X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fyoutube.py;h=5449df8e0d6292a8fe45b504e8eea110b8e6a9ca;hb=23d3c422ab1e38e674ed9f0e6efebb41d5ff63fc;hp=1f3aa4322c274e226b6e5945874e8627724a79cf;hpb=e399853d0c5784257ffcb6fba147d0b47d3f9bb6;p=youtube-dl.git diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 1f3aa4322..5449df8e0 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -223,6 +223,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): '246': {'ext': 'webm', 'height': 480, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40}, '247': {'ext': 'webm', 'height': 720, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40}, '248': {'ext': 'webm', 'height': 1080, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40}, + '271': {'ext': 'webm', 'height': 1440, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40}, + '272': {'ext': 'webm', 'height': 2160, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40}, # Dash webm audio '171': {'ext': 'webm', 'vcodec': 'none', 'format_note': 'DASH audio', 'abr': 48, 'preference': -50}, @@ -242,7 +244,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): u"uploader": u"Philipp Hagemeister", u"uploader_id": u"phihag", u"upload_date": u"20121002", - u"description": u"test chars: \"'/\\ä↭𝕐\ntest URL: https://github.com/rg3/youtube-dl/issues/1892\n\nThis is a test video for youtube-dl.\n\nFor more information, contact phihag@phihag.de ." + u"description": u"test chars: \"'/\\ä↭𝕐\ntest URL: https://github.com/rg3/youtube-dl/issues/1892\n\nThis is a test video for youtube-dl.\n\nFor more information, contact phihag@phihag.de .", + u"categories": [u'Science & Technology'], } }, { @@ -344,8 +347,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): self.to_screen(u'RTMP download detected') def _extract_signature_function(self, video_id, player_url, slen): - id_m = re.match(r'.*-(?P[a-zA-Z0-9_-]+)\.(?P[a-z]+)$', - player_url) + id_m = re.match( + r'.*-(?P[a-zA-Z0-9_-]+)(?:/watch_as3)?\.(?P[a-z]+)$', + player_url) player_type = id_m.group('ext') player_id = id_m.group('id') @@ -438,7 +442,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): def _parse_sig_js(self, jscode): funcname = self._search_regex( - r'signature=([a-zA-Z]+)', jscode, + r'signature=([$a-zA-Z]+)', jscode, u'Initial JS player signature function name') jsi = JSInterpreter(jscode) @@ -862,71 +866,26 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): def _decrypt_signature(self, s, video_id, player_url, age_gate=False): """Turn the encrypted s field into a working signature""" - if player_url is not None: - if player_url.startswith(u'//'): - player_url = u'https:' + player_url - try: - player_id = (player_url, len(s)) - if player_id not in self._player_cache: - func = self._extract_signature_function( - video_id, player_url, len(s) - ) - self._player_cache[player_id] = func - func = self._player_cache[player_id] - if self._downloader.params.get('youtube_print_sig_code'): - self._print_sig_code(func, len(s)) - return func(s) - except Exception: - tb = traceback.format_exc() - self._downloader.report_warning( - u'Automatic signature extraction failed: ' + tb) - - self._downloader.report_warning( - u'Warning: Falling back to static signature algorithm') - - return self._static_decrypt_signature( - s, video_id, player_url, age_gate) - - def _static_decrypt_signature(self, s, video_id, player_url, age_gate): - if age_gate: - # The videos with age protection use another player, so the - # algorithms can be different. - if len(s) == 86: - return s[2:63] + s[82] + s[64:82] + s[63] - - if len(s) == 93: - return s[86:29:-1] + s[88] + s[28:5:-1] - elif len(s) == 92: - return s[25] + s[3:25] + s[0] + s[26:42] + s[79] + s[43:79] + s[91] + s[80:83] - elif len(s) == 91: - return s[84:27:-1] + s[86] + s[26:5:-1] - elif len(s) == 90: - return s[25] + s[3:25] + s[2] + s[26:40] + s[77] + s[41:77] + s[89] + s[78:81] - elif len(s) == 89: - return s[84:78:-1] + s[87] + s[77:60:-1] + s[0] + s[59:3:-1] - elif len(s) == 88: - return s[7:28] + s[87] + s[29:45] + s[55] + s[46:55] + s[2] + s[56:87] + s[28] - elif len(s) == 87: - return s[6:27] + s[4] + s[28:39] + s[27] + s[40:59] + s[2] + s[60:] - elif len(s) == 86: - return s[80:72:-1] + s[16] + s[71:39:-1] + s[72] + s[38:16:-1] + s[82] + s[15::-1] - elif len(s) == 85: - return s[3:11] + s[0] + s[12:55] + s[84] + s[56:84] - elif len(s) == 84: - return s[78:70:-1] + s[14] + s[69:37:-1] + s[70] + s[36:14:-1] + s[80] + s[:14][::-1] - elif len(s) == 83: - return s[80:63:-1] + s[0] + s[62:0:-1] + s[63] - elif len(s) == 82: - return s[80:37:-1] + s[7] + s[36:7:-1] + s[0] + s[6:0:-1] + s[37] - elif len(s) == 81: - return s[56] + s[79:56:-1] + s[41] + s[55:41:-1] + s[80] + s[40:34:-1] + s[0] + s[33:29:-1] + s[34] + s[28:9:-1] + s[29] + s[8:0:-1] + s[9] - elif len(s) == 80: - return s[1:19] + s[0] + s[20:68] + s[19] + s[69:80] - elif len(s) == 79: - return s[54] + s[77:54:-1] + s[39] + s[53:39:-1] + s[78] + s[38:34:-1] + s[0] + s[33:29:-1] + s[34] + s[28:9:-1] + s[29] + s[8:0:-1] + s[9] + if player_url is None: + raise ExtractorError(u'Cannot decrypt signature without player_url') - else: - raise ExtractorError(u'Unable to decrypt signature, key length %d not supported; retrying might work' % (len(s))) + if player_url.startswith(u'//'): + player_url = u'https:' + player_url + try: + player_id = (player_url, len(s)) + if player_id not in self._player_cache: + func = self._extract_signature_function( + video_id, player_url, len(s) + ) + self._player_cache[player_id] = func + func = self._player_cache[player_id] + if self._downloader.params.get('youtube_print_sig_code'): + self._print_sig_code(func, len(s)) + return func(s) + except Exception as e: + tb = traceback.format_exc() + raise ExtractorError( + u'Automatic signature extraction failed: ' + tb, cause=e) def _get_available_subtitles(self, video_id, webpage): try: @@ -1136,11 +1095,24 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): # upload date upload_date = None - mobj = re.search(r'id="eow-date.*?>(.*?)', video_webpage, re.DOTALL) + mobj = re.search(r'(?s)id="eow-date.*?>(.*?)', video_webpage) + if mobj is None: + mobj = re.search( + r'(?s)id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live) on (.*?)', + video_webpage) if mobj is not None: upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split()) upload_date = unified_strdate(upload_date) + m_cat_container = get_element_by_id("eow-category", video_webpage) + if m_cat_container: + category = self._html_search_regex( + r'(?s)(.*?)', m_cat_container, 'category', + default=None) + video_categories = None if category is None else [category] + else: + video_categories = None + # description video_description = get_element_by_id("eow-description", video_webpage) if video_description: @@ -1249,31 +1221,38 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): url += '&signature=' + url_data['sig'][0] elif 's' in url_data: encrypted_sig = url_data['s'][0] + + if not age_gate: + jsplayer_url_json = self._search_regex( + r'"assets":.+?"js":\s*("[^"]+")', + video_webpage, u'JS player URL') + player_url = json.loads(jsplayer_url_json) + if player_url is None: + player_url_json = self._search_regex( + r'ytplayer\.config.*?"url"\s*:\s*("[^"]+")', + video_webpage, u'age gate player URL') + player_url = json.loads(player_url_json) + if self._downloader.params.get('verbose'): - if age_gate: - if player_url is None: - player_version = 'unknown' - else: + if player_url is None: + player_version = 'unknown' + player_desc = 'unknown' + else: + if player_url.endswith('swf'): player_version = self._search_regex( r'-(.+)\.swf$', player_url, u'flash player', fatal=False) - player_desc = 'flash player %s' % player_version - else: - player_version = self._search_regex( - r'html5player-(.+?)\.js', video_webpage, - 'html5 player', fatal=False) - player_desc = u'html5 player %s' % player_version + player_desc = 'flash player %s' % player_version + else: + player_version = self._search_regex( + r'html5player-(.+?)\.js', video_webpage, + 'html5 player', fatal=False) + player_desc = u'html5 player %s' % player_version parts_sizes = u'.'.join(compat_str(len(part)) for part in encrypted_sig.split('.')) self.to_screen(u'encrypted signature length %d (%s), itag %s, %s' % (len(encrypted_sig), parts_sizes, url_data['itag'][0], player_desc)) - if not age_gate: - jsplayer_url_json = self._search_regex( - r'"assets":.+?"js":\s*("[^"]+")', - video_webpage, u'JS player URL') - player_url = json.loads(jsplayer_url_json) - signature = self._decrypt_signature( encrypted_sig, video_id, player_url, age_gate) url += '&signature=' + signature @@ -1347,6 +1326,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): 'title': video_title, 'thumbnail': video_thumbnail, 'description': video_description, + 'categories': video_categories, 'subtitles': video_subtitles, 'duration': video_duration, 'age_limit': 18 if age_gate else 0, @@ -1370,13 +1350,13 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): | p/ ) ( - (?:PL|EC|UU|FL|RD)?[0-9A-Za-z-_]{10,} + (?:PL|LL|EC|UU|FL|RD)?[0-9A-Za-z-_]{10,} # Top tracks, they can also include dots |(?:MC)[\w\.]* ) .* | - ((?:PL|EC|UU|FL|RD)[0-9A-Za-z-_]{10,}) + ((?:PL|LL|EC|UU|FL|RD)[0-9A-Za-z-_]{10,}) )""" _TEMPLATE_URL = 'https://www.youtube.com/playlist?list=%s' _MORE_PAGES_INDICATOR = r'data-link-type="next"' @@ -1399,11 +1379,9 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): title_span = (search_title('playlist-title') or search_title('title long-title') or search_title('title')) title = clean_html(title_span) - video_re = r'''(?x)data-video-username="(.*?)".*? + video_re = r'''(?x)data-video-username=".*?".*? href="/watch\?v=([0-9A-Za-z_-]{11})&[^"]*?list=%s''' % re.escape(playlist_id) - matches = orderedSet(re.findall(video_re, webpage, flags=re.DOTALL)) - # Some of the videos may have been deleted, their username field is empty - ids = [video_id for (username, video_id) in matches if username] + ids = orderedSet(re.findall(video_re, webpage, flags=re.DOTALL)) url_results = self._ids_to_results(ids) return self.playlist_result(url_results, playlist_id, title) @@ -1683,14 +1661,14 @@ class YoutubeSearchURLIE(InfoExtractor): webpage = self._download_webpage(url, query) result_code = self._search_regex( - r'(?s)
    ', webpage, u'result HTML') + r'(?s)
      ', webpage, u'result HTML') part_codes = re.findall( r'(?s)

      (.*?)

      ', result_code) entries = [] for part_code in part_codes: part_title = self._html_search_regex( - r'(?s)title="([^"]+)"', part_code, 'item title', fatal=False) + [r'(?s)title="([^"]+)"', r'>([^<]+)'], part_code, 'item title', fatal=False) part_url_snippet = self._html_search_regex( r'(?s)href="([^"]+)"', part_code, 'item URL') part_url = compat_urlparse.urljoin( @@ -1760,9 +1738,12 @@ class YoutubeFeedsInfoExtractor(YoutubeBaseInfoExtractor): feed_entries.extend( self.url_result(video_id, 'Youtube', video_id=video_id) for video_id in ids) - if info['paging'] is None: + mobj = re.search( + r'data-uix-load-more-href="/?[^"]+paging=(?P\d+)', + feed_html) + if mobj is None: break - paging = info['paging'] + paging = mobj.group('paging') return self.playlist_result(feed_entries, playlist_title=self._PLAYLIST_TITLE) class YoutubeSubscriptionsIE(YoutubeFeedsInfoExtractor): @@ -1807,10 +1788,21 @@ class YoutubeTruncatedURLIE(InfoExtractor): IE_NAME = 'youtube:truncated_url' IE_DESC = False # Do not list _VALID_URL = r'''(?x) - (?:https?://)?[^/]+/watch\?(?:feature=[a-z_]+)?$| + (?:https?://)?[^/]+/watch\?(?: + feature=[a-z_]+| + annotation_id=annotation_[^&]+ + )?$| (?:https?://)?(?:www\.)?youtube\.com/attribution_link\?a=[^&]+$ ''' + _TESTS = [{ + 'url': 'http://www.youtube.com/watch?annotation_id=annotation_3951667041', + 'only_matching': True, + }, { + 'url': 'http://www.youtube.com/watch?', + 'only_matching': True, + }] + def _real_extract(self, url): raise ExtractorError( u'Did you forget to quote the URL? Remember that & is a meta '