X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/6d835d4d6903cf5e18ad01844736929e06d16004..aebf9c8f14c9f5d45f8c6aa523ee135fdfd421b2:/youtube_dl/extractor/abc.py diff --git a/youtube_dl/extractor/abc.py b/youtube_dl/extractor/abc.py index 69f8932..6a29e58 100644 --- a/youtube_dl/extractor/abc.py +++ b/youtube_dl/extractor/abc.py @@ -1,41 +1,89 @@ from __future__ import unicode_literals import re -import json from .common import InfoExtractor +from ..utils import ( + ExtractorError, + js_to_json, + int_or_none, +) class ABCIE(InfoExtractor): IE_NAME = 'abc.net.au' - _VALID_URL = r'http://www\.abc\.net\.au/news/[^/]+/[^/]+/(?P\d+)' + _VALID_URL = r'http://www\.abc\.net\.au/news/(?:[^/]+/){1,2}(?P\d+)' - _TEST = { - 'url': 'http://www.abc.net.au/news/2014-07-25/bringing-asylum-seekers-to-australia-would-give/5624716', - 'md5': 'dad6f8ad011a70d9ddf887ce6d5d0742', + _TESTS = [{ + 'url': 'http://www.abc.net.au/news/2014-11-05/australia-to-staff-ebola-treatment-centre-in-sierra-leone/5868334', + 'md5': 'cb3dd03b18455a661071ee1e28344d9f', 'info_dict': { - 'id': '5624716', + 'id': '5868334', 'ext': 'mp4', - 'title': 'Bringing asylum seekers to Australia would give them right to asylum claims: professor', - 'description': 'md5:ba36fa5e27e5c9251fd929d339aea4af', + 'title': 'Australia to help staff Ebola treatment centre in Sierra Leone', + 'description': 'md5:809ad29c67a05f54eb41f2a105693a67', }, - } + 'skip': 'this video has expired', + }, { + 'url': 'http://www.abc.net.au/news/2015-08-17/warren-entsch-introduces-same-sex-marriage-bill/6702326', + 'md5': 'db2a5369238b51f9811ad815b69dc086', + 'info_dict': { + 'id': 'NvqvPeNZsHU', + 'ext': 'mp4', + 'upload_date': '20150816', + 'uploader': 'ABC News (Australia)', + 'description': 'Government backbencher Warren Entsch introduces a cross-party sponsored bill to legalise same-sex marriage, saying the bill is designed to promote "an inclusive Australia, not a divided one.". Read more here: http://ab.co/1Mwc6ef', + 'uploader_id': 'NewsOnABC', + 'title': 'Marriage Equality: Warren Entsch introduces same sex marriage bill', + }, + 'add_ie': ['Youtube'], + 'skip': 'Not accessible from Travis CI server', + }, { + 'url': 'http://www.abc.net.au/news/2015-10-23/nab-lifts-interest-rates-following-westpac-and-cba/6880080', + 'md5': 'b96eee7c9edf4fc5a358a0252881cc1f', + 'info_dict': { + 'id': '6880080', + 'ext': 'mp3', + 'title': 'NAB lifts interest rates, following Westpac and CBA', + 'description': 'md5:f13d8edc81e462fce4a0437c7dc04728', + }, + }, { + 'url': 'http://www.abc.net.au/news/2015-10-19/6866214', + 'only_matching': True, + }] def _real_extract(self, url): video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) - urls_info_json = self._search_regex( - r'inlineVideoData\.push\((.*?)\);', webpage, 'video urls', - flags=re.DOTALL) - urls_info = json.loads(urls_info_json.replace('\'', '"')) + mobj = re.search( + r'inline(?PVideo|Audio|YouTube)Data\.push\((?P[^)]+)\);', + webpage) + if mobj is None: + expired = self._html_search_regex(r'(?s)class="expired-(?:video|audio)".+?(.+?)', webpage, 'expired', None) + if expired: + raise ExtractorError('%s said: %s' % (self.IE_NAME, expired), expected=True) + raise ExtractorError('Unable to extract video urls') + + urls_info = self._parse_json( + mobj.group('json_data'), video_id, transform_source=js_to_json) + + if not isinstance(urls_info, list): + urls_info = [urls_info] + + if mobj.group('type') == 'YouTube': + return self.playlist_result([ + self.url_result(url_info['url']) for url_info in urls_info]) + formats = [{ 'url': url_info['url'], - 'width': int(url_info['width']), - 'height': int(url_info['height']), - 'tbr': int(url_info['bitrate']), - 'filesize': int(url_info['filesize']), + 'vcodec': url_info.get('codec') if mobj.group('type') == 'Video' else 'none', + 'width': int_or_none(url_info.get('width')), + 'height': int_or_none(url_info.get('height')), + 'tbr': int_or_none(url_info.get('bitrate')), + 'filesize': int_or_none(url_info.get('filesize')), } for url_info in urls_info] + self._sort_formats(formats) return {