You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

191 lines
6.7 KiB

  1. # coding: utf-8
  2. from __future__ import unicode_literals
  3. import re
  4. from .common import InfoExtractor
  5. from ..compat import compat_str
  6. from ..utils import (
  7. float_or_none,
  8. try_get,
  9. unified_timestamp,
  10. )
  11. class CCTVIE(InfoExtractor):
  12. IE_DESC = '央视网'
  13. _VALID_URL = r'https?://(?:(?:[^/]+)\.(?:cntv|cctv)\.(?:com|cn)|(?:www\.)?ncpa-classic\.com)/(?:[^/]+/)*?(?P<id>[^/?#&]+?)(?:/index)?(?:\.s?html|[?#&]|$)'
  14. _TESTS = [{
  15. # fo.addVariable("videoCenterId","id")
  16. 'url': 'http://sports.cntv.cn/2016/02/12/ARTIaBRxv4rTT1yWf1frW2wi160212.shtml',
  17. 'md5': 'd61ec00a493e09da810bf406a078f691',
  18. 'info_dict': {
  19. 'id': '5ecdbeab623f4973b40ff25f18b174e8',
  20. 'ext': 'mp4',
  21. 'title': '[NBA]二少联手砍下46分 雷霆主场击败鹈鹕(快讯)',
  22. 'description': 'md5:7e14a5328dc5eb3d1cd6afbbe0574e95',
  23. 'duration': 98,
  24. 'uploader': 'songjunjie',
  25. 'timestamp': 1455279956,
  26. 'upload_date': '20160212',
  27. },
  28. }, {
  29. # var guid = "id"
  30. 'url': 'http://tv.cctv.com/2016/02/05/VIDEUS7apq3lKrHG9Dncm03B160205.shtml',
  31. 'info_dict': {
  32. 'id': 'efc5d49e5b3b4ab2b34f3a502b73d3ae',
  33. 'ext': 'mp4',
  34. 'title': '[赛车]“车王”舒马赫恢复情况成谜(快讯)',
  35. 'description': '2月4日,蒙特泽莫罗透露了关于“车王”舒马赫恢复情况,但情况是否属实遭到了质疑。',
  36. 'duration': 37,
  37. 'uploader': 'shujun',
  38. 'timestamp': 1454677291,
  39. 'upload_date': '20160205',
  40. },
  41. 'params': {
  42. 'skip_download': True,
  43. },
  44. }, {
  45. # changePlayer('id')
  46. 'url': 'http://english.cntv.cn/special/four_comprehensives/index.shtml',
  47. 'info_dict': {
  48. 'id': '4bb9bb4db7a6471ba85fdeda5af0381e',
  49. 'ext': 'mp4',
  50. 'title': 'NHnews008 ANNUAL POLITICAL SEASON',
  51. 'description': 'Four Comprehensives',
  52. 'duration': 60,
  53. 'uploader': 'zhangyunlei',
  54. 'timestamp': 1425385521,
  55. 'upload_date': '20150303',
  56. },
  57. 'params': {
  58. 'skip_download': True,
  59. },
  60. }, {
  61. # loadvideo('id')
  62. 'url': 'http://cctv.cntv.cn/lm/tvseries_russian/yilugesanghua/index.shtml',
  63. 'info_dict': {
  64. 'id': 'b15f009ff45c43968b9af583fc2e04b2',
  65. 'ext': 'mp4',
  66. 'title': 'Путь,усыпанный космеями Серия 1',
  67. 'description': 'Путь, усыпанный космеями',
  68. 'duration': 2645,
  69. 'uploader': 'renxue',
  70. 'timestamp': 1477479241,
  71. 'upload_date': '20161026',
  72. },
  73. 'params': {
  74. 'skip_download': True,
  75. },
  76. }, {
  77. # var initMyAray = 'id'
  78. 'url': 'http://www.ncpa-classic.com/2013/05/22/VIDE1369219508996867.shtml',
  79. 'info_dict': {
  80. 'id': 'a194cfa7f18c426b823d876668325946',
  81. 'ext': 'mp4',
  82. 'title': '小泽征尔音乐塾 音乐梦想无国界',
  83. 'duration': 2173,
  84. 'timestamp': 1369248264,
  85. 'upload_date': '20130522',
  86. },
  87. 'params': {
  88. 'skip_download': True,
  89. },
  90. }, {
  91. # var ids = ["id"]
  92. 'url': 'http://www.ncpa-classic.com/clt/more/416/index.shtml',
  93. 'info_dict': {
  94. 'id': 'a8606119a4884588a79d81c02abecc16',
  95. 'ext': 'mp3',
  96. 'title': '来自维也纳的新年贺礼',
  97. 'description': 'md5:f13764ae8dd484e84dd4b39d5bcba2a7',
  98. 'duration': 1578,
  99. 'uploader': 'djy',
  100. 'timestamp': 1482942419,
  101. 'upload_date': '20161228',
  102. },
  103. 'params': {
  104. 'skip_download': True,
  105. },
  106. 'expected_warnings': ['Failed to download m3u8 information'],
  107. }, {
  108. 'url': 'http://ent.cntv.cn/2016/01/18/ARTIjprSSJH8DryTVr5Bx8Wb160118.shtml',
  109. 'only_matching': True,
  110. }, {
  111. 'url': 'http://tv.cntv.cn/video/C39296/e0210d949f113ddfb38d31f00a4e5c44',
  112. 'only_matching': True,
  113. }, {
  114. 'url': 'http://english.cntv.cn/2016/09/03/VIDEhnkB5y9AgHyIEVphCEz1160903.shtml',
  115. 'only_matching': True,
  116. }, {
  117. 'url': 'http://tv.cctv.com/2016/09/07/VIDE5C1FnlX5bUywlrjhxXOV160907.shtml',
  118. 'only_matching': True,
  119. }, {
  120. 'url': 'http://tv.cntv.cn/video/C39296/95cfac44cabd3ddc4a9438780a4e5c44',
  121. 'only_matching': True,
  122. }]
  123. def _real_extract(self, url):
  124. video_id = self._match_id(url)
  125. webpage = self._download_webpage(url, video_id)
  126. video_id = self._search_regex(
  127. [r'var\s+guid\s*=\s*["\']([\da-fA-F]+)',
  128. r'videoCenterId["\']\s*,\s*["\']([\da-fA-F]+)',
  129. r'changePlayer\s*\(\s*["\']([\da-fA-F]+)',
  130. r'load[Vv]ideo\s*\(\s*["\']([\da-fA-F]+)',
  131. r'var\s+initMyAray\s*=\s*["\']([\da-fA-F]+)',
  132. r'var\s+ids\s*=\s*\[["\']([\da-fA-F]+)'],
  133. webpage, 'video id')
  134. data = self._download_json(
  135. 'http://vdn.apps.cntv.cn/api/getHttpVideoInfo.do', video_id,
  136. query={
  137. 'pid': video_id,
  138. 'url': url,
  139. 'idl': 32,
  140. 'idlr': 32,
  141. 'modifyed': 'false',
  142. })
  143. title = data['title']
  144. formats = []
  145. video = data.get('video')
  146. if isinstance(video, dict):
  147. for quality, chapters_key in enumerate(('lowChapters', 'chapters')):
  148. video_url = try_get(
  149. video, lambda x: x[chapters_key][0]['url'], compat_str)
  150. if video_url:
  151. formats.append({
  152. 'url': video_url,
  153. 'format_id': 'http',
  154. 'quality': quality,
  155. 'preference': -1,
  156. })
  157. hls_url = try_get(data, lambda x: x['hls_url'], compat_str)
  158. if hls_url:
  159. hls_url = re.sub(r'maxbr=\d+&?', '', hls_url)
  160. formats.extend(self._extract_m3u8_formats(
  161. hls_url, video_id, 'mp4', entry_protocol='m3u8_native',
  162. m3u8_id='hls', fatal=False))
  163. self._sort_formats(formats)
  164. uploader = data.get('editer_name')
  165. description = self._html_search_meta(
  166. 'description', webpage, default=None)
  167. timestamp = unified_timestamp(data.get('f_pgmtime'))
  168. duration = float_or_none(try_get(video, lambda x: x['totalLength']))
  169. return {
  170. 'id': video_id,
  171. 'title': title,
  172. 'description': description,
  173. 'uploader': uploader,
  174. 'timestamp': timestamp,
  175. 'duration': duration,
  176. 'formats': formats,
  177. }