story\.php
)\?(?:.*?)(?:v|video_id|story_fbid)=|
[^/]+/videos/(?:[^/]+/)?|
- [^/]+/posts/
+ [^/]+/posts/|
+ groups/[^/]+/permalink/
)|
facebook:
)
}, {
'url': 'facebook:544765982287235',
'only_matching': True,
+ }, {
+ 'url': 'https://www.facebook.com/groups/164828000315060/permalink/764967300301124/',
+ 'only_matching': True,
}]
+ @staticmethod
+ def _extract_url(webpage):
+ mobj = re.search(
+ r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
+ if mobj is not None:
+ return mobj.group('url')
+
+ # Facebook API embed
+ # see https://developers.facebook.com/docs/plugins/embedded-video-player
+ mobj = re.search(r'''(?x)<div[^>]+
+ class=(?P<q1>[\'"])[^\'"]*\bfb-(?:video|post)\b[^\'"]*(?P=q1)[^>]+
+ data-href=(?P<q2>[\'"])(?P<url>(?:https?:)?//(?:www\.)?facebook.com/.+?)(?P=q2)''', webpage)
+ if mobj is not None:
+ return mobj.group('url')
+
def _login(self):
(useremail, password) = self._get_login_info()
if useremail is None:
BEFORE = '{swf.addParam(param[0], param[1]);});'
AFTER = '.forEach(function(variable) {swf.addVariable(variable[0], variable[1]);});'
- m = re.search(re.escape(BEFORE) + '(?:\n|\\\\n)(.*?)' + re.escape(AFTER), webpage)
- if m:
- swf_params = m.group(1).replace('\\\\', '\\').replace('\\"', '"')
+ PATTERN = re.escape(BEFORE) + '(?:\n|\\\\n)(.*?)' + re.escape(AFTER)
+
+ for m in re.findall(PATTERN, webpage):
+ swf_params = m.replace('\\\\', '\\').replace('\\"', '"')
data = dict(json.loads(swf_params))
params_raw = compat_urllib_parse_unquote(data['params'])
- video_data = json.loads(params_raw)['video_data']
+ video_data_candidate = json.loads(params_raw)['video_data']
+ for _, f in video_data_candidate.items():
+ if not f:
+ continue
+ if isinstance(f, dict):
+ f = [f]
+ if isinstance(f, list):
+ continue
+ if f[0].get('video_id') == video_id:
+ video_data = video_data_candidate
+ break
def video_data_list2dict(video_data):
ret = {}
formats = []
for format_id, f in video_data.items():
+ if f and isinstance(f, dict):
+ f = [f]
if not f or not isinstance(f, list):
continue
for quality in ('sd', 'hd'):
if '/posts/' in url:
entries = [
- self.url_result('facebook:%s' % video_id, FacebookIE.ie_key())
- for video_id in self._parse_json(
+ self.url_result('facebook:%s' % vid, FacebookIE.ie_key())
+ for vid in self._parse_json(
self._search_regex(
r'(["\'])video_ids\1\s*:\s*(?P<ids>\[.+?\])',
webpage, 'video ids', group='ids'),