yt_dlp/extractor/sapo.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     parse_duration,
   6     unified_strdate,
   7 )
   8
   9
  10 class SapoIE(InfoExtractor):
  11     IE_DESC = 'SAPO Vídeos'
  12     _VALID_URL = r'https?://(?:(?:v2|www)\.)?videos\.sapo\.(?:pt|cv|ao|mz|tl)/(?P<id>[\da-zA-Z]{20})'
  13
  14     _TESTS = [
  15         {
  16             'url': 'http://videos.sapo.pt/UBz95kOtiWYUMTA5Ghfi',
  17             'md5': '79ee523f6ecb9233ac25075dee0eda83',
  18             'note': 'SD video',
  19             'info_dict': {
  20                 'id': 'UBz95kOtiWYUMTA5Ghfi',
  21                 'ext': 'mp4',
  22                 'title': 'Benfica - Marcas na Hitória',
  23                 'description': 'md5:c9082000a128c3fd57bf0299e1367f22',
  24                 'duration': 264,
  25                 'uploader': 'tiago_1988',
  26                 'upload_date': '20080229',
  27                 'categories': ['benfica', 'cabral', 'desporto', 'futebol', 'geovanni', 'hooijdonk', 'joao', 'karel', 'lisboa', 'miccoli'],
  28             },
  29         },
  30         {
  31             'url': 'http://videos.sapo.pt/IyusNAZ791ZdoCY5H5IF',
  32             'md5': '90a2f283cfb49193fe06e861613a72aa',
  33             'note': 'HD video',
  34             'info_dict': {
  35                 'id': 'IyusNAZ791ZdoCY5H5IF',
  36                 'ext': 'mp4',
  37                 'title': 'Codebits VII - Report',
  38                 'description': 'md5:6448d6fd81ce86feac05321f354dbdc8',
  39                 'duration': 144,
  40                 'uploader': 'codebits',
  41                 'upload_date': '20140427',
  42                 'categories': ['codebits', 'codebits2014'],
  43             },
  44         },
  45         {
  46             'url': 'http://v2.videos.sapo.pt/yLqjzPtbTimsn2wWBKHz',
  47             'md5': 'e5aa7cc0bdc6db9b33df1a48e49a15ac',
  48             'note': 'v2 video',
  49             'info_dict': {
  50                 'id': 'yLqjzPtbTimsn2wWBKHz',
  51                 'ext': 'mp4',
  52                 'title': 'Hipnose Condicionativa 4',
  53                 'description': 'md5:ef0481abf8fb4ae6f525088a6dadbc40',
  54                 'duration': 692,
  55                 'uploader': 'sapozen',
  56                 'upload_date': '20090609',
  57                 'categories': ['condicionativa', 'heloisa', 'hipnose', 'miranda', 'sapo', 'zen'],
  58             },
  59         },
  60     ]
  61
  62     def _real_extract(self, url):
  63         mobj = self._match_valid_url(url)
  64         video_id = mobj.group('id')
  65
  66         item = self._download_xml(
  67             f'http://rd3.videos.sapo.pt/{video_id}/rss2', video_id).find('./channel/item')
  68
  69         title = item.find('./title').text
  70         description = item.find('./{http://videos.sapo.pt/mrss/}synopse').text
  71         thumbnail = item.find('./{http://search.yahoo.com/mrss/}content').get('url')
  72         duration = parse_duration(item.find('./{http://videos.sapo.pt/mrss/}time').text)
  73         uploader = item.find('./{http://videos.sapo.pt/mrss/}author').text
  74         upload_date = unified_strdate(item.find('./pubDate').text)
  75         view_count = int(item.find('./{http://videos.sapo.pt/mrss/}views').text)
  76         comment_count = int(item.find('./{http://videos.sapo.pt/mrss/}comment_count').text)
  77         tags = item.find('./{http://videos.sapo.pt/mrss/}tags').text
  78         categories = tags.split() if tags else []
  79         age_limit = 18 if item.find('./{http://videos.sapo.pt/mrss/}m18').text == 'true' else 0
  80
  81         video_url = item.find('./{http://videos.sapo.pt/mrss/}videoFile').text
  82         video_size = item.find('./{http://videos.sapo.pt/mrss/}videoSize').text.split('x')
  83
  84         formats = [{
  85             'url': video_url,
  86             'ext': 'mp4',
  87             'format_id': 'sd',
  88             'width': int(video_size[0]),
  89             'height': int(video_size[1]),
  90         }]
  91
  92         if item.find('./{http://videos.sapo.pt/mrss/}HD').text == 'true':
  93             formats.append({
  94                 'url': re.sub(r'/mov/1$', '/mov/39', video_url),
  95                 'ext': 'mp4',
  96                 'format_id': 'hd',
  97                 'width': 1280,
  98                 'height': 720,
  99             })
 100
 101         return {
 102             'id': video_id,
 103             'title': title,
 104             'description': description,
 105             'thumbnail': thumbnail,
 106             'duration': duration,
 107             'uploader': uploader,
 108             'upload_date': upload_date,
 109             'view_count': view_count,
 110             'comment_count': comment_count,
 111             'categories': categories,
 112             'age_limit': age_limit,
 113             'formats': formats,
 114         }