aboutsummaryrefslogtreecommitdiffstats
path: root/youtube_dl/extractor/la7.py
diff options
context:
space:
mode:
authorRogério Brito <rbrito@ime.usp.br>2016-08-17 09:19:41 -0300
committerRogério Brito <rbrito@ime.usp.br>2016-08-17 09:19:41 -0300
commitced7488f6d3a519b2c1b1cbd31048743fb8285bd (patch)
tree868396b5d0031b626ea3e2ef822dad6430d70c67 /youtube_dl/extractor/la7.py
parent9dc487f48b50767cf540fa36c3de2c386fd74c04 (diff)
downloadyoutube-dl-ced7488f6d3a519b2c1b1cbd31048743fb8285bd.zip
youtube-dl-ced7488f6d3a519b2c1b1cbd31048743fb8285bd.tar.gz
youtube-dl-ced7488f6d3a519b2c1b1cbd31048743fb8285bd.tar.bz2
Imported Upstream version 2016.08.17
Diffstat (limited to 'youtube_dl/extractor/la7.py')
-rw-r--r--youtube_dl/extractor/la7.py85
1 files changed, 45 insertions, 40 deletions
diff --git a/youtube_dl/extractor/la7.py b/youtube_dl/extractor/la7.py
index b08f6e3..da5a5de 100644
--- a/youtube_dl/extractor/la7.py
+++ b/youtube_dl/extractor/la7.py
@@ -1,60 +1,65 @@
+# coding: utf-8
from __future__ import unicode_literals
from .common import InfoExtractor
from ..utils import (
- parse_duration,
+ js_to_json,
+ smuggle_url,
)
class LA7IE(InfoExtractor):
- IE_NAME = 'la7.tv'
- _VALID_URL = r'''(?x)
- https?://(?:www\.)?la7\.tv/
- (?:
- richplayer/\?assetid=|
- \?contentId=
- )
- (?P<id>[0-9]+)'''
-
- _TEST = {
- 'url': 'http://www.la7.tv/richplayer/?assetid=50355319',
- 'md5': 'ec7d1f0224d20ba293ab56cf2259651f',
+ IE_NAME = 'la7.it'
+ _VALID_URL = r'''(?x)(https?://)?(?:
+ (?:www\.)?la7\.it/([^/]+)/(?:rivedila7|video)/|
+ tg\.la7\.it/repliche-tgla7\?id=
+ )(?P<id>.+)'''
+
+ _TESTS = [{
+ # 'src' is a plain URL
+ 'url': 'http://www.la7.it/crozza/video/inccool8-02-10-2015-163722',
+ 'md5': '8b613ffc0c4bf9b9e377169fc19c214c',
'info_dict': {
- 'id': '50355319',
+ 'id': 'inccool8-02-10-2015-163722',
'ext': 'mp4',
- 'title': 'IL DIVO',
- 'description': 'Un film di Paolo Sorrentino con Toni Servillo, Anna Bonaiuto, Giulio Bosetti e Flavio Bucci',
- 'duration': 6254,
+ 'title': 'Inc.Cool8',
+ 'description': 'Benvenuti nell\'incredibile mondo della INC. COOL. 8. dove “INC.” sta per “Incorporated” “COOL” sta per “fashion” ed Eight sta per il gesto atletico',
+ 'thumbnail': 're:^https?://.*',
+ 'uploader_id': 'kdla7pillole@iltrovatore.it',
+ 'timestamp': 1443814869,
+ 'upload_date': '20151002',
},
- 'skip': 'Blocked in the US',
- }
+ }, {
+ # 'src' is a dictionary
+ 'url': 'http://tg.la7.it/repliche-tgla7?id=189080',
+ 'md5': '6b0d8888d286e39870208dfeceaf456b',
+ 'info_dict': {
+ 'id': '189080',
+ 'ext': 'mp4',
+ 'title': 'TG LA7',
+ },
+ }, {
+ 'url': 'http://www.la7.it/omnibus/rivedila7/omnibus-news-02-07-2016-189077',
+ 'only_matching': True,
+ }]
def _real_extract(self, url):
video_id = self._match_id(url)
- xml_url = 'http://www.la7.tv/repliche/content/index.php?contentId=%s' % video_id
- doc = self._download_xml(xml_url, video_id)
-
- video_title = doc.find('title').text
- description = doc.find('description').text
- duration = parse_duration(doc.find('duration').text)
- thumbnail = doc.find('img').text
- view_count = int(doc.find('views').text)
- prefix = doc.find('.//fqdn').text.strip().replace('auto:', 'http:')
+ webpage = self._download_webpage(url, video_id)
- formats = [{
- 'format': vnode.find('quality').text,
- 'tbr': int(vnode.find('quality').text),
- 'url': vnode.find('fms').text.strip().replace('mp4:', prefix),
- } for vnode in doc.findall('.//videos/video')]
- self._sort_formats(formats)
+ player_data = self._parse_json(
+ self._search_regex(r'videoLa7\(({[^;]+})\);', webpage, 'player data'),
+ video_id, transform_source=js_to_json)
return {
+ '_type': 'url_transparent',
+ 'url': smuggle_url('kaltura:103:%s' % player_data['vid'], {
+ 'service_url': 'http://kdam.iltrovatore.it',
+ }),
'id': video_id,
- 'title': video_title,
- 'description': description,
- 'thumbnail': thumbnail,
- 'duration': duration,
- 'formats': formats,
- 'view_count': view_count,
+ 'title': player_data['title'],
+ 'description': self._og_search_description(webpage, default=None),
+ 'thumbnail': player_data.get('poster'),
+ 'ie_key': 'Kaltura',
}