[xtube] Fix formats extraction

This commit is contained in:
Sergey M․ 2014-03-21 23:58:40 +07:00
parent d77650525d
commit aa488e1385

View File

@ -1,11 +1,10 @@
from __future__ import unicode_literals from __future__ import unicode_literals
import os
import re import re
import json
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import ( from ..utils import (
compat_urllib_parse_urlparse,
compat_urllib_request, compat_urllib_request,
parse_duration, parse_duration,
str_to_int, str_to_int,
@ -42,7 +41,6 @@ def _real_extract(self, url):
r'so_s\.addVariable\("owner_u", "([^"]+)', webpage, 'uploader', fatal=False) r'so_s\.addVariable\("owner_u", "([^"]+)', webpage, 'uploader', fatal=False)
video_description = self._html_search_regex( video_description = self._html_search_regex(
r'<p class="fieldsDesc">([^<]+)', webpage, 'description', fatal=False) r'<p class="fieldsDesc">([^<]+)', webpage, 'description', fatal=False)
video_url = self._html_search_regex(r'var videoMp4 = "([^"]+)', webpage, 'video_url').replace('\\/', '/')
duration = parse_duration(self._html_search_regex( duration = parse_duration(self._html_search_regex(
r'<span class="bold">Runtime:</span> ([^<]+)</p>', webpage, 'duration', fatal=False)) r'<span class="bold">Runtime:</span> ([^<]+)</p>', webpage, 'duration', fatal=False))
view_count = self._html_search_regex( view_count = self._html_search_regex(
@ -54,12 +52,18 @@ def _real_extract(self, url):
if comment_count: if comment_count:
comment_count = str_to_int(comment_count) comment_count = str_to_int(comment_count)
path = compat_urllib_parse_urlparse(video_url).path player_quality_option = json.loads(self._html_search_regex(
extension = os.path.splitext(path)[1][1:] r'playerQualityOption = ({.+?});', webpage, 'player quality option'))
format = path.split('/')[5].split('_')[:2]
format[0] += 'p' QUALITIES = ['3gp', 'mp4_normal', 'mp4_high', 'flv', 'mp4_ultra', 'mp4_720', 'mp4_1080']
format[1] += 'k' formats = [
format = "-".join(format) {
'url': url,
'format_id': format_id,
'preference': QUALITIES.index(format_id) if format_id in QUALITIES else -1,
} for format_id, url in player_quality_option.items()
]
self._sort_formats(formats)
return { return {
'id': video_id, 'id': video_id,
@ -69,9 +73,6 @@ def _real_extract(self, url):
'duration': duration, 'duration': duration,
'view_count': view_count, 'view_count': view_count,
'comment_count': comment_count, 'comment_count': comment_count,
'url': video_url, 'formats': formats,
'ext': extension,
'format': format,
'format_id': format,
'age_limit': 18, 'age_limit': 18,
} }