[ard:beta] Improve extraction robustness, fix subtitles extraction, improve geo restricted videos extraction

This commit is contained in:
Sergey M․ 2018-12-17 05:29:59 +07:00
parent 1c82122741
commit 752582183a
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -8,13 +8,16 @@
from ..utils import (
determine_ext,
ExtractorError,
qualities,
int_or_none,
parse_duration,
qualities,
str_or_none,
try_get,
unified_strdate,
xpath_text,
unified_timestamp,
update_url_query,
url_or_none,
xpath_text,
)
from ..compat import compat_etree_fromstring
@ -336,43 +339,62 @@ def _real_extract(self, url):
'display_id': display_id,
}
formats = []
subtitles = {}
geoblocked = False
for widget in data.values():
if widget.get('_geoblocked'):
raise ExtractorError('This video is not available due to geoblocking', expected=True)
if widget.get('_geoblocked') is True:
geoblocked = True
if '_duration' in widget:
res['duration'] = widget['_duration']
res['duration'] = int_or_none(widget['_duration'])
if 'clipTitle' in widget:
res['title'] = widget['clipTitle']
if '_previewImage' in widget:
res['thumbnail'] = widget['_previewImage']
if 'broadcastedOn' in widget:
res['upload_date'] = unified_strdate(widget['broadcastedOn'])
res['timestamp'] = unified_timestamp(widget['broadcastedOn'])
if 'synopsis' in widget:
res['description'] = widget['synopsis']
if '_subtitleUrl' in widget:
res['subtitles'] = {'de': [{
subtitle_url = url_or_none(widget.get('_subtitleUrl'))
if subtitle_url:
subtitles.setdefault('de', []).append({
'ext': 'ttml',
'url': widget['_subtitleUrl'],
}]}
'url': subtitle_url,
})
if '_quality' in widget:
format_url = widget['_stream']['json'][0]
if format_url.endswith('.f4m'):
format_url = url_or_none(try_get(
widget, lambda x: x['_stream']['json'][0]))
if not format_url:
continue
ext = determine_ext(format_url)
if ext == 'f4m':
formats.extend(self._extract_f4m_formats(
format_url + '?hdcore=3.11.0',
video_id, f4m_id='hds', fatal=False))
elif format_url.endswith('m3u8'):
elif ext == 'm3u8':
formats.extend(self._extract_m3u8_formats(
format_url, video_id, 'mp4', m3u8_id='hls', fatal=False))
format_url, video_id, 'mp4', m3u8_id='hls',
fatal=False))
else:
# HTTP formats are not available when geoblocked is True,
# other formats are fine though
if geoblocked:
continue
quality = str_or_none(widget.get('_quality'))
formats.append({
'format_id': 'http-' + widget['_quality'],
'format_id': ('http-' + quality) if quality else 'http',
'url': format_url,
'preference': 10, # Plain HTTP, that's nice
})
if not formats and geoblocked:
self.raise_geo_restricted(
msg='This video is not available due to geoblocking',
countries=['DE'])
self._sort_formats(formats)
res['formats'] = formats
res.update({
'subtitles': subtitles,
'formats': formats,
})
return res