From 594c4d79a5cc988af99cbd0e3c00d1752e277dd1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sat, 23 Jan 2016 01:47:54 +0600 Subject: [PATCH] [svt] Improve subtitles extraction and add test (Closes #8265) --- youtube_dl/extractor/svt.py | 43 ++++++++++++++++--------------------- 1 file changed, 19 insertions(+), 24 deletions(-) diff --git a/youtube_dl/extractor/svt.py b/youtube_dl/extractor/svt.py index 7ff5ee11a..399c3b8ee 100644 --- a/youtube_dl/extractor/svt.py +++ b/youtube_dl/extractor/svt.py @@ -37,11 +37,13 @@ class SVTBaseIE(InfoExtractor): }) self._sort_formats(formats) - # SVT does not tell us the language, so we assume swedish. subtitles = {} - for sr in video_info['subtitleReferences']: - if 'url' in sr: - subtitles.setdefault('sv', []).append({'url': sr['url']}) + subtitle_references = video_info.get('subtitleReferences') + if isinstance(subtitle_references, list): + 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') age_limit = 18 if video_info.get('inappropriateForChildren') else 0 @@ -90,30 +92,23 @@ class SVTIE(SVTBaseIE): class SVTPlayIE(SVTBaseIE): IE_DESC = 'SVT Play and Öppet arkiv' _VALID_URL = r'https?://(?:www\.)?(?Psvtplay|oppetarkiv)\.se/video/(?P[0-9]+)' - _TESTS = [{ - 'url': 'http://www.svtplay.se/video/2609989/sm-veckan/sm-veckan-rally-final-sasong-1-sm-veckan-rally-final', - 'md5': 'ade3def0643fa1c40587a422f98edfd9', - 'info_dict': { - 'id': '2609989', - 'ext': 'flv', - 'title': 'SM veckan vinter, Örebro - Rally, final', - 'duration': 4500, - 'thumbnail': 're:^https?://.*[\.-]jpg$', - 'age_limit': 0, - }, - }, { - 'url': 'http://www.oppetarkiv.se/video/1058509/rederiet-sasong-1-avsnitt-1-av-318', - 'md5': 'c3101a17ce9634f4c1f9800f0746c187', + _TEST = { + 'url': 'http://www.svtplay.se/video/5996901/flygplan-till-haile-selassie/flygplan-till-haile-selassie-2', + 'md5': '2b6704fe4a28801e1a098bbf3c5ac611', 'info_dict': { - 'id': '1058509', - 'ext': 'flv', - 'title': 'Farlig kryssning', - 'duration': 2566, + 'id': '5996901', + 'ext': 'mp4', + 'title': 'Flygplan till Haile Selassie', + 'duration': 3527, 'thumbnail': 're:^https?://.*[\.-]jpg$', 'age_limit': 0, + 'subtitles': { + 'sv': [{ + 'ext': 'wsrt', + }] + }, }, - 'skip': 'Only works from Sweden', - }] + } def _real_extract(self, url): mobj = re.match(self._VALID_URL, url)