r7.py 4.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109
  1. from .common import InfoExtractor
  2. from ..utils import int_or_none
  3. class R7IE(InfoExtractor):
  4. _VALID_URL = r'''(?x)
  5. https?://
  6. (?:
  7. (?:[a-zA-Z]+)\.r7\.com(?:/[^/]+)+/idmedia/|
  8. noticias\.r7\.com(?:/[^/]+)+/[^/]+-|
  9. player\.r7\.com/video/i/
  10. )
  11. (?P<id>[\da-f]{24})
  12. '''
  13. _TESTS = [{
  14. 'url': 'http://videos.r7.com/policiais-humilham-suspeito-a-beira-da-morte-morre-com-dignidade-/idmedia/54e7050b0cf2ff57e0279389.html',
  15. 'md5': '403c4e393617e8e8ddc748978ee8efde',
  16. 'info_dict': {
  17. 'id': '54e7050b0cf2ff57e0279389',
  18. 'ext': 'mp4',
  19. 'title': 'Policiais humilham suspeito à beira da morte: "Morre com dignidade"',
  20. 'description': 'md5:01812008664be76a6479aa58ec865b72',
  21. 'thumbnail': r're:^https?://.*\.jpg$',
  22. 'duration': 98,
  23. 'like_count': int,
  24. 'view_count': int,
  25. },
  26. }, {
  27. 'url': 'http://esportes.r7.com/videos/cigano-manda-recado-aos-fas/idmedia/4e176727b51a048ee6646a1b.html',
  28. 'only_matching': True,
  29. }, {
  30. 'url': 'http://noticias.r7.com/record-news/video/representante-do-instituto-sou-da-paz-fala-sobre-fim-do-estatuto-do-desarmamento-5480fc580cf2285b117f438d/',
  31. 'only_matching': True,
  32. }, {
  33. 'url': 'http://player.r7.com/video/i/54e7050b0cf2ff57e0279389?play=true&video=http://vsh.r7.com/54e7050b0cf2ff57e0279389/ER7_RE_BG_MORTE_JOVENS_570kbps_2015-02-2009f17818-cc82-4c8f-86dc-89a66934e633-ATOS_copy.mp4&linkCallback=http://videos.r7.com/policiais-humilham-suspeito-a-beira-da-morte-morre-com-dignidade-/idmedia/54e7050b0cf2ff57e0279389.html&thumbnail=http://vtb.r7.com/ER7_RE_BG_MORTE_JOVENS_570kbps_2015-02-2009f17818-cc82-4c8f-86dc-89a66934e633-thumb.jpg&idCategory=192&share=true&layout=full&full=true',
  34. 'only_matching': True,
  35. }]
  36. def _real_extract(self, url):
  37. video_id = self._match_id(url)
  38. video = self._download_json(
  39. 'http://player-api.r7.com/video/i/%s' % video_id, video_id)
  40. title = video['title']
  41. formats = []
  42. media_url_hls = video.get('media_url_hls')
  43. if media_url_hls:
  44. formats.extend(self._extract_m3u8_formats(
  45. media_url_hls, video_id, 'mp4', entry_protocol='m3u8_native',
  46. m3u8_id='hls', fatal=False))
  47. media_url = video.get('media_url')
  48. if media_url:
  49. f = {
  50. 'url': media_url,
  51. 'format_id': 'http',
  52. }
  53. # m3u8 format always matches the http format, let's copy metadata from
  54. # one to another
  55. m3u8_formats = list(filter(
  56. lambda f: f.get('vcodec') != 'none', formats))
  57. if len(m3u8_formats) == 1:
  58. f_copy = m3u8_formats[0].copy()
  59. f_copy.update(f)
  60. f_copy['protocol'] = 'http'
  61. f = f_copy
  62. formats.append(f)
  63. description = video.get('description')
  64. thumbnail = video.get('thumb')
  65. duration = int_or_none(video.get('media_duration'))
  66. like_count = int_or_none(video.get('likes'))
  67. view_count = int_or_none(video.get('views'))
  68. return {
  69. 'id': video_id,
  70. 'title': title,
  71. 'description': description,
  72. 'thumbnail': thumbnail,
  73. 'duration': duration,
  74. 'like_count': like_count,
  75. 'view_count': view_count,
  76. 'formats': formats,
  77. }
  78. class R7ArticleIE(InfoExtractor):
  79. _VALID_URL = r'https?://(?:[a-zA-Z]+)\.r7\.com/(?:[^/]+/)+[^/?#&]+-(?P<id>\d+)'
  80. _TEST = {
  81. 'url': 'http://tv.r7.com/record-play/balanco-geral/videos/policiais-humilham-suspeito-a-beira-da-morte-morre-com-dignidade-16102015',
  82. 'only_matching': True,
  83. }
  84. @classmethod
  85. def suitable(cls, url):
  86. return False if R7IE.suitable(url) else super(R7ArticleIE, cls).suitable(url)
  87. def _real_extract(self, url):
  88. display_id = self._match_id(url)
  89. webpage = self._download_webpage(url, display_id)
  90. video_id = self._search_regex(
  91. r'<div[^>]+(?:id=["\']player-|class=["\']embed["\'][^>]+id=["\'])([\da-f]{24})',
  92. webpage, 'video id')
  93. return self.url_result('http://player.r7.com/video/i/%s' % video_id, R7IE.ie_key())