X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Ffacebook.py;h=899b0896b92b5d4ddfc7f291f6b064334f3830af;hb=c1406299959b780e7b86b567bd4eeecc8556dce0;hp=39c481068fbc692ba43651c84b1afb270900fa8a;hpb=a8ae232fa9c24534bd9c838c793f182e6796fe4e;p=youtube-dl.git diff --git a/youtube_dl/extractor/facebook.py b/youtube_dl/extractor/facebook.py index 39c481068..899b0896b 100644 --- a/youtube_dl/extractor/facebook.py +++ b/youtube_dl/extractor/facebook.py @@ -23,15 +23,23 @@ from ..utils import ( class FacebookIE(InfoExtractor): _VALID_URL = r'''(?x) - https?://(?:\w+\.)?facebook\.com/ - (?:[^#]*?\#!/)? - (?: - (?:video/video\.php|photo\.php|video\.php|video/embed)\?(?:.*?) - (?:v|video_id)=| - [^/]+/videos/(?:[^/]+/)? - ) - (?P[0-9]+) - (?:.*)''' + (?: + https?:// + (?:\w+\.)?facebook\.com/ + (?:[^#]*?\#!/)? + (?: + (?: + video/video\.php| + photo\.php| + video\.php| + video/embed + )\?(?:.*?)(?:v|video_id)=| + [^/]+/videos/(?:[^/]+/)? + )| + facebook: + ) + (?P[0-9]+) + ''' _LOGIN_URL = 'https://www.facebook.com/login.php?next=http%3A%2F%2Ffacebook.com%2Fhome.php&login_attempt=1' _CHECKPOINT_URL = 'https://www.facebook.com/checkpoint/?next=http%3A%2F%2Ffacebook.com%2Fhome.php&_fb_noscript=1' _NETRC_MACHINE = 'facebook' @@ -66,6 +74,9 @@ class FacebookIE(InfoExtractor): }, { 'url': 'https://www.facebook.com/ChristyClarkForBC/videos/vb.22819070941/10153870694020942/?type=2&theater', 'only_matching': True, + }, { + 'url': 'facebook:544765982287235', + 'only_matching': True, }] def _login(self): @@ -74,7 +85,7 @@ class FacebookIE(InfoExtractor): return login_page_req = sanitized_Request(self._LOGIN_URL) - login_page_req.add_header('Cookie', 'locale=en_US') + self._set_cookie('facebook.com', 'locale', 'en_US') login_page = self._download_webpage(login_page_req, None, note='Downloading login page', errnote='Unable to download login page') @@ -100,13 +111,25 @@ class FacebookIE(InfoExtractor): login_results = self._download_webpage(request, None, note='Logging in', errnote='unable to fetch login page') if re.search(r'', login_results) is not None: - self._downloader.report_warning('unable to log in: bad username/password, or exceded login rate limit (~3/min). Check credentials or wait.') + error = self._html_search_regex( + r'(?s)]+class=(["\']).*?login_error_box.*?\1[^>]*>]*>.*?]*>(?P.+?)', + login_results, 'login error', default=None, group='error') + if error: + raise ExtractorError('Unable to login: %s' % error, expected=True) + self._downloader.report_warning('unable to log in: bad username/password, or exceeded login rate limit (~3/min). Check credentials or wait.') + return + + fb_dtsg = self._search_regex( + r'name="fb_dtsg" value="(.+?)"', login_results, 'fb_dtsg', default=None) + h = self._search_regex( + r'name="h"\s+(?:\w+="[^"]+"\s+)*?value="([^"]+)"', login_results, 'h', default=None) + + if not fb_dtsg or not h: return check_form = { - 'fb_dtsg': self._search_regex(r'name="fb_dtsg" value="(.+?)"', login_results, 'fb_dtsg'), - 'h': self._search_regex( - r'name="h"\s+(?:\w+="[^"]+"\s+)*?value="([^"]+)"', login_results, 'h'), + 'fb_dtsg': fb_dtsg, + 'h': h, 'name_action_selected': 'dont_save', } check_req = sanitized_Request(self._CHECKPOINT_URL, urlencode_postdata(check_form)) @@ -114,7 +137,7 @@ class FacebookIE(InfoExtractor): check_response = self._download_webpage(check_req, None, note='Confirming login') if re.search(r'id="checkpointSubmitButton"', check_response) is not None: - self._downloader.report_warning('Unable to confirm login, you have to login in your brower and authorize the login.') + self._downloader.report_warning('Unable to confirm login, you have to login in your browser and authorize the login.') except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: self._downloader.report_warning('unable to log in: %s' % error_to_compat_str(err)) return @@ -127,10 +150,32 @@ class FacebookIE(InfoExtractor): url = 'https://www.facebook.com/video/video.php?v=%s' % video_id webpage = self._download_webpage(url, video_id) + video_data = None + BEFORE = '{swf.addParam(param[0], param[1]);});\n' AFTER = '.forEach(function(variable) {swf.addVariable(variable[0], variable[1]);});' m = re.search(re.escape(BEFORE) + '(.*?)' + re.escape(AFTER), webpage) - if not m: + if m: + data = dict(json.loads(m.group(1))) + params_raw = compat_urllib_parse_unquote(data['params']) + video_data = json.loads(params_raw)['video_data'] + + def video_data_list2dict(video_data): + ret = {} + for item in video_data: + format_id = item['stream_type'] + ret.setdefault(format_id, []).append(item) + return ret + + if not video_data: + server_js_data = self._parse_json(self._search_regex( + r'handleServerJS\(({.+})\);', webpage, 'server js data'), video_id) + for item in server_js_data['instances']: + if item[1][0] == 'VideoConfig': + video_data = video_data_list2dict(item[2][0]['videoData']) + break + + if not video_data: m_msg = re.search(r'class="[^"]*uiInterstitialContent[^"]*">
(.*?)
', webpage) if m_msg is not None: raise ExtractorError( @@ -138,12 +183,9 @@ class FacebookIE(InfoExtractor): expected=True) else: raise ExtractorError('Cannot parse data') - data = dict(json.loads(m.group(1))) - params_raw = compat_urllib_parse_unquote(data['params']) - params = json.loads(params_raw) formats = [] - for format_id, f in params['video_data'].items(): + for format_id, f in video_data.items(): if not f or not isinstance(f, list): continue for quality in ('sd', 'hd'): @@ -176,3 +218,33 @@ class FacebookIE(InfoExtractor): 'formats': formats, 'uploader': uploader, } + + +class FacebookPostIE(InfoExtractor): + IE_NAME = 'facebook:post' + _VALID_URL = r'https?://(?:\w+\.)?facebook\.com/[^/]+/posts/(?P\d+)' + _TEST = { + 'url': 'https://www.facebook.com/maxlayn/posts/10153807558977570', + 'md5': '037b1fa7f3c2d02b7a0d7bc16031ecc6', + 'info_dict': { + 'id': '544765982287235', + 'ext': 'mp4', + 'title': '"What are you doing running in the snow?"', + 'uploader': 'FailArmy', + } + } + + def _real_extract(self, url): + post_id = self._match_id(url) + + webpage = self._download_webpage(url, post_id) + + entries = [ + self.url_result('facebook:%s' % video_id, FacebookIE.ie_key()) + for video_id in self._parse_json( + self._search_regex( + r'(["\'])video_ids\1\s*:\s*(?P\[.+?\])', + webpage, 'video ids', group='ids'), + post_id)] + + return self.playlist_result(entries, post_id)