Move charset to the beginning of the list.

[bookmarks_db.git] / Robots / bkmk_rsimple.py

diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py
index f481d66e1bce396d46387571211d010eeea40a52..ba302943d175dd5e969adcf82cf564e0f2cb644b 100644 (file)
--- a/Robots/bkmk_rsimple.py
+++ b/Robots/bkmk_rsimple.py
@@ -1,7 +1,7 @@
 """
    Simple, strightforward robot
 
-   Written by Oleg BroytMann. Copyright (C) 2000-2008 PhiloSoft Design.
+   Written by Oleg Broytman. Copyright (C) 2000-2010 PhiloSoft Design.
 """
 
 import sys, os
@@ -40,6 +40,8 @@ class MyURLopener(urllib.URLopener):
 
    # Error 301 -- also relocated (permanently)
    http_error_301 = http_error_302
+   # Error 307 -- also relocated (temporary)
+   http_error_307 = http_error_302
 
    # Error 401 -- authentication required
    def http_error_401(self, url, fp, errcode, errmsg, headers, data=None): 
@@ -156,6 +158,7 @@ class robot_simple(Robot):
          if headers:
             try:
                content_type = headers["Content-Type"]
+               self.log("   Content-Type: %s" % content_type)
                try:
                   # extract charset from "text/html; foo; charset=UTF-8, bar; baz;"
                   content_type, charset = content_type.split(';', 1)
@@ -165,7 +168,13 @@ class robot_simple(Robot):
                except (ValueError, IndexError):
                   charset = None
                   self.log("   no charset in Content-Type header")
-               if content_type in ("text/html", "application/xhtml+xml"):
+               for ctype in ("text/html", "application/xhtml+xml"):
+                  if content_type.startswith(ctype):
+                      html = True
+                      break
+               else:
+                  html = False
+               if html:
                   parser = parse_html(fname, charset, self.log)
                   bookmark.real_title = parser.title
                   if parser.refresh:
@@ -193,6 +202,7 @@ class robot_simple(Robot):
                   self.log("   looking for icon at: %s" % icon)
                   if icon in icons:
                      if icons[icon]:
+                        bookmark.icon_href = icon
                         content_type, bookmark.icon = icons[icon]
                         self.log("       cached icon: %s" % content_type)
                      else:
@@ -221,9 +231,10 @@ class robot_simple(Robot):
                         content_type = headers["Content-Type"]
                         if content_type.startswith("image/"):
                            icon_file = open(fname, "rb")
-                           icon = icon_file.read()
+                           icon_data = icon_file.read()
                            icon_file.close()
-                           bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon))
+                           bookmark.icon_href = icon
+                           bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon_data))
                            self.log("   got icon       : %s" % content_type)
                            icons[icon] = (content_type, bookmark.icon)
                         else:
@@ -231,28 +242,34 @@ class robot_simple(Robot):
                            icons[icon] = None
                         os.remove(fname)
 
-            except KeyError:
-               pass
+            except KeyError, key:
+               self.log("   no header: %s" % key)
 
       except IOError, msg:
          if (msg[0] == "http error") and (msg[1] == -1):
             bookmark.no_error = "The server did not return any header - it is not an error, actually"
+            self.log(bookmark.no_error)
          else:
             bookmark.error = get_error(msg)
+            self.log(bookmark.error)
 
       except EOFError:
          bookmark.error = "Unexpected EOF (FTP server closed connection)"
+         self.log(bookmark.error)
 
       except RedirectException, msg:
          bookmark.moved = str(msg)
+         self.log(bookmark.moved)
 
       except KeyboardInterrupt:
+         self.log("Keyboard interrupt (^C)")
          return 0
 
       except:
          import traceback
          traceback.print_exc()
          bookmark.error = "Exception!"
+         self.log(bookmark.error)
 
       finally:
          self.finish_check_url(bookmark)