From ab49d7a9fae08763de549f85ba138b22f9122a99 Mon Sep 17 00:00:00 2001 From: Remita Amine Date: Wed, 6 Jul 2016 09:11:46 +0100 Subject: [PATCH] use mimetype2ext to determine manifest ext in multiple extractors --- youtube_dl/extractor/amp.py | 18 ++++++++++++------ youtube_dl/extractor/dailymotion.py | 8 +++++--- youtube_dl/extractor/metacafe.py | 5 ++--- youtube_dl/extractor/onionstudios.py | 6 +++--- youtube_dl/extractor/sixplay.py | 8 ++++++-- youtube_dl/extractor/threeqsdn.py | 9 ++++----- 6 files changed, 32 insertions(+), 22 deletions(-) diff --git a/youtube_dl/extractor/amp.py b/youtube_dl/extractor/amp.py index 8545681bea..283736fc5a 100644 --- a/youtube_dl/extractor/amp.py +++ b/youtube_dl/extractor/amp.py @@ -5,6 +5,8 @@ from ..utils import ( int_or_none, parse_iso8601, + mimetype2ext, + determine_ext, ) @@ -50,21 +52,25 @@ def get_media_node(name, default=None): if isinstance(media_content, dict): media_content = [media_content] for media_data in media_content: - media = media_data['@attributes'] - media_type = media['type'] - if media_type in ('video/f4m', 'application/f4m+xml'): + media = media_data.get('@attributes', {}) + media_url = media.get('url') + if not media_url: + continue + ext = mimetype2ext(media.get('type')) or determne_ext(media_url) + if ext == 'f4m': formats.extend(self._extract_f4m_formats( - media['url'] + '?hdcore=3.4.0&plugin=aasp-3.4.0.132.124', + media_url + '?hdcore=3.4.0&plugin=aasp-3.4.0.132.124', video_id, f4m_id='hds', fatal=False)) - elif media_type == 'application/x-mpegURL': + elif ext == 'm3u8': formats.extend(self._extract_m3u8_formats( - media['url'], video_id, 'mp4', m3u8_id='hls', fatal=False)) + media_url, video_id, 'mp4', m3u8_id='hls', fatal=False)) else: formats.append({ 'format_id': media_data.get('media-category', {}).get('@attributes', {}).get('label'), 'url': media['url'], 'tbr': int_or_none(media.get('bitrate')), 'filesize': int_or_none(media.get('fileSize')), + 'ext': ext, }) self._sort_formats(formats) diff --git a/youtube_dl/extractor/dailymotion.py b/youtube_dl/extractor/dailymotion.py index 2e6226ea07..15bfc59b24 100644 --- a/youtube_dl/extractor/dailymotion.py +++ b/youtube_dl/extractor/dailymotion.py @@ -16,6 +16,7 @@ sanitized_Request, str_to_int, unescapeHTML, + mimetype2ext, ) @@ -153,18 +154,19 @@ def _real_extract(self, url): type_ = media.get('type') if type_ == 'application/vnd.lumberjack.manifest': continue - ext = determine_ext(media_url) - if type_ == 'application/x-mpegURL' or ext == 'm3u8': + ext = mimetype2ext(type_) or determine_ext(media_url) + if ext == 'm3u8': formats.extend(self._extract_m3u8_formats( media_url, video_id, 'mp4', preference=-1, m3u8_id='hls', fatal=False)) - elif type_ == 'application/f4m' or ext == 'f4m': + elif ext == 'f4m': formats.extend(self._extract_f4m_formats( media_url, video_id, preference=-1, f4m_id='hds', fatal=False)) else: f = { 'url': media_url, 'format_id': 'http-%s' % quality, + 'ext': ext, } m = re.search(r'H264-(?P\d+)x(?P\d+)', media_url) if m: diff --git a/youtube_dl/extractor/metacafe.py b/youtube_dl/extractor/metacafe.py index acab81863c..e6e7659a1d 100644 --- a/youtube_dl/extractor/metacafe.py +++ b/youtube_dl/extractor/metacafe.py @@ -234,9 +234,8 @@ def _real_extract(self, url): source_url = source.get('src') if not source_url: continue - mime_type = source.get('type') - ext = mimetype2ext(mime_type) or determine_ext(source_url) - if mime_type == 'application/x-mpegURL' or ext == 'm3u8': + ext = mimetype2ext(source.get('type')) or determine_ext(source_url) + if ext == 'm3u8': video_url.extend(self._extract_m3u8_formats( source_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)) diff --git a/youtube_dl/extractor/onionstudios.py b/youtube_dl/extractor/onionstudios.py index 9fd5b21231..6fb1a3fcc0 100644 --- a/youtube_dl/extractor/onionstudios.py +++ b/youtube_dl/extractor/onionstudios.py @@ -8,6 +8,7 @@ determine_ext, int_or_none, float_or_none, + mimetype2ext, ) @@ -50,9 +51,8 @@ def _real_extract(self, url): source_url = source.get('url') if not source_url: continue - content_type = source.get('content_type') - ext = determine_ext(source_url) - if content_type == 'application/x-mpegURL' or ext == 'm3u8': + ext = mimetype2ext(source.get('content_type')) or determine_ext(source_url) + if ext == 'm3u8': formats.extend(self._extract_m3u8_formats( source_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)) else: diff --git a/youtube_dl/extractor/sixplay.py b/youtube_dl/extractor/sixplay.py index 759a332d28..d3aba58a29 100644 --- a/youtube_dl/extractor/sixplay.py +++ b/youtube_dl/extractor/sixplay.py @@ -5,6 +5,8 @@ from ..utils import ( qualities, int_or_none, + mimetype2ext, + determine_ext, ) @@ -34,19 +36,21 @@ def _real_extract(self, url): source_type, source_url = source.get('type'), source.get('src') if not source_url or source_type == 'hls/primetime': continue - if source_type == 'application/vnd.apple.mpegURL': + ext = mimetype2ext(source_type) or determine_ext(source_url) + if ext == 'm3u8': formats.extend(self._extract_m3u8_formats( source_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)) formats.extend(self._extract_f4m_formats( source_url.replace('.m3u8', '.f4m'), video_id, f4m_id='hds', fatal=False)) - elif source_type == 'video/mp4': + elif ext == 'mp4': quality = source.get('quality') formats.append({ 'url': source_url, 'format_id': quality, 'quality': quality_key(quality), + 'ext': ext, }) self._sort_formats(formats) diff --git a/youtube_dl/extractor/threeqsdn.py b/youtube_dl/extractor/threeqsdn.py index c77a07989a..a0bc12c81f 100644 --- a/youtube_dl/extractor/threeqsdn.py +++ b/youtube_dl/extractor/threeqsdn.py @@ -92,12 +92,11 @@ def extract_formats(item_url, item={}): if not item_url or item_url in urls: return urls.add(item_url) - type_ = item.get('type') - ext = determine_ext(item_url, default_ext=None) - if type_ == 'application/dash+xml' or ext == 'mpd': + ext = mimetype2ext(item.get('type')) or determine_ext(item_url, default_ext=None) + if ext == 'mpd': formats.extend(self._extract_mpd_formats( item_url, video_id, mpd_id='mpd', fatal=False)) - elif type_ in ('application/vnd.apple.mpegURL', 'application/x-mpegurl') or ext == 'm3u8': + elif ext == 'm3u8': formats.extend(self._extract_m3u8_formats( item_url, video_id, 'mp4', entry_protocol='m3u8' if live else 'm3u8_native', @@ -111,7 +110,7 @@ def extract_formats(item_url, item={}): formats.append({ 'url': item_url, 'format_id': item.get('quality'), - 'ext': 'mp4' if item_url.startswith('rtsp') else mimetype2ext(type_) or ext, + 'ext': 'mp4' if item_url.startswith('rtsp') else ext, 'vcodec': 'none' if stream_type == 'audio' else None, })