Changeset 3410 for trunk/gsdl/bin

Timestamp:

2002-08-29T15:53:15+12:00 (22 years ago)

Author:

jrm21

Message:

now post-filter the html so that the filename is not included for anchor
links. (eg <a href="thisdoc.html#4"> is now just <a href="#4">)

File:

• trunk/gsdl/bin/script/pdftohtml.pl (modified) (2 diffs)

trunk/gsdl/bin/script/pdftohtml.pl

@@ -75 +75 @@
     my $input_filename = $ARGV[0];
     my $output_filestem = $ARGV[1];
-    $output_filestem =~ s/\.html$//; # pdftohtml adds this suffix
+    $output_filestem =~ s/\.html$//i; # pdftohtml adds this suffix
 
     my @dir = split (/(\/|\\)/, $input_filename);
-    pop(@dir);
+    my $input_basename = pop(@dir);
+    $input_basename =~ s/\.pdf//i;
     my $dir = join ("", @dir);
 
@@ -144 +145 @@
     }
 
+    # relative hrefs to own document...
+    $line =~ s@href=\"$input_basename\.html\#@href=\"\#@go;
 # escape underscores, but not if they're inside tags (eg img/href names)
     my $inatag = 0; # allow multi-line tags