[tvp] extract all formats and detect erros

This commit is contained in:
Remita Amine 2016-06-23 04:34:07 +01:00
parent 011bd3221b
commit 6e3c2047f8

View file

@ -4,6 +4,12 @@
import re
from .common import InfoExtractor
from ..utils import (
determine_ext,
clean_html,
get_element_by_attribute,
ExtractorError,
)
class TVPIE(InfoExtractor):
@ -21,7 +27,7 @@ class TVPIE(InfoExtractor):
},
}, {
'url': 'http://www.tvp.pl/there-can-be-anything-so-i-shortened-it/17916176',
'md5': 'c3b15ed1af288131115ff17a17c19dda',
'md5': 'b0005b542e5b4de643a9690326ab1257',
'info_dict': {
'id': '17916176',
'ext': 'mp4',
@ -53,6 +59,11 @@ def _real_extract(self, url):
webpage = self._download_webpage(
'http://www.tvp.pl/sess/tvplayer.php?object_id=%s' % video_id, video_id)
error_massage = get_element_by_attribute('class', 'msg error', webpage)
if error_massage:
raise ExtractorError('%s said: %s' % (
self.IE_NAME, clean_html(error_massage)), expected=True)
title = self._search_regex(
r'name\s*:\s*([\'"])Title\1\s*,\s*value\s*:\s*\1(?P<title>.+?)\1',
webpage, 'title', group='title')
@ -66,24 +77,50 @@ def _real_extract(self, url):
r"poster\s*:\s*'([^']+)'", webpage, 'thumbnail', default=None)
video_url = self._search_regex(
r'0:{src:([\'"])(?P<url>.*?)\1', webpage, 'formats', group='url', default=None)
if not video_url:
r'0:{src:([\'"])(?P<url>.*?)\1', webpage,
'formats', group='url', default=None)
if not video_url or 'material_niedostepny.mp4' in video_url:
video_url = self._download_json(
'http://www.tvp.pl/pub/stat/videofileinfo?video_id=%s' % video_id,
video_id)['video_url']
ext = video_url.rsplit('.', 1)[-1]
if ext != 'ism/manifest':
if '/' in ext:
ext = 'mp4'
formats = []
video_url_base = self._search_regex(
r'(https?://.+?/video)(?:\.(?:ism|f4m|m3u8)|-\d+\.mp4)',
video_url, 'video base url', default=None)
if video_url_base:
# TODO: Current DASH formats are broken - $Time$ pattern in
# <SegmentTemplate> not implemented yet
# formats.extend(self._extract_mpd_formats(
# video_url_base + '.ism/video.mpd',
# video_id, mpd_id='dash', fatal=False))
formats.extend(self._extract_f4m_formats(
video_url_base + '.ism/video.f4m',
video_id, f4m_id='hds', fatal=False))
m3u8_formats = self._extract_m3u8_formats(
video_url_base + '.ism/video.m3u8', video_id,
'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)
self._sort_formats(m3u8_formats)
m3u8_formats = list(filter(
lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple',
m3u8_formats))
formats.extend(m3u8_formats)
for i, m3u8_format in enumerate(m3u8_formats, 2):
http_url = '%s-%d.mp4' % (video_url_base, i)
if self._is_valid_url(http_url, video_id):
f = m3u8_format.copy()
f.update({
'url': http_url,
'format_id': f['format_id'].replace('hls', 'http'),
'protocol': 'http',
})
formats.append(f)
else:
formats = [{
'format_id': 'direct',
'url': video_url,
'ext': ext,
'ext': determine_ext(video_url, 'mp4'),
}]
else:
m3u8_url = re.sub('([^/]*)\.ism/manifest', r'\1.ism/\1.m3u8', video_url)
formats = self._extract_m3u8_formats(m3u8_url, video_id, 'mp4')
self._sort_formats(formats)