w/youtube-dl
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

[ted] check for resources validity and extract subtitled downloads(closes #22513)

Browse source
This commit is contained in:
Remita Amine 2019-09-26 11:44:57 +01:00
parent 33c1c7d80f
commit 2a88a0c44d
1 changed files with 47 additions and 35 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

44
youtube_dl/extractor/ted.py
View file

@ -182,20 +182,29 @@ class TEDIE(InfoExtractor):
title = talk_info['title'].strip() title = talk_info['title'].strip()
native_downloads = try_get( downloads = talk_info.get('downloads') or {}
talk_info, native_downloads = downloads.get('nativeDownloads') or talk_info.get('nativeDownloads') or {}
(lambda x: x['downloads']['nativeDownloads'],
lambda x: x['nativeDownloads']),
dict) or {}
formats = [{ formats = [{
'url': format_url, 'url': format_url,
'format_id': format_id, 'format_id': format_id,
'format': format_id,
} for (format_id, format_url) in native_downloads.items() if format_url is not None] } for (format_id, format_url) in native_downloads.items() if format_url is not None]
subtitled_downloads = downloads.get('subtitledDownloads') or {}
for lang, subtitled_download in subtitled_downloads.items():
for q in self._NATIVE_FORMATS:
q_url = subtitled_download.get(q)
if not q_url:
continue
formats.append({
'url': q_url,
'format_id': '%s-%s' % (q, lang),
'language': lang,
})
if formats: if formats:
for f in formats: for f in formats:
finfo = self._NATIVE_FORMATS.get(f['format_id']) finfo = self._NATIVE_FORMATS.get(f['format_id'].split('-')[0])
if finfo: if finfo:
f.update(finfo) f.update(finfo)
@ -215,6 +224,18 @@ class TEDIE(InfoExtractor):
http_url = None http_url = None
for format_id, resources in resources_.items(): for format_id, resources in resources_.items():
if format_id == 'hls':
if not isinstance(resources, dict):
continue
stream_url = url_or_none(resources.get('stream'))
if not stream_url:
continue
formats.extend(self._extract_m3u8_formats(
stream_url, video_name, 'mp4', m3u8_id=format_id,
fatal=False))
else:
if not isinstance(resources, list):
continue
if format_id == 'h264': if format_id == 'h264':
for resource in resources: for resource in resources:
h264_url = resource.get('file') h264_url = resource.get('file')
@ -242,15 +263,6 @@ class TEDIE(InfoExtractor):
'height': int_or_none(resource.get('height')), 'height': int_or_none(resource.get('height')),
'tbr': int_or_none(resource.get('bitrate')), 'tbr': int_or_none(resource.get('bitrate')),
}) })
elif format_id == 'hls':
if not isinstance(resources, dict):
continue
stream_url = url_or_none(resources.get('stream'))
if not stream_url:
continue
formats.extend(self._extract_m3u8_formats(
stream_url, video_name, 'mp4', m3u8_id=format_id,
fatal=False))
m3u8_formats = list(filter( m3u8_formats = list(filter(
lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none', lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none',