Python collects a hand video, 1080p HD without watermark, complete data source analysis + complete code

Knowledge points

• Dynamic packet capture
• Dynamic page analysis
• Requests carries parameters to send requests
• JSON data parsing

development environment

• Python 3.8 newer stable running code
• Pycharm 2021.2 auxiliary code
• Requests third party module

I Data source analysis (train of thought analysis)

1. Open the developer tool to refresh the web page

• Right click Check or F12 to open

• Select network and refresh the web page
  

• Click to open a video
  

• Click to find the content
  

• Expand the view in turn to find the video address we need

2. Determine the URL address, request method, request parameters and request header parameters

• Request header parameters
  

• Request parameters
  

3. Summary

• Request method: Post
• Request header (disguise):

headers = {
'content-type': 'application/json',
'cookie': 'your own cookie',
'Host': 'www.kuaishou.com',
'Origin': 'https://www.kuaishou.com',
'Referer': 'https://www.kuaishou.com/profile/3xv78fxycm35nn4',
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36
(KHTML, like Gecko) Chrome/95.0.4638.69 Safari/537.36'
}

• Request parameters:

data = {
'operationName': "visionProfilePhotoList",
'query': "query visionProfilePhotoList($pcursor: String, $userId: String, $page:
String, $webPageArea: String) {\n visionProfilePhotoList(pcursor: $pcursor, userId:
$userId, page: $page, webPageArea: $webPageArea) {\n result\n llsid\n
webPageArea\n feeds {\n type\n author {\n id\n name\n
following\n headerUrl\n headerUrls {\n cdn\n url\n
__typename\n }\n __typename\n }\n tags {\n type\n
name\n __typename\n }\n photo {\n id\n
duration\n caption\n likeCount\n realLikeCount\n
coverUrl\n coverUrls {\n cdn\n url\n __typename\n
}\n photoUrls {\n cdn\n url\n __typename\n
}\n photoUrl\n liked\n timestamp\n expTag\n
animatedCoverUrl\n stereoType\n videoRatio\n
profileUserTopPhoto\n __typename\n }\n canAddComment\n
currentPcursor\n llsid\n status\n __typename\n }\n hostName\n
pcursor\n __typename\n }\n}\n",
'variables': {'userId': "3x9dquvtb9n9fps", 'pcursor': "", 'page': "profile"}
}

• If you need to turn pages and crawl later, you need to use recursive implementation
  

II code implementation

1. Send a request to visit the website

url = 'https://www.kuaishou.com/graphql'
#Disguise
headers = {
    #Control data type JSON type string
    'content-type': 'application/json',
    'Cookie': 'kpf=PC_WEB; kpn=KUAISHOU_VISION; clientid=3; did=web_ea128125517a46bd491ae9ccb255e242; client_key=65890b29; userId=270932146; kuaishou.server.web_st=ChZrdWFpc2hvdS5zZXJ2ZXIud2ViLnN0EqABnjkpJPZ-QanEQnI0XWMVZxXtIqPj-hwjsXBn9DHaTzispQcLjGR-5Xr-rY4VFaIC-egxv508oQoRYdgafhxSBpZYqLnApsaeuAaoLj2xMbRoytYGCrTLF6vVWJvzz3nzBVzNSyrXyhz-RTlRJP4xe1VjSp7XLNLRnVFVEtGPuBz0xkOnemy7-1-k6FEwoPIbOau9qgO5mukNg0qQ2NLz_xoSKS0sDuL1vMmNDXbwL4KX-qDmIiCWJ_fVUQoL5jjg3553H5iUdvpNxx97u6I6MkKEzwOaSigFMAE; kuaishou.server.web_ph=b282f9af819333f3d13e9c45765ed62560a1',
    'Host': 'www.kuaishou.com',
    'Origin': 'https://www.kuaishou.com',
    'Referer': 'https://www.kuaishou.com/profile/3xauthkq46ftgkg',
    'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/97.0.4692.71 Safari/537.36',
}
#< response [200] >: send request success result
response = requests.post(url=url, headers=headers, json=data)

2. Obtain data

json_data = response.json()

3. Parse the data to remove unwanted content

feeds = json_data['data']['visionProfilePhotoList']['feeds']
#Parameters required on the next page
pcursor = json_data['data']['visionProfilePhotoList']['pcursor']
# print(pcursor)
for feed in feeds:
    Caption = feed ['photo '] ['caption'] # title
    Photourl = feed ['photo '] ['photourl'] # video link
    #\: escape character, write directly \ cannot match\
    #\ \ to match\
    #Using CSS and XPath is necessary. The data you get is a web page source code
    caption = re.sub('[\\/:*?"<>|\n\t]', '', caption)
    print(caption, photoUrl)

5. The video data obtained is binary video data

video_data = requests.get(url=photoUrl).content

6. Save the video in binary mode

with open(f'video/{caption}.mp4', mode='wb') as f:
    f.write(video_data)
Print (caption, 'download complete!')

Page crawling

def get_page(pcursor):
    #The required data must be specified
    #Recursion, call yourself and jump out of recursion
    data = {
        'operationName': "visionProfilePhotoList",
        'query': "query visionProfilePhotoList($pcursor: String, $userId: String, $page: String, $webPageArea: String) {\n  visionProfilePhotoList(pcursor: $pcursor, userId: $userId, page: $page, webPageArea: $webPageArea) {\n    result\n    llsid\n    webPageArea\n    feeds {\n      type\n      author {\n        id\n        name\n        following\n        headerUrl\n        headerUrls {\n          cdn\n          url\n          __typename\n        }\n        __typename\n      }\n      tags {\n        type\n        name\n        __typename\n      }\n      photo {\n        id\n        duration\n        caption\n        likeCount\n        realLikeCount\n        coverUrl\n        coverUrls {\n          cdn\n          url\n          __typename\n        }\n        photoUrls {\n          cdn\n          url\n          __typename\n        }\n        photoUrl\n        liked\n        timestamp\n        expTag\n        animatedCoverUrl\n        stereoType\n        videoRatio\n        profileUserTopPhoto\n        __typename\n      }\n      canAddComment\n      currentPcursor\n      llsid\n      status\n      __typename\n    }\n    hostName\n    pcursor\n    __typename\n  }\n}\n",
        'variables': {'userId': "3xauthkq46ftgkg", 'pcursor': pcursor, 'page': "profile"}
    }
    if pcursor == None:
        Print ('download complete ')
        return 0
    
    get_page(pcursor)

get_page('')

Effect display