package Apache::Tidy; use strict; use warnings; use vars qw($VERSION); $VERSION = "0.1"; use Apache::Constants qw(OK DECLINED NOT_FOUND); use Apache::File; sub handler { my $r = shift; # we only care about html return DECLINED unless $r->content_type eq 'text/html'; my $fh = undef; if (lc $r->dir_config('Filter') eq 'on') { # register as a filter $r = $r->filter_register; # get input from any previous filters ($fh,my $status) = $r->filter_input; return $status unless $status == OK; } else { $fh = Apache::File->new($r->filename); return DECLINED unless $fh; } my $dirty = do {local $/; <$fh>}; my $tidy_path = $r->dir_config('TidyPath') || "/usr/bin/tidy"; my $temp_dir = $r->dir_config('TidyTempDir') || "/tmp"; my $options = join ' ', $r->dir_config->get('TidyOptions'); $options = $options || "-q -asxhtml"; $r->send_http_header('text/html'); # clean up the path so we can run in taint mode delete $ENV{PATH}; eval { # write a tempfile open(TMP,">$temp_dir/tidy_$$.html") or die "couldn't write to tempfile: $!"; print TMP $dirty; close TMP; # run tidy over it system("$tidy_path $options $temp_dir/tidy_$$.html > $temp_dir/tidy_out_$$.html"); # read in results open(OUT,"<$temp_dir/tidy_out_$$.html") or die "couldn't read tempfile: $!"; my @results = ; close OUT; # clean up unlink "$temp_dir/tidy_$$.html"; unlink "$temp_dir/tidy_out_$$.html"; print @results; }; if ($@) { # if something generated an error, # we default to just passing the content on unchanged. print $dirty; } return OK; } 1; __END__ =head1 NAME Apache::Tidy - htmltidy as an apache filter =head1 SYNOPSIS PerlModule Apache::Filter PerlModule Apache::Tidy SetHandler perl-script PerlHandler Apache::Tidy =head1 ABSTRACT Cleans up and fixes invalid HTML on the fly. =head1 DESCRIPTION Wrapper for the htmltidy program (L) using the Apache::Filter framework. Fixes HTML/XHTML validation issues on the fly. Dave Raggett's HTML Tidy is a free command-line utility for cleaning up messy and invalid HTML or XHTML code. It will correct missing or mismatched end tags, clean up Microsoft Word generated HTML, convert pages to XHTML, and format markup for easier reading. Apache::Tidy uses the Apache::Filter framework to allow you to automatically run tidy over web content as it is being served. This can be very useful if you have editors or CMSes that produce invalid markup. To filter static content add the following to your httpd.conf: PerlModule Apache::Tidy SetHandler perl-script PerlHandler Apache::Tidy Apache::Tidy can also work as part of an Apache::Filter chain: PerlModule Apache::Filter PerlModule Apache::RegistryFilter PerlModule Apache::Tidy PerlSetVar Filter On SetHandler perl-script PerlHandler Apache::RegistryFilter Apache::Tidy Apache::Tidy supports all of htmltidy's command-line options by setting TidyOptions: SetHandler perl-script PerlHandler Apache::Tidy PerlSetVar TidyOptions '-wrap 60' PerlSetVar TidyOptions -clean PerlSetVar TidyOptions -asxhtml It defaults to '-q -asxhtml' if no options are explicitly set. You can also specify a different path to the tidy executable (necessary if you've installed it anywhere but in /usr/bin/) and the temp directory used can also be specified (defaults to /tmp): SetHandler perl-script PerlHandler Apache::Tidy PerlSetVar TidyPath /opt/local/bin/tidy PerlSetVar TidyTempDir /some/other/temp/dir =head1 NOTES You must have htmltidy installed on your system. if it is installed anywhere other than in /usr/bin/, you'll have to specify the full path with PerlSetVar TidyPath /path/to/tidy I've only tested Apache::Tidy on unix systems. It may run on other platforms, but you will probably have to change the path, temp directory, and options. Since Apache::Tidy just jumps out to the shell to call the external tidy program, it probably isn't very efficient. I'd like to reimplement this someday with an XS or SWIG wrapped tidylib. =head1 SEE ALSO L, L, L =head1 AUTHOR Anders Pearson, Eanders@columbia.eduE =head1 COPYRIGHT AND LICENSE Copyright 2003 by Anders Pearson This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself. =cut