MarcoBuster/yt-dlp
1
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-27 15:16:52 +01:00
Code Issues Projects Releases Wiki Activity

pbs: fix vague 'Full Episode' titles; prepend name of show

Browse Source 
Many videos are given the title 'Full Episode' or 'Episode x', etc;
in the info json 'title' is the episode title (for oneshots this is
'Full Episode') while the program name is 'program':{'title'. This
changes the title from '(episode title)' to
 '(program name) - (episode title)'

The following urls demonstrate useless titles

    http://video.pbs.org/video/2365367186/
    http://video.pbs.org/video/2365519307/
    http://video.pbs.org/video/2365527039/
    http://video.pbs.org/video/2365530605/

Before change

    Full Episode
    Episode 5 | Preview
    Season 3 | Episode 4 Preview
    July 17, 2015

After change

    To Catch A Comet - Full Episode
    The Crimson Field - Episode 5 | Preview
    Last Tango in Halifax - Season 3 | Episode 4 Preview
    Charlie Rose The Week - July 17, 2015
This commit is contained in:
fnord 2015-07-19 11:47:58 -05:00
parent 1d18e26eca
commit a5dd9a0c5d
1 changed files with 27 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
youtube_dl/extractor/pbs.py
View File

@ -32,7 +32,7 @@ class PBSIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '2365006249', 'id': '2365006249',
'ext': 'mp4', 'ext': 'mp4',
'title': 'A More Perfect Union', 'title': 'Constitution USA with Peter Sagal - A More Perfect Union',
'description': 'md5:ba0c207295339c8d6eced00b7c363c6a', 'description': 'md5:ba0c207295339c8d6eced00b7c363c6a',
'duration': 3190, 'duration': 3190,
}, },
@ -46,7 +46,7 @@ class PBSIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '2365297690', 'id': '2365297690',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Losing Iraq', 'title': 'FRONTLINE - Losing Iraq',
'description': 'md5:f5bfbefadf421e8bb8647602011caf8e', 'description': 'md5:f5bfbefadf421e8bb8647602011caf8e',
'duration': 5050, 'duration': 5050,
}, },
@ -60,7 +60,7 @@ class PBSIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '2201174722', 'id': '2201174722',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Cyber Schools Gain Popularity, but Quality Questions Persist', 'title': 'PBS NewsHour - Cyber Schools Gain Popularity, but Quality Questions Persist',
'description': 'md5:5871c15cba347c1b3d28ac47a73c7c28', 'description': 'md5:5871c15cba347c1b3d28ac47a73c7c28',
'duration': 801, 'duration': 801,
}, },
@ -72,7 +72,7 @@ class PBSIE(InfoExtractor):
'id': '2365297708', 'id': '2365297708',
'ext': 'mp4', 'ext': 'mp4',
'description': 'md5:68d87ef760660eb564455eb30ca464fe', 'description': 'md5:68d87ef760660eb564455eb30ca464fe',
'title': 'Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full', 'title': 'Great Performances - Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full',
'duration': 6559, 'duration': 6559,
'thumbnail': 're:^https?://.*\.jpg$', 'thumbnail': 're:^https?://.*\.jpg$',
}, },
@ -88,7 +88,7 @@ class PBSIE(InfoExtractor):
'display_id': 'killer-typhoon', 'display_id': 'killer-typhoon',
'ext': 'mp4', 'ext': 'mp4',
'description': 'md5:c741d14e979fc53228c575894094f157', 'description': 'md5:c741d14e979fc53228c575894094f157',
'title': 'Killer Typhoon', 'title': 'NOVA - Killer Typhoon',
'duration': 3172, 'duration': 3172,
'thumbnail': 're:^https?://.*\.jpg$', 'thumbnail': 're:^https?://.*\.jpg$',
'upload_date': '20140122', 'upload_date': '20140122',
@ -110,7 +110,7 @@ class PBSIE(InfoExtractor):
'id': '2280706814', 'id': '2280706814',
'display_id': 'player', 'display_id': 'player',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Death and the Civil War', 'title': 'American Experience - Death and the Civil War',
'description': 'American Experience, TV’s most-watched history series, brings to life the compelling stories from our past that inform our understanding of the world today.', 'description': 'American Experience, TV’s most-watched history series, brings to life the compelling stories from our past that inform our understanding of the world today.',
'duration': 6705, 'duration': 6705,
'thumbnail': 're:^https?://.*\.jpg$', 'thumbnail': 're:^https?://.*\.jpg$',
@ -118,6 +118,21 @@ class PBSIE(InfoExtractor):
'params': { 'params': {
'skip_download': True, # requires ffmpeg 'skip_download': True, # requires ffmpeg
}, },
},
{
'url': 'http://video.pbs.org/video/2365367186/',
'info_dict': {
'id': '2365367186',
'display_id': '2365367186',
'ext': 'mp4',
'title': 'To Catch A Comet - Full Episode',
'description': 'On November 12, 2014, billions of kilometers from Earth, spacecraft orbiter Rosetta and lander Philae did what no other had dared to attempt \u2014 land on the volatile surface of a comet as it zooms around the sun at 67,000 km/hr. The European Space Agency hopes this mission can help peer into our past and unlock secrets of our origins.',
'duration': 3342,
'thumbnail': 're:^https?://.*\.jpg$',
},
'params': {
'skip_download': True, # requires ffmpeg
},
} }
] ]
@ -232,6 +247,12 @@ def _real_extract(self, url):
'url': closed_captions_url, 'url': closed_captions_url,
}] }]
# video.pbs.org video.pbs.org/videoInfo/... frequently provides an obscure 'title' value, like
# 'Full Episode', 'Episode 5', etc. prepend program->title
alt_title = info.get('program', {}).get('title', '')
if alt_title != '':
info['title'] = alt_title + ' - ' + re.sub(r'^' + alt_title + '[\s\-\:]+', '', info['title'])
return { return {
'id': video_id, 'id': video_id,
'display_id': display_id, 'display_id': display_id,