X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Futils.py;h=01c8c017d53a90477871f0ad4b6ef15250cc5e5b;hb=79bc27b53ab57da4a3067972283be3baddde1725;hp=ff124d9e8cbd42d04904e527750b7c3d94e74f59;hpb=b7ab05908440915c6c5faa541abe00c62a88bc27;p=youtube-dl.git

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index ff124d9e8..01c8c017d 100644
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -225,7 +225,7 @@ if sys.version_info >= (2,7):
     def find_xpath_attr(node, xpath, key, val):
         """ Find the xpath xpath[@key=val] """
         assert re.match(r'^[a-zA-Z]+$', key)
-        assert re.match(r'^[a-zA-Z0-9@\s]*$', val)
+        assert re.match(r'^[a-zA-Z0-9@\s:._]*$', val)
         expr = xpath + u"[@%s='%s']" % (key, val)
         return node.find(expr)
 else:
@@ -751,13 +751,14 @@ class YoutubeDLHandler(compat_urllib_request.HTTPHandler):
     https_request = http_request
     https_response = http_response
 
+
 def unified_strdate(date_str):
     """Return a string with the date in the format YYYYMMDD"""
     upload_date = None
     #Replace commas
     date_str = date_str.replace(',',' ')
     # %z (UTC offset) is only supported in python>=3.2
-    date_str = re.sub(r' (\+|-)[\d]*$', '', date_str)
+    date_str = re.sub(r' ?(\+|-)[0-9:]*$', '', date_str)
     format_expressions = [
         '%d %B %Y',
         '%B %d %Y',
@@ -771,11 +772,12 @@ def unified_strdate(date_str):
         '%Y-%m-%dT%H:%M:%S.%fZ',
         '%Y-%m-%dT%H:%M:%S.%f0Z',
         '%Y-%m-%dT%H:%M:%S',
+        '%Y-%m-%dT%H:%M',
     ]
     for expression in format_expressions:
         try:
             upload_date = datetime.datetime.strptime(date_str, expression).strftime('%Y%m%d')
-        except:
+        except ValueError:
             pass
     if upload_date is None:
         timetuple = email.utils.parsedate_tz(date_str)
@@ -1093,9 +1095,12 @@ def month_by_name(name):
         return None
 
 
-def fix_xml_all_ampersand(xml_str):
+def fix_xml_ampersands(xml_str):
     """Replace all the '&' by '&' in XML"""
-    return xml_str.replace(u'&', u'&')
+    return re.sub(
+        r'&(?!amp;|lt;|gt;|apos;|quot;|#x[0-9a-fA-F]{,4};|#[0-9]{,4};)',
+        u'&',
+        xml_str)
 
 
 def setproctitle(title):
@@ -1129,8 +1134,8 @@ class HEADRequest(compat_urllib_request.Request):
         return "HEAD"
 
 
-def int_or_none(v):
-    return v if v is None else int(v)
+def int_or_none(v, scale=1):
+    return v if v is None else (int(v) // scale)
 
 
 def parse_duration(s):
@@ -1169,6 +1174,10 @@ class PagedList(object):
         self._pagefunc = pagefunc
         self._pagesize = pagesize
 
+    def __len__(self):
+        # This is only useful for tests
+        return len(self.getslice())
+
     def getslice(self, start=0, end=None):
         res = []
         for pagenum in itertools.count(start // self._pagesize):