Document and test categories (#2923)

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

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 1f3aa4322c274e226b6e5945874e8627724a79cf..3c8f7f7a2a65720e72c9efeb0fbf017202902de5 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -242,7 +242,8 @@ 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'],
             }
         },
         {
@@ -1136,11 +1137,20 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
 
         # upload date
         upload_date = None
-        mobj = re.search(r'id="eow-date.*?>(.*?)</span>', video_webpage, re.DOTALL)
+        mobj = re.search(r'(?s)id="eow-date.*?>(.*?)</span>', 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)<a[^<]+>(.*?)</a>', m_cat_container, 'cateory',
+                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:
@@ -1347,6 +1357,7 @@ 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,