X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/415fdb62500dca2e22067a05008dfbf87c75b662..d4ff594119bc679aa175947eb59a97bee8f966f4:/youtube_dl/extractor/appletrailers.py diff --git a/youtube_dl/extractor/appletrailers.py b/youtube_dl/extractor/appletrailers.py index dc8657b..287f71e 100644 --- a/youtube_dl/extractor/appletrailers.py +++ b/youtube_dl/extractor/appletrailers.py @@ -4,8 +4,9 @@ import re import json from .common import InfoExtractor +from ..compat import compat_urlparse from ..utils import ( - compat_urlparse, + int_or_none, ) @@ -69,15 +70,17 @@ class AppleTrailersIE(InfoExtractor): uploader_id = mobj.group('company') playlist_url = compat_urlparse.urljoin(url, 'includes/playlists/itunes.inc') + def fix_html(s): s = re.sub(r'(?s).*?', '', s) s = re.sub(r'', r'', s) # The ' in the onClick attributes are not escaped, it couldn't be parsed # like: http://trailers.apple.com/trailers/wb/gravity/ + def _clean_json(m): return 'iTunes.playURL(%s);' % m.group(1).replace('\'', ''') s = re.sub(self._JSON_RE, _clean_json, s) - s = '' + s + u'' + s = '%s' % s return s doc = self._download_xml(playlist_url, movie, transform_source=fix_html) @@ -85,7 +88,7 @@ class AppleTrailersIE(InfoExtractor): for li in doc.findall('./div/ul/li'): on_click = li.find('.//a').attrib['onClick'] trailer_info_json = self._search_regex(self._JSON_RE, - on_click, 'trailer info') + on_click, 'trailer info') trailer_info = json.loads(trailer_info_json) title = trailer_info['title'] video_id = movie + '-' + re.sub(r'[^a-zA-Z0-9]', '', title).lower() @@ -110,8 +113,8 @@ class AppleTrailersIE(InfoExtractor): formats.append({ 'url': format_url, 'format': format['type'], - 'width': format['width'], - 'height': int(format['height']), + 'width': int_or_none(format['width']), + 'height': int_or_none(format['height']), }) self._sort_formats(formats) @@ -119,14 +122,15 @@ class AppleTrailersIE(InfoExtractor): playlist.append({ '_type': 'video', 'id': video_id, - 'title': title, 'formats': formats, 'title': title, 'duration': duration, 'thumbnail': thumbnail, 'upload_date': upload_date, 'uploader_id': uploader_id, - 'user_agent': 'QuickTime compatible (youtube-dl)', + 'http_headers': { + 'User-Agent': 'QuickTime compatible (youtube-dl)', + }, }) return {