From 5d51a883c2049e0186074ded9405b01f79470d57 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Wed, 26 Jun 2013 11:03:44 +0200 Subject: Use a dictionary for storing the subtitles The errors while getting the subtitles are reported as warnings, if no subtitles are found return and empty dict. --- youtube_dl/extractor/youtube.py | 67 +++++++++++++++++++++-------------------- 1 file changed, 34 insertions(+), 33 deletions(-) (limited to 'youtube_dl/extractor/youtube.py') diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 12e8fc25d..78500b0f7 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -209,11 +209,13 @@ class YoutubeIE(InfoExtractor): try: sub_list = compat_urllib_request.urlopen(request).read().decode('utf-8') except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: - return (u'unable to download video subtitles: %s' % compat_str(err), None) + self._downloader.report_warning(u'unable to download video subtitles: %s' % compat_str(err)) + return {} sub_lang_list = re.findall(r'name="([^"]*)"[^>]+lang_code="([\w\-]+)"', sub_list) sub_lang_list = dict((l[1], l[0]) for l in sub_lang_list) if not sub_lang_list: - return (u'video doesn\'t have subtitles', None) + self._downloader.report_warning(u'video doesn\'t have subtitles') + return {} return sub_lang_list def _list_available_subtitles(self, video_id): @@ -222,8 +224,7 @@ class YoutubeIE(InfoExtractor): def _request_subtitle(self, sub_lang, sub_name, video_id, format): """ - Return tuple: - (error_message, sub_lang, sub) + Return the subtitle as a string or None if they are not found """ self.report_video_subtitles_request(video_id, sub_lang, format) params = compat_urllib_parse.urlencode({ @@ -236,10 +237,12 @@ class YoutubeIE(InfoExtractor): try: sub = compat_urllib_request.urlopen(url).read().decode('utf-8') except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: - return (u'unable to download video subtitles: %s' % compat_str(err), None, None) + self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err))) + return if not sub: - return (u'Did not fetch video subtitles', None, None) - return (None, sub_lang, sub) + self._downloader.report_warning(u'Did not fetch video subtitles') + return + return sub def _request_automatic_caption(self, video_id, webpage): """We need the webpage for getting the captions url, pass it as an @@ -250,7 +253,8 @@ class YoutubeIE(InfoExtractor): mobj = re.search(r';ytplayer.config = ({.*?});', webpage) err_msg = u'Couldn\'t find automatic captions for "%s"' % sub_lang if mobj is None: - return [(err_msg, None, None)] + self._downloader.report_warning(err_msg) + return {} player_config = json.loads(mobj.group(1)) try: args = player_config[u'args'] @@ -265,19 +269,20 @@ class YoutubeIE(InfoExtractor): }) subtitles_url = caption_url + '&' + params sub = self._download_webpage(subtitles_url, video_id, u'Downloading automatic captions') - return [(None, sub_lang, sub)] + return {sub_lang: sub} except KeyError: - return [(err_msg, None, None)] + self._downloader.report_warning(err_msg) + return {} def _extract_subtitle(self, video_id): """ - Return a list with a tuple: - [(error_message, sub_lang, sub)] + Return a dictionary: {language: subtitles} or {} if the subtitles + couldn't be found """ sub_lang_list = self._get_available_subtitles(video_id) sub_format = self._downloader.params.get('subtitlesformat') - if isinstance(sub_lang_list,tuple): #There was some error, it didn't get the available subtitles - return [(sub_lang_list[0], None, None)] + if not sub_lang_list: #There was some error, it didn't get the available subtitles + return {} if self._downloader.params.get('subtitleslang', False): sub_lang = self._downloader.params.get('subtitleslang') elif 'en' in sub_lang_list: @@ -285,20 +290,28 @@ class YoutubeIE(InfoExtractor): else: sub_lang = list(sub_lang_list.keys())[0] if not sub_lang in sub_lang_list: - return [(u'no closed captions found in the specified language "%s"' % sub_lang, None, None)] + self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang) + return {} subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format) - return [subtitle] + if subtitle: + self.to_screen('sub %s' % subtitle[:20]) + return {sub_lang: subtitle} + else: + return {} def _extract_all_subtitles(self, video_id): + """ + Return a dicitonary: {language: subtitles} or {} if the subtitles + couldn't be found + """ sub_lang_list = self._get_available_subtitles(video_id) sub_format = self._downloader.params.get('subtitlesformat') - if isinstance(sub_lang_list,tuple): #There was some error, it didn't get the available subtitles - return [(sub_lang_list[0], None, None)] - subtitles = [] + subtitles = {} for sub_lang in sub_lang_list: subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format) - subtitles.append(subtitle) + if subtitle: + subtitles[sub_lang] = subtitle return subtitles def _print_formats(self, formats): @@ -523,23 +536,11 @@ class YoutubeIE(InfoExtractor): if self._downloader.params.get('writesubtitles', False): video_subtitles = self._extract_subtitle(video_id) - if video_subtitles: - (sub_error, sub_lang, sub) = video_subtitles[0] - if sub_error: - self._downloader.report_warning(sub_error) - - if self._downloader.params.get('writeautomaticsub', False): + elif self._downloader.params.get('writeautomaticsub', False): video_subtitles = self._request_automatic_caption(video_id, video_webpage) - (sub_error, sub_lang, sub) = video_subtitles[0] - if sub_error: - self._downloader.report_warning(sub_error) if self._downloader.params.get('allsubtitles', False): video_subtitles = self._extract_all_subtitles(video_id) - for video_subtitle in video_subtitles: - (sub_error, sub_lang, sub) = video_subtitle - if sub_error: - self._downloader.report_warning(sub_error) if self._downloader.params.get('listsubtitles', False): self._list_available_subtitles(video_id) -- cgit v1.2.3 From 88ae5991cd777f05b437dbe7b4399f1ff25d6b85 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Wed, 26 Jun 2013 11:39:34 +0200 Subject: YoutubeIE: use the same function for getting the subtitles for the "--write-sub" and "--all-sub" options --- youtube_dl/extractor/youtube.py | 46 ++++++++++++++--------------------------- 1 file changed, 15 insertions(+), 31 deletions(-) (limited to 'youtube_dl/extractor/youtube.py') diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 78500b0f7..30036524f 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -273,8 +273,8 @@ class YoutubeIE(InfoExtractor): except KeyError: self._downloader.report_warning(err_msg) return {} - - def _extract_subtitle(self, video_id): + + def _extract_subtitles(self, video_id): """ Return a dictionary: {language: subtitles} or {} if the subtitles couldn't be found @@ -283,30 +283,17 @@ class YoutubeIE(InfoExtractor): sub_format = self._downloader.params.get('subtitlesformat') if not sub_lang_list: #There was some error, it didn't get the available subtitles return {} - if self._downloader.params.get('subtitleslang', False): - sub_lang = self._downloader.params.get('subtitleslang') - elif 'en' in sub_lang_list: - sub_lang = 'en' - else: - sub_lang = list(sub_lang_list.keys())[0] - if not sub_lang in sub_lang_list: - self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang) - return {} - - subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format) - if subtitle: - self.to_screen('sub %s' % subtitle[:20]) - return {sub_lang: subtitle} - else: - return {} - - def _extract_all_subtitles(self, video_id): - """ - Return a dicitonary: {language: subtitles} or {} if the subtitles - couldn't be found - """ - sub_lang_list = self._get_available_subtitles(video_id) - sub_format = self._downloader.params.get('subtitlesformat') + if self._downloader.params.get('writesubtitles', False): + if self._downloader.params.get('subtitleslang', False): + sub_lang = self._downloader.params.get('subtitleslang') + elif 'en' in sub_lang_list: + sub_lang = 'en' + else: + sub_lang = list(sub_lang_list.keys())[0] + if not sub_lang in sub_lang_list: + self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang) + return {} + sub_lang_list = {sub_lang: sub_lang_list[sub_lang]} subtitles = {} for sub_lang in sub_lang_list: subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format) @@ -534,14 +521,11 @@ class YoutubeIE(InfoExtractor): # subtitles video_subtitles = None - if self._downloader.params.get('writesubtitles', False): - video_subtitles = self._extract_subtitle(video_id) + if self._downloader.params.get('writesubtitles', False) or self._downloader.params.get('allsubtitles', False): + video_subtitles = self._extract_subtitles(video_id) elif self._downloader.params.get('writeautomaticsub', False): video_subtitles = self._request_automatic_caption(video_id, video_webpage) - if self._downloader.params.get('allsubtitles', False): - video_subtitles = self._extract_all_subtitles(video_id) - if self._downloader.params.get('listsubtitles', False): self._list_available_subtitles(video_id) return -- cgit v1.2.3 From 2f799533ae680dc788c8b4f6ce41272cf89689cc Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Sat, 29 Jun 2013 22:11:18 +0200 Subject: YoutubeIE: don't crash when trying to get automatic captions if the videos has standard subtitles. --- youtube_dl/extractor/youtube.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'youtube_dl/extractor/youtube.py') diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 30036524f..2b03226f6 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -270,7 +270,9 @@ class YoutubeIE(InfoExtractor): subtitles_url = caption_url + '&' + params sub = self._download_webpage(subtitles_url, video_id, u'Downloading automatic captions') return {sub_lang: sub} - except KeyError: + # An extractor error can be raise by the download process if there are + # no automatic captions but there are subtitles + except (KeyError, ExtractorError): self._downloader.report_warning(err_msg) return {} -- cgit v1.2.3 From 066090dd3f15886be40310f5a0702fe485c1512e Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Fri, 23 Aug 2013 11:33:56 +0200 Subject: [youtube] add algo for length 80 and update player info --- youtube_dl/extractor/youtube.py | 2 ++ 1 file changed, 2 insertions(+) (limited to 'youtube_dl/extractor/youtube.py') diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index e402ef17f..2a3e5031c 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -434,6 +434,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): return s[1:19] + s[0] + s[20:68] + s[19] + s[69:82] 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] -- cgit v1.2.3 From df3df7fb6458132a05a4cb64ab37701e7b6a39f2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Fri, 23 Aug 2013 13:14:22 +0200 Subject: [youtube] Fix download of subtitles with '--all-subs' If _extract_subtitles is called the option 'write subtitles' is always true. --- youtube_dl/extractor/youtube.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'youtube_dl/extractor/youtube.py') diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index c4d763ee1..446d53f64 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -534,7 +534,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor): sub_format = self._downloader.params.get('subtitlesformat') if not sub_lang_list: #There was some error, it didn't get the available subtitles return {} - if self._downloader.params.get('writesubtitles', False): + if self._downloader.params.get('allsubtitles', False): + pass + else: if self._downloader.params.get('subtitleslang', False): sub_lang = self._downloader.params.get('subtitleslang') elif 'en' in sub_lang_list: -- cgit v1.2.3 From aa6a10c44a8e2e86f709c5301f9ea6ac3f01f002 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Fri, 23 Aug 2013 18:34:57 +0200 Subject: Allow to specify multiple subtitles languages separated by commas (closes #518) --- youtube_dl/extractor/youtube.py | 29 ++++++++++++++++------------- 1 file changed, 16 insertions(+), 13 deletions(-) (limited to 'youtube_dl/extractor/youtube.py') diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 446d53f64..5f843a871 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -496,7 +496,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): def _request_automatic_caption(self, video_id, webpage): """We need the webpage for getting the captions url, pass it as an argument to speed up the process.""" - sub_lang = self._downloader.params.get('subtitleslang') or 'en' + sub_lang = (self._downloader.params.get('subtitleslangs') or ['en'])[0] sub_format = self._downloader.params.get('subtitlesformat') self.to_screen(u'%s: Looking for automatic captions' % video_id) mobj = re.search(r';ytplayer.config = ({.*?});', webpage) @@ -530,23 +530,26 @@ class YoutubeIE(YoutubeBaseInfoExtractor): Return a dictionary: {language: subtitles} or {} if the subtitles couldn't be found """ - sub_lang_list = self._get_available_subtitles(video_id) + available_subs_list = self._get_available_subtitles(video_id) sub_format = self._downloader.params.get('subtitlesformat') - if not sub_lang_list: #There was some error, it didn't get the available subtitles + if not available_subs_list: #There was some error, it didn't get the available subtitles return {} if self._downloader.params.get('allsubtitles', False): - pass + sub_lang_list = available_subs_list else: - if self._downloader.params.get('subtitleslang', False): - sub_lang = self._downloader.params.get('subtitleslang') - elif 'en' in sub_lang_list: - sub_lang = 'en' + if self._downloader.params.get('subtitleslangs', False): + reqested_langs = self._downloader.params.get('subtitleslangs') + elif 'en' in available_subs_list: + reqested_langs = ['en'] else: - sub_lang = list(sub_lang_list.keys())[0] - if not sub_lang in sub_lang_list: - self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang) - return {} - sub_lang_list = {sub_lang: sub_lang_list[sub_lang]} + reqested_langs = [list(available_subs_list.keys())[0]] + + sub_lang_list = {} + for sub_lang in reqested_langs: + if not sub_lang in available_subs_list: + self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang) + continue + sub_lang_list[sub_lang] = available_subs_list[sub_lang] subtitles = {} for sub_lang in sub_lang_list: subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format) -- cgit v1.2.3 From 306170518f016ac7ebf8311d7d2a23937aeaf1b3 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Fri, 23 Aug 2013 22:36:59 +0200 Subject: [youtube] update algo for length 86 (fixes #1302) --- youtube_dl/extractor/youtube.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'youtube_dl/extractor/youtube.py') diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 5f843a871..e4987b2b3 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -423,7 +423,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 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[5:20] + s[2] + s[21:] + return s[5:40] + s[3] + s[41:48] + s[0] + s[49:86] elif len(s) == 85: return s[83:34:-1] + s[0] + s[33:27:-1] + s[3] + s[26:19:-1] + s[34] + s[18:3:-1] + s[27] elif len(s) == 84: -- cgit v1.2.3