trunk/spider/filter.pm

sub filter {
        my $contents = shift || return;
        # if you don't want content to be indexed, include it in
        # <noindex> foobar </noindex> tags or surround it with comments
        # <!-- noindex --> foobar <!-- /noindex -->
        # <!-- noindex --> foobar <!-- index --> (also supported by swish)
        $contents =~ s,<noindex>.+?</noindex>,,isg;
        $contents =~ s,<!--\s*noindex\s*-->.+?<!--\s*/noindex\s*-->,,isg;
        $contents =~ s,<!--\s*noindex\s*-->.+?<!--\s*index\s*-->,,isg;
        # this will remove all script from indexing content
        $contents =~ s,<script[^>]*>.+?</script>,,isg;
        # remap Windows charset to ISO-8859-2
        $contents =~ tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2
        # this will fix badly formatted html in form:
        # <head><title>some text</title
        # ></head>
        # which will confuse indexer (or libxml2?)
        $contents =~ s/[\n\r]^(>)/$1\n/msg;
        # remove comments between <html> and <head> texi2html inserts them
        # there and swish can't find document title then (libxml or swish bug?)
        while ($contents =~ s/(<html>.*)<!--.*?-->(.*<head>)/$1$2/msi) { };

        # remote TPJ left column
        if ($contents =~ s,<!-- BEGIN LEFT SIDE BAR CELL -->.+?<!-- END LEFT SIDE BAR CELL -->,,isg) {
                my $title;
                # extract title and add to title
                if ($contents =~ m,<!-- the article goes here -->\s*<h2[^>]*>(.+?)</h2>,si) {
                        $title = $1;
                } elsif ($contents =~ m,<h1[^>]*>(.+?)</h1>,is) {
                        $title = $1;
                } elsif ($contents =~ m,<h2[^>]*>(.+?)</h2>,is) {
                        $title = $1;
                } else {
                        $title = "no detail title";
                }
                $contents =~ s,(<title>)([^<]+)(</title>),$1$2: $title$3,gsi if ($title);

        }

        return $contents;
}

1;
1	dpavlin	46	sub filter {
2			my $contents = shift \|\| return;
3			# if you don't want content to be indexed, include it in
4			# <noindex> foobar </noindex> tags or surround it with comments
5			# <!-- noindex --> foobar <!-- /noindex -->
6			# <!-- noindex --> foobar <!-- index --> (also supported by swish)
7			$contents =~ s,<noindex>.+?</noindex>,,isg;
8			$contents =~ s,<!--\snoindex\s-->.+?<!--\s/noindex\s-->,,isg;
9			$contents =~ s,<!--\snoindex\s-->.+?<!--\sindex\s-->,,isg;
10			# this will remove all script from indexing content
11	dpavlin	51	$contents =~ s,<script[^>]*>.+?</script>,,isg;
12	dpavlin	46	# remap Windows charset to ISO-8859-2
13			$contents =~ tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2
14			# this will fix badly formatted html in form:
15			# <head><title>some text</title
16			# ></head>
17			# which will confuse indexer (or libxml2?)
18			$contents =~ s/[\n\r]^(>)/$1\n/msg;
19			# remove comments between <html> and <head> texi2html inserts them
20			# there and swish can't find document title then (libxml or swish bug?)
21			while ($contents =~ s/(<html>.)<!--.?-->(.*<head>)/$1$2/msi) { };
22
23			# remote TPJ left column
24			if ($contents =~ s,<!-- BEGIN LEFT SIDE BAR CELL -->.+?<!-- END LEFT SIDE BAR CELL -->,,isg) {
25			my $title;
26			# extract title and add to title
27			if ($contents =~ m,<!-- the article goes here -->\s<h2[^>]>(.+?)</h2>,si) {
28			$title = $1;
29			} elsif ($contents =~ m,<h1[^>]*>(.+?)</h1>,is) {
30			$title = $1;
31			} elsif ($contents =~ m,<h2[^>]*>(.+?)</h2>,is) {
32			$title = $1;
33			} else {
34			$title = "no detail title";
35			}
36			$contents =~ s,(<title>)([^<]+)(</title>),$1$2: $title$3,gsi if ($title);
37
38			}
39	dpavlin	51
40	dpavlin	46	return $contents;
41			}
42
43			1;