From: remitamine <remitamine@gmail.com>
Date: Tue, 29 Dec 2015 09:40:32 +0000 (+0100)
Subject: Merge branch 'daum' of https://github.com/remitamine/youtube-dl into remitamine-daum
X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=commitdiff_plain;h=126d7701b0f88c33b5fbda6fd07ef056b96c9367;hp=-c;p=youtube-dl

Merge branch 'daum' of https://github.com/remitamine/youtube-dl into remitamine-daum
---

126d7701b0f88c33b5fbda6fd07ef056b96c9367
diff --combined youtube_dl/extractor/__init__.py
index 68c9c1288,99b4be002..ec1f80155
--- a/youtube_dl/extractor/__init__.py
+++ b/youtube_dl/extractor/__init__.py
@@@ -3,18 -3,13 +3,18 @@@ from __future__ import unicode_literal
  from .abc import ABCIE
  from .abc7news import Abc7NewsIE
  from .academicearth import AcademicEarthCourseIE
 +from .acast import (
 +    ACastIE,
 +    ACastChannelIE,
 +)
  from .addanime import AddAnimeIE
  from .adobetv import (
      AdobeTVIE,
 +    AdobeTVShowIE,
 +    AdobeTVChannelIE,
      AdobeTVVideoIE,
  )
  from .adultswim import AdultSwimIE
 -from .aftenposten import AftenpostenIE
  from .aftonbladet import AftonbladetIE
  from .airmozilla import AirMozillaIE
  from .aljazeera import AlJazeeraIE
@@@ -25,10 -20,7 +25,10 @@@ from .aol import AolI
  from .allocine import AllocineIE
  from .aparat import AparatIE
  from .appleconnect import AppleConnectIE
 -from .appletrailers import AppleTrailersIE
 +from .appletrailers import (
 +    AppleTrailersIE,
 +    AppleTrailersSectionIE,
 +)
  from .archiveorg import ArchiveOrgIE
  from .ard import (
      ARDIE,
@@@ -46,7 -38,6 +46,7 @@@ from .arte import 
  )
  from .atresplayer import AtresPlayerIE
  from .atttechchannel import ATTTechChannelIE
 +from .audimedia import AudiMediaIE
  from .audiomack import AudiomackIE, AudiomackAlbumIE
  from .azubu import AzubuIE
  from .baidu import BaiduVideoIE
@@@ -63,19 -54,13 +63,19 @@@ from .beatportpro import BeatportProI
  from .bet import BetIE
  from .bild import BildIE
  from .bilibili import BiliBiliIE
 +from .bleacherreport import (
 +    BleacherReportIE,
 +    BleacherReportCMSIE,
 +)
  from .blinkx import BlinkxIE
 -from .bliptv import BlipTVIE, BlipTVUserIE
  from .bloomberg import BloombergIE
  from .bpb import BpbIE
  from .br import BRIE
  from .breakcom import BreakIE
 -from .brightcove import BrightcoveIE
 +from .brightcove import (
 +    BrightcoveLegacyIE,
 +    BrightcoveNewIE,
 +)
  from .buzzfeed import BuzzFeedIE
  from .byutv import BYUtvIE
  from .c56 import C56IE
@@@ -83,6 -68,7 +83,6 @@@ from .camdemy import 
      CamdemyIE,
      CamdemyFolderIE
  )
 -from .canal13cl import Canal13clIE
  from .canalplus import CanalplusIE
  from .canalc2 import Canalc2IE
  from .cbs import CBSIE
@@@ -133,22 -119,18 +133,25 @@@ from .dailymotion import 
      DailymotionUserIE,
      DailymotionCloudIE,
  )
- from .daum import DaumIE
+ from .daum import (
+     DaumIE,
+     DaumClipIE,
+ )
  from .dbtv import DBTVIE
 -from .dcn import DCNIE
 +from .dcn import (
 +    DCNIE,
 +    DCNVideoIE,
 +    DCNLiveIE,
 +    DCNSeasonIE,
 +)
  from .dctp import DctpTvIE
  from .deezer import DeezerPlaylistIE
 +from .democracynow import DemocracynowIE
  from .dfb import DFBIE
  from .dhm import DHMIE
  from .dotsub import DotsubIE
  from .douyutv import DouyuTVIE
 +from .dplay import DPlayIE
  from .dramafever import (
      DramaFeverIE,
      DramaFeverSeriesIE,
@@@ -203,10 -185,7 +206,10 @@@ from .fourtube import FourTubeI
  from .foxgay import FoxgayIE
  from .foxnews import FoxNewsIE
  from .foxsports import FoxSportsIE
 -from .franceculture import FranceCultureIE
 +from .franceculture import (
 +    FranceCultureIE,
 +    FranceCultureEmissionIE,
 +)
  from .franceinter import FranceInterIE
  from .francetv import (
      PluzzIE,
@@@ -218,9 -197,7 +221,9 @@@
  from .freesound import FreesoundIE
  from .freespeech import FreespeechIE
  from .freevideo import FreeVideoIE
 +from .funimation import FunimationIE
  from .funnyordie import FunnyOrDieIE
 +from .gameinformer import GameInformerIE
  from .gamekings import GamekingsIE
  from .gameone import (
      GameOneIE,
@@@ -237,18 -214,14 +240,18 @@@ from .gfycat import GfycatI
  from .giantbomb import GiantBombIE
  from .giga import GigaIE
  from .glide import GlideIE
 -from .globo import GloboIE
 +from .globo import (
 +    GloboIE,
 +    GloboArticleIE,
 +)
  from .godtube import GodTubeIE
  from .goldenmoustache import GoldenMoustacheIE
  from .golem import GolemIE
 +from .googledrive import GoogleDriveIE
  from .googleplus import GooglePlusIE
  from .googlesearch import GoogleSearchIE
 -from .gorillavid import GorillaVidIE
  from .goshgay import GoshgayIE
 +from .gputechconf import GPUTechConfIE
  from .groupon import GrouponIE
  from .hark import HarkIE
  from .hearthisat import HearThisAtIE
@@@ -261,17 -234,12 +264,17 @@@ from .history import HistoryI
  from .hitbox import HitboxIE, HitboxLiveIE
  from .hornbunny import HornBunnyIE
  from .hotnewhiphop import HotNewHipHopIE
 +from .hotstar import HotStarIE
  from .howcast import HowcastIE
  from .howstuffworks import HowStuffWorksIE
  from .huffpost import HuffPostIE
  from .hypem import HypemIE
  from .iconosquare import IconosquareIE
 -from .ign import IGNIE, OneUPIE
 +from .ign import (
 +    IGNIE,
 +    OneUPIE,
 +    PCMagIE,
 +)
  from .imdb import (
      ImdbIE,
      ImdbListIE
@@@ -300,7 -268,6 +303,7 @@@ from .jadorecettepub import JadoreCette
  from .jeuxvideo import JeuxVideoIE
  from .jove import JoveIE
  from .jukebox import JukeboxIE
 +from .jwplatform import JWPlatformIE
  from .jpopsukitv import JpopsukiIE
  from .kaltura import KalturaIE
  from .kanalplay import KanalPlayIE
@@@ -355,7 -322,6 +358,7 @@@ from .lynda import 
  from .m6 import M6IE
  from .macgamestore import MacGameStoreIE
  from .mailru import MailRuIE
 +from .makertv import MakerTVIE
  from .malemotion import MalemotionIE
  from .mdr import MDRIE
  from .metacafe import MetacafeIE
@@@ -379,6 -345,7 +382,6 @@@ from .motherless import MotherlessI
  from .motorsport import MotorsportIE
  from .movieclips import MovieClipsIE
  from .moviezine import MoviezineIE
 -from .movshare import MovShareIE
  from .mtv import (
      MTVIE,
      MTVServicesEmbeddedIE,
@@@ -444,22 -411,14 +447,22 @@@ from .noco import NocoI
  from .normalboots import NormalbootsIE
  from .nosvideo import NosVideoIE
  from .nova import NovaIE
 -from .novamov import NovaMovIE
 +from .novamov import (
 +    NovaMovIE,
 +    WholeCloudIE,
 +    NowVideoIE,
 +    VideoWeedIE,
 +    CloudTimeIE,
 +)
  from .nowness import (
      NownessIE,
      NownessPlaylistIE,
      NownessSeriesIE,
  )
 -from .nowtv import NowTVIE
 -from .nowvideo import NowVideoIE
 +from .nowtv import (
 +    NowTVIE,
 +    NowTVListIE,
 +)
  from .npo import (
      NPOIE,
      NPOLiveIE,
@@@ -496,7 -455,10 +499,7 @@@ from .orf import 
  from .parliamentliveuk import ParliamentLiveUKIE
  from .patreon import PatreonIE
  from .pbs import PBSIE
 -from .periscope import (
 -    PeriscopeIE,
 -    QuickscopeIE,
 -)
 +from .periscope import PeriscopeIE
  from .philharmoniedeparis import PhilharmonieDeParisIE
  from .phoenix import PhoenixIE
  from .photobucket import PhotobucketIE
@@@ -540,10 -502,7 +543,10 @@@ from .radiode import RadioDeI
  from .radiojavan import RadioJavanIE
  from .radiobremen import RadioBremenIE
  from .radiofrance import RadioFranceIE
 -from .rai import RaiIE
 +from .rai import (
 +    RaiTVIE,
 +    RaiIE,
 +)
  from .rbmaradio import RBMARadioIE
  from .rds import RDSIE
  from .redtube import RedTubeIE
@@@ -591,10 -550,6 +594,10 @@@ from .shahid import ShahidI
  from .shared import SharedIE
  from .sharesix import ShareSixIE
  from .sina import SinaIE
 +from .skynewsarabia import (
 +    SkyNewsArabiaIE,
 +    SkyNewsArabiaArticleIE,
 +)
  from .slideshare import SlideshareIE
  from .slutload import SlutloadIE
  from .smotri import (
@@@ -609,12 -564,15 +612,12 @@@ from .snagfilms import 
  )
  from .snotr import SnotrIE
  from .sohu import SohuIE
 -from .soompi import (
 -    SoompiIE,
 -    SoompiShowIE,
 -)
  from .soundcloud import (
      SoundcloudIE,
      SoundcloudSetIE,
      SoundcloudUserIE,
 -    SoundcloudPlaylistIE
 +    SoundcloudPlaylistIE,
 +    SoundcloudSearchIE
  )
  from .soundgasm import (
      SoundgasmIE,
@@@ -667,7 -625,6 +670,7 @@@ from .teachingchannel import TeachingCh
  from .teamcoco import TeamcocoIE
  from .techtalks import TechTalksIE
  from .ted import TEDIE
 +from .tele13 import Tele13IE
  from .telebruxelles import TeleBruxellesIE
  from .telecinco import TelecincoIE
  from .telegraaf import TelegraafIE
@@@ -677,7 -634,6 +680,7 @@@ from .tenplay import TenPlayI
  from .testurl import TestURLIE
  from .testtube import TestTubeIE
  from .tf1 import TF1IE
 +from .theintercept import TheInterceptIE
  from .theonion import TheOnionIE
  from .theplatform import (
      ThePlatformIE,
@@@ -697,7 -653,6 +700,7 @@@ from .tnaflix import 
      EMPFlixIE,
      MovieFapIE,
  )
 +from .toggle import ToggleIE
  from .thvideo import (
      THVideoIE,
      THVideoPlaylistIE
@@@ -711,13 -666,7 +714,13 @@@ from .tube8 import Tube8I
  from .tubitv import TubiTvIE
  from .tudou import TudouIE
  from .tumblr import TumblrIE
 -from .tunein import TuneInIE
 +from .tunein import (
 +    TuneInClipIE,
 +    TuneInStationIE,
 +    TuneInProgramIE,
 +    TuneInTopicIE,
 +    TuneInShortenerIE,
 +)
  from .turbo import TurboIE
  from .tutv import TutvIE
  from .tv2 import (
@@@ -774,15 -723,16 +777,15 @@@ from .vh1 import VH1I
  from .vice import ViceIE
  from .viddler import ViddlerIE
  from .videodetective import VideoDetectiveIE
 -from .videolecturesnet import VideoLecturesNetIE
  from .videofyme import VideofyMeIE
  from .videomega import VideoMegaIE
  from .videopremium import VideoPremiumIE
  from .videott import VideoTtIE
 -from .videoweed import VideoWeedIE
  from .vidme import VidmeIE
  from .vidzi import VidziIE
  from .vier import VierIE, VierVideosIE
  from .viewster import ViewsterIE
 +from .viidea import ViideaIE
  from .vimeo import (
      VimeoIE,
      VimeoAlbumIE,
@@@ -835,7 -785,6 +838,7 @@@ from .wrzuta import WrzutaI
  from .wsj import WSJIE
  from .xbef import XBefIE
  from .xboxclips import XboxClipsIE
 +from .xfileshare import XFileShareIE
  from .xhamster import (
      XHamsterIE,
      XHamsterEmbedIE,
@@@ -879,7 -828,6 +882,7 @@@ from .youtube import 
      YoutubeTruncatedIDIE,
      YoutubeTruncatedURLIE,
      YoutubeUserIE,
 +    YoutubePlaylistsIE,
      YoutubeWatchLaterIE,
  )
  from .zapiks import ZapiksIE
diff --combined youtube_dl/extractor/daum.py
index 9a94cf361,0f5686e07..a083cc0dc
--- a/youtube_dl/extractor/daum.py
+++ b/youtube_dl/extractor/daum.py
@@@ -2,71 -2,45 +2,51 @@@
  
  from __future__ import unicode_literals
  
- import re
- 
  from .common import InfoExtractor
- from ..compat import (
-     compat_urllib_parse,
+ from ..compat import compat_urllib_parse
 -from ..utils import int_or_none
++from ..utils import (
++    int_or_none,
++    str_to_int,
++    xpath_text,
 +)
  
  
  class DaumIE(InfoExtractor):
-     _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/(?:v/|.*?clipid=)(?P<id>[^?#&]+)'
+     _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/v/(?P<id>[^?#&]+)'
      IE_NAME = 'daum.net'
  
      _TESTS = [{
-         'url': 'http://tvpot.daum.net/clip/ClipView.do?clipid=52554690',
-         'info_dict': {
-             'id': '52554690',
-             'ext': 'mp4',
-             'title': 'DOTA 2GETHER 시즌2 6회 - 2부',
-             'description': 'DOTA 2GETHER 시즌2 6회 - 2부',
-             'upload_date': '20130831',
-             'duration': 3868,
-         },
-     }, {
-         # Test for https://github.com/rg3/youtube-dl/issues/7949
-         'url': 'http://tvpot.daum.net/mypot/View.do?ownerid=M1O35s8HPOo0&clipid=73147290',
-         'md5': 'c92d78bcee4424451f1667f275c1dc97',
+         'url': 'http://tvpot.daum.net/v/vab4dyeDBysyBssyukBUjBz',
          'info_dict': {
-             'id': '73147290',
+             'id': 'vab4dyeDBysyBssyukBUjBz',
              'ext': 'mp4',
-             'title': '싸이 - 나팔바지 [유희열의 스케치북] 299회 20151218',
-             'description': '싸이 - 나팔바지',
-             'upload_date': '20151219',
-             'duration': 232,
+             'title': '마크 헌트 vs 안토니오 실바',
+             'description': 'Mark Hunt vs Antonio Silva',
+             'upload_date': '20131217',
+             'duration': 2117,
++            'view_count': int,
++            'comment_count': int,
          },
-     }, {
-         'url': 'http://tvpot.daum.net/v/vab4dyeDBysyBssyukBUjBz',
-         'only_matching': True,
      }, {
          'url': 'http://tvpot.daum.net/v/07dXWRka62Y%24',
          'only_matching': True,
      }]
  
      def _real_extract(self, url):
-         mobj = re.match(self._VALID_URL, url)
-         video_id = mobj.group('id')
-         canonical_url = 'http://tvpot.daum.net/v/%s' % video_id
-         webpage = self._download_webpage(canonical_url, video_id)
-         og_url = self._og_search_url(webpage, default=None) or self._search_regex(
-             r'<link[^>]+rel=(["\'])canonical\1[^>]+href=(["\'])(?P<url>.+?)\2',
-             webpage, 'canonical url', group='url')
-         full_id = self._search_regex(
-             r'tvpot\.daum\.net/v/([^/]+)', og_url, 'full id')
-         query = compat_urllib_parse.urlencode({'vid': full_id})
+         video_id = self._match_id(url)
+         query = compat_urllib_parse.urlencode({'vid': video_id})
          info = self._download_xml(
              'http://tvpot.daum.net/clip/ClipInfoXml.do?' + query, video_id,
              'Downloading video info')
-         urls = self._download_xml(
-             'http://videofarm.daum.net/controller/api/open/v1_2/MovieData.apixml?' + query,
+         movie_data = self._download_json(
+             'http://videofarm.daum.net/controller/api/closed/v1_2/IntegratedMovieData.json?' + query,
              video_id, 'Downloading video formats info')
  
          formats = []
-         for format_el in urls.findall('result/output_list/output_list'):
-             profile = format_el.attrib['profile']
+         for format_el in movie_data['output_list']['output_list']:
+             profile = format_el['profile']
              format_query = compat_urllib_parse.urlencode({
-                 'vid': full_id,
+                 'vid': video_id,
                  'profile': profile,
              })
              url_doc = self._download_xml(
@@@ -76,14 -50,53 +56,57 @@@
              formats.append({
                  'url': format_url,
                  'format_id': profile,
+                 'width': int_or_none(format_el.get('width')),
+                 'height': int_or_none(format_el.get('height')),
+                 'filesize': int_or_none(format_el.get('filesize')),
              })
+         self._sort_formats(formats)
  
          return {
              'id': video_id,
              'title': info.find('TITLE').text,
              'formats': formats,
-             'thumbnail': self._og_search_thumbnail(webpage),
 -            'thumbnail': info.find('THUMB_URL').text,
--            'description': info.find('CONTENTS').text,
-             'duration': int(info.find('DURATION').text),
 -            'duration': int_or_none(info.find('DURATION').text),
++            'thumbnail': xpath_text(info, 'THUMB_URL'),
++            'description': xpath_text(info, 'CONTENTS'),
++            'duration': int_or_none(xpath_text(info, 'DURATION')),
              'upload_date': info.find('REGDTTM').text[:8],
++            'view_count': str_to_int(xpath_text(info, 'PLAY_CNT')),
++            'comment_count': str_to_int(xpath_text(info, 'COMMENT_CNT')),
+         }
+ 
+ 
+ class DaumClipIE(InfoExtractor):
 -    _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/.*?clipid=(?P<id>\d+)'
++    _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/(?:clip/ClipView.do|mypot/View.do)\?.*?clipid=(?P<id>\d+)'
+     IE_NAME = 'daum.net'
+ 
+     _TESTS = [{
+         'url': 'http://tvpot.daum.net/clip/ClipView.do?clipid=52554690',
+         'info_dict': {
+             'id': '52554690',
+             'ext': 'mp4',
+             'title': 'DOTA 2GETHER 시즌2 6회 - 2부',
+             'description': 'DOTA 2GETHER 시즌2 6회 - 2부',
+             'upload_date': '20130831',
+             'duration': 3868,
+             'view_count': int,
+         },
+     }]
+ 
+     def _real_extract(self, url):
+         video_id = self._match_id(url)
 -        clip_info = self._download_json('http://tvpot.daum.net/mypot/json/GetClipInfo.do?clipid=%s' % video_id, video_id)['clip_bean']
++        clip_info = self._download_json(
++            'http://tvpot.daum.net/mypot/json/GetClipInfo.do?clipid=%s' % video_id,
++            video_id, 'Downloading clip info')['clip_bean']
+ 
+         return {
+             '_type': 'url_transparent',
+             'id': video_id,
+             'url': 'http://tvpot.daum.net/v/%s' % clip_info['vid'],
+             'title': clip_info['title'],
+             'thumbnail': clip_info.get('thumb_url'),
+             'description': clip_info.get('contents'),
+             'duration': int_or_none(clip_info.get('duration')),
+             'upload_date': clip_info.get('up_date')[:8],
+             'view_count': int_or_none(clip_info.get('play_count')),
+             'ie_key': 'Daum',
          }