paradizelost/youtube-dl
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[svt] Improve subtitles extraction and add test (Closes #8265)

Browse Source
This commit is contained in:
Sergey M․ 2016-01-23 01:47:54 +06:00
parent 1f16b958b1
commit 594c4d79a5
1 changed files with 19 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
youtube_dl/extractor/svt.py
View File

@ -37,11 +37,13 @@ class SVTBaseIE(InfoExtractor):
}) })
self._sort_formats(formats) self._sort_formats(formats)
# SVT does not tell us the language, so we assume swedish.
subtitles = {} subtitles = {}
for sr in video_info['subtitleReferences']: subtitle_references = video_info.get('subtitleReferences')
if 'url' in sr: if isinstance(subtitle_references, list):
subtitles.setdefault('sv', []).append({'url': sr['url']}) for sr in subtitle_references:
subtitle_url = sr.get('url')
if subtitle_url:
subtitles.setdefault('sv', []).append({'url': subtitle_url})
duration = video_info.get('materialLength') duration = video_info.get('materialLength')
age_limit = 18 if video_info.get('inappropriateForChildren') else 0 age_limit = 18 if video_info.get('inappropriateForChildren') else 0
@ -90,30 +92,23 @@ class SVTIE(SVTBaseIE):
class SVTPlayIE(SVTBaseIE): class SVTPlayIE(SVTBaseIE):
IE_DESC = 'SVT Play and Öppet arkiv' IE_DESC = 'SVT Play and Öppet arkiv'
_VALID_URL = r'https?://(?:www\.)?(?P<host>svtplay|oppetarkiv)\.se/video/(?P<id>[0-9]+)' _VALID_URL = r'https?://(?:www\.)?(?P<host>svtplay|oppetarkiv)\.se/video/(?P<id>[0-9]+)'
_TESTS = [{ _TEST = {
'url': 'http://www.svtplay.se/video/2609989/sm-veckan/sm-veckan-rally-final-sasong-1-sm-veckan-rally-final', 'url': 'http://www.svtplay.se/video/5996901/flygplan-till-haile-selassie/flygplan-till-haile-selassie-2',
'md5': 'ade3def0643fa1c40587a422f98edfd9', 'md5': '2b6704fe4a28801e1a098bbf3c5ac611',
'info_dict': { 'info_dict': {
'id': '2609989', 'id': '5996901',
'ext': 'flv', 'ext': 'mp4',
'title': 'SM veckan vinter, Örebro - Rally, final', 'title': 'Flygplan till Haile Selassie',
'duration': 4500, 'duration': 3527,
'thumbnail': 're:^https?://.*[\.-]jpg$', 'thumbnail': 're:^https?://.*[\.-]jpg$',
'age_limit': 0, 'age_limit': 0,
'subtitles': {
'sv': [{
'ext': 'wsrt',
}]
},
}, },
}, { }
'url': 'http://www.oppetarkiv.se/video/1058509/rederiet-sasong-1-avsnitt-1-av-318',
'md5': 'c3101a17ce9634f4c1f9800f0746c187',
'info_dict': {
'id': '1058509',
'ext': 'flv',
'title': 'Farlig kryssning',
'duration': 2566,
'thumbnail': 're:^https?://.*[\.-]jpg$',
'age_limit': 0,
},
'skip': 'Only works from Sweden',
}]
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)