No need to split bookmark.href - a robot will split it itself.

[bookmarks_db.git] / Robots / bkmk_rsimple.py

diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py
index c34a12728d136307da774e5a6ac086622bf51515..1c4795c11d6aecb8bfc6a460a260be5b56a65d8b 100644 (file)
--- a/Robots/bkmk_rsimple.py
+++ b/Robots/bkmk_rsimple.py
@@ -5,8 +5,9 @@
 """
 
 
-import string, os
+import sys, os, string
 import time, urllib
+from base64 import b64encode
 from urlparse import urljoin
 from m_lib.net.www.util import parse_time
 from m_lib.md5wrapper import md5wrapper
@@ -82,13 +83,14 @@ from bkmk_objects import Robot
 from parse_html import parse_html
 
 class robot_simple(Robot):
-   def check_url(self, bookmark, url_type, url_rest):
+   def check_url(self, bookmark):
       if not self.tempfname:
          self.tempfname = bookmark.tempfname
 
       try:
          try:
             self.start = int(time.time())
+            url_type, url_rest = urllib.splittype(bookmark.href)
             url_host, url_path = urllib.splithost(url_rest)
             url_path, url_tag  = urllib.splittag(url_path)
 
@@ -158,11 +160,29 @@ class robot_simple(Robot):
                            raise RedirectException("html", "Bad redirect to %s (%s)" % (url, refresh))
                         else:
                            raise RedirectException("html", "%s (%d sec)" % (url, timeout))
+
+                     # Get favicon.ico
                      icon = parser.icon
                      if not icon:
                         icon = "/favicon.ico"
-                     icon = urljoin("%s://%s" % (url_type, url_host), icon)
+                     icon = urljoin("%s://%s%s" % (url_type, url_host, url_path), icon)
                      if self.log: self.log("   icon           : %s" % icon)
+                     try:
+                        fname, headers = urllib.urlretrieve(icon)
+                     except:
+                        etype, emsg, tb = sys.exc_info()
+                        if self.log: self.log("   no icon        : %s %s" % (etype, emsg))
+                        etype = None
+                        emsg = None
+                        tb = None
+                     else:
+                        icon_file = open(fname, "rb")
+                        icon = icon_file.read()
+                        icon_file.close()
+                        os.remove(fname)
+                        content_type = headers["Content-Type"]
+                        bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon))
+
                except KeyError:
                   pass