[mit] Fix ocw tests

This commit is contained in:
Philipp Hagemeister 2014-02-26 00:29:45 +01:00
parent 2437fbca64
commit 7216de55d6

View File

@ -4,6 +4,7 @@ import re
import json import json
from .common import InfoExtractor from .common import InfoExtractor
from .youtube import YoutubeIE
from ..utils import ( from ..utils import (
compat_urlparse, compat_urlparse,
clean_html, clean_html,
@ -89,40 +90,47 @@ class MITIE(TechTVMITIE):
r'<iframe .*?src="(.+?)"', webpage, 'embed url') r'<iframe .*?src="(.+?)"', webpage, 'embed url')
return self.url_result(embed_url, ie='TechTVMIT') return self.url_result(embed_url, ie='TechTVMIT')
class OCWMITIE(InfoExtractor): class OCWMITIE(InfoExtractor):
IE_NAME = u'ocw.mit.edu' IE_NAME = 'ocw.mit.edu'
_VALID_URL = r'^http://ocw\.mit\.edu/courses/(?P<topic>[a-z0-9\-]+)' _VALID_URL = r'^http://ocw\.mit\.edu/courses/(?P<topic>[a-z0-9\-]+)'
_BASE_URL = u'http://ocw.mit.edu/' _BASE_URL = 'http://ocw.mit.edu/'
_TESTS = [ _TESTS = [
{ {
u'url': u'http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/video-lectures/lecture-7-multiple-variables-expectations-independence/', 'url': 'http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/video-lectures/lecture-7-multiple-variables-expectations-independence/',
u'md5': u'348bef727b573c0bd9ad8a7c08c89ebd', 'info_dict': {
u'info_dict': { 'id': 'EObHWIEKGjA',
u'title': u'7. Discrete Random Variables III', 'ext': 'mp4',
u'description': u'In this lecture, the professor discussed multiple random variables, expectations, and binomial distribution.', 'title': 'Lecture 7: Multiple Discrete Random Variables: Expectations, Conditioning, Independence',
u'subtitles': u'http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/video-lectures/lecture-7-multiple-variables-expectations-independence/MIT6_041F11_lec07_300k.mp4.srt' 'description': 'In this lecture, the professor discussed multiple random variables, expectations, and binomial distribution.',
#'subtitles': 'http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/video-lectures/lecture-7-multiple-variables-expectations-independence/MIT6_041F11_lec07_300k.mp4.srt'
} }
}, },
{ {
u'url': u'http://ocw.mit.edu/courses/mathematics/18-01sc-single-variable-calculus-fall-2010/1.-differentiation/part-a-definition-and-basic-rules/session-1-introduction-to-derivatives/', 'url': 'http://ocw.mit.edu/courses/mathematics/18-01sc-single-variable-calculus-fall-2010/1.-differentiation/part-a-definition-and-basic-rules/session-1-introduction-to-derivatives/',
u'md5': u'f4a434f08f15e581eb67cec0b57bcf6f', 'info_dict': {
u'info_dict': { 'id': '7K1sB05pE0A',
u'title': u'Lec 1 _ MIT 18.01 Single Variable Calculus, Fall 2007', 'ext': 'mp4',
u'subtitles': u'http://ocw.mit.edu//courses/mathematics/18-01sc-single-variable-calculus-fall-2010/ocw-18.01-f07-lec01_300k.SRT' 'title': 'Session 1: Introduction to Derivatives',
'description': 'This section contains lecture video excerpts, lecture notes, an interactive mathlet with supporting documents, and problem solving videos.',
#'subtitles': 'http://ocw.mit.edu//courses/mathematics/18-01sc-single-variable-calculus-fall-2010/ocw-18.01-f07-lec01_300k.SRT'
} }
} }
] ]
def _real_extract(self, url): def _real_extract(self, url):
webpage = self._download_webpage(url, self.IE_NAME) mobj = re.match(self._VALID_URL, url)
title = self._html_search_meta(u'WT.cg_s', webpage) topic = mobj.group('topic')
description = self._html_search_meta(u'Description', webpage)
webpage = self._download_webpage(url, topic)
title = self._html_search_meta('WT.cg_s', webpage)
description = self._html_search_meta('Description', webpage)
# search for call to ocw_embed_chapter_media(container_id, media_url, provider, page_url, image_url, start, stop, captions_file) # search for call to ocw_embed_chapter_media(container_id, media_url, provider, page_url, image_url, start, stop, captions_file)
embed_chapter_media = re.search(r'ocw_embed_chapter_media\((.+?)\)', webpage) embed_chapter_media = re.search(r'ocw_embed_chapter_media\((.+?)\)', webpage)
if embed_chapter_media: if embed_chapter_media:
metadata = re.sub(r'[\'"]', u'', embed_chapter_media.group(1)) metadata = re.sub(r'[\'"]', '', embed_chapter_media.group(1))
metadata = re.split(r', ?', metadata) metadata = re.split(r', ?', metadata)
yt = metadata[1] yt = metadata[1]
subs = compat_urlparse.urljoin(self._BASE_URL, metadata[7]) subs = compat_urlparse.urljoin(self._BASE_URL, metadata[7])
@ -130,14 +138,21 @@ class OCWMITIE(InfoExtractor):
# search for call to ocw_embed_chapter_media(container_id, media_url, provider, page_url, image_url, captions_file) # search for call to ocw_embed_chapter_media(container_id, media_url, provider, page_url, image_url, captions_file)
embed_media = re.search(r'ocw_embed_media\((.+?)\)', webpage) embed_media = re.search(r'ocw_embed_media\((.+?)\)', webpage)
if embed_media: if embed_media:
metadata = re.sub(r'[\'"]', u'', embed_media.group(1)) metadata = re.sub(r'[\'"]', '', embed_media.group(1))
metadata = re.split(r', ?', metadata) metadata = re.split(r', ?', metadata)
yt = metadata[1] yt = metadata[1]
subs = compat_urlparse.urljoin(self._BASE_URL, metadata[5]) subs = compat_urlparse.urljoin(self._BASE_URL, metadata[5])
else: else:
raise ExtractorError('Unable to find embedded YouTube video.') raise ExtractorError('Unable to find embedded YouTube video.')
video_id = YoutubeIE.extract_id(yt)
data = self.url_result(yt, 'Youtube') return {
data['subtitles'] = subs '_type': 'url_transparent',
'id': video_id,
return data 'title': title,
'description': description,
'url': yt,
'url_transparent'
'subtitles': subs,
'ie_key': 'Youtube',
}