Changeset 12456

Timestamp:

2006-08-17T13:31:53+12:00 (18 years ago)

Author:

mdewsnip

Message:

Much improved RTF to HTML conversion.

File:

• trunk/cic-hcap/perllib/plugins/CICPlug.pm (modified) (3 diffs)

trunk/cic-hcap/perllib/plugins/CICPlug.pm

@@ -965 +965 @@
     $metadata_value = &unicode::ascii2utf8(\$metadata_value);
 
-    # Remove '#' characters around links
+    # Remove '#' characters around links (!! BAD)
     if ($metadata_value =~ /\#(.*?)\#/) {
     $metadata_value = $1;
@@ -990 +990 @@
     # Split the references and try to parse title and author
     my $place_references_html = "";
-    $place_references_html_raw =~ s/(\r|\n)//g;
-    my @place_references = split(/<br \/><br \/>/, $place_references_html_raw);
+    $place_references_html_raw =~ s/(\r|\n)//g;  # Remove all newlines
+    $place_references_html_raw =~ s/<br \/><i><br \/>/<br \/><br \/><i>/g;  # Move italic tags
+    my @place_references = split(/<br \/>\s*<br \/>/, $place_references_html_raw);
     foreach my $place_reference (@place_references) {
     $place_reference =~ s/^(<br \/>\s*)*//;
@@ -1043 +1044 @@
     }
     }
-    $rtf_string =~ s/\\ldblquote /"/g;
-    $rtf_string =~ s/\\rdblquote /"/g;
-    $rtf_string =~ s/\\rquote /'/g;  # ' # (for Emacs)
+    $rtf_string =~ s/\\ldblquote\s/"/g;
+    $rtf_string =~ s/\\ldblquote\b/"/g;
+    $rtf_string =~ s/\\rdblquote\s/"/g;
+    $rtf_string =~ s/\\rdblquote\b/"/g;
+    $rtf_string =~ s/\\rquote\s/'/g;  # ' # (for Emacs)
+    $rtf_string =~ s/\\rquote\b/'/g;  # ' # (for Emacs)
     $rtf_string =~ s/\\pard//g;
     $rtf_string =~ s/\\par/<br \/>/g;
-    $rtf_string =~ s/\\ul /<i>/g;
-    $rtf_string =~ s/\\ulnone /<\/i>/g;
-    $rtf_string =~ s/\\i0 /<\/i>/g;
-    $rtf_string =~ s/\\i0\\/<\/i>\\/g;
-    $rtf_string =~ s/\\i /<i>/g;
-    $rtf_string =~ s/\\i\\/<i>\\/g;
+    $rtf_string =~ s/\\ul\s/<i>/g;
+    $rtf_string =~ s/\\ul\b/<i>/g;
+    $rtf_string =~ s/\\ulnone\s/<\/i>/g;
+    $rtf_string =~ s/\\ulnone\b/<\/i>/g;
+    $rtf_string =~ s/\\i\s/<i>/g;
+    $rtf_string =~ s/\\i\b/<i>/g;
+    $rtf_string =~ s/\\i0\s/<\/i>/g;
+    $rtf_string =~ s/\\i0\b/<\/i>/g;
     $rtf_string =~ s/\\~/ /g;
     $rtf_string =~ s/\\([A-Za-z0-9\-]+)//g;