mirror of
https://github.com/Evil0ctal/Douyin_TikTok_Download_API.git
synced 2025-04-22 02:10:24 +08:00
✨增加对TikTok图集的支持以及修复bug
This commit is contained in:
Evil0ctal
GitHub
parent
b522d9b796
commit
adb42994c9
295
scraper.py
295
scraper.py
@ -2,7 +2,7 @@
|
|||||||
# -*- encoding: utf-8 -*-
|
# -*- encoding: utf-8 -*-
|
||||||
# @Author: https://github.com/Evil0ctal/
|
# @Author: https://github.com/Evil0ctal/
|
||||||
# @Time: 2021/11/06
|
# @Time: 2021/11/06
|
||||||
# @Update: 2022/06/03
|
# @Update: 2022/06/05
|
||||||
# @Function:
|
# @Function:
|
||||||
# 核心代码，估值1块(๑•̀ㅂ•́)و✧
|
# 核心代码，估值1块(๑•̀ㅂ•́)و✧
|
||||||
# 用于爬取Douyin/TikTok数据并以字典形式返回。
|
# 用于爬取Douyin/TikTok数据并以字典形式返回。
|
||||||
@ -190,7 +190,9 @@ class Scraper:
|
|||||||
vid = str(js['item_list'][0]['video']['vid'])
|
vid = str(js['item_list'][0]['video']['vid'])
|
||||||
# 无水印1080p视频链接
|
# 无水印1080p视频链接
|
||||||
try:
|
try:
|
||||||
r = requests.get("https://aweme.snssdk.com/aweme/v1/play/?video_id={}&radio=1080p&line=0".format(vid), headers=headers, allow_redirects=False)
|
r = requests.get(
|
||||||
|
"https://aweme.snssdk.com/aweme/v1/play/?video_id={}&radio=1080p&line=0".format(vid),
|
||||||
|
headers=headers, allow_redirects=False)
|
||||||
nwm_video_url_1080p = r.headers['Location']
|
nwm_video_url_1080p = r.headers['Location']
|
||||||
except:
|
except:
|
||||||
nwm_video_url_1080p = "None"
|
nwm_video_url_1080p = "None"
|
||||||
@ -302,109 +304,192 @@ class Scraper:
|
|||||||
# 获取视频ID
|
# 获取视频ID
|
||||||
video_id = re.findall('video/(\d+)?', original_url)[0]
|
video_id = re.findall('video/(\d+)?', original_url)[0]
|
||||||
print('获取到的TikTok视频ID是{}'.format(video_id))
|
print('获取到的TikTok视频ID是{}'.format(video_id))
|
||||||
# 从TikTok网页获取部分视频数据
|
# 尝试从TikTok网页获取部分视频数据，失败后判断为图集
|
||||||
tiktok_headers = self.tiktok_headers
|
|
||||||
html = requests.get(url=original_url, headers=tiktok_headers)
|
|
||||||
# 正则检索网页中存在的JSON信息
|
|
||||||
resp = re.search('"ItemModule":{(.*)},"UserModule":', html.text).group(1)
|
|
||||||
resp_info = ('{"ItemModule":{' + resp + '}}')
|
|
||||||
result = json.loads(resp_info)
|
|
||||||
# 从网页中获得的视频JSON数据
|
|
||||||
video_info = result["ItemModule"][video_id]
|
|
||||||
# 从TikTok官方API获取部分视频数据
|
|
||||||
tiktok_api_link = 'https://api.tiktokv.com/aweme/v1/multi/aweme/detail/?aweme_ids=%5B{}%5D'.format(video_id)
|
|
||||||
print('正在请求API链接:{}'.format(tiktok_api_link))
|
|
||||||
response = requests.get(url=tiktok_api_link, headers=headers).text
|
|
||||||
# 将API获取到的内容格式化为JSON
|
|
||||||
result = json.loads(response)
|
|
||||||
# 类型为视频
|
|
||||||
url_type = 'video'
|
|
||||||
# 无水印视频链接
|
|
||||||
nwm_video_url = result["aweme_details"][0]["video"]["play_addr"]["url_list"][0]
|
|
||||||
try:
|
try:
|
||||||
# 有水印视频链接
|
tiktok_headers = self.tiktok_headers
|
||||||
wm_video_url = result["aweme_details"][0]["video"]['download_addr']['url_list'][0]
|
html = requests.get(url=original_url, headers=tiktok_headers)
|
||||||
except Exception:
|
# 正则检索网页中存在的JSON信息
|
||||||
# 有水印视频链接
|
resp = re.search('"ItemModule":{(.*)},"UserModule":', html.text).group(1)
|
||||||
wm_video_url = 'None'
|
resp_info = ('{"ItemModule":{' + resp + '}}')
|
||||||
# 视频标题
|
result = json.loads(resp_info)
|
||||||
video_title = result["aweme_details"][0]["desc"]
|
# 从网页中获得的视频JSON数据
|
||||||
# 视频作者昵称
|
video_info = result["ItemModule"][video_id]
|
||||||
video_author_nickname = result["aweme_details"][0]['author']["nickname"]
|
# 从TikTok官方API获取部分视频数据
|
||||||
# 视频作者ID
|
tiktok_api_link = 'https://api.tiktokv.com/aweme/v1/multi/aweme/detail/?aweme_ids=%5B{}%5D'.format(
|
||||||
video_author_id = result["aweme_details"][0]['author']["unique_id"]
|
video_id)
|
||||||
# 上传时间戳
|
print('正在请求API链接:{}'.format(tiktok_api_link))
|
||||||
video_create_time = result["aweme_details"][0]['create_time']
|
response = requests.get(url=tiktok_api_link, headers=headers).text
|
||||||
# 视频ID
|
# 将API获取到的内容格式化为JSON
|
||||||
video_aweme_id = result["aweme_details"][0]['statistics']['aweme_id']
|
result = json.loads(response)
|
||||||
# 视频BGM标题
|
# 类型为视频
|
||||||
video_music_title = result["aweme_details"][0]['music']['title']
|
url_type = 'video'
|
||||||
# 视频BGM作者
|
# 无水印视频链接
|
||||||
video_music_author = result["aweme_details"][0]['music']['author']
|
nwm_video_url = result["aweme_details"][0]["video"]["play_addr"]["url_list"][0]
|
||||||
# 视频BGM ID
|
try:
|
||||||
video_music_id = result["aweme_details"][0]['music']['id']
|
# 有水印视频链接
|
||||||
# 视频BGM链接
|
wm_video_url = result["aweme_details"][0]["video"]['download_addr']['url_list'][0]
|
||||||
video_music_url = result["aweme_details"][0]['music']['play_url']['url_list'][0]
|
except Exception:
|
||||||
# 评论数量
|
# 有水印视频链接
|
||||||
video_comment_count = result["aweme_details"][0]['statistics']['comment_count']
|
wm_video_url = 'None'
|
||||||
# 获赞数量
|
# 视频标题
|
||||||
video_digg_count = result["aweme_details"][0]['statistics']['digg_count']
|
video_title = result["aweme_details"][0]["desc"]
|
||||||
# 播放次数
|
# 视频作者昵称
|
||||||
video_play_count = result["aweme_details"][0]['statistics']['play_count']
|
video_author_nickname = result["aweme_details"][0]['author']["nickname"]
|
||||||
# 下载次数
|
# 视频作者ID
|
||||||
video_download_count = result["aweme_details"][0]['statistics']['download_count']
|
video_author_id = result["aweme_details"][0]['author']["unique_id"]
|
||||||
# 分享次数
|
# 上传时间戳
|
||||||
video_share_count = result["aweme_details"][0]['statistics']['share_count']
|
video_create_time = result["aweme_details"][0]['create_time']
|
||||||
# 作者粉丝数量
|
# 视频ID
|
||||||
video_author_followerCount = video_info['authorStats']['followerCount']
|
video_aweme_id = result["aweme_details"][0]['statistics']['aweme_id']
|
||||||
# 作者关注数量
|
try:
|
||||||
video_author_followingCount = video_info['authorStats']['followingCount']
|
# 视频BGM标题
|
||||||
# 作者获赞数量
|
video_music_title = result["aweme_details"][0]['music']['title']
|
||||||
video_author_heartCount = video_info['authorStats']['heartCount']
|
# 视频BGM作者
|
||||||
# 作者视频数量
|
video_music_author = result["aweme_details"][0]['music']['author']
|
||||||
video_author_videoCount = video_info['authorStats']['videoCount']
|
# 视频BGM ID
|
||||||
# 作者已赞作品数量
|
video_music_id = result["aweme_details"][0]['music']['id']
|
||||||
video_author_diggCount = video_info['authorStats']['diggCount']
|
# 视频BGM链接
|
||||||
# 将话题保存在列表中
|
video_music_url = result["aweme_details"][0]['music']['play_url']['url_list'][0]
|
||||||
video_hashtags = []
|
except:
|
||||||
for tag in video_info['challenges']:
|
video_music_title, video_music_author, video_music_id, video_music_url = "None", "None", "None", "None"
|
||||||
video_hashtags.append(tag['title'])
|
# 评论数量
|
||||||
# 结束时间
|
video_comment_count = result["aweme_details"][0]['statistics']['comment_count']
|
||||||
end = time.time()
|
# 获赞数量
|
||||||
# 解析时间
|
video_digg_count = result["aweme_details"][0]['statistics']['digg_count']
|
||||||
analyze_time = format((end - start), '.4f')
|
# 播放次数
|
||||||
# 储存数据
|
video_play_count = result["aweme_details"][0]['statistics']['play_count']
|
||||||
video_date = {'status': 'success',
|
# 下载次数
|
||||||
'analyze_time': (analyze_time + 's'),
|
video_download_count = result["aweme_details"][0]['statistics']['download_count']
|
||||||
'url_type': url_type,
|
# 分享次数
|
||||||
'api_url': tiktok_api_link,
|
video_share_count = result["aweme_details"][0]['statistics']['share_count']
|
||||||
'original_url': original_url,
|
# 作者粉丝数量
|
||||||
'platform': 'tiktok',
|
video_author_followerCount = video_info['authorStats']['followerCount']
|
||||||
'video_title': video_title,
|
# 作者关注数量
|
||||||
'nwm_video_url': nwm_video_url,
|
video_author_followingCount = video_info['authorStats']['followingCount']
|
||||||
'wm_video_url': wm_video_url,
|
# 作者获赞数量
|
||||||
'video_author_nickname': video_author_nickname,
|
video_author_heartCount = video_info['authorStats']['heartCount']
|
||||||
'video_author_id': video_author_id,
|
# 作者视频数量
|
||||||
'video_create_time': video_create_time,
|
video_author_videoCount = video_info['authorStats']['videoCount']
|
||||||
'video_aweme_id': video_aweme_id,
|
# 作者已赞作品数量
|
||||||
'video_music_title': video_music_title,
|
video_author_diggCount = video_info['authorStats']['diggCount']
|
||||||
'video_music_author': video_music_author,
|
# 将话题保存在列表中
|
||||||
'video_music_id': video_music_id,
|
video_hashtags = []
|
||||||
'video_music_url': video_music_url,
|
for tag in video_info['challenges']:
|
||||||
'video_comment_count': video_comment_count,
|
video_hashtags.append(tag['title'])
|
||||||
'video_digg_count': video_digg_count,
|
# 结束时间
|
||||||
'video_play_count': video_play_count,
|
end = time.time()
|
||||||
'video_share_count': video_share_count,
|
# 解析时间
|
||||||
'video_download_count': video_download_count,
|
analyze_time = format((end - start), '.4f')
|
||||||
'video_author_followerCount': video_author_followerCount,
|
# 储存数据
|
||||||
'video_author_followingCount': video_author_followingCount,
|
video_data = {'status': 'success',
|
||||||
'video_author_heartCount': video_author_heartCount,
|
'analyze_time': (analyze_time + 's'),
|
||||||
'video_author_videoCount': video_author_videoCount,
|
'url_type': url_type,
|
||||||
'video_author_diggCount': video_author_diggCount,
|
'api_url': tiktok_api_link,
|
||||||
'video_hashtags': video_hashtags
|
'original_url': original_url,
|
||||||
}
|
'platform': 'tiktok',
|
||||||
# 返回包含数据的字典
|
'video_title': video_title,
|
||||||
return video_date
|
'nwm_video_url': nwm_video_url,
|
||||||
|
'wm_video_url': wm_video_url,
|
||||||
|
'video_author_nickname': video_author_nickname,
|
||||||
|
'video_author_id': video_author_id,
|
||||||
|
'video_create_time': video_create_time,
|
||||||
|
'video_aweme_id': video_aweme_id,
|
||||||
|
'video_music_title': video_music_title,
|
||||||
|
'video_music_author': video_music_author,
|
||||||
|
'video_music_id': video_music_id,
|
||||||
|
'video_music_url': video_music_url,
|
||||||
|
'video_comment_count': video_comment_count,
|
||||||
|
'video_digg_count': video_digg_count,
|
||||||
|
'video_play_count': video_play_count,
|
||||||
|
'video_share_count': video_share_count,
|
||||||
|
'video_download_count': video_download_count,
|
||||||
|
'video_author_followerCount': video_author_followerCount,
|
||||||
|
'video_author_followingCount': video_author_followingCount,
|
||||||
|
'video_author_heartCount': video_author_heartCount,
|
||||||
|
'video_author_videoCount': video_author_videoCount,
|
||||||
|
'video_author_diggCount': video_author_diggCount,
|
||||||
|
'video_hashtags': video_hashtags
|
||||||
|
}
|
||||||
|
# 返回包含数据的字典
|
||||||
|
return video_data
|
||||||
|
except:
|
||||||
|
# 判断链接是图集链接
|
||||||
|
# https://www.tiktok.com/@tamm6636/video/7105440975878655278
|
||||||
|
video_id = re.findall('video/(\d+)?', original_url)[0]
|
||||||
|
print('视频ID为: {}'.format(video_id))
|
||||||
|
# 从TikTok官方API获取部分视频数据
|
||||||
|
tiktok_api_link = 'https://api.tiktokv.com/aweme/v1/multi/aweme/detail/?aweme_ids=%5B{}%5D'.format(
|
||||||
|
video_id)
|
||||||
|
print('正在请求API链接:{}'.format(tiktok_api_link))
|
||||||
|
response = requests.get(url=tiktok_api_link, headers=headers).text
|
||||||
|
# 将API获取到的内容格式化为JSON
|
||||||
|
result = json.loads(response)
|
||||||
|
# 类型为视频
|
||||||
|
url_type = 'album'
|
||||||
|
# 视频标题
|
||||||
|
album_title = result["aweme_details"][0]["desc"]
|
||||||
|
# 视频作者昵称
|
||||||
|
album_author_nickname = result["aweme_details"][0]['author']["nickname"]
|
||||||
|
# 视频作者ID
|
||||||
|
album_author_id = result["aweme_details"][0]['author']["unique_id"]
|
||||||
|
# 上传时间戳
|
||||||
|
album_create_time = result["aweme_details"][0]['create_time']
|
||||||
|
# 视频ID
|
||||||
|
album_aweme_id = result["aweme_details"][0]['statistics']['aweme_id']
|
||||||
|
try:
|
||||||
|
# 视频BGM标题
|
||||||
|
album_music_title = result["aweme_details"][0]['music']['title']
|
||||||
|
# 视频BGM作者
|
||||||
|
album_music_author = result["aweme_details"][0]['music']['author']
|
||||||
|
# 视频BGM ID
|
||||||
|
album_music_id = result["aweme_details"][0]['music']['id']
|
||||||
|
# 视频BGM链接
|
||||||
|
album_music_url = result["aweme_details"][0]['music']['play_url']['url_list'][0]
|
||||||
|
except:
|
||||||
|
album_music_title, album_music_author, album_music_id, album_music_url = "None", "None", "None", "None"
|
||||||
|
# 评论数量
|
||||||
|
album_comment_count = result["aweme_details"][0]['statistics']['comment_count']
|
||||||
|
# 获赞数量
|
||||||
|
album_digg_count = result["aweme_details"][0]['statistics']['digg_count']
|
||||||
|
# 播放次数
|
||||||
|
album_play_count = result["aweme_details"][0]['statistics']['play_count']
|
||||||
|
# 下载次数
|
||||||
|
album_download_count = result["aweme_details"][0]['statistics']['download_count']
|
||||||
|
# 分享次数
|
||||||
|
album_share_count = result["aweme_details"][0]['statistics']['share_count']
|
||||||
|
# 无水印图集
|
||||||
|
album_list = []
|
||||||
|
for i in result["aweme_details"][0]['image_post_info']['images'][0]['display_image']['url_list']:
|
||||||
|
album_list.append(i)
|
||||||
|
# 结束时间
|
||||||
|
end = time.time()
|
||||||
|
# 解析时间
|
||||||
|
analyze_time = format((end - start), '.4f')
|
||||||
|
# 储存数据
|
||||||
|
album_data = {'status': 'success',
|
||||||
|
'analyze_time': (analyze_time + 's'),
|
||||||
|
'url_type': url_type,
|
||||||
|
'api_url': tiktok_api_link,
|
||||||
|
'original_url': original_url,
|
||||||
|
'platform': 'tiktok',
|
||||||
|
'album_title': album_title,
|
||||||
|
'album_list': album_list,
|
||||||
|
'album_author_nickname': album_author_nickname,
|
||||||
|
'album_author_id': album_author_id,
|
||||||
|
'album_create_time': album_create_time,
|
||||||
|
'album_aweme_id': album_aweme_id,
|
||||||
|
'album_music_title': album_music_title,
|
||||||
|
'album_music_author': album_music_author,
|
||||||
|
'album_music_id': album_music_id,
|
||||||
|
'album_music_url': album_music_url,
|
||||||
|
'album_comment_count': album_comment_count,
|
||||||
|
'album_digg_count': album_digg_count,
|
||||||
|
'album_play_count': album_play_count,
|
||||||
|
'album_share_count': album_share_count,
|
||||||
|
'album_download_count': album_download_count
|
||||||
|
}
|
||||||
|
# 返回包含数据的字典
|
||||||
|
return album_data
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
# 异常捕获
|
# 异常捕获
|
||||||
return {'status': 'failed', 'reason': e, 'function': 'Scraper.tiktok()', 'value': original_url}
|
return {'status': 'failed', 'reason': e, 'function': 'Scraper.tiktok()', 'value': original_url}
|
||||||
@ -425,5 +510,3 @@ if __name__ == '__main__':
|
|||||||
print(tiktok_date)
|
print(tiktok_date)
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
print("Error: " + str(e))
|
print("Error: " + str(e))
|
||||||
|
|
||||||
|
|
||||||
|
|||||||
Loading…
x
Reference in New Issue
Block a user