_ Git - youtube-dl/blob - youtube_dl/extractor/daum.py

   1 # encoding: utf-8
   2
   3 from __future__ import unicode_literals
   4
   5 from .common import InfoExtractor
   6 from ..compat import compat_urllib_parse
   7 from ..utils import (
   8     int_or_none,
   9     str_to_int,
  10     xpath_text,
  11 )
  12
  13
  14 class DaumIE(InfoExtractor):
  15     _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/v/(?P<id>[^?#&]+)'
  16     IE_NAME = 'daum.net'
  17
  18     _TESTS = [{
  19         'url': 'http://tvpot.daum.net/v/vab4dyeDBysyBssyukBUjBz',
  20         'info_dict': {
  21             'id': 'vab4dyeDBysyBssyukBUjBz',
  22             'ext': 'mp4',
  23             'title': '마크 헌트 vs 안토니오 실바',
  24             'description': 'Mark Hunt vs Antonio Silva',
  25             'upload_date': '20131217',
  26             'duration': 2117,
  27             'view_count': int,
  28             'comment_count': int,
  29         },
  30     }, {
  31         'url': 'http://tvpot.daum.net/v/07dXWRka62Y%24',
  32         'only_matching': True,
  33     }]
  34
  35     def _real_extract(self, url):
  36         video_id = self._match_id(url)
  37         query = compat_urllib_parse.urlencode({'vid': video_id})
  38         info = self._download_xml(
  39             'http://tvpot.daum.net/clip/ClipInfoXml.do?' + query, video_id,
  40             'Downloading video info')
  41         movie_data = self._download_json(
  42             'http://videofarm.daum.net/controller/api/closed/v1_2/IntegratedMovieData.json?' + query,
  43             video_id, 'Downloading video formats info')
  44
  45         formats = []
  46         for format_el in movie_data['output_list']['output_list']:
  47             profile = format_el['profile']
  48             format_query = compat_urllib_parse.urlencode({
  49                 'vid': video_id,
  50                 'profile': profile,
  51             })
  52             url_doc = self._download_xml(
  53                 'http://videofarm.daum.net/controller/api/open/v1_2/MovieLocation.apixml?' + format_query,
  54                 video_id, note='Downloading video data for %s format' % profile)
  55             format_url = url_doc.find('result/url').text
  56             formats.append({
  57                 'url': format_url,
  58                 'format_id': profile,
  59                 'width': int_or_none(format_el.get('width')),
  60                 'height': int_or_none(format_el.get('height')),
  61                 'filesize': int_or_none(format_el.get('filesize')),
  62             })
  63         self._sort_formats(formats)
  64
  65         return {
  66             'id': video_id,
  67             'title': info.find('TITLE').text,
  68             'formats': formats,
  69             'thumbnail': xpath_text(info, 'THUMB_URL'),
  70             'description': xpath_text(info, 'CONTENTS'),
  71             'duration': int_or_none(xpath_text(info, 'DURATION')),
  72             'upload_date': info.find('REGDTTM').text[:8],
  73             'view_count': str_to_int(xpath_text(info, 'PLAY_CNT')),
  74             'comment_count': str_to_int(xpath_text(info, 'COMMENT_CNT')),
  75         }
  76
  77
  78 class DaumClipIE(InfoExtractor):
  79     _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/(?:clip/ClipView.do|mypot/View.do)\?.*?clipid=(?P<id>\d+)'
  80     IE_NAME = 'daum.net'
  81
  82     _TESTS = [{
  83         'url': 'http://tvpot.daum.net/clip/ClipView.do?clipid=52554690',
  84         'info_dict': {
  85             'id': '52554690',
  86             'ext': 'mp4',
  87             'title': 'DOTA 2GETHER 시즌2 6회 - 2부',
  88             'description': 'DOTA 2GETHER 시즌2 6회 - 2부',
  89             'upload_date': '20130831',
  90             'duration': 3868,
  91             'view_count': int,
  92         },
  93     }]
  94
  95     def _real_extract(self, url):
  96         video_id = self._match_id(url)
  97         clip_info = self._download_json(
  98             'http://tvpot.daum.net/mypot/json/GetClipInfo.do?clipid=%s' % video_id,
  99             video_id, 'Downloading clip info')['clip_bean']
 100
 101         return {
 102             '_type': 'url_transparent',
 103             'id': video_id,
 104             'url': 'http://tvpot.daum.net/v/%s' % clip_info['vid'],
 105             'title': clip_info['title'],
 106             'thumbnail': clip_info.get('thumb_url'),
 107             'description': clip_info.get('contents'),
 108             'duration': int_or_none(clip_info.get('duration')),
 109             'upload_date': clip_info.get('up_date')[:8],
 110             'view_count': int_or_none(clip_info.get('play_count')),
 111             'ie_key': 'Daum',
 112         }