JD'  order by docID desc limit 10000"
-    query_spider_sql = f"SELECT id,url,mallId FROM DMP_SEARCH_SPIDER WHERE ID NOT IN (SELECT ID FROM DMP_SEARCH_DATA) and mallId = 'DS-JD' order by docID desc limit 10000"
+    # query_spider_sql = f"SELECT id,url,mallId FROM DMP_SEARCH_SPIDER WHERE ID NOT IN (SELECT ID FROM DMP_SEARCH_DATA) and mallId = 'DS-JD' and  data_batch ='202403051815460001' order by id" \
+    #                    f" limit 100000"
+    query_spider_sql = f"SELECT id,url,mallId FROM DMP_SEARCH_SPIDER WHERE ID NOT IN (SELECT ID FROM DMP_SEARCH_DATA) and mallId = 'DS-JD' and project_id = '100' limit 10000"
     query_spider = mysql_zdzs_proxy(query_spider_sql)
+    index = 0
+    len_ = len(query_spider)
     for row in query_spider:
+        index += 1
+        print(f"len={len_},index={index}")
         # while True:
         #     row = {
         #         'url':'https://search.jd.com/search?keyword=COMIX%20B3749&enc=utf-8',
@@ -489,7 +497,7 @@ def get_cookie():
                 # print("未查找到验证图片背景图，可能是旋转图片，换一种方式获取div")
                 try:
                     while_count_2 = 0
-                    while while_count_2 < 10:
+                    while while_count_2 < 30:
                         while_count_2 += 1
                         background = driver.find_element(by=By.XPATH, value='//*[@id="img-back-div"]')
                         background_url = background.get_attribute("style").replace('"); height: 181px;', "").replace(
@@ -536,7 +544,7 @@ def get_cookie():
                             break
                         except:
                             try:
-                                WebDriverWait(driver, 5).until(
+                                WebDriverWait(driver, 2).until(
                                     EC.presence_of_element_located((By.CLASS_NAME, "verifyBtn"))
                                 )
                             except NoSuchElementException:
@@ -595,13 +603,20 @@ def get_cookie():
         mysql_zdzs_proxy(insert_sql)
 
         print("保存成功")
+
+        updateCookie(driver)
+
         pass
     # 登录成功
     cookie = ''
+
     for i in driver.get_cookies():
+
         cookie = cookie + i['name'] + '=' + i['value'] + ';'
+
     driver.quit()
-    return cookie
+
+    return get_cookie()
 
 
 if __name__ == "__main__":

天猫价格.py

+#!/usr/bin/env python
+# -*- coding:utf-8 -*-
+# @Time  : 2022/9/6 0006 16:05
+import json
+import random
+import re
+from DrissionPage import WebPage,ChromiumPage, ChromiumOptions
+import time
+import pymysql
+from DrissionPage.common import ActionChains
+
+def returnDB(host,user,password,database):
+
+    Conn = pymysql.connect(
+        host=host,
+        user=user,
+        password=password,
+        database=database,
+        charset="utf8",
+        cursorclass=pymysql.cursors.DictCursor,
+        autocommit=True
+
+    )
+
+    Cur = Conn.cursor()
+
+    return Cur,Conn
+#返回IP
+def returnIP():
+
+    Cur,Conn = returnDB("59.110.219.171","zgcindex","zgcprice2019","zdzs_proxy")
+
+    Cur.execute('SELECT * FROM proxy')
+
+    IP_List = Cur.fetchall()
+
+    Cur.close()
+
+    Conn.close()
+
+    return IP_List
+#初始化浏览器对象
+def initDriver():
+
+
+    path = r'D:\ChromeMaYi\guge\chrome.exe'  # 请改为你电脑内Chrome可执行文件路径
+
+    co = ChromiumOptions()
+
+    #co.auto_port(True)
+    # 隐身无头模式
+    #co.set_argument('--headless')
+    # 无图片 静音
+    #co.set_no_imgs(True).set_mute(True)
+    # 是否禁用JS
+    #co.set_no_js(True)
+    # 链接 加载 超时时间
+    co.set_timeouts(implicit=5, pageLoad=5)
+
+    #co.set_argument('--user-agent',random.choice(MY_USER_AGENT))
+    # 启动参数代理
+    #IP = returnIP()[0]['ip_port']
+
+    #print(IP)
+
+    #co.set_proxy(f"http://{IP}")
+    # 以该配置创建页面对象
+    Driver = WebPage(driver_or_options=co)
+
+    return Driver
+
+
+def insertDBxpath(Driver,Cur,Key):
+
+    try:
+
+        XJ = 0
+
+        q = 0
+
+        SupName = ''
+
+        Name = ''
+
+        Price = ''
+
+        if '找不到了！' not in Driver.html and '下架' not in Driver.html:
+
+            Price = Driver.s_ele('tag:span@@class:Price--priceText').text
+
+            Name = Driver.s_ele('tag:h1@@class:ItemHeader--mainTitle').text
+
+            SupName = Driver.s_ele('tag:div@@class:ShopHeader--title').text if 'chaoshi' not in Driver.url else '天猫超市'
+
+            #img = Driver.s_ele('tag:img@@class:PicGallery--thumbnailPic').src
+
+        else:
+
+            Price = -1
+
+            XJ = 1
+
+        if SupName and '旗舰' in SupName:
+
+            q = 4
+
+        Sql = f"""UPDATE product_url SET status = 2 WHERE id = {Key['ID']}"""
+
+        print(Sql)
+
+        Cur.execute(Sql)
+
+        Sql = f"""INSERT INTO product_url_data(sku,goodsName,Price,qijian,Url,status,XJ,supName)VALUES('{Key['ID']}','{Name}','{Price}','{q}','{Key['Url']}',2,{XJ},'{SupName}')"""
+
+        print(Sql)
+
+        Cur.execute(Sql)
+
+    except Exception as e:
+
+        print(e)
+
+        if 'id' not in Driver.url or 'shop' in Driver.url:
+
+            Sql = f"""UPDATE product_url SET status = 2 WHERE id = {Key['ID']}"""
+
+            print(Sql)
+
+            Cur.execute(Sql)
+
+            Sql = f"""INSERT INTO product_url_data(sku,goodsName,Price,qijian,Url,status,XJ,supName)VALUES('{Key['sku']}','','','','{Key['Url']}',2,1,'{SupName}')"""
+
+            print(Sql)
+
+            Cur.execute(Sql)
+
+
+def Main():
+
+    Cur, Conn = returnDB("39.105.1.55", "root", "*#9U7P61R!657", "ZD_GT_pro")
+
+    Cur.execute("""SELECT ID,sku,sku_url as Url FROM `product_url` WHERE sku_url like '%tmall%' AND Status is null and (id like "%7" ) order by id desc limit 1000""")
+
+    Driver = initDriver()
+
+    Driver.get('https://s.taobao.com/search?page=1&q=%E6%97%A0%20CTL-350HK', timeout=5)
+
+    while True:
+
+        J = int(input("是否已经滑动验证并登录,完成请按:【1】\n"))
+
+        if J == 1:
+
+            break
+
+        else:
+
+            pass
+
+    searchList = Cur.fetchall()
+
+    searchNum = len(searchList)
+
+    for K in searchList:
+
+        try:
+
+            Url = f"""https://detail.tmall.com/item.htm?id={K['Url'].split('id=')[-1]}""" if 'm.tmal' in K['Url'] else K['Url']
+
+            if Driver.ele("xpath://span[@id='nc_1_n1z']"):
+
+                ac = ActionChains(Driver)
+
+                ac.move_to(Driver.ele("xpath://span[@id='nc_1_n1z']")).hold().right(300).wait(2).release()
+
+                Driver.get(Url,timeout=5)
+
+            else:
+
+                Driver.get(Url, timeout=5)
+
+                searchNum -= 1
+
+                print(K,searchNum)
+
+                # Driver.ele("xpath://input[@id='q']").input(parse.unquote(brandName))
+                #
+                #time.sleep(random.randint(2,5))
+                #
+                # Driver.ele("xpath://button[@id='button']").click()
+
+                #time.sleep(random.randint(0,3))
+
+            #if Driver.s_ele('tag:img@@class:Title--iconPic'):
+
+            time.sleep(2)
+
+            insertDBxpath(Driver,Cur,K)
+
+            # else:
+            #
+            #     insertDBre(Driver, Cur, K,brandName)
+
+        except:
+
+            pass
+
+    Driver.quit()
+
+Main()
\ No newline at end of file