From 1e108029907ca28b75f37d2cf0bf25bcabbfbdac Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sun, 25 Jan 2015 05:21:39 +0600 Subject: [PATCH] [krasview] Fix extraction --- youtube_dl/extractor/krasview.py | 22 +++++++++------------- 1 file changed, 9 insertions(+), 13 deletions(-) diff --git a/youtube_dl/extractor/krasview.py b/youtube_dl/extractor/krasview.py index 6f3d2345b6..e46954b474 100644 --- a/youtube_dl/extractor/krasview.py +++ b/youtube_dl/extractor/krasview.py @@ -2,18 +2,17 @@ from __future__ import unicode_literals import json -import re from .common import InfoExtractor from ..utils import ( int_or_none, - unescapeHTML, + js_to_json, ) class KrasViewIE(InfoExtractor): IE_DESC = 'Красвью' - _VALID_URL = r'https?://krasview\.ru/video/(?P\d+)' + _VALID_URL = r'https?://krasview\.ru/(?:video|embed)/(?P\d+)' _TEST = { 'url': 'http://krasview.ru/video/512228', @@ -29,20 +28,18 @@ class KrasViewIE(InfoExtractor): } def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') + video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) - flashvars = json.loads(self._search_regex( - r'flashvars\s*:\s*({.+?})\s*}\);', webpage, 'flashvars')) + flashvars = json.loads(js_to_json(self._search_regex( + r'video_Init\(({.+?})', webpage, 'flashvars'))) video_url = flashvars['url'] - title = unescapeHTML(flashvars['title']) - description = unescapeHTML(flashvars.get('subtitle') or self._og_search_description(webpage, default=None)) - thumbnail = flashvars['image'] - duration = int(flashvars['duration']) - filesize = int(flashvars['size']) + title = self._og_search_title(webpage) + description = self._og_search_description(webpage, default=None) + thumbnail = flashvars.get('image') or self._og_search_thumbnail(webpage) + duration = int_or_none(flashvars.get('duration')) width = int_or_none(self._og_search_property('video:width', webpage, 'video width')) height = int_or_none(self._og_search_property('video:height', webpage, 'video height')) @@ -53,7 +50,6 @@ def _real_extract(self, url): 'description': description, 'thumbnail': thumbnail, 'duration': duration, - 'filesize': filesize, 'width': width, 'height': height, }