pepepper/youtube-dl
1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl synced 2024-11-05 09:57:59 +09:00
Code Issues Projects Releases Wiki Activity

[francetv] Use unicode_literals

Browse Source
This commit is contained in:
Philipp Hagemeister 2014-01-30 06:13:57 +01:00
parent 24b3d5e538
commit 3c1b4669d0
1 changed files with 73 additions and 70 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

143
youtube_dl/extractor/francetv.py
View File

@ -1,4 +1,7 @@
# encoding: utf-8 # encoding: utf-8
from __future__ import unicode_literals
import re import re
import json import json
@ -30,7 +33,7 @@ class FranceTVBaseInfoExtractor(InfoExtractor):
class PluzzIE(FranceTVBaseInfoExtractor): class PluzzIE(FranceTVBaseInfoExtractor):
IE_NAME = u'pluzz.francetv.fr' IE_NAME = 'pluzz.francetv.fr'
_VALID_URL = r'https?://pluzz\.francetv\.fr/videos/(.*?)\.html' _VALID_URL = r'https?://pluzz\.francetv\.fr/videos/(.*?)\.html'
# Can't use tests, videos expire in 7 days # Can't use tests, videos expire in 7 days
@ -44,17 +47,17 @@ class PluzzIE(FranceTVBaseInfoExtractor):
class FranceTvInfoIE(FranceTVBaseInfoExtractor): class FranceTvInfoIE(FranceTVBaseInfoExtractor):
IE_NAME = u'francetvinfo.fr' IE_NAME = 'francetvinfo.fr'
_VALID_URL = r'https?://www\.francetvinfo\.fr/replay.*/(?P<title>.+)\.html' _VALID_URL = r'https?://www\.francetvinfo\.fr/replay.*/(?P<title>.+)\.html'
_TEST = { _TEST = {
u'url': u'http://www.francetvinfo.fr/replay-jt/france-3/soir-3/jt-grand-soir-3-lundi-26-aout-2013_393427.html', 'url': 'http://www.francetvinfo.fr/replay-jt/france-3/soir-3/jt-grand-soir-3-lundi-26-aout-2013_393427.html',
u'file': u'84981923.mp4', 'file': '84981923.mp4',
u'info_dict': { 'info_dict': {
u'title': u'Soir 3', 'title': 'Soir 3',
}, },
u'params': { 'params': {
u'skip_download': True, 'skip_download': True,
}, },
} }
@ -62,13 +65,13 @@ class FranceTvInfoIE(FranceTVBaseInfoExtractor):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
page_title = mobj.group('title') page_title = mobj.group('title')
webpage = self._download_webpage(url, page_title) webpage = self._download_webpage(url, page_title)
video_id = self._search_regex(r'id-video=(\d+?)[@"]', webpage, u'video id') video_id = self._search_regex(r'id-video=(\d+?)[@"]', webpage, 'video id')
return self._extract_video(video_id) return self._extract_video(video_id)
class FranceTVIE(FranceTVBaseInfoExtractor): class FranceTVIE(FranceTVBaseInfoExtractor):
IE_NAME = u'francetv' IE_NAME = 'francetv'
IE_DESC = u'France 2, 3, 4, 5 and Ô' IE_DESC = 'France 2, 3, 4, 5 and Ô'
_VALID_URL = r'''(?x)https?://www\.france[2345o]\.fr/ _VALID_URL = r'''(?x)https?://www\.france[2345o]\.fr/
(?: (?:
emissions/.*?/(videos|emissions)/(?P<id>[^/?]+) emissions/.*?/(videos|emissions)/(?P<id>[^/?]+)
@ -78,73 +81,73 @@ class FranceTVIE(FranceTVBaseInfoExtractor):
_TESTS = [ _TESTS = [
# france2 # france2
{ {
u'url': u'http://www.france2.fr/emissions/13h15-le-samedi-le-dimanche/videos/75540104', 'url': 'http://www.france2.fr/emissions/13h15-le-samedi-le-dimanche/videos/75540104',
u'file': u'75540104.mp4', 'file': '75540104.mp4',
u'info_dict': { 'info_dict': {
u'title': u'13h15, le samedi...', 'title': '13h15, le samedi...',
u'description': u'md5:2e5b58ba7a2d3692b35c792be081a03d', 'description': 'md5:2e5b58ba7a2d3692b35c792be081a03d',
}, },
u'params': { 'params': {
# m3u8 download # m3u8 download
u'skip_download': True, 'skip_download': True,
}, },
}, },
# france3 # france3
{ {
u'url': u'http://www.france3.fr/emissions/pieces-a-conviction/diffusions/13-11-2013_145575', 'url': 'http://www.france3.fr/emissions/pieces-a-conviction/diffusions/13-11-2013_145575',
u'info_dict': { 'info_dict': {
u'id': u'000702326_CAPP_PicesconvictionExtrait313022013_120220131722_Au', 'id': '000702326_CAPP_PicesconvictionExtrait313022013_120220131722_Au',
u'ext': u'flv', 'ext': 'flv',
u'title': u'Le scandale du prix des médicaments', 'title': 'Le scandale du prix des médicaments',
u'description': u'md5:1384089fbee2f04fc6c9de025ee2e9ce', 'description': 'md5:1384089fbee2f04fc6c9de025ee2e9ce',
}, },
u'params': { 'params': {
# rtmp download # rtmp download
u'skip_download': True, 'skip_download': True,
}, },
}, },
# france4 # france4
{ {
u'url': u'http://www.france4.fr/emissions/hero-corp/videos/rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4', 'url': 'http://www.france4.fr/emissions/hero-corp/videos/rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4',
u'info_dict': { 'info_dict': {
u'id': u'rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4', 'id': 'rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4',
u'ext': u'flv', 'ext': 'flv',
u'title': u'Hero Corp Making of - Extrait 1', 'title': 'Hero Corp Making of - Extrait 1',
u'description': u'md5:c87d54871b1790679aec1197e73d650a', 'description': 'md5:c87d54871b1790679aec1197e73d650a',
}, },
u'params': { 'params': {
# rtmp download # rtmp download
u'skip_download': True, 'skip_download': True,
}, },
}, },
# france5 # france5
{ {
u'url': u'http://www.france5.fr/emissions/c-a-dire/videos/92837968', 'url': 'http://www.france5.fr/emissions/c-a-dire/videos/92837968',
u'info_dict': { 'info_dict': {
u'id': u'92837968', 'id': '92837968',
u'ext': u'mp4', 'ext': 'mp4',
u'title': u'C à dire ?!', 'title': 'C à dire ?!',
u'description': u'md5:fb1db1cbad784dcce7c7a7bd177c8e2f', 'description': 'md5:fb1db1cbad784dcce7c7a7bd177c8e2f',
}, },
u'params': { 'params': {
# m3u8 download # m3u8 download
u'skip_download': True, 'skip_download': True,
}, },
}, },
# franceo # franceo
{ {
u'url': u'http://www.franceo.fr/jt/info-afrique/04-12-2013', 'url': 'http://www.franceo.fr/jt/info-afrique/04-12-2013',
u'info_dict': { 'info_dict': {
u'id': u'92327925', 'id': '92327925',
u'ext': u'mp4', 'ext': 'mp4',
u'title': u'Infô-Afrique', 'title': 'Infô-Afrique',
u'description': u'md5:ebf346da789428841bee0fd2a935ea55', 'description': 'md5:ebf346da789428841bee0fd2a935ea55',
}, },
u'params': { 'params': {
# m3u8 download # m3u8 download
u'skip_download': True, 'skip_download': True,
}, },
u'skip': u'The id changes frequently', 'skip': 'The id changes frequently',
}, },
] ]
@ -160,26 +163,26 @@ class FranceTVIE(FranceTVBaseInfoExtractor):
'\.fr/\?id-video=([^"/&]+)'), '\.fr/\?id-video=([^"/&]+)'),
(r'<a class="video" id="ftv_player_(.+?)"'), (r'<a class="video" id="ftv_player_(.+?)"'),
] ]
video_id = self._html_search_regex(id_res, webpage, u'video ID') video_id = self._html_search_regex(id_res, webpage, 'video ID')
else: else:
video_id = mobj.group('id') video_id = mobj.group('id')
return self._extract_video(video_id) return self._extract_video(video_id)
class GenerationQuoiIE(InfoExtractor): class GenerationQuoiIE(InfoExtractor):
IE_NAME = u'france2.fr:generation-quoi' IE_NAME = 'france2.fr:generation-quoi'
_VALID_URL = r'https?://generation-quoi\.france2\.fr/portrait/(?P<name>.*)(\?|$)' _VALID_URL = r'https?://generation-quoi\.france2\.fr/portrait/(?P<name>.*)(\?|$)'
_TEST = { _TEST = {
u'url': u'http://generation-quoi.france2.fr/portrait/garde-a-vous', 'url': 'http://generation-quoi.france2.fr/portrait/garde-a-vous',
u'file': u'k7FJX8VBcvvLmX4wA5Q.mp4', 'file': 'k7FJX8VBcvvLmX4wA5Q.mp4',
u'info_dict': { 'info_dict': {
u'title': u'Génération Quoi - Garde à Vous', 'title': 'Génération Quoi - Garde à Vous',
u'uploader': u'Génération Quoi', 'uploader': 'Génération Quoi',
}, },
u'params': { 'params': {
# It uses Dailymotion # It uses Dailymotion
u'skip_download': True, 'skip_download': True,
}, },
} }
@ -194,20 +197,20 @@ class GenerationQuoiIE(InfoExtractor):
class CultureboxIE(FranceTVBaseInfoExtractor): class CultureboxIE(FranceTVBaseInfoExtractor):
IE_NAME = u'culturebox.francetvinfo.fr' IE_NAME = 'culturebox.francetvinfo.fr'
_VALID_URL = r'https?://culturebox\.francetvinfo\.fr/(?P<name>.*?)(\?|$)' _VALID_URL = r'https?://culturebox\.francetvinfo\.fr/(?P<name>.*?)(\?|$)'
_TEST = { _TEST = {
u'url': u'http://culturebox.francetvinfo.fr/einstein-on-the-beach-au-theatre-du-chatelet-146813', 'url': 'http://culturebox.francetvinfo.fr/einstein-on-the-beach-au-theatre-du-chatelet-146813',
u'info_dict': { 'info_dict': {
u'id': u'EV_6785', 'id': 'EV_6785',
u'ext': u'mp4', 'ext': 'mp4',
u'title': u'Einstein on the beach au Théâtre du Châtelet', 'title': 'Einstein on the beach au Théâtre du Châtelet',
u'description': u'md5:9ce2888b1efefc617b5e58b3f6200eeb', 'description': 'md5:9ce2888b1efefc617b5e58b3f6200eeb',
}, },
u'params': { 'params': {
# m3u8 download # m3u8 download
u'skip_download': True, 'skip_download': True,
}, },
} }
@ -215,5 +218,5 @@ class CultureboxIE(FranceTVBaseInfoExtractor):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
name = mobj.group('name') name = mobj.group('name')
webpage = self._download_webpage(url, name) webpage = self._download_webpage(url, name)
video_id = self._search_regex(r'"http://videos\.francetv\.fr/video/(.*?)"', webpage, u'video id') video_id = self._search_regex(r'"http://videos\.francetv\.fr/video/(.*?)"', webpage, 'video id')
return self._extract_video(video_id) return self._extract_video(video_id)