X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fbkmk_rsimple.py;h=8bfa4298299f0c3b9842cdc0fa071b361ed8cd48;hb=a01b9869e595d0c1cc5ebeb6f5a1da66d45c3d22;hp=c4a16644a468925a2a68248fdc195973b2cc9dfe;hpb=09ecb9e8c3d908c67a0f43a5623f574f39fba1eb;p=bookmarks_db.git

diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py
index c4a1664..8bfa429 100644
--- a/Robots/bkmk_rsimple.py
+++ b/Robots/bkmk_rsimple.py
@@ -1,9 +1,15 @@
-"""
-   Simple, strightforward robot
+"""Simple, strightforward robot
 
-   Written by Oleg Broytman. Copyright (C) 2000-2010 PhiloSoft Design.
+This file is a part of Bookmarks database and Internet robot.
 """
 
+__version__ = "$Revision$"[11:-2]
+__revision__ = "$Id$"[5:-2]
+__date__ = "$Date$"[7:-2]
+__author__ = "Oleg Broytman <phd@phdru.name>"
+__copyright__ = "Copyright (C) 2000-2011 PhiloSoft Design"
+__license__ = "GNU GPL"
+
 import sys, os
 import time, urllib
 from base64 import b64encode
@@ -176,24 +182,6 @@ class robot_simple(Robot):
                   parser = parse_html(fname, charset, self.log)
                   if parser:
                       bookmark.real_title = parser.title
-                      if parser.refresh:
-                         refresh = parser.refresh
-                         try:
-                            url = refresh.split('=', 1)[1]
-                         except IndexError:
-                            url = "self"
-                         try:
-                            timeout = float(refresh.split(';')[0])
-                         except (IndexError, ValueError):
-                            raise RedirectException("html", "Bad redirect to %s (%s)" % (url, refresh))
-                         else:
-                            try:
-                               timeout = int(refresh.split(';')[0])
-                            except ValueError:
-                               pass # float timeout
-                            raise RedirectException("html", "%s (%s sec)" % (url, timeout))
-
-                      # Get favicon.ico
                       icon = parser.icon
                   else:
                      icon = None
@@ -224,23 +212,43 @@ class robot_simple(Robot):
                      except:
                         etype, emsg, tb = sys.exc_info()
                         self.log("   no icon        : %s %s" % (etype, emsg))
-                        etype = None
-                        emsg = None
-                        tb = None
+                        etype = emsg = tb = None
                         icons[icon] = None
                      else:
                         content_type = headers["Content-Type"]
-                        if content_type.startswith("image/"):
+                        if content_type.startswith("application/") \
+                              or content_type.startswith("image/") \
+                              or content_type.startswith("text/plain"):
                            icon_file = open(icon_fname, "rb")
                            icon_data = icon_file.read()
                            icon_file.close()
                            bookmark.icon_href = icon
-                           bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon_data))
                            self.log("   got icon       : %s" % content_type)
+                           if content_type.startswith("application/") \
+                                 or content_type.startswith("text/plain"):
+                              self.log("       non-image content type, assume x-icon")
+                              content_type = 'image/x-icon'
+                           bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon_data))
                            icons[icon] = (content_type, bookmark.icon)
                         else:
                            self.log("   no icon        : bad content type '%s'" % content_type)
                            icons[icon] = None
+                  if parser and parser.refresh:
+                     refresh = parser.refresh
+                     try:
+                        url = refresh.split('=', 1)[1]
+                     except IndexError:
+                        url = "self"
+                     try:
+                        timeout = float(refresh.split(';')[0])
+                     except (IndexError, ValueError):
+                        raise RedirectException("html", "Bad redirect to %s (%s)" % (url, refresh))
+                     else:
+                        try:
+                           timeout = int(refresh.split(';')[0])
+                        except ValueError:
+                           pass # float timeout
+                        raise RedirectException("html", "%s (%s sec)" % (url, timeout))
 
             except KeyError, key:
                self.log("   no header: %s" % key)
@@ -291,3 +299,4 @@ class robot_simple(Robot):
 
       now = int(time.time())
       bookmark.test_time = str(now - start)
+      urllib.urlcleanup()