Use a dictionary for storing the subtitles

[youtube-dl] / youtube_dl / extractor / youtube.py

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 12e8fc25d6f5c19f2eca21dae9b0b1caec508650..78500b0f72586a1dfd5e581f014dc5a5c3ad6b30 100644 (file)
--- 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)