‘大案牍术’告诉你为什么《长安十二时辰》这么火！

一、功能描述

想用我们现代的大案牍术来分析一下《长安十二时称》这部电视剧到底为什么会火，大家都对这部电视剧的评价是什么样的？（所有弹幕最高频的900个词）

二、技术方案

1、分析优酷弹幕的加载方式然后使用requests库爬取

2、抓取整部剧所有集的所有弹幕

3、重点做数据清洗，比如：剧情、人名、高能君弹幕等等

4、将弹幕做成词云

三、技术实现

1、分析并获取弹幕接口的url

①打开优酷网站，点击电视剧播放，在页面鼠标右键选择检查（或F12）,调出浏览器的调试窗口。

②复制任意一条弹幕，然后点击调试窗口按Control+F进行搜索！

 ③点击该请求的Headers按钮，查看请求url，并且注意请求头中的Referer和User-Agent参数。

 通过上面的步骤，我们就可以轻松加愉快的获取到弹幕加载的url

https://service.danmu.youku.com/list?jsoncallback=jQuery111208275429479734884_1569416025133&mat=0&mcount=1&ct=1001&iid=1061156738&aid=322943&cid=97&lid=0&ouid=0&_=1569416025149

2、爬取弹幕数据

URL找到之后我们便可以开始coding了，还是老规矩：先从一条数据的抓取、提取、保存，这些都没问题之后我们再研究批量抓取。

import requests

def spider_danmu():
    '''爬取优酷指定页的弹幕'''
    url = 'https://service.danmu.youku.com/list?jsoncallback=jQuery111208275429479734884_1569416025133&mat=0&mcount=1&ct=1001&iid=1061156738&aid=322943&cid=97&lid=0&ouid=0&_=1569416025149'
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36'
    }
    try:
        r = requests.get(url, headers = kv)
        r.raise_for_status()
        print(r.text)
    except Exception as e:
        print('爬取失败')


if __name__ == '__main__':
    spider_danmu()

3、提取数据

①提取json数据

我们观察返回的数据会发现，跨域请求用的是jsonp，所以我们需要对返回的数据进行稍微的截取，就是将jQuery111208275429479734884_1569416025133(和最后的)去掉，只保留中间的json数据。

import json

import requests

def spider_danmu():
    '''爬取优酷指定页的弹幕'''
    url = 'https://service.danmu.youku.com/list?jsoncallback=jQuery111208275429479734884_1569416025133&mat=0&mcount=1&ct=1001&iid=1061156738&aid=322943&cid=97&lid=0&ouid=0&_=1569416025149'
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36'
    }
    try:
        r = requests.get(url, headers = kv)
        r.raise_for_status()
        # 找到jsonp数据的左括号位置并加1
        json_start_index = r.text.index('(') + 1
        # 截取json数据字符串
        r_json_str = r.text[json_start_index:-2]
        # 字符串转换为json对象
        r_json_obj = json.loads(r_json_str)
        print(r_json_obj)
    except Exception as e:
        print('爬取失败')


if __name__ == '__main__':
    spider_danmu()

②提取弹幕数据

得到json之后，我们就来分析弹幕数据在哪里，我们可以在浏览器的调试窗口的Preview里面查看

可以看到result字段里面便是弹幕数据，而且它的数据格式是一个列表，列表中是每个弹幕对象，弹幕对象中的content字段就是实际的弹幕内容，好那我们用json把他们提取并打印出来。

import json

import requests

def spider_danmu():
    '''爬取优酷指定页的弹幕'''
    url = 'https://service.danmu.youku.com/list?jsoncallback=jQuery111208275429479734884_1569416025133&mat=0&mcount=1&ct=1001&iid=1061156738&aid=322943&cid=97&lid=0&ouid=0&_=1569416025149'
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36'
    }
    try:
        r = requests.get(url, headers = kv)
        r.raise_for_status()
        # 找到jsonp数据的左括号位置并加1
        json_start_index = r.text.index('(') + 1
        # 截取json数据字符串
        r_json_str = r.text[json_start_index:-2]
        # 字符串转换为json对象
        r_json_obj = json.loads(r_json_str)
        r_json_result = r_json_obj['result']
        for r_json_danmu in r_json_result:
            print(r_json_danmu['content'])
    except Exception as e:
        print('爬取失败')

4、存储数据

想要的数据提取出来之后，我们就可以把数据保存。数据保存我们还是使用文件来保存，原因是操作方便，满足需求。

import json

import requests

def spider_danmu():
    '''爬取优酷指定页的弹幕'''
    url = 'https://service.danmu.youku.com/list?jsoncallback=jQuery111208275429479734884_1569416025133&mat=0&mcount=1&ct=1001&iid=1061156738&aid=322943&cid=97&lid=0&ouid=0&_=1569416025149'
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36'
    }
    try:
        r = requests.get(url, headers = kv)
        r.raise_for_status()
        # 找到jsonp数据的左括号位置并加1
        json_start_index = r.text.index('(') + 1
        # 截取json数据字符串
        r_json_str = r.text[json_start_index:-2]
        # 字符串转换为json对象
        r_json_obj = json.loads(r_json_str)
        r_json_result = r_json_obj['result']
        for r_json_danmu in r_json_result:
            with open('danmu.txt','a+', encoding='utf-8') as f:
                f.write(r_json_danmu['content']+'
')
            print(r_json_danmu)
    except Exception as e:
        print(e)


if __name__ == '__main__':
    spider_danmu()

5、批量爬取

完成一次请求请求的爬取、提取、保存之后，我们来研究下如何批量保存数据。这里和其他批量爬取有所区别：如何爬取多集的批量数据？

这里我们把批量爬取分为两步：第一步批量爬取一集的所有弹幕；第二步爬取多集的弹幕！ 

第一步：爬取某一集所有弹幕
批量爬取的关键就在于找到分页参数，找分页的技巧就是：比较两个请求url的参数，看看有何不同。

https://service.danmu.youku.com/list?jsoncallback=jQuery111203042330950003598_1569417439433&mat=0&mcount=1&ct=1001&iid=1061156738&aid=322943&cid=97&lid=0&ouid=0&_=1569417439449



https://service.danmu.youku.com/list?jsoncallback=jQuery111203042330950003598_1569417439433&mat=1&mcount=1&ct=1001&iid=1061156738&aid=322943&cid=97&lid=0&ouid=0&_=1569417439457

我们比较同一集的第一次请求与第二次请求的url发现mat参数不同，而且还是依次递增的趋势，这个参数便是我们寻找的分页参数（其实mat参数表示分钟数，表示获取第几分钟的弹幕），找到分页参数后我们就可以对原方法改造，改造思路：

将原url中分页参数变为可变参数，由方法传入。然后新建一个批量爬取的方法，循环调用单次爬取方法，每次调用传入页数即可

import json
import os
import random
import time

import requests

def spider_danmu(i):
    '''爬取优酷指定页的弹幕'''
    url = 'https://service.danmu.youku.com/list?jsoncallback=jQuery111208275429479734884_1569416025133&mat=%s&mcount=1&ct=1001&iid=1061156738&aid=322943&cid=97&lid=0&ouid=0&_=1569416025149'%i
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36'
    }
    try:
        r = requests.get(url, headers = kv)
        r.raise_for_status()
        # 找到jsonp数据的左括号位置并加1
        json_start_index = r.text.index('(') + 1
        # 截取json数据字符串
        r_json_str = r.text[json_start_index:-2]
        # 字符串转换为json对象
        r_json_obj = json.loads(r_json_str)
        # 如果请求总数count = 0 则说明弹幕爬取完成
        if not r_json_obj['count']:
            return 0
        r_json_result = r_json_obj['result']
        for r_json_danmu in r_json_result:
            with open('danmu.txt','a+', encoding='utf-8') as f:
                f.write(r_json_danmu['content']+'
')

        return 1
    except Exception as e:
        print(e)

def batch_spider_commnet():
    if os.path.exists('danmu.txt'):
        os.remove('danmu.txt')

    i = 0
    while spider_danmu(i):  # 判断返回值，如果爬完了，则返回0,
        time.sleep(random.random()*5)
        i += 1


if __name__ == '__main__':
    batch_spider_commnet()

第二步：爬取多集的所有弹幕
这一步的关键在于找到代表集数的参数，我们可以同样可以使用对比的方法：比较第一集与第二集的第一个弹幕请求URL，从而找到不同参数！

我们发现第一集的iid=1061156738、第二集的iid=1061112026，但是这个iid参数并不是递增，如何找到规律？

这时候我们还是要回到网页中寻找答案，我们复制第一集的iid值1061156738到浏览器的调试窗口搜索，找到iid就是某接口的vid值。

 找到集数参数之后，我们就可以写一个函数将所有集数参数爬取到。

import requests


def spider_jishu():
    '''爬取集数'''
    url = 'https://acs.youku.com/h5/mtop.youku.play.ups.appinfo.get/1.1/?jsv=2.4.16&appKey=24679788&t=1569420420811&sign=20948aee5c6366f01ac4f9e920621ad3&api=mtop.youku.play.ups.appinfo.get&v=1.1&timeout=20000&YKPid=20160317PLF000211&YKLoginRequest=true&AntiFlood=true&AntiCreep=true&type=jsonp&dataType=jsonp&callback=mtopjsonp1&data=%7B%22steal_params%22%3A%22%7B%5C%22ccode%5C%22%3A%5C%220502%5C%22%2C%5C%22client_ip%5C%22%3A%5C%22192.168.1.1%5C%22%2C%5C%22utid%5C%22%3A%5C%22ve1CFYo4KygCAa%2B%2Fy0oetkRW%5C%22%2C%5C%22client_ts%5C%22%3A1569420420%2C%5C%22version%5C%22%3A%5C%221.9.2%5C%22%2C%5C%22ckey%5C%22%3A%5C%22120%23bX1bSVkn%2BmDVauQx6cTUzzU%2F%2FZoQpqQ6kmvsjBzcWpZkZ8iV6Qg3s%2FNi%2FxKz6g6rWs2FwYkKucaKY0N4gWl52SLy8Xw6dlGUq%2FMki7B%2Fnp1cvGv94wDRpK08zqV9LiBboedoZCYVN2eozwi7gMqaTpYs9IVMfCbwDEc%2FbbYt7aPyyE0ACZ6P%2B174vgvIb1G74Jii4C4Rej%2FIVsTlXHizN1Jppz3DaDcAziGiThDKE9EFi0Gif3Hv%2BnP76OP1pwmOb6MkeoxBB8WEDXyiJjzuouAMB8FugAMXAFfIh5G8AKKsp1RCpnhF55S2duN3suo8TIYlsQTGuebMSNcHTeRtxjTTrXMnlNPUcmSi1Bw5luaBp5SETdw2rySpDkNXiywgtH2NPewbWu83FCNC9oe8LB47fgYR7SXQlv38xGqMkHZ6XkdLiQ8fSLH5NhwztQ6OYmMR8vFBFAg7AaZgFADF6gfZYquxh%2FWnrE02rphGNv7kXebRKNgCUYy%2BMSeXuUOCRku8rHPJlj1%2F%2BwzdIh%2BgYFJ2swhuOVhnv%2BD3cxg6hpWGMjdkUlvc0qTwjFse6VRV0J6OohmFvaD81x%2FJoMyV1NSMJIkQFfX%2Fb5bCXufxxFPDdTDI%2FzjoNwC2BeX6s5rPStajSEkSXBM3SKoxkhFujzO4GEa103YcourXbJno6a1PmqRNVUTFNvP9qE%2FubnQHNU2dNprm%2BHWBiro%2FmPDw4nme1DxfXcc1UU2e2%2FQ9dDdmZsuCOea5EK%2BA%2Fb%3D%3D%5C%22%7D%22%2C%22biz_params%22%3A%22%7B%5C%22vid%5C%22%3A%5C%22XNDI0NDQ0ODEwNA%3D%3D%5C%22%2C%5C%22play_ability%5C%22%3A5376%2C%5C%22current_showid%5C%22%3A%5C%22322943%5C%22%2C%5C%22preferClarity%5C%22%3A3%2C%5C%22master_m3u8%5C%22%3A1%2C%5C%22media_type%5C%22%3A%5C%22standard%2Csubtitle%5C%22%2C%5C%22app_ver%5C%22%3A%5C%221.9.2%5C%22%7D%22%2C%22ad_params%22%3A%22%7B%5C%22vs%5C%22%3A%5C%221.0%5C%22%2C%5C%22pver%5C%22%3A%5C%221.9.2%5C%22%2C%5C%22sver%5C%22%3A%5C%222.0%5C%22%2C%5C%22site%5C%22%3A1%2C%5C%22aw%5C%22%3A%5C%22w%5C%22%2C%5C%22fu%5C%22%3A0%2C%5C%22d%5C%22%3A%5C%220%5C%22%2C%5C%22bt%5C%22%3A%5C%22pc%5C%22%2C%5C%22os%5C%22%3A%5C%22win%5C%22%2C%5C%22osv%5C%22%3A%5C%2210%5C%22%2C%5C%22dq%5C%22%3A%5C%22auto%5C%22%2C%5C%22atm%5C%22%3A%5C%22%5C%22%2C%5C%22partnerid%5C%22%3A%5C%22null%5C%22%2C%5C%22wintype%5C%22%3A%5C%22interior%5C%22%2C%5C%22isvert%5C%22%3A0%2C%5C%22vip%5C%22%3A0%2C%5C%22emb%5C%22%3A%5C%22AjEwNjExMTIwMjYCdi55b3VrdS5jb20CL3Zfc2hvdy9pZF9YTkRJME5EWXlOamsxTWc9PS5odG1s%5C%22%2C%5C%22p%5C%22%3A1%2C%5C%22rst%5C%22%3A%5C%22mp4%5C%22%2C%5C%22needbf%5C%22%3A2%7D%22%7D'
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36'
    }
    try:
        r = requests.get(url, headers=kv)
        r.raise_for_status()
        print(r.text)
    except Exception as e:
        print(e)

if __name__ == '__main__':
    spider_jishu()

执行结果：

mtopjsonp1({"api":"mtop.youku.play.ups.appinfo.get","data":{},"ret":["FAIL_SYS_TOKEN_EMPTY::令牌为空"],"v":"1.1"})

令牌为空？很奇怪，URL和headers我们都填了为什么还是不行？而浏览器却可以？

这里需要引入另一个请求头：Cookie，Cookie是干什么的？

因为HTTP协议是无状态协议，也就是说下次再请求服务器并不知道你是谁，所以就用Cookie和Seesion来记录状态，最简单的例子就是用户登录后，服务器就给浏览器遗传一串加密字符串（key），然后服务器自己缓存一个key-value，这样浏览器每次来请求都带上这个key，服务器就知道你是哪个用户！

那我们去哪里找Cookie呢？答案当然是浏览器咯！

那这么多Cookie到底哪个才是我们要找的那个？这个谁也不知道，也不用找，我们直接把所有Cookie复制到代码里面就可以。

但是这种表格形式的根本无法复制，有没有什么小技巧能方便我们复制Cookie吗？当然有，我们点击浏览器调试窗口的Console按钮，然后输入document.cookie就可以看到全部Cookie啦，直接复制出来就可以，是不是很方便！

 我们把Cookie复制到代码里试试吧，注意Cookie有过期时间，尤其是这个token大概十几分钟可能就会过期，过期之后在浏览器中重新复制即可！

import requests


def spider_jishu():
    '''爬取集数'''
    url = 'https://acs.youku.com/h5/mtop.youku.play.ups.appinfo.get/1.1/?jsv=2.4.16&appKey=24679788&t=1569420420811&sign=20948aee5c6366f01ac4f9e920621ad3&api=mtop.youku.play.ups.appinfo.get&v=1.1&timeout=20000&YKPid=20160317PLF000211&YKLoginRequest=true&AntiFlood=true&AntiCreep=true&type=jsonp&dataType=jsonp&callback=mtopjsonp1&data=%7B%22steal_params%22%3A%22%7B%5C%22ccode%5C%22%3A%5C%220502%5C%22%2C%5C%22client_ip%5C%22%3A%5C%22192.168.1.1%5C%22%2C%5C%22utid%5C%22%3A%5C%22ve1CFYo4KygCAa%2B%2Fy0oetkRW%5C%22%2C%5C%22client_ts%5C%22%3A1569420420%2C%5C%22version%5C%22%3A%5C%221.9.2%5C%22%2C%5C%22ckey%5C%22%3A%5C%22120%23bX1bSVkn%2BmDVauQx6cTUzzU%2F%2FZoQpqQ6kmvsjBzcWpZkZ8iV6Qg3s%2FNi%2FxKz6g6rWs2FwYkKucaKY0N4gWl52SLy8Xw6dlGUq%2FMki7B%2Fnp1cvGv94wDRpK08zqV9LiBboedoZCYVN2eozwi7gMqaTpYs9IVMfCbwDEc%2FbbYt7aPyyE0ACZ6P%2B174vgvIb1G74Jii4C4Rej%2FIVsTlXHizN1Jppz3DaDcAziGiThDKE9EFi0Gif3Hv%2BnP76OP1pwmOb6MkeoxBB8WEDXyiJjzuouAMB8FugAMXAFfIh5G8AKKsp1RCpnhF55S2duN3suo8TIYlsQTGuebMSNcHTeRtxjTTrXMnlNPUcmSi1Bw5luaBp5SETdw2rySpDkNXiywgtH2NPewbWu83FCNC9oe8LB47fgYR7SXQlv38xGqMkHZ6XkdLiQ8fSLH5NhwztQ6OYmMR8vFBFAg7AaZgFADF6gfZYquxh%2FWnrE02rphGNv7kXebRKNgCUYy%2BMSeXuUOCRku8rHPJlj1%2F%2BwzdIh%2BgYFJ2swhuOVhnv%2BD3cxg6hpWGMjdkUlvc0qTwjFse6VRV0J6OohmFvaD81x%2FJoMyV1NSMJIkQFfX%2Fb5bCXufxxFPDdTDI%2FzjoNwC2BeX6s5rPStajSEkSXBM3SKoxkhFujzO4GEa103YcourXbJno6a1PmqRNVUTFNvP9qE%2FubnQHNU2dNprm%2BHWBiro%2FmPDw4nme1DxfXcc1UU2e2%2FQ9dDdmZsuCOea5EK%2BA%2Fb%3D%3D%5C%22%7D%22%2C%22biz_params%22%3A%22%7B%5C%22vid%5C%22%3A%5C%22XNDI0NDQ0ODEwNA%3D%3D%5C%22%2C%5C%22play_ability%5C%22%3A5376%2C%5C%22current_showid%5C%22%3A%5C%22322943%5C%22%2C%5C%22preferClarity%5C%22%3A3%2C%5C%22master_m3u8%5C%22%3A1%2C%5C%22media_type%5C%22%3A%5C%22standard%2Csubtitle%5C%22%2C%5C%22app_ver%5C%22%3A%5C%221.9.2%5C%22%7D%22%2C%22ad_params%22%3A%22%7B%5C%22vs%5C%22%3A%5C%221.0%5C%22%2C%5C%22pver%5C%22%3A%5C%221.9.2%5C%22%2C%5C%22sver%5C%22%3A%5C%222.0%5C%22%2C%5C%22site%5C%22%3A1%2C%5C%22aw%5C%22%3A%5C%22w%5C%22%2C%5C%22fu%5C%22%3A0%2C%5C%22d%5C%22%3A%5C%220%5C%22%2C%5C%22bt%5C%22%3A%5C%22pc%5C%22%2C%5C%22os%5C%22%3A%5C%22win%5C%22%2C%5C%22osv%5C%22%3A%5C%2210%5C%22%2C%5C%22dq%5C%22%3A%5C%22auto%5C%22%2C%5C%22atm%5C%22%3A%5C%22%5C%22%2C%5C%22partnerid%5C%22%3A%5C%22null%5C%22%2C%5C%22wintype%5C%22%3A%5C%22interior%5C%22%2C%5C%22isvert%5C%22%3A0%2C%5C%22vip%5C%22%3A0%2C%5C%22emb%5C%22%3A%5C%22AjEwNjExMTIwMjYCdi55b3VrdS5jb20CL3Zfc2hvdy9pZF9YTkRJME5EWXlOamsxTWc9PS5odG1s%5C%22%2C%5C%22p%5C%22%3A1%2C%5C%22rst%5C%22%3A%5C%22mp4%5C%22%2C%5C%22needbf%5C%22%3A2%7D%22%7D'
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36',
        'Cookie': '"P_F=1; u_l_v_t=2477; P_T=1569428606; __ysuid=1558850451468kWf; UM_distinctid=16cc149cbe9200-0c10118752256b-7373e61-100200-16cc149cbea15d; cna=ve1CFYo4KygCAa+/y0oetkRW; __ayft=1569414353794; __aysid=1569414353795JuO; __ayscnt=1; juid=01dlk65ssf1bbe; yseid=1569414382716nCOPtB; yseidcount=1; ycid=0; l=AklJpyXfn2sDf789h/ro64KZ2X6jFj3I; __arycid=dz-3-00; __arcms=dz-3-00; CNZZDATA1277955961=2129516980-1569411834-https%253A%252F%252Fso.youku.com%252F%7C1569417234; _m_h5_tk=d32919f4ff891b7d9282fb593fee1ca0_1569424127226; _m_h5_tk_enc=a6e8267c83d20cdbe05ba5d00b204ee9; seid=01dlkbfqbj26qo; referhost=https%3A%2F%2Fv.youku.com; __arpvid=1569421380032DKZSGN-1569421380058; __aypstp=17; __ayspstp=17; seidtimeout=1569423184746; ypvid=1569421389178o8j8T2; ysestep=11; yseidtimeout=1569428589180; ystep=11; __ayvstp=86; __aysvstp=86; isg=BIuL34lGUtlxHY5meEB35JHSGi91IJ-i94W_wv2KR0ohHK5-hPCn8k-u9FxXO_ea"'
    }
    try:
        r = requests.get(url, headers=kv)
        r.raise_for_status()
        print(r.text)
    except Exception as e:
        print(e)

if __name__ == '__main__':
    spider_jishu()

 我们可以观察到返回的数据同样是个jsonp函数，同样需要提取内部的json数据，所以我们可以封装一个公用方法，用于提取jsonp返回的数据转为json对象，这样提高了复用性！

import json

import requests


def jsonp_func_to_json_obj(jsonp_func):
    """
    jsonp返回函数提取json并转换为对象
    :param jsonp_func: jsonp请求返回的数据，格式xxx(json)
    :return: json对象
    """
    # 找到jsonp数据的左括号位置并加1
    json_start_index = jsonp_func.index("(") + 1
    json_end_index = jsonp_func.index(')')
    # 截取json数据字符串
    r_json_str = jsonp_func[json_start_index:json_end_index]
    return json.loads(r_json_str)

def spider_jishu():
    '''爬取集数'''
    url = 'https://acs.youku.com/h5/mtop.youku.play.ups.appinfo.get/1.1/?jsv=2.4.16&appKey=24679788&t=1569420420811&sign=20948aee5c6366f01ac4f9e920621ad3&api=mtop.youku.play.ups.appinfo.get&v=1.1&timeout=20000&YKPid=20160317PLF000211&YKLoginRequest=true&AntiFlood=true&AntiCreep=true&type=jsonp&dataType=jsonp&callback=mtopjsonp1&data=%7B%22steal_params%22%3A%22%7B%5C%22ccode%5C%22%3A%5C%220502%5C%22%2C%5C%22client_ip%5C%22%3A%5C%22192.168.1.1%5C%22%2C%5C%22utid%5C%22%3A%5C%22ve1CFYo4KygCAa%2B%2Fy0oetkRW%5C%22%2C%5C%22client_ts%5C%22%3A1569420420%2C%5C%22version%5C%22%3A%5C%221.9.2%5C%22%2C%5C%22ckey%5C%22%3A%5C%22120%23bX1bSVkn%2BmDVauQx6cTUzzU%2F%2FZoQpqQ6kmvsjBzcWpZkZ8iV6Qg3s%2FNi%2FxKz6g6rWs2FwYkKucaKY0N4gWl52SLy8Xw6dlGUq%2FMki7B%2Fnp1cvGv94wDRpK08zqV9LiBboedoZCYVN2eozwi7gMqaTpYs9IVMfCbwDEc%2FbbYt7aPyyE0ACZ6P%2B174vgvIb1G74Jii4C4Rej%2FIVsTlXHizN1Jppz3DaDcAziGiThDKE9EFi0Gif3Hv%2BnP76OP1pwmOb6MkeoxBB8WEDXyiJjzuouAMB8FugAMXAFfIh5G8AKKsp1RCpnhF55S2duN3suo8TIYlsQTGuebMSNcHTeRtxjTTrXMnlNPUcmSi1Bw5luaBp5SETdw2rySpDkNXiywgtH2NPewbWu83FCNC9oe8LB47fgYR7SXQlv38xGqMkHZ6XkdLiQ8fSLH5NhwztQ6OYmMR8vFBFAg7AaZgFADF6gfZYquxh%2FWnrE02rphGNv7kXebRKNgCUYy%2BMSeXuUOCRku8rHPJlj1%2F%2BwzdIh%2BgYFJ2swhuOVhnv%2BD3cxg6hpWGMjdkUlvc0qTwjFse6VRV0J6OohmFvaD81x%2FJoMyV1NSMJIkQFfX%2Fb5bCXufxxFPDdTDI%2FzjoNwC2BeX6s5rPStajSEkSXBM3SKoxkhFujzO4GEa103YcourXbJno6a1PmqRNVUTFNvP9qE%2FubnQHNU2dNprm%2BHWBiro%2FmPDw4nme1DxfXcc1UU2e2%2FQ9dDdmZsuCOea5EK%2BA%2Fb%3D%3D%5C%22%7D%22%2C%22biz_params%22%3A%22%7B%5C%22vid%5C%22%3A%5C%22XNDI0NDQ0ODEwNA%3D%3D%5C%22%2C%5C%22play_ability%5C%22%3A5376%2C%5C%22current_showid%5C%22%3A%5C%22322943%5C%22%2C%5C%22preferClarity%5C%22%3A3%2C%5C%22master_m3u8%5C%22%3A1%2C%5C%22media_type%5C%22%3A%5C%22standard%2Csubtitle%5C%22%2C%5C%22app_ver%5C%22%3A%5C%221.9.2%5C%22%7D%22%2C%22ad_params%22%3A%22%7B%5C%22vs%5C%22%3A%5C%221.0%5C%22%2C%5C%22pver%5C%22%3A%5C%221.9.2%5C%22%2C%5C%22sver%5C%22%3A%5C%222.0%5C%22%2C%5C%22site%5C%22%3A1%2C%5C%22aw%5C%22%3A%5C%22w%5C%22%2C%5C%22fu%5C%22%3A0%2C%5C%22d%5C%22%3A%5C%220%5C%22%2C%5C%22bt%5C%22%3A%5C%22pc%5C%22%2C%5C%22os%5C%22%3A%5C%22win%5C%22%2C%5C%22osv%5C%22%3A%5C%2210%5C%22%2C%5C%22dq%5C%22%3A%5C%22auto%5C%22%2C%5C%22atm%5C%22%3A%5C%22%5C%22%2C%5C%22partnerid%5C%22%3A%5C%22null%5C%22%2C%5C%22wintype%5C%22%3A%5C%22interior%5C%22%2C%5C%22isvert%5C%22%3A0%2C%5C%22vip%5C%22%3A0%2C%5C%22emb%5C%22%3A%5C%22AjEwNjExMTIwMjYCdi55b3VrdS5jb20CL3Zfc2hvdy9pZF9YTkRJME5EWXlOamsxTWc9PS5odG1s%5C%22%2C%5C%22p%5C%22%3A1%2C%5C%22rst%5C%22%3A%5C%22mp4%5C%22%2C%5C%22needbf%5C%22%3A2%7D%22%7D'
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36',
        'Cookie': '"P_F=1; u_l_v_t=2477; P_T=1569428606; __ysuid=1558850451468kWf; UM_distinctid=16cc149cbe9200-0c10118752256b-7373e61-100200-16cc149cbea15d; cna=ve1CFYo4KygCAa+/y0oetkRW; __ayft=1569414353794; __aysid=1569414353795JuO; __ayscnt=1; juid=01dlk65ssf1bbe; yseid=1569414382716nCOPtB; yseidcount=1; ycid=0; l=AklJpyXfn2sDf789h/ro64KZ2X6jFj3I; __arycid=dz-3-00; __arcms=dz-3-00; CNZZDATA1277955961=2129516980-1569411834-https%253A%252F%252Fso.youku.com%252F%7C1569417234; _m_h5_tk=d32919f4ff891b7d9282fb593fee1ca0_1569424127226; _m_h5_tk_enc=a6e8267c83d20cdbe05ba5d00b204ee9; seid=01dlkbfqbj26qo; referhost=https%3A%2F%2Fv.youku.com; __arpvid=1569421380032DKZSGN-1569421380058; __aypstp=17; __ayspstp=17; seidtimeout=1569423184746; ypvid=1569421389178o8j8T2; ysestep=11; yseidtimeout=1569428589180; ystep=11; __ayvstp=86; __aysvstp=86; isg=BIuL34lGUtlxHY5meEB35JHSGi91IJ-i94W_wv2KR0ohHK5-hPCn8k-u9FxXO_ea"'
    }
    try:
        r = requests.get(url, headers=kv)
        r.raise_for_status()
        # 提取json数据并转换为json对象
        r_json_obj = jsonp_func_to_json_obj(r.text)
        # 获取电视剧简介集数
        video_list = r_json_obj['data']['data']['videos']['list']
        # 创建一个生成器并返回

        return (video['vid'] for video in video_list)
    except Exception as e:
        print(e)




if __name__ == '__main__':
    for i in spider_jishu():
        print(i)

得到json数据之后没我们通过观察可以得到知其数据结构，然后将vid提取出来并返回.

代表集数的id拿到了，现在我们就可以双层循环去爬所有的弹幕啦，上代码。

import json
import os
import random
import time

import requests


def jsonp_func_to_json_obj(jsonp_func):
    """
    jsonp返回函数提取json并转换为对象
    :param jsonp_func: jsonp请求返回的数据，格式xxx(json)
    :return: json对象
    """
    # 找到jsonp数据的左括号位置并加1
    json_start_index = jsonp_func.index("(") + 1
    json_end_index = jsonp_func.index(')')
    # 截取json数据字符串
    r_json_str = jsonp_func[json_start_index:json_end_index]
    return json.loads(r_json_str)


def spider_jishu():
    '''爬取集数'''
    url = 'https://acs.youku.com/h5/mtop.youku.play.ups.appinfo.get/1.1/?jsv=2.4.16&appKey=24679788&t=1569420420811&sign=20948aee5c6366f01ac4f9e920621ad3&api=mtop.youku.play.ups.appinfo.get&v=1.1&timeout=20000&YKPid=20160317PLF000211&YKLoginRequest=true&AntiFlood=true&AntiCreep=true&type=jsonp&dataType=jsonp&callback=mtopjsonp1&data=%7B%22steal_params%22%3A%22%7B%5C%22ccode%5C%22%3A%5C%220502%5C%22%2C%5C%22client_ip%5C%22%3A%5C%22192.168.1.1%5C%22%2C%5C%22utid%5C%22%3A%5C%22ve1CFYo4KygCAa%2B%2Fy0oetkRW%5C%22%2C%5C%22client_ts%5C%22%3A1569420420%2C%5C%22version%5C%22%3A%5C%221.9.2%5C%22%2C%5C%22ckey%5C%22%3A%5C%22120%23bX1bSVkn%2BmDVauQx6cTUzzU%2F%2FZoQpqQ6kmvsjBzcWpZkZ8iV6Qg3s%2FNi%2FxKz6g6rWs2FwYkKucaKY0N4gWl52SLy8Xw6dlGUq%2FMki7B%2Fnp1cvGv94wDRpK08zqV9LiBboedoZCYVN2eozwi7gMqaTpYs9IVMfCbwDEc%2FbbYt7aPyyE0ACZ6P%2B174vgvIb1G74Jii4C4Rej%2FIVsTlXHizN1Jppz3DaDcAziGiThDKE9EFi0Gif3Hv%2BnP76OP1pwmOb6MkeoxBB8WEDXyiJjzuouAMB8FugAMXAFfIh5G8AKKsp1RCpnhF55S2duN3suo8TIYlsQTGuebMSNcHTeRtxjTTrXMnlNPUcmSi1Bw5luaBp5SETdw2rySpDkNXiywgtH2NPewbWu83FCNC9oe8LB47fgYR7SXQlv38xGqMkHZ6XkdLiQ8fSLH5NhwztQ6OYmMR8vFBFAg7AaZgFADF6gfZYquxh%2FWnrE02rphGNv7kXebRKNgCUYy%2BMSeXuUOCRku8rHPJlj1%2F%2BwzdIh%2BgYFJ2swhuOVhnv%2BD3cxg6hpWGMjdkUlvc0qTwjFse6VRV0J6OohmFvaD81x%2FJoMyV1NSMJIkQFfX%2Fb5bCXufxxFPDdTDI%2FzjoNwC2BeX6s5rPStajSEkSXBM3SKoxkhFujzO4GEa103YcourXbJno6a1PmqRNVUTFNvP9qE%2FubnQHNU2dNprm%2BHWBiro%2FmPDw4nme1DxfXcc1UU2e2%2FQ9dDdmZsuCOea5EK%2BA%2Fb%3D%3D%5C%22%7D%22%2C%22biz_params%22%3A%22%7B%5C%22vid%5C%22%3A%5C%22XNDI0NDQ0ODEwNA%3D%3D%5C%22%2C%5C%22play_ability%5C%22%3A5376%2C%5C%22current_showid%5C%22%3A%5C%22322943%5C%22%2C%5C%22preferClarity%5C%22%3A3%2C%5C%22master_m3u8%5C%22%3A1%2C%5C%22media_type%5C%22%3A%5C%22standard%2Csubtitle%5C%22%2C%5C%22app_ver%5C%22%3A%5C%221.9.2%5C%22%7D%22%2C%22ad_params%22%3A%22%7B%5C%22vs%5C%22%3A%5C%221.0%5C%22%2C%5C%22pver%5C%22%3A%5C%221.9.2%5C%22%2C%5C%22sver%5C%22%3A%5C%222.0%5C%22%2C%5C%22site%5C%22%3A1%2C%5C%22aw%5C%22%3A%5C%22w%5C%22%2C%5C%22fu%5C%22%3A0%2C%5C%22d%5C%22%3A%5C%220%5C%22%2C%5C%22bt%5C%22%3A%5C%22pc%5C%22%2C%5C%22os%5C%22%3A%5C%22win%5C%22%2C%5C%22osv%5C%22%3A%5C%2210%5C%22%2C%5C%22dq%5C%22%3A%5C%22auto%5C%22%2C%5C%22atm%5C%22%3A%5C%22%5C%22%2C%5C%22partnerid%5C%22%3A%5C%22null%5C%22%2C%5C%22wintype%5C%22%3A%5C%22interior%5C%22%2C%5C%22isvert%5C%22%3A0%2C%5C%22vip%5C%22%3A0%2C%5C%22emb%5C%22%3A%5C%22AjEwNjExMTIwMjYCdi55b3VrdS5jb20CL3Zfc2hvdy9pZF9YTkRJME5EWXlOamsxTWc9PS5odG1s%5C%22%2C%5C%22p%5C%22%3A1%2C%5C%22rst%5C%22%3A%5C%22mp4%5C%22%2C%5C%22needbf%5C%22%3A2%7D%22%7D'
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36',
        'Cookie': '"P_F=1; u_l_v_t=2477; P_T=1569428606; __ysuid=1558850451468kWf; UM_distinctid=16cc149cbe9200-0c10118752256b-7373e61-100200-16cc149cbea15d; cna=ve1CFYo4KygCAa+/y0oetkRW; __ayft=1569414353794; __aysid=1569414353795JuO; __ayscnt=1; juid=01dlk65ssf1bbe; yseid=1569414382716nCOPtB; yseidcount=1; ycid=0; l=AklJpyXfn2sDf789h/ro64KZ2X6jFj3I; __arycid=dz-3-00; __arcms=dz-3-00; CNZZDATA1277955961=2129516980-1569411834-https%253A%252F%252Fso.youku.com%252F%7C1569417234; _m_h5_tk=d32919f4ff891b7d9282fb593fee1ca0_1569424127226; _m_h5_tk_enc=a6e8267c83d20cdbe05ba5d00b204ee9; seid=01dlkbfqbj26qo; referhost=https%3A%2F%2Fv.youku.com; __arpvid=1569421380032DKZSGN-1569421380058; __aypstp=17; __ayspstp=17; seidtimeout=1569423184746; ypvid=1569421389178o8j8T2; ysestep=11; yseidtimeout=1569428589180; ystep=11; __ayvstp=86; __aysvstp=86; isg=BIuL34lGUtlxHY5meEB35JHSGi91IJ-i94W_wv2KR0ohHK5-hPCn8k-u9FxXO_ea"'
    }
    try:
        r = requests.get(url, headers=kv)
        r.raise_for_status()
        # 提取json数据并转换为json对象
        r_json_obj = jsonp_func_to_json_obj(r.text)
        # 获取电视剧简介集数
        video_list = r_json_obj['data']['data']['videos']['list']
        # 创建一个生成器并返回
        return (video['vid'] for video in video_list)
    except Exception as e:
        print(e)



def spider_danmu(i,vid):
    '''爬取优酷弹幕'''
    url = 'https://service.danmu.youku.com/list?jsoncallback=jQuery111208275429479734884_1569416025133&mat=%s&mcount=1&ct=1001&iid=%s&aid=322943&cid=97&lid=0&ouid=0&_=1569416025149'%(i,vid)
    kv = {
        'Referer': 'https://v.youku.com/v_show/id_XNDI0NDYyNjk1Mg==.html?spm=a2h0k.11417342.soresults.dselectbutton&s=efbfbd78efbfbd5cefbf',
        'Sec-Fetch-Mode': 'no-cors',
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/77.0.3865.90 Safari/537.36'
    }
    try:
        r = requests.get(url, headers=kv)
        r.raise_for_status()
        #r_json_obj = jsonp_func_to_json_obj(r.text)
        # 找到jsonp数据的左括号位置并加1
        json_start_index = r.text.index('(') + 1
        # 截取json数据字符串
        r_json_str = r.text[json_start_index:-2]
        # 字符串转换为json对象
        r_json_obj = json.loads(r_json_str)
        # 如果请求总数count = 0 则说明弹幕爬取完成
        if not r_json_obj['count']:
            return 0
        r_json_result = r_json_obj['result']
        for r_json_danmu in r_json_result:
            with open('danmu.txt','a+', encoding='utf-8') as f:
                f.write(r_json_danmu['content']+'
')

        return 1
    except Exception as e:
        print(e)

def batch_spider_commnet():
    if os.path.exists('danmu.txt'):
        os.remove('danmu.txt')

    # 爬取所有集数的vid
    vids = spider_jishu()
    for vid in vids:
        print(vid)
        i = 0
        while spider_danmu(i, vid):  # 判断返回值，如果爬完了，则返回0,
            time.sleep(random.random()*5)
            i += 1
    print('爬取完毕')

if __name__ == '__main__':
    batch_spider_commnet()

一共爬取了近30万条的数据，大概用了40分钟，中间空挡去斗地主赢了一万金币哈哈哈，然如果你觉得时间间隔太长也可缩短，但是建议不要太频繁，不然对人家服务器或者被监控到就不好。

6、数据清洗+生成词云

我们要清洗什么数据？其实这个事先很难猜到，所以我们不做数据清洗直接生成云词看看会是什么效果，然后再做调整。

import jieba
import numpy as np
from PIL import Image
from wordcloud import WordCloud
from matplotlib import pyplot as plt

def cut_word():
    with open('danmu.txt', 'r', encoding='utf-8') as f:
        comment_txt = f.read()
        wordlist = jieba.cut(comment_txt, cut_all=False)
        wl = " ".join(wordlist)
        print(wl)
        return wl


def create_word_cloud():
    # 设置词云形状
    wc_mask = np.array(Image.open('bige.jpg'))
    # 设置词云的一些配置，如字体，背景色，词云形状，大小
    wc = WordCloud(background_color='white', max_words=2000, width=940, height=2000,max_font_size=50, random_state=42,
                  font_path='C:WindowsFontsmsyhbd.ttc')

    # 生成词云
    wc.generate(cut_word())
    # 在只设置mask情况下，会得到一个拥有图片形状的词云
    plt.imshow(wc, interpolation='bilinear')
    plt.axis('off')
    plt.figure()
    plt.show()
  

if __name__ == '__main__':
    create_word_cloud(

 我们看到像：哈哈、不是、这个、什么，会员等这些词比较多，这种单词没有太大的分析价值，所以我们数据清洗便有了方向。（ps：某酷的弹幕没啥内涵。。。）

import jieba
import numpy as np
from PIL import Image
from wordcloud import WordCloud
from matplotlib import pyplot as plt

def cut_word():
    with open('danmu.txt', 'r', encoding='utf-8') as f:
        comment_txt = f.read()
        wordlist = jieba.cut(comment_txt, cut_all=False)
        wl = " ".join(wordlist)
        print(wl)
        return wl


def create_word_cloud():
    stop_words = ['哈哈','哈哈哈','哈哈哈哈','会员', '什么', '为什么','不是','就是','还是','真是','这是','是不是','应该','不能','这个',
                  '电视','电视剧','怎么', '这么','那么','没有','不能','不知','知道', '这样', '还有']
    # 设置词云形状
    wc_mask = np.array(Image.open('bige.jpg'))
    # 设置词云的一些配置，如字体，背景色，词云形状，大小
    wc = WordCloud(background_color='white', max_words=900, mask=wc_mask, width=940, height=400,scale=10,max_font_size=50, random_state=42,
                   stopwords=stop_words,font_path='C:WindowsFontsmsyhbd.ttc')

    # 生成词云
    wc.generate(cut_word())
    # 在只设置mask情况下，会得到一个拥有图片形状的词云
    plt.imshow(wc, interpolation='bilinear')
    plt.axis('off')
    plt.figure()
    plt.show()


if __name__ == '__main__':
    create_word_cloud()

7.分析词云图

从上面的词云图中我们可以分析出：

1. 此电视剧中的一些主角：张小敬、李必、崔器、龙波、徐斌、竟然还有人喜欢曹破延。

2. 有人说好看，有人说看不懂，说明剧情可能有点深度

3. 画风可能有点像刺客信条

4. 四字弟弟、千玺，说明剧中有易烊千玺

5. 片尾曲可能有惊喜

6. 大唐、长安说明了故事背景

7. 弹幕、智商，可能大家在提醒你：关弹幕，保智商！