实战｜手把手教你用Python爬取存储数据，还能自动在Excel中可视化！

from bs4 import BeautifulSoup
import requests
import xlsxwriter
import os

def Teamlists(url):
    TeamName=[] 
    TeamURL=[] 
    GET=requests.get(URL1)
    soup=BeautifulSoup(GET.content,'lxml')
    lables=soup.select('html body div div div ul li span a') 
    for lable in lables:
        ballname=lable.get_text()
        TeamName.append(ballname)
        print(ballname)
    teamname=input("请输入想查询的球队名：")#此处可变为GUI界面中的按键值
    c=TeamName.index(teamname)
    for item in lables:
     HREF=item.get('href')
     TeamURL.append(HREF)
    URL2=TeamURL[c] 
    return URL2

#自定义函数获取队员列表和对应的URL
def playerlists(URL2):
    PlayerName=[] 
    PlayerURL=[] 
    GET2=requests.get(URL1)
    soup2=BeautifulSoup(GET2.content,'lxml')
    lables2=soup2.select('html body div div table tbody tr td b a')
    for lable2 in lables2:
        playername=lable2.get_text()
        PlayerName.append(playername)
        print(playername)
    name=input("请输入球员名：") #此处可变为GUI界面中的按键值
    d=PlayerName.index(name)
    for item2 in lables2:
     HREF2=item2.get('href')
     PlayerURL.append(HREF2)
    URL3=PlayerURL[d]
    return URL3,name

def Competition(URL3):
    data=[]
    GET3=requests.get(URL3)
    soup3=BeautifulSoup(GET3.content,'lxml')
    lables3=soup3.select('html body div div div div div div div div p')
    lables4=soup3.select('div div table tbody tr td')
    for lable3 in lables3:
     introduction=lable3.get_text() 
     print(introduction)  #球员基本信息
    for lable4 in lables4:
        competition=lable4.get_text()
        data.append(competition) 
    for i in range(len(data)):
        if data[i]=='职业生涯常规赛平均数据':
            a=data[i+31]
            a=data.index(a)
    del(data[:a]) 
    for x in range(len(data)):
        if data[x]=='职业生涯季后赛平均数据':
            b=data[x]
            b=data.index(b)
    del(data[b:])
    return data

def file_add(path):  #此时的内函数path可与GUI界面的Statictext绑定
    creatpath=path+'\\Basketball' 
    try:
     if not os.path.isdir(creatpath):
      os.makedirs(creatpath)       
    except:
     print("文件夹存在")
    return creatpath

def player_chart(name,data,creatpath):
    #此为表格名称——球员名称+chart
    EXCEL=xlsxwriter.Workbook(creatpath+'\\'+name+'chart.xlsx')
    worksheet=EXCEL.add_worksheet(name) 
    bold=EXCEL.add_format({'bold':1}) 
    headings=data[:18]
    worksheet.write_row('A1',headings,bold) #写入表头
    num=(len(data))//18
    a=0
    for i in range(num):
        a=a+18
        c=a+18
        i=i+1
        worksheet.write_row('A'+str(i+1),data[a:c]) #写入数据
    chart_col = EXCEL.add_chart({'type': 'line'}) #创建一个折线图
    chart_col.add_series({
        'name': '='+name+'!$R$1', #设置折线描述名称
        'categories':'='+name+'!$A$2:$A$'+str(num), #设置图表类别标签范围
        'values': '='+name+'!$R$2:$R$'+str(num-1),    #设置图表数据范围
        'line': {'color': 'red'}, })   #设置图表线条属性
    #设置图标的标题和想x，y轴信息
    chart_col.set_title({'name': name+'生涯常规赛平均得分'}) 
    chart_col.set_x_axis({'name': '年份 (年)'}) 
    chart_col.set_y_axis({'name': '平均得分(分)'})
    chart_col.set_style(1) #设置图表风格
    worksheet.insert_chart('A14', chart_col, {'x_offset':25, 'y_offset':3,}) #把图标插入工作台中并设置偏移
    EXCEL.close()