aboutsummaryrefslogtreecommitdiff
path: root/youtube_dl
diff options
context:
space:
mode:
authorSergey M․ <dstftw@gmail.com>2020-03-01 03:09:19 +0700
committerSergey M․ <dstftw@gmail.com>2020-03-01 03:09:19 +0700
commit6d475d01d8f1fc9f5e92ac2caa986453bee95ae1 (patch)
treeae7079bc46aeeafb0ea5b933c41ca66863c102f5 /youtube_dl
parentf8cbd8c96367871db974621d5a92b324acb57938 (diff)
[telecinco] Add support for article opening videos
Diffstat (limited to 'youtube_dl')
-rw-r--r--youtube_dl/extractor/telecinco.py42
1 files changed, 37 insertions, 5 deletions
diff --git a/youtube_dl/extractor/telecinco.py b/youtube_dl/extractor/telecinco.py
index d79ab1e82..9ba3da341 100644
--- a/youtube_dl/extractor/telecinco.py
+++ b/youtube_dl/extractor/telecinco.py
@@ -11,6 +11,7 @@ from ..utils import (
determine_ext,
int_or_none,
str_or_none,
+ try_get,
urljoin,
)
@@ -24,7 +25,7 @@ class TelecincoIE(InfoExtractor):
'info_dict': {
'id': '1876350223',
'title': 'Bacalao con kokotxas al pil-pil',
- 'description': 'md5:1382dacd32dd4592d478cbdca458e5bb',
+ 'description': 'md5:716caf5601e25c3c5ab6605b1ae71529',
},
'playlist': [{
'md5': 'adb28c37238b675dad0f042292f209a7',
@@ -56,6 +57,26 @@ class TelecincoIE(InfoExtractor):
'duration': 50,
},
}, {
+ # video in opening's content
+ 'url': 'https://www.telecinco.es/vivalavida/fiorella-sobrina-edmundo-arrocet-entrevista_18_2907195140.html',
+ 'info_dict': {
+ 'id': '2907195140',
+ 'title': 'La surrealista entrevista a la sobrina de Edmundo Arrocet: "No puedes venir aquí y tomarnos por tontos"',
+ 'description': 'md5:73f340a7320143d37ab895375b2bf13a',
+ },
+ 'playlist': [{
+ 'md5': 'adb28c37238b675dad0f042292f209a7',
+ 'info_dict': {
+ 'id': 'TpI2EttSDAReWpJ1o0NVh2',
+ 'ext': 'mp4',
+ 'title': 'La surrealista entrevista a la sobrina de Edmundo Arrocet: "No puedes venir aquí y tomarnos por tontos"',
+ 'duration': 1015,
+ },
+ }],
+ 'params': {
+ 'skip_download': True,
+ },
+ }, {
'url': 'http://www.telecinco.es/informativos/nacional/Pablo_Iglesias-Informativos_Telecinco-entrevista-Pedro_Piqueras_2_1945155182.html',
'only_matching': True,
}, {
@@ -138,14 +159,25 @@ class TelecincoIE(InfoExtractor):
r'window\.\$REACTBASE_STATE\.article(?:_multisite)?\s*=\s*({.+})',
webpage, 'article'), display_id)['article']
title = article.get('title')
- description = clean_html(article.get('leadParagraph'))
+ description = clean_html(article.get('leadParagraph')) or ''
if article.get('editorialType') != 'VID':
entries = []
- for p in article.get('body', []):
+ body = [article.get('opening')]
+ body.extend(try_get(article, lambda x: x['body'], list) or [])
+ for p in body:
+ if not isinstance(p, dict):
+ continue
content = p.get('content')
- if p.get('type') != 'video' or not content:
+ if not content:
+ continue
+ type_ = p.get('type')
+ if type_ == 'paragraph':
+ content_str = str_or_none(content)
+ if content_str:
+ description += content_str
continue
- entries.append(self._parse_content(content, url))
+ if type_ == 'video' and isinstance(content, dict):
+ entries.append(self._parse_content(content, url))
return self.playlist_result(
entries, str_or_none(article.get('id')), title, description)
content = article['opening']['content']