[varzesh3] Fix metadata extraction (closes #9197)

This commit is contained in:
Yen Chi Hsuan 2016-04-16 17:13:22 +08:00
parent 759e37c9e6
commit ee94e7e66d
No known key found for this signature in database
GPG Key ID: 3FDDD575826C5C30

View File

@ -2,11 +2,19 @@
from __future__ import unicode_literals from __future__ import unicode_literals
from .common import InfoExtractor from .common import InfoExtractor
from ..compat import (
compat_urllib_parse_urlparse,
compat_parse_qs,
)
from ..utils import (
clean_html,
remove_start,
)
class Varzesh3IE(InfoExtractor): class Varzesh3IE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?video\.varzesh3\.com/(?:[^/]+/)+(?P<id>[^/]+)/?' _VALID_URL = r'https?://(?:www\.)?video\.varzesh3\.com/(?:[^/]+/)+(?P<id>[^/]+)/?'
_TEST = { _TESTS = [{
'url': 'http://video.varzesh3.com/germany/bundesliga/5-%D9%88%D8%A7%DA%A9%D9%86%D8%B4-%D8%A8%D8%B1%D8%AA%D8%B1-%D8%AF%D8%B1%D9%88%D8%A7%D8%B2%D9%87%E2%80%8C%D8%A8%D8%A7%D9%86%D8%A7%D9%86%D8%9B%D9%87%D9%81%D8%AA%D9%87-26-%D8%A8%D9%88%D9%86%D8%AF%D8%B3/', 'url': 'http://video.varzesh3.com/germany/bundesliga/5-%D9%88%D8%A7%DA%A9%D9%86%D8%B4-%D8%A8%D8%B1%D8%AA%D8%B1-%D8%AF%D8%B1%D9%88%D8%A7%D8%B2%D9%87%E2%80%8C%D8%A8%D8%A7%D9%86%D8%A7%D9%86%D8%9B%D9%87%D9%81%D8%AA%D9%87-26-%D8%A8%D9%88%D9%86%D8%AF%D8%B3/',
'md5': '2a933874cb7dce4366075281eb49e855', 'md5': '2a933874cb7dce4366075281eb49e855',
'info_dict': { 'info_dict': {
@ -15,8 +23,18 @@ class Varzesh3IE(InfoExtractor):
'title': '۵ واکنش برتر دروازه‌بانان؛هفته ۲۶ بوندسلیگا', 'title': '۵ واکنش برتر دروازه‌بانان؛هفته ۲۶ بوندسلیگا',
'description': 'فصل ۲۰۱۵-۲۰۱۴', 'description': 'فصل ۲۰۱۵-۲۰۱۴',
'thumbnail': 're:^https?://.*\.jpg$', 'thumbnail': 're:^https?://.*\.jpg$',
} },
} 'skip': 'HTTP 404 Error',
}, {
'url': 'http://video.varzesh3.com/video/112785/%D8%AF%D9%84%D9%87-%D8%B9%D9%84%DB%8C%D8%9B-%D8%B3%D8%AA%D8%A7%D8%B1%D9%87-%D9%86%D9%88%D8%B8%D9%87%D9%88%D8%B1-%D9%84%DB%8C%DA%AF-%D8%A8%D8%B1%D8%AA%D8%B1-%D8%AC%D8%B2%DB%8C%D8%B1%D9%87',
'info_dict': {
'id': '112785',
'ext': 'mp4',
'title': 'دله علی؛ ستاره نوظهور لیگ برتر جزیره',
'description': 'فوتبال 120',
},
'expected_warnings': ['description'],
}]
def _real_extract(self, url): def _real_extract(self, url):
display_id = self._match_id(url) display_id = self._match_id(url)
@ -26,15 +44,30 @@ class Varzesh3IE(InfoExtractor):
video_url = self._search_regex( video_url = self._search_regex(
r'<source[^>]+src="([^"]+)"', webpage, 'video url') r'<source[^>]+src="([^"]+)"', webpage, 'video url')
title = self._og_search_title(webpage) title = remove_start(self._html_search_regex(
r'<title>([^<]+)</title>', webpage, 'title'), 'ویدیو ورزش 3 | ')
description = self._html_search_regex( description = self._html_search_regex(
r'(?s)<div class="matn">(.+?)</div>', r'(?s)<div class="matn">(.+?)</div>',
webpage, 'description', fatal=False) webpage, 'description', default=None)
thumbnail = self._og_search_thumbnail(webpage) if description is None:
description = clean_html(self._html_search_meta('description', webpage))
thumbnail = self._og_search_thumbnail(webpage, default=None)
if thumbnail is None:
fb_sharer_url = self._search_regex(
r'<a[^>]+href="(https?://www\.facebook\.com/sharer/sharer\.php?[^"]+)"',
webpage, 'facebook sharer URL', fatal=False)
sharer_params = compat_parse_qs(compat_urllib_parse_urlparse(fb_sharer_url).query)
thumbnail = sharer_params.get('p[images][0]', [None])[0]
video_id = self._search_regex( video_id = self._search_regex(
r"<link[^>]+rel='(?:canonical|shortlink)'[^>]+href='/\?p=([^']+)'", r"<link[^>]+rel='(?:canonical|shortlink)'[^>]+href='/\?p=([^']+)'",
webpage, display_id, default=display_id) webpage, display_id, default=None)
if video_id is None:
video_id = self._search_regex(
'var\s+VideoId\s*=\s*(\d+);', webpage, 'video id',
default=display_id)
return { return {
'url': video_url, 'url': video_url,