Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/moevideo.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5 import re
   6
   7 from .common import InfoExtractor
   8 from ..utils import (
   9     ExtractorError,
  10     int_or_none,
  11     sanitized_Request,
  12     urlencode_postdata,
  13 )
  14
  15
  16 class MoeVideoIE(InfoExtractor):
  17     IE_DESC = 'LetitBit video services: moevideo.net, playreplay.net and videochart.net'
  18     _VALID_URL = r'''(?x)
  19         https?://(?P<host>(?:www\.)?
  20         (?:(?:moevideo|playreplay|videochart)\.net))/
  21         (?:video|framevideo)/(?P<id>[0-9]+\.[0-9A-Za-z]+)'''
  22     _API_URL = 'http://api.letitbit.net/'
  23     _API_KEY = 'tVL0gjqo5'
  24     _TESTS = [
  25         {
  26             'url': 'http://moevideo.net/video/00297.0036103fe3d513ef27915216fd29',
  27             'md5': '129f5ae1f6585d0e9bb4f38e774ffb3a',
  28             'info_dict': {
  29                 'id': '00297.0036103fe3d513ef27915216fd29',
  30                 'ext': 'flv',
  31                 'title': 'Sink cut out machine',
  32                 'description': 'md5:f29ff97b663aefa760bf7ca63c8ca8a8',
  33                 'thumbnail': 're:^https?://.*\.jpg$',
  34                 'width': 540,
  35                 'height': 360,
  36                 'duration': 179,
  37                 'filesize': 17822500,
  38             }
  39         },
  40         {
  41             'url': 'http://playreplay.net/video/77107.7f325710a627383d40540d8e991a',
  42             'md5': '74f0a014d5b661f0f0e2361300d1620e',
  43             'info_dict': {
  44                 'id': '77107.7f325710a627383d40540d8e991a',
  45                 'ext': 'flv',
  46                 'title': 'Operacion Condor.',
  47                 'description': 'md5:7e68cb2fcda66833d5081c542491a9a3',
  48                 'thumbnail': 're:^https?://.*\.jpg$',
  49                 'width': 480,
  50                 'height': 296,
  51                 'duration': 6027,
  52                 'filesize': 588257923,
  53             },
  54             'skip': 'Video has been removed',
  55         },
  56     ]
  57
  58     def _real_extract(self, url):
  59         mobj = re.match(self._VALID_URL, url)
  60         video_id = mobj.group('id')
  61
  62         webpage = self._download_webpage(
  63             'http://%s/video/%s' % (mobj.group('host'), video_id),
  64             video_id, 'Downloading webpage')
  65
  66         title = self._og_search_title(webpage)
  67         thumbnail = self._og_search_thumbnail(webpage)
  68         description = self._og_search_description(webpage)
  69
  70         r = [
  71             self._API_KEY,
  72             [
  73                 'preview/flv_link',
  74                 {
  75                     'uid': video_id,
  76                 },
  77             ],
  78         ]
  79         r_json = json.dumps(r)
  80         post = urlencode_postdata({'r': r_json})
  81         req = sanitized_Request(self._API_URL, post)
  82         req.add_header('Content-type', 'application/x-www-form-urlencoded')
  83
  84         response = self._download_json(req, video_id)
  85         if response['status'] != 'OK':
  86             raise ExtractorError(
  87                 '%s returned error: %s' % (self.IE_NAME, response['data']),
  88                 expected=True
  89             )
  90         item = response['data'][0]
  91         video_url = item['link']
  92         duration = int_or_none(item['length'])
  93         width = int_or_none(item['width'])
  94         height = int_or_none(item['height'])
  95         filesize = int_or_none(item['convert_size'])
  96
  97         formats = [{
  98             'format_id': 'sd',
  99             'http_headers': {'Range': 'bytes=0-'},  # Required to download
 100             'url': video_url,
 101             'width': width,
 102             'height': height,
 103             'filesize': filesize,
 104         }]
 105
 106         return {
 107             'id': video_id,
 108             'title': title,
 109             'thumbnail': thumbnail,
 110             'description': description,
 111             'duration': duration,
 112             'formats': formats,
 113         }