Python 爬取数据并存入SQL Server数据库

import requests
import re
import time
from SQLServers import MSSQL
import  pymssql
from pyquery import PyQuery as pq

#抓取页面Html

def GetHtml(url,page):

    time.sleep(3)
    print("当前页%d"%(int(page)+1))
    headers = {
        'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36'
    }
    fromContent = {
        '__EVENTTARGET': 'ctl00$ContentPlaceHolder1$GridViewData$ctl13$ddlJumpToPageNum',
        '__VIEWSTATE': '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',
        'ctl00$ContentPlaceHolder1$GridViewData$ctl13$ddlJumpToPageNum': page
    }
    try:
        response = requests.post(url=url, data=fromContent,headers=headers, timeout=10)
        response.encoding=response.apparent_encoding
        if response.status_code==200:
            print("获取源码成功")
    except Exception as e:
        print('获取源码失败:%s'%e)
    return response.text

#获取子页面HTML

def Get_ZIHtml(url):
    try:
        time.sleep(2)
        headers = {
            'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36'
        }
        response=requests.get(url=url,headers=headers)
        response.encoding = response.apparent_encoding
        print('获取子页面成功')
    except Exception as e:
        print('获取子页面失败%s'%e)
    return  response.text

#获取页面总页数

def Get_Pagenumber(url,page):
    try:
        html = GetHtml(url, page)
        doc = pq(html)
        pageItm = doc('#ctl00_ContentPlaceHolder1_GridViewData .td0 tr td')
        page = re.findall(r",共(.+?)页", pageItm.text())[0]
        print('获取页数成功')
    except Exception as e:
        print('获取页数失败%s'%e)
    return page

#获取Html遍历所有数据

def Get_Data(page,url):
    i = 0
    LXXXList = []   # 立项信息
    FZRList=[]      # 项目负责人及课题组成员
    JFYSList=[]     # 经费预算及执行
    JFLJList=[]     # 经费累计收支情况
    JFZCList=[]     # 经费支出
    KYCGList=[]     # 科研成果
    DESPList=[]     # 大额设备和材料名称和价格
    XMYSList=[]     # 结题项目验收
    SYList = []     # 首页数据
    while i < int(page):
        html = GetHtml(url, i)
        doc = pq(html)
        print(i)
        # 项目主表标题
        bodyItem = doc('#ctl00_ContentPlaceHolder1_GridViewData .hashover')
        for x in bodyItem('tr'):

            #子页面数据
            ZIurl = 'http://kyjf.zjedu.gov.cn/kyjfgk/%s' % pq(x).find('td a').attr('href')
            ZIhtml = Get_ZIHtml(ZIurl)
            ZIdoc = pq(ZIhtml,parser="html")

            #立项信息
            LXXX=ZIdoc('.tablelist1').children().children().children()
            Tup_LXXX = (
                        pq(x).find('td').eq(0).text(),#所属项目的项目编号
                        LXXX.eq(1).text(), #项目名称
                        LXXX.eq(3).text(), #立项部门
                        LXXX.eq(5).text(), #项目承担单位
                        LXXX.eq(7).text(), #项目协作单位
                        LXXX.eq(9).text(), #项目负责人
                        LXXX.eq(11).text(),#项目领域
                        LXXX.eq(13).text(),#立项时间
                        LXXX.eq(15).text(),#项目开始时间
                        LXXX.eq(17).text(),#项目结束时间
                        LXXX.eq(19).text(),#项目总额
                        LXXX.eq(21).text())#备注
            LXXXList.append(Tup_LXXX)

            #项目负责人以及课题组成员
            FZRTr = ZIdoc('#ctl00_ContentPlaceHolder1_GridViewData1 .hashover')
            for Ftr in FZRTr:
                Tup_FZR=(
                        pq(x).find('td').eq(0).text(),#所属项目的项目编号
                        pq(Ftr).children().eq(1).text(),#姓名
                        pq(Ftr).children().eq(2).text(),#职称
                        pq(Ftr).children().eq(3).text(),#工作单位
                        pq(Ftr).children().eq(4).text()#承担任务
                        )
                FZRList.append(Tup_FZR)

            #经费预算及执行
            JFYSTr = ZIdoc('#ctl00_ContentPlaceHolder1_GridViewData9 .hashover')
            for JFYSTh in JFYSTr:
                Tup_JFYS=(
                    pq(x).find('td').eq(0).text(),#所属项目的项目编号
                    pq(JFYSTh).children().eq(1).text(),#科目编号
                    pq(JFYSTh).children().eq(2).text(),#科目名称
                    pq(JFYSTh).children().eq(3).text(),#支出用途
                    pq(JFYSTh).children().eq(4).text(),#预算经费(元)
                    pq(JFYSTh).children().eq(5).text())#实际支出(元)
                JFYSList.append(Tup_JFYS)


            #经费累计收支情况
            JFLJTr = ZIdoc('#ctl00_ContentPlaceHolder1_GridViewData2 .hashover')
            for  JFLJTh in  JFLJTr:
                Tup_JFLJ=(
                    pq(x).find('td').eq(0).text(),#所属项目的项目编号
                    pq(JFLJTh).children().eq(1).text(),#项目经费名称
                    pq(JFLJTh).children().eq(2).text(),#经费类型
                    pq(JFLJTh).children().eq(3).text(),#经费来源
                    pq(JFLJTh).children().eq(4).text(),#已收入(元)
                    pq(JFLJTh).children().eq(5).text())#实际支出(元)
                JFLJList.append(Tup_JFLJ)



            #经费支出
            JFZCTitle = ZIdoc('ul[class="listtab"] li')
            y = 1
            for JFZC in JFZCTitle.items():
                JFZCTr = ZIdoc('#ctl00_ContentPlaceHolder1_UcJf%d_GridViewData .hashover' % y)
                y = y + 1
                for JFZCTh in JFZCTr:
                    Tup_JFZC=(
                        pq(x).find('td').eq(0).text(),# 所属项目的项目编号
                        JFZC.text(),#Table Title
                        pq(JFZCTh).children().eq(1).text(),#支出科目名称
                        pq(JFZCTh).children().eq(2).text()#实际支出(元)
                    )
                    JFZCList.append(Tup_JFZC)


            #科研成果
            KYCGTr = ZIdoc('#ctl00_ContentPlaceHolder1_GridViewData4 .hashover')
            for KYCGTh in KYCGTr:
                Tup_KYCG=(
                    pq(x).find('td').eq(0).text(),  # 所属项目的项目编号
                    pq(KYCGTh).children().eq(1).text(),  # 类型
                    pq(KYCGTh).children().eq(2).text(),  # 编号
                    pq(KYCGTh).children().eq(3).text(),  # 名称
                    pq(KYCGTh).children().eq(4).text()  # 日期
                )
                KYCGList.append(Tup_KYCG)

            #大额设备和材料名称和价格
            DESPTr = ZIdoc('#ctl00_ContentPlaceHolder1_GridViewData7 .hashover')
            for DESPTh in DESPTr:
                Tup_DESP=(
                    pq(x).find('td').eq(0).text(),
                    pq(DESPTh).children().eq(1).text(),
                    pq(DESPTh).children().eq(2).text(),
                    pq(DESPTh).children().eq(3).text(),
                    pq(DESPTh).children().eq(4).text(),
                    pq(DESPTh).children().eq(5).text(),
                    pq(DESPTh).children().eq(6).text(),
                    pq(DESPTh).children().eq(7).text()
                )
                DESPList.append(Tup_DESP)

                # 结题项目验收
                XMYS = ZIdoc('#ctl00_ContentPlaceHolder1_div8 .tablelist1 td')
                if XMYS.html() is not None:
                    Tup_XMYS = (
                        pq(x).find('td').eq(0).text(),  # 所属项目的项目编号
                        XMYS.eq(1).text(),  # 验收组织单位
                        XMYS.eq(3).text(),  # 验收类型
                        XMYS.eq(5).text(),  # 验收日期
                        XMYS.eq(7).text(),  # 验收组成员
                        XMYS.eq(9).text(),  # 验收意见
                        XMYS.eq(11).text()  # 备注
                    )
                    XMYSList.append(Tup_XMYS)

            #首页数据
            Tup_SY=(
                i + 1,
                "http://kyjf.zjedu.gov.cn/kyjfgk/" + pq(x).find('td a').attr('href'),
                pq(x).find('td').eq(0).text(),
                pq(x).find('td').eq(1).text(),
                pq(x).find('td').eq(2).text(),
                pq(x).find('td').eq(3).text(),
                pq(x).find('td').eq(4).text(),
                pq(x).find('td').eq(5).text(),
                pq(x).find('td').eq(6).text(),
                pq(x).find('td').eq(7).text()
            )
            SYList.append(Tup_SY)


        print("---第%d页完成---" % (int(i)+1))
        i += 1
    with pymssql.connect(".", "sa", "sa", "test") as conn:
        with conn.cursor(as_dict=True) as cursor:
            if len(LXXXList):
                cursor.executemany("INSERT INTO Tb_Article_LXXX " \
                                   " (FID,ProjectName,ProjectUnit,ProjectDepartment,CollaborationUnit,ProjectLeader,ProjectArea,ProjectData" \
                                   ",ProjectStartData,ProjectEndData ,ProjectTotal,Remarks)" \
                                   " VALUES (%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)", LXXXList)
                conn.commit()
            if len(DESPList):
                cursor.executemany("INSERT INTO Tb_Article_DESP"\
                               "(FID,SubjectName,Specifications,Unit,Number,Price,Total,Remarks)"\
                               "  VALUES(%s,%s,%s,%s,%s,%s,%s,%s)",DESPList)
                conn.commit()
            if len(FZRList):
                cursor.executemany("INSERT INTO [test].[dbo].[Tb_Article_FZR]"
                               " ([FID]"
                               " ,[Name]"
                               " ,[ZhiCheng]"
                               " ,[Unit]"
                               " ,[Task])"
                               " VALUES(%s,%s,%s,%s,%s)", FZRList)
                conn.commit()
            if len(JFLJList):
                cursor.executemany("INSERT INTO [test].[dbo].[Tb_Article_JFLJ]"
                               "   ([FID]"
                               "  ,[Name]"
                               "  ,[Type]"
                               "  ,[Come]"
                               "  ,[Income]"
                               "  ,[Funds])"
                               "   VALUES(%s,%s,%s,%s,%s,%s)", JFLJList)
                conn.commit()
            if len(JFYSList):
                cursor.executemany("INSERT INTO [test].[dbo].[Tb_Article_JFYS]"
                               "  ([FID]"
                               "  ,[Code]"
                               "  ,[Name]"
                               "  ,[Purpose]"
                               "  ,[Funds]"
                               "  ,[ActualFunds])"
                               "   VALUES(%s,%s,%s,%s,%s,%s)", JFYSList)
                conn.commit()
            if len(JFZCList):
                 cursor.executemany("INSERT INTO [test].[dbo].[Tb_Article_JFZC]"
                               "  ([FID]"
                               "  ,[Title]"
                               "  ,[SubjectName]"
                               "  ,[Actual])"
                               "   VALUES(%s,%s,%s,%s)", JFZCList)
                 conn.commit()
            if len(KYCGList):
                cursor.executemany("INSERT INTO [test].[dbo].[Tb_Article_KYCG]"
                               "  ([FID]"
                               "  ,[Type]"
                               "  ,[Number]"
                               "  ,[Name]"
                               "  ,[Data])"
                               "   VALUES(%s,%s,%s,%s,%s)", KYCGList)
                conn.commit()
            if len(SYList):
                 cursor.executemany("INSERT INTO [test].[dbo].[Tb_Article_SY]"
                               " ([AP_Page]"
                               "  ,[AP_Url]"
                               "  ,[AP_projectCode]"
                               "  ,[AP_projectName]"
                               "  ,[AP_projectType]"
                               "  ,[AP_LXBM]"
                               "  ,[AP_projectNuit]"
                               "  ,[AP_Data]"
                               "  ,[AP_Amount]"
                               "  ,[AP_Uname])"
                               "  VALUES(%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)", SYList)
                 conn.commit()
            if len(XMYSList):
                 cursor.executemany("INSERT INTO [test].[dbo].[Tb_Article_XMYS] "
                               "  ([FID]"
                               "  ,[AcceptanceUnit]"
                               "  ,[AcceptanceType]"
                               "  ,[AcceptanceData]"
                               "  ,[AcceptanceMember]"
                               "  ,[AcceptanceOpinion]"
                               "  ,[Remarks])"
                               "  VALUES(%s,%s,%s,%s,%s,%s,%s)", XMYSList)
                 conn.commit()
    print('----数据获取成功!----')


if __name__ == '__main__':

    try:
        url = "http://kyjf.zjedu.gov.cn/kyjfgk/Item_Xm_query.aspx"
        page=Get_Pagenumber(url,1)
        print(page)
        Get_Data(page,url)
        print('获取完成')
    except Exception as e:
        print('数据获取失败%s'%e)

猜你喜欢

转载自www.cnblogs.com/zldqpm/p/10857195.html