Changeset 3720 for trunk/gsdl/bin/script/gsConvert.pl

Timestamp:

2003-01-28T16:22:26+13:00 (21 years ago)

Author:

sjboddie

Message:

Added options to PDFPlug to take advantage of the improvements in
version 0.34 of pdftohtml. It now works much better for non latin
input documents (producing UTF-8 encoded HTML).

File:

• trunk/gsdl/bin/script/gsConvert.pl (modified) (4 diffs)

trunk/gsdl/bin/script/gsConvert.pl

@@ -59 +59 @@
 
 my $use_strings;
+my $pdf_complex;
+my $pdf_zoom;
+my $pdf_ignore_images;
 
 sub print_usage
@@ -70 +73 @@
     print STDERR "\t-output\thtml|text\n";
     print STDERR "\t-timeout\t<max cpu seconds>\t(ulimit on unix systems)\n";
-    print STDERR "\t-use_strings\t(use strings to extract text if conversion fails)\n";
+    print STDERR "\t-use_strings\tuse strings to extract text if conversion fails\n";
+    print STDERR "\t-pdf_complex\tuse complex output when converting PDF to HTML\n";
+    print STDERR "\t-pdf_ignore_images\tdon't attempt to extract images when\n";
+    print STDERR "\t\tconverting PDF to HTML\n";
+    print STDERR "\t-pdf_zoom\tfactor by which to zoom PDF (only useful if\n";
+    print STDERR "\t\t-pdf_complex is set\n";
     exit(1);
 }
@@ -89 +97 @@
              'timeout/\d+/0',\$timeout,
              'verbose/\d+/0',   \$verbose,
-             'use_strings', \$use_strings))
+             'use_strings', \$use_strings,
+             'pdf_complex', \$pdf_complex,
+             'pdf_zoom/\d+/2', \$pdf_zoom
+             ))
     {
     print_usage();
@@ -628 +639 @@
     $cmd = "";
     if ($timeout) {$cmd = "ulimit -t $timeout;";}
-    $cmd .= "perl -S pdftohtml.pl ";
+    $cmd .= "perl -S pdftohtml.pl -zoom $pdf_zoom";
+    $cmd .= " -c" if ($pdf_complex);
+    $cmd .= " -i" if ($pdf_ignore_images);
     $cmd .= " \"$input_filename\" \"$output_filestem\"";