爬取个人知乎收藏夹内容生成网站
2017-02-24 本文已影响1670人
treelake
- 无聊的时候,习惯刷刷知乎,但是新的有价值的内容也不多,多的是不断涌入的营销号、推广和某些知乎live。于是乎,想着不如刷自己的收藏夹吧,很多优秀回答其实看了不久就忘了,静静地躺在收藏夹里,一直没被翻过,何况我收藏频率虽然不高,几年下来积累也不少,这样刷起来也能消磨不少时光了,还能美其名曰温故而知新了。虽然前端改版,但是知乎的收藏夹用起来感觉还是不那么方便。自己动手,丰衣足食。
效果
- 利用python爬虫爬取了自己的所有收藏夹,利用flask做后端api和vuejs做前端显示,前后端分离,实现效果如下
爬虫
- 一开始想着github上有许多开源的知乎爬虫,可以省去不少麻烦,结果找了一下,高赞的多已不再维护,知乎又改版了,新的项目有一点,但是功能不太完善,只有自己上手,毕竟需求很简单明确,就是收集自己的所有收藏夹内容。(使用python3)
- 针对此次需求,爬虫的逻辑很简单。知乎在个人常用机上直接post用户名和密码无需验证码就可以登录,利用request.Session保存请求状态,按照url中?page=num的页码规则直接爬取所有收藏夹页面,解析出所有收藏夹的url,然后依次请求获取所有收藏夹下的问答列表,解析出相关信息。由于内容不多,为了方便,直接存为json文件。而且由于收藏夹内容不会很多,直接使用requests库单线程爬取即可。
- 以下为爬虫代码,生成两个json文件,一个是所有收藏夹及其下问答的相关信息
知乎收藏文章.json
,一个是所有问题的回答数据url_answer.json
,这样处理,在前端请求时可以先获取前者,在要阅读某个问题的回答时再异步请求后者,只获取对应的答案即可。 - 使用了requests_cache库,仅两行代码,使得请求在意外中断后要重新开始时,直接就从缓存数据库中提取已经请求过的页面,节省时间,省去了自己编码处理请求失败的麻烦。
import os
import json
from bs4 import BeautifulSoup
import requests
from requests.packages.urllib3.exceptions import InsecureRequestWarning
requests.packages.urllib3.disable_warnings(InsecureRequestWarning)
# 参考 http://stackoverflow.com/questions/27981545/suppress-insecurerequestwarning-unverified-https-request-is-being-made-in-pytho
import requests_cache
requests_cache.install_cache('demo_cache')
Cookie_FilePlace = r'.'
Default_Header = {'User-Agent': "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/42.0.2311.135 Safari/537.36",
'Host': "www.zhihu.com",
'Origin': "http://www.zhihu.com",
'Pragma': "no-cache",
'Referer': "http://www.zhihu.com/"}
Zhihu_URL = 'https://www.zhihu.com'
Login_URL = Zhihu_URL + '/login/email'
Profile_URL = 'https://www.zhihu.com/settings/profile'
Collection_URL = 'https://www.zhihu.com/collection/%d'
Cookie_Name = 'cookies.json'
os.chdir(Cookie_FilePlace)
r = requests.Session()
#--------------------Prepare--------------------------------#
r.headers.update(Default_Header)
if os.path.isfile(Cookie_Name):
with open(Cookie_Name, 'r') as f:
cookies = json.load(f)
r.cookies.update(cookies)
def login(r):
print('====== zhihu login =====')
email = input('email: ')
password = input("password: ")
print('====== logging.... =====')
data = {'email': email, 'password': password, 'remember_me': 'true'}
value = r.post(Login_URL, data=data).json()
print('====== result:', value['r'], '-', value['msg'])
if int(value['r']) == 0:
with open(Cookie_Name, 'w') as f:
json.dump(r.cookies.get_dict(), f)
def isLogin(r):
url = Profile_URL
value = r.get(url, allow_redirects=False, verify=False)
status_code = int(value.status_code)
if status_code == 301 or status_code == 302:
print("未登录")
return False
elif status_code == 200:
return True
else:
print(u"网络故障")
return False
if not isLogin(r):
login(r)
#---------------------------------------------------------------------#
url_answer_dict= {}
# 单独生成一个答案的url和答案文本之间的字典,便于后台提供api服务,与123行相关
#-----------------------get collections-------------------------------#
def getCollectionsList():
collections_list = []
content = r.get(Profile_URL).content
soup = BeautifulSoup(content, 'lxml')
own_collections_url = 'http://' + soup.select('#js-url-preview')[0].text + '/collections'
page_num = 0
while True:
page_num += 1
url = own_collections_url + '?page=%d'% page_num
content = r.get(url).content
soup = BeautifulSoup(content, 'lxml')
data = soup.select_one('#data').attrs['data-state']
collections_dict_raw = json.loads(data)['entities']['favlists'].values()
if not collections_dict_raw:
# if len(collections_dict_raw) == 0:
break
for i in collections_dict_raw:
# print(i['id'],' -- ', i['title'])
collections_list.append({
'title': i['title'],
'url': Collection_URL % i['id'],
})
print('====== prepare Collections Done =====')
return collections_list
#-------------------------
def getQaDictListFromOneCollection(collection_url = 'https://www.zhihu.com/collection/71534108'):
qa_dict_list = []
page_num = 0
while True:
page_num += 1
url = collection_url + '?page=%d'% page_num
content = r.get(url).content
soup = BeautifulSoup(content, 'lxml')
titles = soup.select('.zm-item-title a') # .text ; ['href']
if len(titles) == 0:
break
votes = soup.select('.js-vote-count') # .text
answer_urls = soup.select('.toggle-expand') # ['href']
answers = soup.select('textarea') # .text
authors = soup.select('.author-link-line .author-link') # .text ; ['href']
for title, vote, answer_url, answer, author \
in zip(titles, votes, answer_urls, answers, authors):
author_img = getAthorImage(author['href'])
qa_dict_list.append({
'title': title.text,
'question_url': title['href'],
'answer_vote': vote.text,
'answer_url': answer_url['href'],
#'answer': answer.text,
'author': author.text,
'author_url': author['href'],
'author_img': author_img,
})
url_answer_dict[
answer_url['href'][1:]
] = answer.text
# print(title.text, ' - ', author.text)
return qa_dict_list
def getAthorImage(author_url):
url = Zhihu_URL+author_url
content = r.get(url).content
soup = BeautifulSoup(content, 'lxml')
return soup.select_one('.AuthorInfo-avatar')['src']
def getAllQaDictList():
''' 最终结果要是列表和字典的嵌套形式,以便前端解析'''
all_qa_dict_list = []
collections_list = getCollectionsList()
for collection in collections_list:
all_qa_dict_list.append({
'ctitle': collection['title'],
'clist': getQaDictListFromOneCollection(collection['url'])
})
print('====== getQa from %s Done =====' % collection['title'])
return all_qa_dict_list
with open(u'知乎收藏文章.json', 'w', encoding='utf-8') as f:
json.dump(getAllQaDictList(), f)
with open(u'url_answer.json', 'w', encoding='utf-8') as f:
json.dump(url_answer_dict, f)
#---------------------utils------------------------------#
# with open('1.html', 'w', encoding='utf-8') as f:
# f.write(soup.prettify())
# import os
# Cookie_FilePlace = r'.'
# os.chdir(Cookie_FilePlace)
# import json
# dict_ = {}
# with open(u'知乎收藏文章.json', 'r', encoding='utf-8') as f:
# dict_ = json.load(f)
前端
- 前端要求不高,就是单页显示,要简洁漂亮,而且要便于我查找和翻看问题和答案。其次是对于我这种html和css战五渣,js列表遍历代码都要现谷歌的人来说,一定要简单好操作,我选择了vuejs前端框架(因为简单,也没有使用webpack)。
- 前端发展很快,框架和工具让人应接不暇,从我个人经验看,首先是不要害怕,框架和工具是为了帮助我们解决问题的,也就是说,使用框架和工具可以让我们更简单更快地开发,不少有效的框架和工具的学习成本并不高,掌握了基础部分,加上利用开源代码,可以方便地解决不少问题。此外,搜集好工具真是必备技能,大家面对的困难相似,说不定就有人开发了工具来解决你的痛点呢。
- 首先网站的基本构图采用bootstrap的一个基本模板,省了不少麻烦。vuejs的组件化特性使得我可以轻松地利用各种开源UI组件,像搭积木一样把他们拼接起来构成我的页面。在awesome-vue上我找到了符合我审美且简单易用的UI框架iView,虽然它暂时还只适用于vue1.x,不过由于我的应用简单,差异不大,就是它了。
- 以下为html代码,使用vue-resource异步请求数据,同步到页面。为了开发方便,直接采用了jsonp跨域请求的形式。代码质量仅供参考。组件里的template查看不方便,可以复制出来使用去掉两边单引号和对单引号的转义,利用美化html代码的工具查看。这样写是权宜之计。
<!DOCTYPE html>
<html lang="zh-CN">
<!--view-source:http://v3.bootcss.com/examples/jumbotron-narrow/#-->
<head>
<meta charset="utf-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>知乎个人收藏</title>
<link rel="stylesheet" href="https://cdn.bootcss.com/bootstrap/3.3.7/css/bootstrap.min.css">
<link rel="stylesheet" href="http://v3.bootcss.com/examples/jumbotron-narrow/jumbotron-narrow.css">
<link rel="stylesheet" type="text/css" href="http://unpkg.com/iview/dist/styles/iview.css">
</head>
<body>
<div id="app">
<div class="container">
<div class="header clearfix">
<h3 class="text-muted">知乎个人收藏</h3>
</div>
<div class="jumbotron">
<h1>栏目总览</h1>
<p class="lead">{{ description }}</p>
<my-carousel></my-carousel>
</div>
<div class="row marketing">
<div class="col-lg-6">
<my-card :collection="collection" v-for="collection in left"></my-card>
</div>
<div class="col-lg-6">
<my-card :collection="collection" v-for="collection in right"></my-card>
</div>
</div>
<i-button @click="showLeave" style: "" long>That's all!</i-button>
<Modal :visible.sync="visible" :title="modalTitle"> {{ modalMessage }}
<div v-html="rawHtml" id="inner-content"></div>
</Modal>
<footer class="footer">
<p>© 2017 treelake.</p>
</footer>
</div>
<!-- /container -->
</div>
<script type="text/javascript" src="http://v1.vuejs.org/js/vue.min.js"></script>
<script src="https://cdn.jsdelivr.net/vue.resource/1.2.0/vue-resource.min.js"></script>
<script type="text/javascript" src="http://unpkg.com/iview/dist/iview.min.js"></script>
<script>
Vue.component('my-carousel', {
template: '<div class="showimage"><Carousel arrow="never" autoplay><Carousel-item>![](https://n2-s.mafengwo.net/fl_progressive,q_mini/s10/M00/74/B6/wKgBZ1irpQ-Afw_uAAepw3nE8w884.jpeg)</Carousel-item><Carousel-item>![](https://c4-q.mafengwo.net/s10/M00/21/50/wKgBZ1imrvqAafuJAAeRHcfhBBg66.jpeg?imageMogr2%2Finterlace%2F1)</Carousel-item></Carousel></div>'
})
Vue.component('my-ul', {
template: '<ul id="list"><li v-for="item in items | limitBy limitNum limitFrom"><Badge :count="item.answer_vote" overflow-count="9999"> <a @click="simpleContent(item)" class="author-badge" :style="{ background: \'url(\'+ item.author_img +\') no-repeat\', backgroundSize:\'cover\'}"></a></Badge> <a :href=" \'https://www.zhihu.com\' + item.answer_url" target="_blank" style="font-size: 10px">     {{ item.title }}</a><a :href=" \'https://www.zhihu.com\' + item.question_url" target="_blank"><Icon type="chatbubbles"></Icon></a><hr> </li></ul>',
props: ['items'],
methods: {
changeLimit() {
if (this.limitFrom > this.items.length - this.limitNum) {
this.limitFrom = 0;
} else {
this.limitFrom += this.limitNum;
}
if (this.limitFrom == this.items.length) {
this.limitFrom = 0
}
console.log(this.limitFrom)
},
simpleContent(msg) {
this.$dispatch('child-msg', msg)
// 使用 $dispatch() 派发事件,事件沿着父链冒泡
},
},
data() {
return {
limitNum: 5,
limitFrom: 0,
}
},
events: {
'parent-msg': function () {
this.changeLimit()
}
},
})
Vue.component('my-card', {
template: '<Card style="width:auto; margin-bottom:15px" ><p slot="title"><Icon type="ios-pricetags"></Icon>{{ collection.ctitle }}</p><a v-if="collection.clist.length>5" slot="extra" @click="notify"><Icon type="ios-loop-strong"></Icon>换一换</a> <my-ul :items="collection.clist"></my-ul> </Card>',
props: ['collection'],
methods: {
notify: function () {
this.$broadcast('parent-msg')
// 使用 $broadcast() 广播事件,事件向下传导给所有的后代
}
}
})
var shuju, answer;
new Vue({
el: '#app',
data: {
description: '',
visible: false,
// ctitle: '',
allqa: [],
collection: {
'clist': [],
'ctitle': '',
},
left: [],
right: [],
modalMessage: '旧时光回忆完毕!',
modalTitle: 'Welcome!',
rawHtml: '<a href="https://treeinlake.github.io"> treelake </a>'
},
methods: {
show() {
this.visible = true;
},
showLeave() {
this.rawHtml = '';
this.modalMessage = '旧时光回忆完毕!';
this.show();
}
},
events: {
'child-msg': function (msg) {
this.$http.jsonp('/find' + msg.answer_url, {}, { // 单文件测试:http://localhost:5000/find
headers: {},
emulateJSON: true
}).then(function (response) {
// 这里是处理正确的回调
answer = response.data;
this.rawHtml = answer.answer;
}, function (response) {
// 这里是处理错误的回调
console.log(response);
});
this.modalMessage = '';
this.modalTitle = msg.title;
this.show();
}
},
ready: function () {
this.$http.jsonp('/collections', {}, { // 单文件测试 http://localhost:5000/collections/
headers: {},
emulateJSON: true
}).then(function (response) {
// 这里是处理正确的回调
shuju = response.data
for (i in shuju) {
this.description += (shuju[i].ctitle + ' ');
// console.log(shuju[i])
}
// this.ctitle = shuju[0].ctitle
// this.collection = shuju[0]
this.allqa = shuju
half = parseInt(shuju.length / 2) + 1
this.left = shuju.slice(0, half)
this.right = shuju.slice(half, shuju.length)
console.log(this.collection)
}, function (response) {
// 这里是处理错误的回调
console.log(response);
});
}
})
</script>
<style>
#list {
padding: 10px
}
#list li {
margin-bottom: 10px;
padding-bottom: 10px;
}
.jumbotron img {
width: 100%;
}
.author-badge {
width: 38px;
height: 38px;
border-radius: 6px;
display: inline-block;
}
#inner-content img {
width: 100%;
}
</style>
</body>
</html>
后端
- 后端主要提供api,使用了简洁易用的Flask,但是返回jsonp还需要一层封装,不过开源世界就是强大,直接找到了Flask-Jsonpify库,一句话搞定。主要逻辑就是先从本地加载之前爬下来的数据,然后提供api服务。
/find/<path:answer_url>
路由提供了根据回答的url查找回答文本内容的服务。 - 最后,想让flask在根目录提供html文件,直接访问ip就可以在手机上使用。为了不让flask本身的模板渲染和vuejs的模板特性冲突,直接返回了原本的html文件,避过了flask的模板渲染。
- 以下为服务端代码,连同上面两个文件放在一起,在爬取资料完毕后,
python xxx.py
运行服务即可。
# -*- coding: utf-8 -*-
from flask import Flask
import json
from flask_jsonpify import jsonpify
app = Flask(__name__)
collections = []
with open(u'知乎收藏文章.json', 'r', encoding='utf-8') as f:
collections = json.load(f)
qa_dict = {}
with open('url_answer.json', 'r', encoding='utf-8') as f:
qa_dict = json.load(f)
# print(qa_dict['question/31116099/answer/116025931'])
index_html = ''
with open('zhihuCollection.html', 'r', encoding='utf-8') as f:
index_html = f.read()
@app.route('/')
def index():
return index_html
@app.route('/collections')
def collectionsApi():
return jsonpify(collections)
@app.route('/find/<path:answer_url>') # 使用path修正斜杠的副作用,参见http://flask.pocoo.org/snippets/76/
def answersApi(answer_url):
# show the post with the given id, the id is an integer
return jsonpify({'answer': qa_dict[answer_url]})
@app.route('/test')
def test():
# show the post with the given id, the id is an integer
return jsonpify(qa_dict)
if __name__ == '__main__':
app.run(host='0.0.0.0')