X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fvevo.py;h=eada13ce920b9f4e892f952242ef87bfac504600;hb=9168308579db01dbd65131d963f5721df9106e1e;hp=f0673972c4632248a367c5ed5ca3de3a1f0a093b;hpb=ff6b7b049b14a061a362e701df8ab4b2f7b82456;p=youtube-dl diff --git a/youtube_dl/extractor/vevo.py b/youtube_dl/extractor/vevo.py index f0673972c..eada13ce9 100644 --- a/youtube_dl/extractor/vevo.py +++ b/youtube_dl/extractor/vevo.py @@ -1,9 +1,7 @@ from __future__ import unicode_literals import re -import json import xml.etree.ElementTree -import datetime from .common import InfoExtractor from ..utils import ( @@ -18,22 +16,55 @@ class VevoIE(InfoExtractor): (currently used by MTVIE) """ _VALID_URL = r'''(?x) - (?:https?://www\.vevo\.com/watch/(?:[^/]+/[^/]+/)?| + (?:https?://www\.vevo\.com/watch/(?:[^/]+/(?:[^/]+/)?)?| https?://cache\.vevo\.com/m/html/embed\.html\?video=| https?://videoplayer\.vevo\.com/embed/embedded\?videoId=| vevo:) (?P[^&?#]+)''' + _TESTS = [{ 'url': 'http://www.vevo.com/watch/hurts/somebody-to-die-for/GB1101300280', - 'file': 'GB1101300280.mp4', "md5": "06bea460acb744eab74a9d7dcb4bfd61", 'info_dict': { + 'id': 'GB1101300280', + 'ext': 'mp4', "upload_date": "20130624", "uploader": "Hurts", "title": "Somebody to Die For", "duration": 230.12, "width": 1920, "height": 1080, + # timestamp and upload_date are often incorrect; seem to change randomly + 'timestamp': int, + } + }, { + 'note': 'v3 SMIL format', + 'url': 'http://www.vevo.com/watch/cassadee-pope/i-wish-i-could-break-your-heart/USUV71302923', + 'md5': '893ec0e0d4426a1d96c01de8f2bdff58', + 'info_dict': { + 'id': 'USUV71302923', + 'ext': 'mp4', + 'upload_date': '20140219', + 'uploader': 'Cassadee Pope', + 'title': 'I Wish I Could Break Your Heart', + 'duration': 226.101, + 'age_limit': 0, + 'timestamp': int, + } + }, { + 'note': 'Age-limited video', + 'url': 'https://www.vevo.com/watch/justin-timberlake/tunnel-vision-explicit/USRV81300282', + 'info_dict': { + 'id': 'USRV81300282', + 'ext': 'mp4', + 'age_limit': 18, + 'title': 'Tunnel Vision (Explicit)', + 'uploader': 'Justin Timberlake', + 'upload_date': 're:2013070[34]', + 'timestamp': int, + }, + 'params': { + 'skip_download': 'true', } }] _SMIL_BASE_URL = 'http://smil.lvl3.vevo.com/' @@ -103,12 +134,40 @@ class VevoIE(InfoExtractor): video_id = mobj.group('id') json_url = 'http://videoplayer.vevo.com/VideoService/AuthenticateVideo?isrc=%s' % video_id - video_info = self._download_json(json_url, video_id)['video'] + response = self._download_json(json_url, video_id) + video_info = response['video'] + + if not video_info: + if 'statusMessage' in response: + raise ExtractorError('%s said: %s' % (self.IE_NAME, response['statusMessage']), expected=True) + raise ExtractorError('Unable to extract videos') formats = self._formats_from_json(video_info) + + is_explicit = video_info.get('isExplicit') + if is_explicit is True: + age_limit = 18 + elif is_explicit is False: + age_limit = 0 + else: + age_limit = None + + # Download SMIL + smil_blocks = sorted(( + f for f in video_info['videoVersions'] + if f['sourceType'] == 13), + key=lambda f: f['version']) + + smil_url = '%s/Video/V2/VFILE/%s/%sr.smil' % ( + self._SMIL_BASE_URL, video_id, video_id.lower()) + if smil_blocks: + smil_url_m = self._search_regex( + r'url="([^"]+)"', smil_blocks[-1]['data'], 'SMIL URL', + fatal=False) + if smil_url_m is not None: + smil_url = smil_url_m + try: - smil_url = '%s/Video/V2/VFILE/%s/%sr.smil' % ( - self._SMIL_BASE_URL, video_id, video_id.lower()) smil_xml = self._download_webpage(smil_url, video_id, 'Downloading SMIL info') formats.extend(self._formats_from_smil(smil_xml)) @@ -120,13 +179,14 @@ class VevoIE(InfoExtractor): timestamp_ms = int(self._search_regex( r'/Date\((\d+)\)/', video_info['launchDate'], 'launch date')) - upload_date = datetime.datetime.fromtimestamp(timestamp_ms // 1000) + return { 'id': video_id, 'title': video_info['title'], 'formats': formats, 'thumbnail': video_info['imageUrl'], - 'upload_date': upload_date.strftime('%Y%m%d'), + 'timestamp': timestamp_ms // 1000, 'uploader': video_info['mainArtists'][0]['artistName'], 'duration': video_info['duration'], + 'age_limit': age_limit, }