Convert redirect timeout to int if possible.

[bookmarks_db.git] / Robots / bkmk_rsimple.py

diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py
index 1842b4d7ff94e73327f127afa2b71a74c61f7ed8..421d677530d88fefa0239e2aa388db70533517b2 100644 (file)
--- a/Robots/bkmk_rsimple.py
+++ b/Robots/bkmk_rsimple.py
@@ -44,8 +44,7 @@ class MyURLopener(urllib.URLopener):
 
 urllib._urlopener = MyURLopener()
 
-# Some sites allow only Mozilla-compatible browsers; way to stop robots?
-server_version = "Mozilla/3.0 (compatible; Python-urllib/%s)" % urllib.__version__
+server_version = "bookmarks_db (Python-urllib/%s)" % urllib.__version__
 urllib._urlopener.addheaders[0] = ('User-agent', server_version)
 
 
@@ -145,9 +144,7 @@ class robot_simple(Robot):
                      if self.log: self.log("   no charset in Content-Type header")
                   if content_type == "text/html":
                      parser = parse_html(fname, charset, self.log)
-                     title = parser.title.replace('\r', '').replace('\n', ' ').strip()
-                     bookmark.real_title = parser.unescape(title)
-                     if self.log: self.log("   final title    : %s" % bookmark.real_title)
+                     bookmark.real_title = parser.title
                      if parser.refresh:
                         refresh = parser.refresh
                         try:
@@ -155,13 +152,15 @@ class robot_simple(Robot):
                         except IndexError:
                            url = "self"
                         try:
-                           timeout = int(refresh.split(';')[0])
+                           timeout = float(refresh.split(';')[0])
                         except (IndexError, ValueError):
-                           timeout = None
-                        if timeout is None:
                            raise RedirectException("html", "Bad redirect to %s (%s)" % (url, refresh))
                         else:
-                           raise RedirectException("html", "%s (%d sec)" % (url, timeout))
+                           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
@@ -178,13 +177,15 @@ class robot_simple(Robot):
                         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"]
                         if content_type.startswith("image/"):
+                           icon_file = open(fname, "rb")
+                           icon = icon_file.read()
+                           icon_file.close()
                            bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon))
+                        else:
+                           if self.log: self.log("   no icon        : %s" % content_type)
+                        os.remove(fname)
 
                except KeyError:
                   pass