[abc:iview] Unescape title and series meta fields

This commit is contained in:
xofe 2018-03-27 15:08:40 +00:00 committed by Sergey M
parent 99c3091850
commit 9e6a418015
1 changed files with 11 additions and 9 deletions

View File

@ -13,6 +13,7 @@ from ..utils import (
int_or_none, int_or_none,
parse_iso8601, parse_iso8601,
try_get, try_get,
unescapeHTML,
update_url_query, update_url_query,
) )
@ -109,16 +110,17 @@ class ABCIViewIE(InfoExtractor):
# ABC iview programs are normally available for 14 days only. # ABC iview programs are normally available for 14 days only.
_TESTS = [{ _TESTS = [{
'url': 'http://iview.abc.net.au/programs/call-the-midwife/ZW0898A003S00', 'url': 'https://iview.abc.net.au/programs/ben-and-hollys-little-kingdom/ZY9247A021S00',
'md5': 'cde42d728b3b7c2b32b1b94b4a548afc', 'md5': 'cde42d728b3b7c2b32b1b94b4a548afc',
'info_dict': { 'info_dict': {
'id': 'ZW0898A003S00', 'id': 'ZY9247A021S00',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Series 5 Ep 3', 'title': "Gaston's Visit",
'description': 'md5:e0ef7d4f92055b86c4f33611f180ed79', 'series': "Ben And Holly's Little Kingdom",
'upload_date': '20171228', 'description': 'md5:18db170ad71cf161e006a4c688e33155',
'uploader_id': 'abc1', 'upload_date': '20180318',
'timestamp': 1514499187, 'uploader_id': 'abc4kids',
'timestamp': 1521400959,
}, },
'params': { 'params': {
'skip_download': True, 'skip_download': True,
@ -169,12 +171,12 @@ class ABCIViewIE(InfoExtractor):
return { return {
'id': video_id, 'id': video_id,
'title': title, 'title': unescapeHTML(title),
'description': self._html_search_meta(['og:description', 'twitter:description'], webpage), 'description': self._html_search_meta(['og:description', 'twitter:description'], webpage),
'thumbnail': self._html_search_meta(['og:image', 'twitter:image:src'], webpage), 'thumbnail': self._html_search_meta(['og:image', 'twitter:image:src'], webpage),
'duration': int_or_none(video_params.get('eventDuration')), 'duration': int_or_none(video_params.get('eventDuration')),
'timestamp': parse_iso8601(video_params.get('pubDate'), ' '), 'timestamp': parse_iso8601(video_params.get('pubDate'), ' '),
'series': video_params.get('seriesTitle'), 'series': unescapeHTML(video_params.get('seriesTitle')),
'series_id': video_params.get('seriesHouseNumber') or video_id[:7], 'series_id': video_params.get('seriesHouseNumber') or video_id[:7],
'episode_number': int_or_none(self._html_search_meta('episodeNumber', webpage, default=None)), 'episode_number': int_or_none(self._html_search_meta('episodeNumber', webpage, default=None)),
'episode': self._html_search_meta('episode_title', webpage, default=None), 'episode': self._html_search_meta('episode_title', webpage, default=None),