一个简单的百度贴吧爬虫&&百度搜索爬虫&&模拟登录菜鸟踩坑记（requests、lxml）

这几天在学爬虫，试了下简单的，不涉及scrapy框架，库用的lxml、requests，python3，不涉及网页界面交互。

1、百度贴吧爬虫

爬取贴吧贴子标题、发贴人信息（性别、关注贴吧）等，后来根据数据做了个词云，代码跟词云戳下：

import os
import urllib
from lxml import etree
from bs4 import BeautifulSoup
import re
import json
import requests
import csv
import codecs
import time

def baidu_tieba(url, begin_page, end_page):
    for i in range(begin_page, end_page + 1):
        print('正在下载第', str(i), '个页面')
        time.sleep(1)
        if i > 1:
            suffix = '&&pn=' + str(50 * (i - 1))
        else:
            suffix = ''
        html = etree.HTML(requests.get(url + suffix).text)

        parse_html(html)

def parse_html(html):
    for e in html.xpath("//li[@class=' j_thread_list clearfix']"):
        # get info
        info = json.loads(e.get('data-field'))
        reply_num = info['reply_num']
        author = info['author_name']
        is_top = 1 if info['is_top'] else 0
        is_good = 1 if info['is_good'] else 0
        # get title & author & author info
        temp = e.xpath(
            ".//div[@class='col2_right j_threadlist_li_right ']//div[contains(@class,'threadlist_title pull_left j_th_tit ')]//a")[
            0]
        title = temp.get('title')
        content_url = 'http://tieba.baidu.com' + temp.get('href')

        content = etree.HTML(requests.get(content_url).text)
        print(content_url)
        try:
            content_text = content.xpath("//div[@class='d_post_content_main d_post_content_firstfloor']//cc/div")[
                1].text
            print(content_text)
        except:
            continue

        try:
            temp = e.xpath(".//span[@class='frs-author-name-wrap']//a")[0]
            author_url = 'http://tieba.baidu.com' + temp.get('href')
            print(author_url)
            # get content and user accurate_info
            author_content = requests.get(author_url).text
            author_page = etree.HTML(author_content)

            temp = author_page.xpath("//div[@class='userinfo_userdata']//span")
            gender = temp[0].get('class')
            if re.match('.*female', gender):
                gender = 1
            else:
                gender = 0
            temp = temp[1].xpath("./span/span")

            online_time = re.findall('吧龄:(.*)', temp[1].text)
            post_num = re.findall('发贴:(.*)', temp[3].text)

            temp = author_page.xpath(".//div[@class='ihome_forum_group ihome_section clearfix']")

            follow = []
            if temp:
                for t in temp[0].getchildren()[1].xpath('.//a'):
                    follow.append(t.xpath("./span")[0].text)
            else:
                pass

        except:
            gender = None
            online_time = None
            letter_num = None
            follow = []
            post_num = None
            author_url = None

        f_tieba.writerow([title, author, reply_num, is_top, is_good, content_url, content_text])
        f_author.writerow([author, gender, online_time, post_num, follow, author_url])



def createWordCloud(f_1, f_2):
    import pandas as pd
    import jieba.analyse
    from scipy.misc import imread
    import matplotlib as mpl
    import matplotlib.pyplot as plt
    from wordcloud import WordCloud, STOPWORDS, ImageColorGenerator

    title = pd.read_csv(f_1)
    author = pd.read_csv(f_2)

    author['online_time'] = author['online_time'].apply(lambda x: re.findall("\['(.*)'\]", str(x)))
    author['online_time'] = author['online_time'].apply(lambda x: None if len(x) < 1 else x[0])
    author['online_time'].fillna(0, inplace=True)

    author = author.groupby(['author', 'gender', 'online_time']).first().reset_index()
    title = title.groupby(['title', 'author', 'url']).first().reset_index()

    title['sents'] = title.apply(
        lambda x: x['content'] if len(x['content'].strip()) > len(x['title'].strip()) else x['title'], axis=1)
    sents = list(title['sents'])
    jieba.analyse.set_stop_words('../stopword/stopwords.txt')

    tags = []
    for s in sents:
        tags.extend(jieba.analyse.extract_tags(s, topK=100, withWeight=False))

    text = " ".join(tags)
    text = str(text)
    # read the mask
    wc = WordCloud(font_path='/Users/czw/Library/Fonts/PingFang.ttc',
                   background_color="white", max_words=100, width=1600, height=800,
                   max_font_size=500, random_state=42, min_font_size=40)
    print(text)
    # generate word cloud
    wc.generate(text)
    # wc.to_file('word.jpg')
    # generate 关键词 词云

    plt.imshow(wc)
    plt.axis("off")
    plt.show()

    print(title)
    print(author)

    temp = list(author['follow'].apply(lambda x: re.findall("\[(.*)\]", x)[0]))
    temp = [x.split(',') for x in temp if x]
    follow_word = []
    for f in temp:
        if len(f) > 0:
            f = [i.strip()[1:-1] + '吧' for i in f]
            follow_word.extend(f)

    wc = WordCloud(font_path='/Users/czw/Library/Fonts/PingFang.ttc',
                   background_color="white", max_words=100, width=200, height=1200,
                   max_font_size=400, random_state=42, min_font_size=40)

    wc.generate(str(" ".join(follow_word)))

    # 用户词云
    plt.axis('off')
    plt.imshow(wc)
    plt.savefig('word_2.jpg', dpi=150)
    plt.show()


if __name__ == '__main__':
    url = 'http://tieba.baidu.com/f?kw=%E5%8D%8E%E4%B8%AD%E7%A7%91%E6%8A%80%E5%A4%A7%E5%AD%A6&ie=utf-8'

    f_1 = open('title_try.csv', 'w', encoding='utf-8-sig')
    f_2 = open('author_try.csv', 'w', encoding='utf-8-sig')
    f_tieba = csv.writer(f_1, dialect='excel')
    f_author = csv.writer(f_2, dialect='excel')
    f_tieba.writerow(["title", "author", "reply_num", "is_top", "is_good", "url", "content"])
    f_author.writerow(["author","gender","online_time","post_num","follow","author_url"])

    # page 从1到20页的内容
    baidu_tieba(url, 1, 2)

    f_1.close()
    f_2.close()

    createWordCloud('title_try.csv','author_try.csv')

这里做的是hust贴吧的词云，如下：

爬的数据不太多，随便弄的，第一张是关键词，第二张是附近贴吧。just for fun  :-)

---

2、百度搜索爬虫

这里不涉及网页交互，只是用url的形式搜索，所以不用selenium或者其他通过url_open的方式搜索，想知道的同学请期待下一篇博客（嘻嘻）orz。

这个代码写的比之前的稍微好一点，因为要爬的内容不是很多，就没写队列或者连接db之类的。

class Downloader():
    def __init__(self, url):
        self.url = url
        s = requests.session()
        page = s.get(self.url)
        self.html = lxml.html.fromstring(page.text)
        self.contents = self.html.xpath("//div[contains(@class,'c-container')]")

    def __call__(self, number):
        content = self.contents[number]
        title = str(lxml.html.tostring(content.xpath(".//h3[contains(@class,'t')]")[0]),encoding = "utf-8")
        title = title.replace('','')
        title = title.replace('','')
        title = lxml.html.fromstring(title).xpath(".//a")[0].text
        content_url = content.xpath(".//h3[contains(@class,'t')]/a")[0].get('href')
        try:
            content = content.xpath(".//div[contains(@class,'c-abstract')]")[0]
            content = str(lxml.html.tostring(content))
            content = content.replace('','')
            content = content.replace('','')

            # ems = content.xpath(".//em")
            # for e in ems:
            #     text += e.text
            content = lxml.html.fromstring(content).xpath('.//div')[0].text
            # print(content.xpath('.//div')[0].text)
        except:
            content = None
        # content = content.replace('','')
        # content = content.replace('','')
        # print(content)
        # content = lxml.html.fromstring(content)
        # except:
        #     content = None
        return {'title':title, 'content':content, 'url':content_url}

import urllib.request as up

if __name__ == '__main__':
    keyword = input('the keyword you want search: \n')
    url = 'http://www.baidu.com/s?wd=' + keyword + '&rsv_bp=0&rsv_spt=3&rsv_n=2&inputT=6391'
    # print(url)
    # url = 'http://www.baidu.com/s?wd=ff&rsv_bp=0&rsv_spt=3&rsv_n=2&inputT=6391'
    # url = 'http://www.baidu.com'

    D = Downloader(url)
    import tkinter as tk
    import tkinter.messagebox

    text = ''

    for i in range(6):
        d = D(i)
        text += '==============第{}个搜索结果===========\n'.format(i+1)
        text += '标题为：{}\n\n'.format(d['title'])
        # text += 'url为：{}\n\n'.format(d['url'])
        print('==============第{}个搜索结果==========='.format(i+1))
        print('标题为：',d['title'])
        print('url为：',d['url'])
        if d['content']:
            text += '内容概要为：{}\n\n'.format(d['content'])
            print('内容概要为：',d['content'])
        else:
            text += '没有内容概要\n\n'
            print('没有内容概要')
        text += '\n\n'
        # print(d['content'])
    root = tk.Tk()
    root.title('GUI')
    root.geometry('800x600')

    tkinter.Label(root, text=text, justify ='left',wraplength=450).pack()
    root.mainloop()

结果大概就是

怎么说呢，没有内容概要是因为html里面的文字不太好提取，其实也可以提取出来，只是博主略懒，（有些坑，比如百度是http而不是https，还有其他的就忘记了，看代码吧）

---

3、模拟登录

这里只给一个小网站，是《用python写网络爬虫》这本书里的，不涉及验证码或者比较麻烦的加密登录，要是想知道的话也继续期待下一篇博文哦~

login_url = 'http://example.webscraping.com/places/default/edit/Algeria-4'

login_email = '[email protected]'
login_password = 'example'
# data = {'email':login_user, 'password':login_password}

def parse_form(html, content='//input'):
    tree = lxml.html.fromstring(html)
    data = {}
    for e in tree.xpath(content):
        if e.get('name'):
            data[e.get('name')] = e.get('value')
    return data

r_session = requests.Session()
page = r_session.get(login_url)
data = parse_form(page.content)
print(data)
login_url = page.url

data['email'] = login_email
data['password'] = login_password
# encoded_data = urllib.urlencode(data)
print(data)
resp = r_session.post(login_url, data=data)
print(resp.url)
# print(resp.text)
data = parse_form(resp.text)
print(data)
print(resp)

data['population'] = int(data['population'])+1
resp = r_session.post(resp.url, data=data)
html = lxml.html.fromstring(resp.text)
print(html.xpath("//table//tr[@id='places_population__row']/td[@class='w2p_fw']")[0].text)

坑就是，需要标注cookie，创建一个session对象即可~很简单，看代码就懂啦~

bye~