Changeset 10827

Timestamp:

2005-11-02T15:50:56+13:00 (18 years ago)

Author:

jrm21

Message:

1) include %xx bits when making hrefs out of urls
2) test if text is valid utf-8 first instead of just defaulting to

iso-8859-1 latin charset

File:

• trunk/gsdl/perllib/plugins/EMAILPlug.pm (modified) (4 diffs)

trunk/gsdl/perllib/plugins/EMAILPlug.pm

@@ -487 +487 @@
     # assume hostnames are \.\w\- only, then might have a trailing '/.*'
     # assume URI doesn't finish with a '.'
-    $text =~ s@((http|ftp|https)://[\w\-]+(\.[\w\-]+)*/?((&amp;|\.)?[\w\?\=\-_/~]+)*(\#[\w\.\-_]*)?)@<a href=\"$1\">$1<\/a>@g;
+    $text =~ s@((http|ftp|https)://[\w\-]+(\.[\w\-]+)*/?((&amp;|\.|\%[a-f0-9]{2})?[\w\?\=\-_/~]+)*(\#[\w\.\-_]*)?)@<a href=\"$1\">$1<\/a>@gi;
 
 
@@ -700 +700 @@
     $return_text.="\n<p><hr><strong>&lt;&lt;attachment&gt;&gt;";
     # add part info header
-    my $header_text="<br>Type: $part_content_type<br>\n";
+    my $header_text = "<br>Type: $part_content_type<br>\n";
     if ($filename ne "") {
-        $header_text.="Filename: $filename\n";
+        $header_text .= "Filename: $filename\n";
     }
     $header_text =~ s@_@\\_@g;
-    $return_text.=$header_text . "</strong></p>\n<p>\n";
+    $return_text .= $header_text . "</strong></p>\n<p>\n";
     }
 
     if ($part_content_type =~ m@text/@)
     {
-    my $part_text= $self->text_from_part($message_part);
+    # $message_part includes the mime part headers
+    my $part_text = $self->text_from_part($message_part);
     if ($part_content_type !~ m@text/(ht|x)ml@) {
         $part_text = text_into_html($part_text);
@@ -864 +865 @@
 
     my $type="text/plain"; # default, overridden from part header
-    my $charset="ascii"; # default, overridden from part header
+    my $charset=undef;     # convert2unicode() will guess if necessary
 
     if (! $part_header) { # no header argument was given. check the body
@@ -982 +983 @@
   }
 
+  if (! defined $charset) {
+      # check if we have valid utf-8
+      if ($$textref =~ /^(?: [\0-\x7f]          | # ascii
+              [\xc0-\xdf][\x80-\xbf]    | # 2 byte utf-8
+              [\xe0-\xef][\x80-\xbf]{2} | # 3 byte
+              [\xf0-\xf7][\x80-\xbf]{3} | # 4 byte
+              [\xf8-\xfb][\x80-\xbf]{4} | # 5 byte
+              [\xfc-\xfd][\x80-\xbf]{5} | # 6 byte
+              )+ /x) {
+      $charset = "utf8";
+      }
+
+
+      # default to latin
+      $charset = "iso_8859_1" if ! defined($charset);
+  }
+
   # first get our character encoding name in the right form.
-  $charset = "iso_8859_1" unless defined $charset;
   $charset =~ tr/A-Z/a-z/; # lowercase
   $charset =~ s/\-/_/g;