1 files changed, 64 insertions, 0 deletions
diff --git a/scripts/fix-groff-xhtml.pl b/scripts/fix-groff-xhtml.pl
new file mode 100755
index 0000000..00d587a
--- /dev/null
+++ b/scripts/fix-groff-xhtml.pl
@@ -0,0 +1,64 @@
+#!/usr/bin/perl
+#
+# Format XHTML generated by groff -Thtml (via tidy) for websites
+#
+# Usage: groff -Thtml -P-l something.man | tidy -asxml ... | perl fix-groff-xhtml.pl OUTPUT-FILE
+#
+# (C) Copyright 2003-2014 Dave Beckett
+#
+
+use strict;
+use File::Basename;
+
+my $progname=basename $0;
+
+my $raptor_title="Raptor RDF Parser Toolkit";
+my $redland_title="Redland RDF Application Framework";
+my $rasqal_title="Rasqal RDF Query Library";
+
+die "USAGE: $progname OUTPUT-FILE\n" if @ARGV < 1;
+
+my $doc_title;
+
+my($file)=@ARGV;
+
+open(OUT, ">$file") or die "$progname: Cannot create $file - $!\n";
+open(IN, "-");
+while(<IN>) {
+
+  s%<title>libraptor</title>%<title>$raptor_title - Raptor API</title>%;
+  s%<h1 align="center">libraptor</h1>%<h1>$raptor_title - Raptor API</h1>%;
+
+  s%<title>rapper</title>%<title>$raptor_title - Raptor RDF parser utility</title>%;
+  s%<h1 align="center">rapper</h1>%<h1>$raptor_title - Raptor RDF parser utility</h1>%;
+
+  s%<title>rdfproc</title>%<title>$redland_title - Redland RDF processor utility</title>%;
+  s%<h1 align="center">rdfproc</h1>%<h1>$redland_title - Redland RDF processor utility</h1>%;
+
+  s%<title>librasqal</title>%<title>$rasqal_title - Rasqal API</title>%;
+  s%<h1 align="center">librasqal</h1>%<h1>$rasqal_title - Rasqal API</h1>%;
+
+  s%<title>roqet</title>%<title>$rasqal_title - Rasqal RDF parser utility</title>%;
+  s%<h1 align="center">roqet</h1>%<h1>$rasqal_title - Rasqal RDF parser utility</h1>%;
+
+  next if /^<link|meta/i;
+
+  s%^<body[^>]*>%<body>%;
+
+  # This is not xhtml
+  s% cols="\d+" % %;
+
+  s%(name|id)="([^"]+)"%my($at,$val)=($1,$2); $val =~ s/ /_/g; qq{$at="$val"};%eg;
+
+  s%(Dave Beckett|Institute for Learning and Research Technology .ILRT.|University of Bristol) (?:- |)(http://[^<]+)%<a href="$2">$1</a>%;
+
+  my $year=1900+(localtime)[5];
+  print OUT <<"EOT" if m%^</body>%;
+
+<p>Copyright 2002-$year <a href="http://www.dajobe.org/">Dave Beckett</a><br />2002-2005 <a href="http://www.bristol.ac.uk/">University of Bristol</a></p>
+
+EOT
+  print OUT;
+}
+close(IN);
+close(OUT);