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.

173 lines
6.9 KiB

10 years ago
10 years ago
10 years ago
10 years ago
  1. from __future__ import unicode_literals
  2. import re
  3. from .common import InfoExtractor
  4. from ..utils import (
  5. parse_duration,
  6. parse_iso8601,
  7. )
  8. class MLBIE(InfoExtractor):
  9. _VALID_URL = r'''(?x)
  10. https?://
  11. (?:[\da-z_-]+\.)*mlb\.com/
  12. (?:
  13. (?:
  14. (?:.*?/)?video/(?:topic/[\da-z_-]+/)?v|
  15. (?:
  16. shared/video/embed/(?:embed|m-internal-embed)\.html|
  17. (?:[^/]+/)+(?:play|index)\.jsp|
  18. )\?.*?\bcontent_id=
  19. )
  20. (?P<id>n?\d+)|
  21. (?:[^/]+/)*(?P<path>[^/]+)
  22. )
  23. '''
  24. _TESTS = [
  25. {
  26. 'url': 'http://m.mlb.com/sea/video/topic/51231442/v34698933/nymsea-ackley-robs-a-home-run-with-an-amazing-catch/?c_id=sea',
  27. 'md5': 'ff56a598c2cf411a9a38a69709e97079',
  28. 'info_dict': {
  29. 'id': '34698933',
  30. 'ext': 'mp4',
  31. 'title': "Ackley's spectacular catch",
  32. 'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
  33. 'duration': 66,
  34. 'timestamp': 1405980600,
  35. 'upload_date': '20140721',
  36. 'thumbnail': 're:^https?://.*\.jpg$',
  37. },
  38. },
  39. {
  40. 'url': 'http://m.mlb.com/video/topic/81536970/v34496663/mianym-stanton-practices-for-the-home-run-derby',
  41. 'md5': 'd9c022c10d21f849f49c05ae12a8a7e9',
  42. 'info_dict': {
  43. 'id': '34496663',
  44. 'ext': 'mp4',
  45. 'title': 'Stanton prepares for Derby',
  46. 'description': 'md5:d00ce1e5fd9c9069e9c13ab4faedfa57',
  47. 'duration': 46,
  48. 'timestamp': 1405105800,
  49. 'upload_date': '20140711',
  50. 'thumbnail': 're:^https?://.*\.jpg$',
  51. },
  52. },
  53. {
  54. 'url': 'http://m.mlb.com/video/topic/vtp_hrd_sponsor/v34578115/hrd-cespedes-wins-2014-gillette-home-run-derby',
  55. 'md5': '0e6e73d509321e142409b695eadd541f',
  56. 'info_dict': {
  57. 'id': '34578115',
  58. 'ext': 'mp4',
  59. 'title': 'Cespedes repeats as Derby champ',
  60. 'description': 'md5:08df253ce265d4cf6fb09f581fafad07',
  61. 'duration': 488,
  62. 'timestamp': 1405399936,
  63. 'upload_date': '20140715',
  64. 'thumbnail': 're:^https?://.*\.jpg$',
  65. },
  66. },
  67. {
  68. 'url': 'http://m.mlb.com/video/v34577915/bautista-on-derby-captaining-duties-his-performance',
  69. 'md5': 'b8fd237347b844365d74ea61d4245967',
  70. 'info_dict': {
  71. 'id': '34577915',
  72. 'ext': 'mp4',
  73. 'title': 'Bautista on Home Run Derby',
  74. 'description': 'md5:b80b34031143d0986dddc64a8839f0fb',
  75. 'duration': 52,
  76. 'timestamp': 1405390722,
  77. 'upload_date': '20140715',
  78. 'thumbnail': 're:^https?://.*\.jpg$',
  79. },
  80. },
  81. {
  82. 'url': 'http://m.mlb.com/news/article/118550098/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer',
  83. 'md5': 'b190e70141fb9a1552a85426b4da1b5d',
  84. 'info_dict': {
  85. 'id': '75609783',
  86. 'ext': 'mp4',
  87. 'title': 'Must C: Pillar climbs for catch',
  88. 'description': '4/15/15: Blue Jays outfielder Kevin Pillar continues his defensive dominance by climbing the wall in left to rob Tim Beckham of a home run',
  89. 'timestamp': 1429124820,
  90. 'upload_date': '20150415',
  91. }
  92. },
  93. {
  94. 'url': 'http://m.mlb.com/shared/video/embed/embed.html?content_id=35692085&topic_id=6479266&width=400&height=224&property=mlb',
  95. 'only_matching': True,
  96. },
  97. {
  98. 'url': 'http://mlb.mlb.com/shared/video/embed/embed.html?content_id=36599553',
  99. 'only_matching': True,
  100. },
  101. {
  102. 'url': 'http://mlb.mlb.com/es/video/play.jsp?content_id=36599553',
  103. 'only_matching': True,
  104. },
  105. {
  106. 'url': 'http://m.cardinals.mlb.com/stl/video/v51175783/atlstl-piscotty-makes-great-sliding-catch-on-line/?partnerId=as_mlb_20150321_42500876&adbid=579409712979910656&adbpl=tw&adbpr=52847728',
  107. 'only_matching': True,
  108. },
  109. {
  110. # From http://m.mlb.com/news/article/118550098/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer
  111. 'url': 'http://mlb.mlb.com/shared/video/embed/m-internal-embed.html?content_id=75609783&property=mlb&autoplay=true&hashmode=false&siteSection=mlb/multimedia/article_118550098/article_embed&club=mlb',
  112. 'only_matching': True,
  113. },
  114. {
  115. 'url': 'http://washington.nationals.mlb.com/mlb/gameday/index.jsp?c_id=was&gid=2015_05_09_atlmlb_wasmlb_1&lang=en&content_id=108309983&mode=video#',
  116. 'only_matching': True,
  117. }
  118. ]
  119. def _real_extract(self, url):
  120. mobj = re.match(self._VALID_URL, url)
  121. video_id = mobj.group('id')
  122. if not video_id:
  123. video_path = mobj.group('path')
  124. webpage = self._download_webpage(url, video_path)
  125. video_id = self._search_regex(
  126. [r'data-video-?id="(\d+)"', r'content_id=(\d+)'], webpage, 'video id')
  127. detail = self._download_xml(
  128. 'http://m.mlb.com/gen/multimedia/detail/%s/%s/%s/%s.xml'
  129. % (video_id[-3], video_id[-2], video_id[-1], video_id), video_id)
  130. title = detail.find('./headline').text
  131. description = detail.find('./big-blurb').text
  132. duration = parse_duration(detail.find('./duration').text)
  133. timestamp = parse_iso8601(detail.attrib['date'][:-5])
  134. thumbnails = [{
  135. 'url': thumbnail.text,
  136. } for thumbnail in detail.findall('./thumbnailScenarios/thumbnailScenario')]
  137. formats = []
  138. for media_url in detail.findall('./url'):
  139. playback_scenario = media_url.attrib['playback_scenario']
  140. fmt = {
  141. 'url': media_url.text,
  142. 'format_id': playback_scenario,
  143. }
  144. m = re.search(r'(?P<vbr>\d+)K_(?P<width>\d+)X(?P<height>\d+)', playback_scenario)
  145. if m:
  146. fmt.update({
  147. 'vbr': int(m.group('vbr')) * 1000,
  148. 'width': int(m.group('width')),
  149. 'height': int(m.group('height')),
  150. })
  151. formats.append(fmt)
  152. self._sort_formats(formats)
  153. return {
  154. 'id': video_id,
  155. 'title': title,
  156. 'description': description,
  157. 'duration': duration,
  158. 'timestamp': timestamp,
  159. 'formats': formats,
  160. 'thumbnails': thumbnails,
  161. }