Merge pull request #235 from family-guy/video-metadata-#195

pytube/__main__.py

@@ -117,6 +117,7 @@ class YouTube(object):
                 self.watch_html,
             )['args']
 
+        self.vid_descr = extract.get_vid_descr(self.watch_html)
         # https://github.com/nficano/pytube/issues/165
         stream_maps = ['url_encoded_fmt_stream_map']
         if 'adaptive_fmts' in self.player_config_args:
@@ -252,6 +253,42 @@ class YouTube(object):
         """
         return self.player_config_args['title']
 
+    @property
+    def description(self):
+        """Get the video description.
+
+        :rtype: str
+
+        """
+        return self.vid_descr
+
+    @property
+    def rating(self):
+        """Get the video average rating.
+
+        :rtype: str
+
+        """
+        return self.player_config_args['avg_rating']
+
+    @property
+    def length(self):
+        """Get the video length in seconds.
+
+        :rtype: str
+
+        """
+        return self.player_config_args['length_seconds']
+
+    @property
+    def views(self):
+        """Get the number of the times the video has been viewed.
+
+        :rtype: str
+
+        """
+        return self.player_config_args['view_count']
+
     def register_on_progress_callback(self, func):
         """Register a download progress callback function post initialization.
 

pytube/compat.py

@@ -4,12 +4,14 @@
 """Python 2/3 compatibility support."""
 import sys
 
+
 PY2 = sys.version_info[0] == 2
 PY3 = sys.version_info[0] == 3
 PY33 = sys.version_info[0:2] >= (3, 3)
 
-
 if PY2:
+    reload(sys)
+    sys.setdefaultencoding('utf8')
     import urllib2
     from urllib import urlencode
     from urllib2 import URLError

pytube/extract.py

@@ -3,12 +3,32 @@
 import json
 from collections import OrderedDict
 
+from pytube.compat import HTMLParser
 from pytube.compat import quote
 from pytube.compat import urlencode
 from pytube.exceptions import RegexMatchError
 from pytube.helpers import regex_search
 
 
+class PytubeHTMLParser(HTMLParser):
+    in_vid_descr = False
+    vid_descr = ''
+
+    def handle_starttag(self, tag, attrs):
+        if tag == 'p':
+            for attr in attrs:
+                if attr[0] == 'id' and attr[1] == 'eow-description':
+                    self.in_vid_descr = True
+
+    def handle_endtag(self, tag):
+        if tag == 'p' and self.in_vid_descr:
+            self.in_vid_descr = False
+
+    def handle_data(self, data):
+        if self.in_vid_descr:
+            self.vid_descr += data
+
+
 def is_age_restricted(watch_html):
     """Check if content is age restricted.
 
@@ -173,3 +193,9 @@ def get_ytplayer_config(html, age_restricted=False):
         pattern = r';ytplayer\.config\s*=\s*({.*?});'
     yt_player_config = regex_search(pattern, html, group=1)
     return json.loads(yt_player_config)
+
+
+def get_vid_descr(html):
+    html_parser = PytubeHTMLParser()
+    html_parser.feed(html)
+    return html_parser.vid_descr

tests/test_extract.py

@@ -43,3 +43,8 @@ def test_age_restricted(age_restricted):
 
 def test_non_age_restricted(cipher_signature):
     assert not extract.is_age_restricted(cipher_signature.watch_html)
+
+
+def test_get_vid_desc(cipher_signature):
+    expected = "PSY - DADDY(feat. CL of 2NE1) M/V @ https://youtu.be/FrG4TEcSuRgPSY - 나팔바지(NAPAL BAJI) M/V @ https://youtu.be/tF27TNC_4pcPSY - 7TH ALBUM '칠집싸이다' on iTunes @ http://smarturl.it/PSY_7THALBUMPSY - GANGNAM STYLE(강남스타일) on iTunes @ http://smarturl.it/PsyGangnam#PSY #싸이 #GANGNAMSTYLE #강남스타일More about PSY@http://www.psypark.com/http://www.youtube.com/officialpsyhttp://www.facebook.com/officialpsyhttp://twitter.com/psy_oppahttps://www.instagram.com/42psy42http://iTunes.com/PSYhttp://sptfy.com/PSYhttp://weibo.com/psyoppahttp://twitter.com/ygent_official"  # noqa
+    assert extract.get_vid_descr(cipher_signature.watch_html) == expected