X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/9dc487f48b50767cf540fa36c3de2c386fd74c04..d037080638e8dbb0ebb2ac9cf23b9671518340f9:/youtube_dl/extractor/aol.py diff --git a/youtube_dl/extractor/aol.py b/youtube_dl/extractor/aol.py index 42c21bf..e87994a 100644 --- a/youtube_dl/extractor/aol.py +++ b/youtube_dl/extractor/aol.py @@ -4,19 +4,24 @@ from __future__ import unicode_literals import re from .common import InfoExtractor +from ..compat import ( + compat_parse_qs, + compat_urllib_parse_urlparse, +) from ..utils import ( ExtractorError, int_or_none, + url_or_none, ) class AolIE(InfoExtractor): - IE_NAME = 'on.aol.com' - _VALID_URL = r'(?:aol-video:|https?://on\.aol\.com/(?:[^/]+/)*(?:[^/?#&]+-)?)(?P[^/?#&]+)' + IE_NAME = 'aol.com' + _VALID_URL = r'(?:aol-video:|https?://(?:www\.)?aol\.(?:com|ca|co\.uk|de|jp)/video/(?:[^/]+/)*)(?P[0-9a-f]+)' _TESTS = [{ # video with 5min ID - 'url': 'http://on.aol.com/video/u-s--official-warns-of-largest-ever-irs-phone-scam-518167793?icid=OnHomepageC2Wide_MustSee_Img', + 'url': 'https://www.aol.com/video/view/u-s--official-warns-of-largest-ever-irs-phone-scam/518167793/', 'md5': '18ef68f48740e86ae94b98da815eec42', 'info_dict': { 'id': '518167793', @@ -33,7 +38,7 @@ class AolIE(InfoExtractor): } }, { # video with vidible ID - 'url': 'http://on.aol.com/video/netflix-is-raising-rates-5707d6b8e4b090497b04f706?context=PC:homepage:PL1944:1460189336183', + 'url': 'https://www.aol.com/video/view/netflix-is-raising-rates/5707d6b8e4b090497b04f706/', 'info_dict': { 'id': '5707d6b8e4b090497b04f706', 'ext': 'mp4', @@ -48,16 +53,28 @@ class AolIE(InfoExtractor): 'skip_download': True, } }, { - 'url': 'http://on.aol.com/partners/abc-551438d309eab105804dbfe8/sneak-peek-was-haley-really-framed-570eaebee4b0448640a5c944', + 'url': 'https://www.aol.com/video/view/park-bench-season-2-trailer/559a1b9be4b0c3bfad3357a7/', 'only_matching': True, }, { - 'url': 'http://on.aol.com/shows/park-bench-shw518173474-559a1b9be4b0c3bfad3357a7?context=SH:SHW518173474:PL4327:1460619712763', + 'url': 'https://www.aol.com/video/view/donald-trump-spokeswoman-tones-down-megyn-kelly-attacks/519442220/', 'only_matching': True, }, { - 'url': 'http://on.aol.com/video/519442220', + 'url': 'aol-video:5707d6b8e4b090497b04f706', 'only_matching': True, }, { - 'url': 'aol-video:5707d6b8e4b090497b04f706', + 'url': 'https://www.aol.com/video/playlist/PL8245/5ca79d19d21f1a04035db606/', + 'only_matching': True, + }, { + 'url': 'https://www.aol.ca/video/view/u-s-woman-s-family-arrested-for-murder-first-pinned-on-panhandler-police/5c7ccf45bc03931fa04b2fe1/', + 'only_matching': True, + }, { + 'url': 'https://www.aol.co.uk/video/view/-one-dead-and-22-hurt-in-bus-crash-/5cb3a6f3d21f1a072b457347/', + 'only_matching': True, + }, { + 'url': 'https://www.aol.de/video/view/eva-braun-privataufnahmen-von-hitlers-geliebter-werden-digitalisiert/5cb2d49de98ab54c113d3d5d/', + 'only_matching': True, + }, { + 'url': 'https://www.aol.jp/video/playlist/5a28e936a1334d000137da0c/5a28f3151e642219fde19831/', 'only_matching': True, }] @@ -72,12 +89,12 @@ class AolIE(InfoExtractor): video_data = response['data'] formats = [] - m3u8_url = video_data.get('videoMasterPlaylist') + m3u8_url = url_or_none(video_data.get('videoMasterPlaylist')) if m3u8_url: formats.extend(self._extract_m3u8_formats( m3u8_url, video_id, 'mp4', m3u8_id='hls', fatal=False)) for rendition in video_data.get('renditions', []): - video_url = rendition.get('url') + video_url = url_or_none(rendition.get('url')) if not video_url: continue ext = rendition.get('format') @@ -95,6 +112,12 @@ class AolIE(InfoExtractor): 'width': int(mobj.group(1)), 'height': int(mobj.group(2)), }) + else: + qs = compat_parse_qs(compat_urllib_parse_urlparse(video_url).query) + f.update({ + 'width': int_or_none(qs.get('w', [None])[0]), + 'height': int_or_none(qs.get('h', [None])[0]), + }) formats.append(f) self._sort_formats(formats, ('width', 'height', 'tbr', 'format_id')) @@ -108,26 +131,3 @@ class AolIE(InfoExtractor): 'uploader': video_data.get('videoOwner'), 'formats': formats, } - - -class AolFeaturesIE(InfoExtractor): - IE_NAME = 'features.aol.com' - _VALID_URL = r'https?://features\.aol\.com/video/(?P[^/?#]+)' - - _TESTS = [{ - 'url': 'http://features.aol.com/video/behind-secret-second-careers-late-night-talk-show-hosts', - 'md5': '7db483bb0c09c85e241f84a34238cc75', - 'info_dict': { - 'id': '519507715', - 'ext': 'mp4', - 'title': 'What To Watch - February 17, 2016', - }, - 'add_ie': ['FiveMin'], - }] - - def _real_extract(self, url): - display_id = self._match_id(url) - webpage = self._download_webpage(url, display_id) - return self.url_result(self._search_regex( - r'