5.5

Python爬取高德大数据各项数据(拥堵区域信息,拥堵路段信息,所有重点城市信息等)

import datetime
import json
import threading
import traceback
import time
from shlex import join

from selenium import webdriver
from selenium.webdriver import ChromeOptions
from selenium.webdriver.chrome.options import Options
from selenium.webdriver import Chrome
import pymysql
import requests
from concurrent.futures import ThreadPoolExecutor
from lxml import html

import re
option = ChromeOptions()
option.add_experimental_option("excludeSwitches", ["enable-automation"])
option.add_argument("--headless")
option.add_argument("--disable-gpu")
option.add_argument('window-size=1920x3000')  # 指定浏览器分辨率
option.add_argument('--disable-gpu')  # 谷歌文档提到需要加上这个属性来规避bug
option.add_argument('--hide-scrollbars')  # 隐藏滚动条, 应对一些特殊页面
option.add_argument('blink-settings=imagesEnabled=false')  # 不加载图片, 提升速度
option.add_argument('--headless')  # 浏览器不提供可视化页面. linux下如果系统不支持可视化不加这条会启动失败
option.binary_location = r"C:UsersLenovoAppDataLocalGoogleChromeApplicationchrome.exe"  # 手动指定使用的浏览器位置


web = Chrome(options=option)  # 把参数配置设置到浏览器中

etree = html.etree
web.get("https://trp.autonavi.com/diagnosis/rank.do")

ul=1
js=web.page_source
obj1 = re.compile(r'<html><head></head><body><pre style="word-wrap: break-word; white-space: pre-wrap;">(?P<ul>.*?)</pre></body></html>', re.S)
result1 = obj1.finditer(js)
for it in result1:
    ul = it.group('ul')
j=json.loads(ul)
#print(j)
code=[]
cityName=[]
jiankangzhishu=[]
yongduzhishu=[]
speed=[]
for i in range(0,101):
    code.append(j[i]["adcode"])
    cityName.append(j[i]['cityName'])
    jiankangzhishu.append(j[i]["healthValue"])
    yongduzhishu.append(j[i]['idx1'])
    speed.append(j[i]['realSpeed'])



def get_conn():
    conn = pymysql.connect(
        host="localhost",
        user="root",
        password="123456",
        db="jtsk",
        charset="utf8",
        port=3306,
    )
    cursor = conn.cursor()
    return conn, cursor


def close_conn(conn, cursor):
    if cursor:
        cursor.close()
    if conn:
        conn.close()

def cityquyu():
    cursor = None
    conn = None
    try:
        tm = time.strftime("%Y-%m-%d %H:%M", time.localtime())
        conn, cursor = get_conn()
        conn1, cursor1 = get_conn()
        sql = 'delete from cityquyu'
        cursor1.execute(sql)
        conn1.commit()
        close_conn(conn1, cursor1)
        for j in range(0,101):
            url=f'https://trp.autonavi.com/ajax/districtRank.do?linksType=1&cityCode={code[j]}'
            resp=requests.get(url)
            js=resp.json()
            for i in js:
                cityname=cityName[j]
                name=i['name']
                index=float(i['index'])

                speed=float(i['speed'])
                sql = "insert into cityquyu(city,name,zhishu,speed,time ) values(%s,%s,%s,%s,%s)"
                sql1="insert into cityquyu_all(city,name,zhishu,speed,time ) values(%s,%s,%s,%s,%s)"
                cursor.execute(sql, [cityname,name,index,speed,tm])
                conn1, cursor1 = get_conn()
                cursor1.execute(sql1 ,[cityname,name,index,speed,tm])
                conn1.commit()
                conn.commit()  # 提交事务 update delete insert操作
                close_conn(conn1, cursor1)
    except:
            traceback.print_exc()
    finally:
        close_conn(conn, cursor)
def roadyuce():
    cursor = None
    conn = None
    try:
        tm =(datetime.datetime.now()+datetime.timedelta(days=1)).strftime('%Y-%m-%d')#明天
        conn, cursor = get_conn()
        conn1, cursor1 = get_conn()
        sql = 'delete from roadyuce'
        cursor1.execute(sql)
        conn1.commit()
        close_conn(conn1, cursor1)
        for j in range(0, 101):
            url = f'https://trp.autonavi.com/ajax/getCityRoadTop.do?adcode={code[j]}&date={tm}'
            resp = requests.get(url)
            js = resp.json()
            for i in range(0, len(js) - 1):
                sql = """
                        insert into roadyuce(name,roadname,zhishu,speed,dir,zuobiao) values(%s,%s,%s,%s,%s,%s)
                        """
                cursor.execute(sql, [cityName[j], js[i]["cityName"], js[i]["idx"], js[i]["speed"],
                                         js[i]["dir"] ,str(js[i]["lnglats"])])
                conn.commit()  # 提交事务 update delete insert操作
    except:
        traceback.print_exc()
    finally:
        close_conn(conn, cursor)
def qxcity1():
    cursor = None
    conn = None
    try:
        conn, cursor = get_conn()
        conn1, cursor1 = get_conn()
        sql = 'delete from qxcity1'
        cursor1.execute(sql)
        conn1.commit()
        close_conn(conn1, cursor1)
        url='https://trp.autonavi.com/cityTravel/inAndOutCity.do?adcode=100000&dt=2021-05-11&willReal=WILL&inOut=IN&size=50'
        url1='https://trp.autonavi.com/cityTravel/inAndOutCity.do?adcode=100000&dt=2021-05-11&willReal=WILL&size=50&inOut=OUT'
        web.get(url)
        ul = 1
        js = web.page_source
        obj1 = re.compile(
            r'<html><head></head><body><pre style="word-wrap: break-word; white-space: pre-wrap;">(?P<ul>.*?)</pre></body></html>',
            re.S)
        result1 = obj1.finditer(js)
        for it in result1:
            ul = it.group('ul')
        j = json.loads(ul)
        for i in range(0,len(j)):
            sql = "insert into qxcity1(code,name,zhishu) values(%s,%s,%s)"
            cursor.execute(sql,
                           [j[i]["adcode"], j[i]["name"],j[i]["willIdx"]])
            conn.commit()  # 提交事务 update delete insert操作
    except:
        traceback.print_exc()
    finally:
        close_conn(conn, cursor)
def qxcity2():
    cursor = None
    conn = None
    try:
        conn, cursor = get_conn()
        conn1, cursor1 = get_conn()
        sql = 'delete from qxcity2'
        cursor1.execute(sql)
        conn1.commit()
        close_conn(conn1, cursor1)

        url='https://trp.autonavi.com/cityTravel/inAndOutCity.do?adcode=100000&dt=2021-05-11&willReal=WILL&size=50&inOut=OUT'
        web.get(url)
        ul = 1
        js = web.page_source
        obj1 = re.compile(
            r'<html><head></head><body><pre style="word-wrap: break-word; white-space: pre-wrap;">(?P<ul>.*?)</pre></body></html>',
            re.S)
        result1 = obj1.finditer(js)
        for it in result1:
            ul = it.group('ul')
        j = json.loads(ul)
        for i in range(0,len(j)):
            sql = "insert into qxcity2(code,name,zhishu) values(%s,%s,%s)"
            cursor.execute(sql,
                           [j[i]["adcode"], j[i]["name"],j[i]["willIdx"]])
            conn.commit()  # 提交事务 update delete insert操作
    except:
        traceback.print_exc()
    finally:
        close_conn(conn, cursor)
def cityroad():
    cursor = None
    conn = None
    try:
        tm = time.strftime("%Y-%m-%d %H:%M", time.localtime())
        conn, cursor = get_conn()
        conn1, cursor1 = get_conn()
        sql = 'delete from cityroad'
        cursor1.execute(sql)
        conn1.commit()
        close_conn(conn1, cursor1)
        for j in range(0,101):
            url=f'https://trp.autonavi.com/ajax/roadRank.do?roadType=0&timeType=0&cityCode={code[j]}'
            resp=requests.get(url)
            js=resp.json()['tableData']
            for i in range(0,len(js)-1):
                # print(js[i]['name'],js[i]["dir"],js[i]["index"],js[i]["speed"],js[i]["length"])
                # print(type(js[i]['name']))
                # print(type(js[i]["dir"]))
                # print(type(js[i]["index"]))
                # print(type(js[i]["speed"]))
                # print(type(js[i]["length"]))
                x=str(js[i]["coords"])
                sql="""
                insert into cityroad(city,name,dir,zhishu,speed,length,time,coords) values(%s,%s,%s,%s,%s,%s,%s,%s)
                """
                cursor.execute(sql, [cityName[j],js[i]["name"],js[i]["dir"],js[i]["index"],js[i]["speed"],js[i]["length"],tm,x])
                conn.commit()  # 提交事务 update delete insert操作
                conn1, cursor1 = get_conn()
                sql1 ="""
                insert into cityroad_all(city,name,dir,zhishu,speed,length,time,coords) values(%s,%s,%s,%s,%s,%s,%s,%s)
                """
                cursor1.execute(sql1, [cityName[j],js[i]["name"],js[i]["dir"],js[i]["index"],js[i]["speed"],js[i]["length"],tm,x])
                conn1.commit()
                close_conn(conn1, cursor1)
    except:
            traceback.print_exc()
    finally:
        close_conn(conn, cursor)

def qrcity():
    cursor = None
    conn = None
    try:
        conn, cursor = get_conn()
        conn1, cursor1 = get_conn()
        sql = 'delete from qrcity'
        cursor1.execute(sql)
        conn1.commit()
        close_conn(conn1, cursor1)

        url =  'https://trp.autonavi.com/cityTravel/inAndOutCity.do?adcode=100000&dt=2021-05-11&willReal=WILL&inOut=IN&size=50'
        web.get(url)
        ul = 1
        js = web.page_source
        obj1 = re.compile(
            r'<html><head></head><body><pre style="word-wrap: break-word; white-space: pre-wrap;">(?P<ul>.*?)</pre></body></html>',
            re.S)
        result1 = obj1.finditer(js)
        for it in result1:
            ul = it.group('ul')
        j = json.loads(ul)
        for i in range(0, len(j)):
           url1=f'https://trp.autonavi.com/cityTravel/inAndOutCity.do?adcode={j[i]["adcode"]}&dt=2021-05-11&willReal=WILL&size=20&inOut=IN'
           web.get(url1)
           ul1 = 1
           js1 = web.page_source
           obj11 = re.compile(
               r'<html><head></head><body><pre style="word-wrap: break-word; white-space: pre-wrap;">(?P<ul>.*?)</pre></body></html>',
               re.S)
           result11 = obj11.finditer(js1)
           for it in result11:
               ul1 = it.group('ul')
           j1 = json.loads(ul1)
           for k in range(0,20):
               sql = "insert into qrcity(city,name) values(%s,%s)"
               cursor.execute(sql, [j[i]["name"],j1[k]["name"]])
               conn.commit()  # 提交事务 update delete insert操作



    except:
        traceback.print_exc()
    finally:
        close_conn(conn, cursor)
def qccity():
    cursor = None
    conn = None
    try:
        conn, cursor = get_conn()
        conn1, cursor1 = get_conn()
        sql = 'delete from qccity'
        cursor1.execute(sql)
        conn1.commit()
        close_conn(conn1, cursor1)

        url = 'https://trp.autonavi.com/cityTravel/inAndOutCity.do?adcode=100000&dt=2021-05-11&willReal=WILL&size=50&inOut=OUT'

        web.get(url)
        ul = 1
        js = web.page_source
        obj1 = re.compile(
            r'<html><head></head><body><pre style="word-wrap: break-word; white-space: pre-wrap;">(?P<ul>.*?)</pre></body></html>',
            re.S)
        result1 = obj1.finditer(js)
        for it in result1:
            ul = it.group('ul')
        j = json.loads(ul)
        for i in range(0, len(j)):
           url1=f'https://trp.autonavi.com/cityTravel/inAndOutCity.do?adcode={j[i]["adcode"]}&dt=2021-05-11&willReal=WILL&inOut=OUT&size=20'

           web.get(url1)
           ul1 = 1
           js1 = web.page_source
           obj11 = re.compile(
               r'<html><head></head><body><pre style="word-wrap: break-word; white-space: pre-wrap;">(?P<ul>.*?)</pre></body></html>',
               re.S)
           result11 = obj11.finditer(js1)
           for it in result11:
               ul1 = it.group('ul')
           j1 = json.loads(ul1)
           for k in range(0,20):
               sql = "insert into qccity(city,name) values(%s,%s)"
               cursor.execute(sql, [j[i]["name"],j1[k]["name"]])
               conn.commit()  # 提交事务 update delete insert操作



    except:
        traceback.print_exc()
    finally:
        close_conn(conn, cursor)
def update_city():
    cursor = None
    conn = None
    try:
        tm=time.strftime("%Y-%m-%d %H:%M", time.localtime())
        conn, cursor = get_conn()
        conn1,cursor1=get_conn()
        sql='delete from city'
        cursor1.execute(sql)
        conn1.commit()
        close_conn(conn1, cursor1)
        for i in range(0,101):
            sql = "insert into city(code,cityName,jiankangzhishu,yongduzhishu,speed,time) values(%s,%s,%s,%s,%s,%s)"
            cursor.execute(sql,[code[i],cityName[i],jiankangzhishu[i],yongduzhishu[i],speed[i],tm])
            conn.commit()  # 提交事务 update delete insert操作
    except:
        traceback.print_exc()
    finally:
        close_conn(conn, cursor)


def allrw():
    print('===================================================================================')

    thr = threading.Timer(300, allrw)  # 每5分钟
    thr.start()

    now1 = datetime.datetime.now()
    print(f'{now1} ----- 开始执行')
    update_city()
    cityquyu()
    cityroad()
    roadyuce()
    now2 = datetime.datetime.now()
    print(f'{now2} ----- 执行结束                       用时{now2 - now1}')
    print('===================================================================================


')
if __name__ == '__main__':
   allrw()


   print("ok")
原文地址:https://www.cnblogs.com/zhaoyids/p/14905353.html