projects
/
youtube-dl
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
[redtube] Improve formats extraction and extract m3u8 formats (closes #25311, closes...
[youtube-dl]
/
youtube_dl
/
extractor
/
tvn24.py
diff --git
a/youtube_dl/extractor/tvn24.py
b/youtube_dl/extractor/tvn24.py
index 12ed6039cb758af5563232a6f62497446a798cfb..de0fb506316814c8562de514daddc5f99165cf6f 100644
(file)
--- a/
youtube_dl/extractor/tvn24.py
+++ b/
youtube_dl/extractor/tvn24.py
@@
-4,12
+4,13
@@
from __future__ import unicode_literals
from .common import InfoExtractor
from ..utils import (
int_or_none,
from .common import InfoExtractor
from ..utils import (
int_or_none,
+ NO_DEFAULT,
unescapeHTML,
)
class TVN24IE(InfoExtractor):
unescapeHTML,
)
class TVN24IE(InfoExtractor):
- _VALID_URL = r'https?://(?:(?:[^/]+)\.)?tvn24(?:bis)?\.pl/(?:[^/]+/)*(?P<id>[^/]+)
\.html
'
+ _VALID_URL = r'https?://(?:(?:[^/]+)\.)?tvn24(?:bis)?\.pl/(?:[^/]+/)*(?P<id>[^/]+)'
_TESTS = [{
'url': 'http://www.tvn24.pl/wiadomosci-z-kraju,3/oredzie-artura-andrusa,702428.html',
'md5': 'fbdec753d7bc29d96036808275f2130c',
_TESTS = [{
'url': 'http://www.tvn24.pl/wiadomosci-z-kraju,3/oredzie-artura-andrusa,702428.html',
'md5': 'fbdec753d7bc29d96036808275f2130c',
@@
-17,9
+18,21
@@
class TVN24IE(InfoExtractor):
'id': '1584444',
'ext': 'mp4',
'title': '"Święta mają być wesołe, dlatego, ludziska, wszyscy pod jemiołę"',
'id': '1584444',
'ext': 'mp4',
'title': '"Święta mają być wesołe, dlatego, ludziska, wszyscy pod jemiołę"',
- 'description': 'Wyjątkowe orędzie Artura Andrusa, jednego z gości
"Szkła kontaktowego"
.',
- 'thumbnail': 're:http://.*[.]jpeg',
+ 'description': 'Wyjątkowe orędzie Artura Andrusa, jednego z gości
Szkła kontaktowego
.',
+ 'thumbnail': 're:http
s?
://.*[.]jpeg',
}
}
+ }, {
+ # different layout
+ 'url': 'https://tvnmeteo.tvn24.pl/magazyny/maja-w-ogrodzie,13/odcinki-online,1,4,1,0/pnacza-ptaki-i-iglaki-odc-691-hgtv-odc-29,1771763.html',
+ 'info_dict': {
+ 'id': '1771763',
+ 'ext': 'mp4',
+ 'title': 'Pnącza, ptaki i iglaki (odc. 691 /HGTV odc. 29)',
+ 'thumbnail': 're:https?://.*',
+ },
+ 'params': {
+ 'skip_download': True,
+ },
}, {
'url': 'http://fakty.tvn24.pl/ogladaj-online,60/53-konferencja-bezpieczenstwa-w-monachium,716431.html',
'only_matching': True,
}, {
'url': 'http://fakty.tvn24.pl/ogladaj-online,60/53-konferencja-bezpieczenstwa-w-monachium,716431.html',
'only_matching': True,
@@
-29,21
+42,27
@@
class TVN24IE(InfoExtractor):
}, {
'url': 'http://tvn24bis.pl/poranek,146,m/gen-koziej-w-tvn24-bis-wracamy-do-czasow-zimnej-wojny,715660.html',
'only_matching': True,
}, {
'url': 'http://tvn24bis.pl/poranek,146,m/gen-koziej-w-tvn24-bis-wracamy-do-czasow-zimnej-wojny,715660.html',
'only_matching': True,
+ }, {
+ 'url': 'https://www.tvn24.pl/magazyn-tvn24/angie-w-jednej-czwartej-polka-od-szarej-myszki-do-cesarzowej-europy,119,2158',
+ 'only_matching': True,
}]
def _real_extract(self, url):
}]
def _real_extract(self, url):
-
video
_id = self._match_id(url)
+
display
_id = self._match_id(url)
- webpage = self._download_webpage(url,
video
_id)
+ webpage = self._download_webpage(url,
display
_id)
- title = self._og_search_title(webpage)
+ title = self._og_search_title(
+ webpage, default=None) or self._search_regex(
+ r'<h\d+[^>]+class=["\']magazineItemHeader[^>]+>(.+?)</h',
+ webpage, 'title')
- def extract_json(attr, name, fatal=True):
+ def extract_json(attr, name,
default=NO_DEFAULT,
fatal=True):
return self._parse_json(
self._search_regex(
r'\b%s=(["\'])(?P<json>(?!\1).+?)\1' % attr, webpage,
return self._parse_json(
self._search_regex(
r'\b%s=(["\'])(?P<json>(?!\1).+?)\1' % attr, webpage,
- name, group='json', fatal=fatal) or '{}',
-
video
_id, transform_source=unescapeHTML, fatal=fatal)
+ name, group='json',
default=default,
fatal=fatal) or '{}',
+
display
_id, transform_source=unescapeHTML, fatal=fatal)
quality_data = extract_json('data-quality', 'formats')
quality_data = extract_json('data-quality', 'formats')
@@
-56,16
+75,24
@@
class TVN24IE(InfoExtractor):
})
self._sort_formats(formats)
})
self._sort_formats(formats)
- description = self._og_search_description(webpage)
+ description = self._og_search_description(webpage
, default=None
)
thumbnail = self._og_search_thumbnail(
webpage, default=None) or self._html_search_regex(
r'\bdata-poster=(["\'])(?P<url>(?!\1).+?)\1', webpage,
'thumbnail', group='url')
thumbnail = self._og_search_thumbnail(
webpage, default=None) or self._html_search_regex(
r'\bdata-poster=(["\'])(?P<url>(?!\1).+?)\1', webpage,
'thumbnail', group='url')
+ video_id = None
+
share_params = extract_json(
share_params = extract_json(
- 'data-share-params', 'share params',
fatal=Fals
e)
+ 'data-share-params', 'share params',
default=Non
e)
if isinstance(share_params, dict):
if isinstance(share_params, dict):
- video_id = share_params.get('id') or video_id
+ video_id = share_params.get('id')
+
+ if not video_id:
+ video_id = self._search_regex(
+ r'data-vid-id=["\'](\d+)', webpage, 'video id',
+ default=None) or self._search_regex(
+ r',(\d+)\.html', url, 'video id', default=display_id)
return {
'id': video_id,
return {
'id': video_id,