[loc] Extract subtites

This commit is contained in:
Sergey M․ 2016-06-03 23:55:22 +07:00
parent 76e9cd7f24
commit c917106be4
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -15,6 +15,7 @@ class LibraryOfCongressIE(InfoExtractor):
IE_DESC = 'Library of Congress' IE_DESC = 'Library of Congress'
_VALID_URL = r'https?://(?:www\.)?loc\.gov/(?:item/|today/cyberlc/feature_wdesc\.php\?.*\brec=)(?P<id>[0-9]+)' _VALID_URL = r'https?://(?:www\.)?loc\.gov/(?:item/|today/cyberlc/feature_wdesc\.php\?.*\brec=)(?P<id>[0-9]+)'
_TESTS = [{ _TESTS = [{
# embedded via <div class="media-player"
'url': 'http://loc.gov/item/90716351/', 'url': 'http://loc.gov/item/90716351/',
'md5': '353917ff7f0255aa6d4b80a034833de8', 'md5': '353917ff7f0255aa6d4b80a034833de8',
'info_dict': { 'info_dict': {
@ -26,8 +27,19 @@ class LibraryOfCongressIE(InfoExtractor):
'view_count': int, 'view_count': int,
}, },
}, { }, {
# webcast embedded via mediaObjectId
'url': 'https://www.loc.gov/today/cyberlc/feature_wdesc.php?rec=5578', 'url': 'https://www.loc.gov/today/cyberlc/feature_wdesc.php?rec=5578',
'only_matching': True, 'info_dict': {
'id': '5578',
'ext': 'mp4',
'title': 'Help! Preservation Training Needs Here, There & Everywhere',
'duration': 3765,
'view_count': int,
'subtitles': 'mincount:1',
},
'params': {
'skip_download': True,
},
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -76,6 +88,14 @@ def _real_extract(self, url):
duration = float_or_none(data.get('duration')) duration = float_or_none(data.get('duration'))
view_count = int_or_none(data.get('viewCount')) view_count = int_or_none(data.get('viewCount'))
subtitles = {}
cc_url = data.get('ccUrl')
if cc_url:
subtitles.setdefault('en', []).append({
'url': cc_url,
'ext': 'ttml',
})
return { return {
'id': video_id, 'id': video_id,
'title': title, 'title': title,
@ -83,4 +103,5 @@ def _real_extract(self, url):
'duration': duration, 'duration': duration,
'view_count': view_count, 'view_count': view_count,
'formats': formats, 'formats': formats,
'subtitles': subtitles,
} }