爬虫实例：天猫商品评论爬虫

最近被种草SK-II，本着学工科的严谨态度，决定用数据说话 

爬取数据

参数解析

itemId是商品ID， sellerId 是卖家ID， currentPage是当前页码，目标url是https://rate.tmall.com/list_detail_rate.htm?itemId=15332134505&spuId=294841&sellerId=917264765&order=3&currentPage=1

正则解析

1.cnt字符串不要随便换行(否则可能报错：SyntaxError: EOL while scanning string literal)，

2.findall(正则规则，字符串) 方法能够以列表的形式返回能匹配的字符串

#coding=utf-8
import re

cnt = '"aliMallSeller":False,"anony":True,"appendComment":"","attributes":"","attributesMap":"","aucNumId":"","auctionPicUrl":"","auctionPrice":"","auctionSku":"化妆品净含量:75ml","auctionTitle":"","buyCount":0,"carServiceLocation":"","cmsSource":"天猫","displayRatePic":"","displayRateSum":0,"displayUserLink":"","displayUserNick":"t***凯","displayUserNumId":"","displayUserRateLink":"","dsr":0.0,"fromMall":True,"fromMemory":0,"gmtCreateTime":1504930533000,"goldUser":False,"id":322848226237,"pics":["//img.alicdn.com/bao/uploaded/i3/2699329812/TB2hr6keQ.HL1JjSZFlXXaiRFXa_!!0-rate.jpg"],"picsSmall":"","position":"920-11-18,20;","rateContent":"送了面膜 和晶莹水 skii就是A 不错","rateDate":"2017-09-09 12:15:33","reply":"一次偶然的机会，遇见了亲，一次偶然的机会，亲选择了SK-II，生命中有太多的选择，亲的每一次选择都是一种缘分。让SK-II与您形影不离，任岁月洗礼而秀美如初~每日清晨拉开窗帘迎来的不仅止破晓曙光，还有崭新的自己~【SK-II官方旗舰店Lily】","sellerId":917264765,"serviceRateContent":"","structuredRateList":[],"tamllSweetLevel":3,"tmallSweetPic":"tmall-grade-t3-18.png","tradeEndTime":1504847657000,"tradeId":"","useful":True,"userIdEncryption":"","userInfo":"","userVipLevel":0,"userVipPic":""'

nickname = []
regex = re.compile('"displayUserNick":"(.*?)"')
print regex
nk = re.findall(regex,cnt)
for i in nk:
    print i
nickname.extend(nk)
print nickname

ak = re.findall('"auctionSku":"(.*?)"',cnt)
for j in ak:
    print j
    
rc = re.findall('"rateContent":"(.*?)"',cnt)
for n in rc:
    print n
    
rd = re.findall('"rateDate":"(.*?)"',cnt)
for m in rd:
    print m

输出：

完整源码

参考：http://www.jianshu.com/p/632a3d3b15c2

#coding=utf-8
import requests
import re
import sys

reload(sys)
sys.setdefaultencoding('utf-8')

#urls = []
#for i in list(range(1,500)):
#    urls.append('https://rate.tmall.com/list_detail_rate.htm?itemId=15332134505&spuId=294841&sellerId=917264765&order=1&currentPage=%s'%i)

tmpt_url = 'https://rate.tmall.com/list_detail_rate.htm?itemId=15332134505&spuId=294841&sellerId=917264765&order=1&currentPage=%d'
urllist = [tmpt_url%i for i in range(1,100)]
#print urllist

nickname = []
auctionSku = []
ratecontent = []
ratedate = []
headers = ''

for url in urllist:
    content = requests.get(url).text
    nk = re.findall('"displayUserNick":"(.*?)"',content)  #findall(正则规则，字符串) 方法能够以列表的形式返回能匹配的字符串
    #print nk  
    nickname.extend(nk)
    auctionSku.extend(re.findall('"auctionSku":"(.*?)"',content))
    ratecontent.extend(re.findall('"rateContent":"(.*?)"',content))
    ratedate.extend(re.findall('"rateDate":"(.*?)"',content))
    print (nickname,ratedate)


for i in list(range(0,len(nickname))):
    text =','.join((nickname[i],ratedate[i],auctionSku[i],ratecontent[i]))+'
'
    with open(r"C:UsersHPDesktopcodesDATASK-II_TmallContent.csv",'a+') as file:
        file.write(text+' ')
        print("写入成功")

注：url每次遍历，正则匹配的数据都不止一个，所以使用extend追加而不是append

输出：

数据分析

 1.要不要买——评论分析

import pandas as pd
from pandas import Series,DataFrame
import jieba
from collections import Counter

df = pd.read_csv(r'C:/Users/HP/Desktop/codes/DATA/SK-II_TmallContent.csv',encoding='gbk')  #否则中文乱码
#print df.columns
df.columns = ['useName','date','type','content']
#print df[:10]


tlist = Series.as_matrix(df['content']).tolist()
text = [i for i in tlist if type(i)!= float] #if type(i)!= float一定得加不然报错
text = ' '.join(text)
#print text

wordlist_jieba = jieba.cut(text,cut_all=True)
stoplist = {}.fromkeys([u'的', u'了', u'是',u'有'])  #自定义中文停词表，注意得是unicode
print stoplist
wordlist_jieba = [i for i in wordlist_jieba if i not in stoplist]  #and len(i) > 1
#print u"[全模式]: ", "/ ".join(wordlist_jieba)

count = Counter(wordlist_jieba)   #统计出现次数,以字典的键值对形式存储，元素作为key，其计数作为value。
result = sorted(count.items(), key=lambda x: x[1], reverse=True)  #key=lambda x: x[1]在此表示用次数作为关键字

for word in result:
    print word[0], word[1]


from pyecharts import WordCloud

data = dict(result[:100])
wordcloud = WordCloud('高频词云',width = 800,height = 600)
wordcloud.add('ryana',data.keys(),data.values(),word_size_range = [30,300])
wordcloud

输出：

好用的频率占据榜首，只是不明白为什么要切分

2.买什么——类型分析

import pandas as pd
from pandas import Series,DataFrame


df = pd.read_csv(r'C:/Users/HP/Desktop/codes/DATA/SK-II_TmallContent.csv',encoding='gbk')  #否则中文乱码
#print df.columns
df.columns = ['useName','date','type','content']
print df[:5]

from pyecharts import Pie

pie = Pie('净含量购买分布')
v = df['type'].tolist()
print v[:5]

#n1 = v.count(u'u5316u5986u54c1u51c0u542bu91cf:230ml')
n1 = v.count(u'化妆品净含量:75ml')
n2 = v.count(u'化妆品净含量:160ml')
n3 = v.count(u'化妆品净含量:230ml')
n4 = v.count(u'化妆品净含量:330ml')
#print n1,n2,n3,n4  #800 87 808 124

N = [n1,n2,n3,n4]  
#print N  #[800,87,808,124]

attr = ['体验装','畅销经典','忠粉挚爱','屯货之选']
pie.add('ryana',attr,N,is_label_show = True)
pie

输出：