shared.py 3.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116
  1. from __future__ import unicode_literals
  2. import re
  3. from .common import InfoExtractor
  4. from ..compat import compat_b64decode
  5. from ..utils import (
  6. ExtractorError,
  7. int_or_none,
  8. url_or_none,
  9. urlencode_postdata,
  10. unescapeHTML,
  11. )
  12. class SharedBaseIE(InfoExtractor):
  13. def _real_extract(self, url):
  14. video_id = self._match_id(url)
  15. webpage, urlh = self._download_webpage_handle(url, video_id)
  16. if self._FILE_NOT_FOUND in webpage:
  17. raise ExtractorError(
  18. 'Video %s does not exist' % video_id, expected=True)
  19. video_url = self._extract_video_url(webpage, video_id, url)
  20. title = self._extract_title(webpage)
  21. filesize = int_or_none(self._html_search_meta(
  22. 'full:size', webpage, 'file size', fatal=False))
  23. return {
  24. 'id': video_id,
  25. 'url': video_url,
  26. 'ext': 'mp4',
  27. 'filesize': filesize,
  28. 'title': title,
  29. }
  30. def _extract_title(self, webpage):
  31. return compat_b64decode(self._html_search_meta(
  32. 'full:title', webpage, 'title')).decode('utf-8')
  33. class SharedIE(SharedBaseIE):
  34. IE_DESC = 'shared.sx'
  35. _VALID_URL = r'https?://shared\.sx/(?P<id>[\da-z]{10})'
  36. _FILE_NOT_FOUND = '>File does not exist<'
  37. _TEST = {
  38. 'url': 'http://shared.sx/0060718775',
  39. 'md5': '106fefed92a8a2adb8c98e6a0652f49b',
  40. 'info_dict': {
  41. 'id': '0060718775',
  42. 'ext': 'mp4',
  43. 'title': 'Bmp4',
  44. 'filesize': 1720110,
  45. },
  46. }
  47. def _extract_video_url(self, webpage, video_id, url):
  48. download_form = self._hidden_inputs(webpage)
  49. video_page = self._download_webpage(
  50. url, video_id, 'Downloading video page',
  51. data=urlencode_postdata(download_form),
  52. headers={
  53. 'Content-Type': 'application/x-www-form-urlencoded',
  54. 'Referer': url,
  55. })
  56. video_url = self._html_search_regex(
  57. r'data-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
  58. video_page, 'video URL', group='url')
  59. return video_url
  60. class VivoIE(SharedBaseIE):
  61. IE_DESC = 'vivo.sx'
  62. _VALID_URL = r'https?://vivo\.sx/(?P<id>[\da-z]{10})'
  63. _FILE_NOT_FOUND = '>The file you have requested does not exists or has been removed'
  64. _TEST = {
  65. 'url': 'http://vivo.sx/d7ddda0e78',
  66. 'md5': '15b3af41be0b4fe01f4df075c2678b2c',
  67. 'info_dict': {
  68. 'id': 'd7ddda0e78',
  69. 'ext': 'mp4',
  70. 'title': 'Chicken',
  71. 'filesize': 528031,
  72. },
  73. }
  74. def _extract_title(self, webpage):
  75. data_title = self._search_regex(
  76. r'data-name\s*=\s*(["\'])(?P<title>(?:(?!\1).)+)\1', webpage,
  77. 'title', default=None, group='title')
  78. if data_title:
  79. return unescapeHTML(re.sub(r"\.[a-z0-9]{3,4}$", "", data_title))
  80. return self._og_search_title(webpage)
  81. def _extract_video_url(self, webpage, video_id, *args):
  82. def decode_url(encoded_url):
  83. return compat_b64decode(encoded_url).decode('utf-8')
  84. stream_url = url_or_none(decode_url(self._search_regex(
  85. r'data-stream\s*=\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
  86. 'stream url', default=None, group='url')))
  87. if stream_url:
  88. return stream_url
  89. return self._parse_json(
  90. self._search_regex(
  91. r'InitializeStream\s*\(\s*(["\'])(?P<url>(?:(?!\1).)+)\1',
  92. webpage, 'stream', group='url'),
  93. video_id, transform_source=decode_url)[0]