projects
/
youtube-dl
/ commitdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
| commitdiff |
tree
raw
|
patch
|
inline
| side by side (parent:
8da17f9
)
[facebook] Extract view count and update tests (closes #16942)
author
Sergey M․
<dstftw@gmail.com>
Wed, 18 Jul 2018 19:25:19 +0000
(
02:25
+0700)
committer
Sergey M․
<dstftw@gmail.com>
Wed, 18 Jul 2018 19:26:05 +0000
(
02:26
+0700)
youtube_dl/extractor/facebook.py
patch
|
blob
|
history
diff --git
a/youtube_dl/extractor/facebook.py
b/youtube_dl/extractor/facebook.py
index 8a9ed96c264ae37c5de2af472eca4c0ff1f4fa86..f78479b92e89acfac9c46f55a92b942b70fa3317 100644
(file)
--- a/
youtube_dl/extractor/facebook.py
+++ b/
youtube_dl/extractor/facebook.py
@@
-20,6
+20,7
@@
from ..utils import (
int_or_none,
js_to_json,
limit_length,
int_or_none,
js_to_json,
limit_length,
+ parse_count,
sanitized_Request,
try_get,
urlencode_postdata,
sanitized_Request,
try_get,
urlencode_postdata,
@@
-75,7
+76,7
@@
class FacebookIE(InfoExtractor):
'info_dict': {
'id': '274175099429670',
'ext': 'mp4',
'info_dict': {
'id': '274175099429670',
'ext': 'mp4',
- 'title': '
Asif Nawab Butt posted a video to his Timeline.
',
+ 'title': '
re:^Asif Nawab Butt posted a video
',
'uploader': 'Asif Nawab Butt',
'upload_date': '20140506',
'timestamp': 1399398998,
'uploader': 'Asif Nawab Butt',
'upload_date': '20140506',
'timestamp': 1399398998,
@@
-133,7
+134,7
@@
class FacebookIE(InfoExtractor):
}, {
# have 1080P, but only up to 720p in swf params
'url': 'https://www.facebook.com/cnn/videos/10155529876156509/',
}, {
# have 1080P, but only up to 720p in swf params
'url': 'https://www.facebook.com/cnn/videos/10155529876156509/',
- 'md5': '
0d9813160b146b3bc8744e006027fcc6
',
+ 'md5': '
9571fae53d4165bbbadb17a94651dcdc
',
'info_dict': {
'id': '10155529876156509',
'ext': 'mp4',
'info_dict': {
'id': '10155529876156509',
'ext': 'mp4',
@@
-142,6
+143,7
@@
class FacebookIE(InfoExtractor):
'upload_date': '20161030',
'uploader': 'CNN',
'thumbnail': r're:^https?://.*',
'upload_date': '20161030',
'uploader': 'CNN',
'thumbnail': r're:^https?://.*',
+ 'view_count': int,
},
}, {
# bigPipe.onPageletArrive ... onPageletArrive pagelet_group_mall
},
}, {
# bigPipe.onPageletArrive ... onPageletArrive pagelet_group_mall
@@
-149,7
+151,7
@@
class FacebookIE(InfoExtractor):
'info_dict': {
'id': '1417995061575415',
'ext': 'mp4',
'info_dict': {
'id': '1417995061575415',
'ext': 'mp4',
- 'title': 'md5:
a7b86ca673f51800cd54687b7f4012fe
',
+ 'title': 'md5:
1db063d6a8c13faa8da727817339c857
',
'timestamp': 1486648217,
'upload_date': '20170209',
'uploader': 'Yaroslav Korpan',
'timestamp': 1486648217,
'upload_date': '20170209',
'uploader': 'Yaroslav Korpan',
@@
-176,7
+178,7
@@
class FacebookIE(InfoExtractor):
'info_dict': {
'id': '1396382447100162',
'ext': 'mp4',
'info_dict': {
'id': '1396382447100162',
'ext': 'mp4',
- 'title': 'md5:
e2d2700afdf84e121f5d0f999bad13a3
',
+ 'title': 'md5:
19a428bbde91364e3de815383b54a235
',
'timestamp': 1486035494,
'upload_date': '20170202',
'uploader': 'Elisabeth Ahtn',
'timestamp': 1486035494,
'upload_date': '20170202',
'uploader': 'Elisabeth Ahtn',
@@
-426,6
+428,10
@@
class FacebookIE(InfoExtractor):
'timestamp', default=None))
thumbnail = self._og_search_thumbnail(webpage)
'timestamp', default=None))
thumbnail = self._og_search_thumbnail(webpage)
+ view_count = parse_count(self._search_regex(
+ r'\bviewCount\s*:\s*["\']([\d,.]+)', webpage, 'view count',
+ default=None))
+
info_dict = {
'id': video_id,
'title': video_title,
info_dict = {
'id': video_id,
'title': video_title,
@@
-433,6
+439,7
@@
class FacebookIE(InfoExtractor):
'uploader': uploader,
'timestamp': timestamp,
'thumbnail': thumbnail,
'uploader': uploader,
'timestamp': timestamp,
'thumbnail': thumbnail,
+ 'view_count': view_count,
}
return webpage, info_dict
}
return webpage, info_dict