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);

        }

        # is second argument collection?
        my $collection = shift || return $contents;

        # construct new title (from various parts of DocBook if available)
        my $new_title;

        if ($contents =~ m,<!--SafTocEntry="([^"]+)"-->,is) {
                $new_title = $1;
        } elsif ($contents =~ m,<h\d\sclass="docPartTitle"[^>]*>([^<]+)</h\d>,is) {
                $new_title = $1;
        } elsif ($contents =~ m,<h\d\sclass="docChapterTitle"[^>]*>([^<]+)</h\d>,is) {
                $new_title = $1;
        } elsif ($contents =~ m,<h\d\sclass="docSection1Title"[^>]*>([^<]+)</h\d>,is) {
                $new_title = $1;
        } else {
                if ($contents =~ m,<title>([^<]+)</title>,is) {
                        $new_title = $1;
                } elsif ($contents =~ m,<h\d[^>]*>([^<]+)</h\d>,is) {
                        $new_title = $1;
                }
        }

        if ($new_title) {
                # check if new title is same as collection name
                my ($a,$b) = ($new_title,$collection);
                $a =~ s/([^a-zA-Z])+/ /gs;
                $b =~ s/([^a-zA-Z])+/ /gs;
                if ($a =~ m/$b/i) {
                        $new_title = $collection;
                } else {
                        $new_title = "$collection :: $new_title";
                }
        } else {
                # fall-back to collection title
                $new_title = $collection;
        }

        $new_title =~ s/\s\s+/ /g;

        if ($contents =~ s,<title>(.*)</title>,$new_title,is) {
                print STDERR "replace title '$1' with '$new_title'\n";
        } else {
                print STDERR "adding new title '$new_title'\n";
                # try to insert after <head>, <html> or at top
                $contents =~ s,(<head>),$1<title>$new_title</title>,is ||
                $contents =~ s,(<html>),$1<title>$new_title</title>,is ||
                $contents =~ s,^,<title>$new_title</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	56	# is second argument collection?
41			my $collection = shift \|\| return $contents;
42
43			# construct new title (from various parts of DocBook if available)
44			my $new_title;
45
46			if ($contents =~ m,<!--SafTocEntry="([^"]+)"-->,is) {
47			$new_title = $1;
48			} elsif ($contents =~ m,<h\d\sclass="docPartTitle"[^>]*>([^<]+)</h\d>,is) {
49			$new_title = $1;
50			} elsif ($contents =~ m,<h\d\sclass="docChapterTitle"[^>]*>([^<]+)</h\d>,is) {
51			$new_title = $1;
52			} elsif ($contents =~ m,<h\d\sclass="docSection1Title"[^>]*>([^<]+)</h\d>,is) {
53			$new_title = $1;
54			} else {
55			if ($contents =~ m,<title>([^<]+)</title>,is) {
56			$new_title = $1;
57			} elsif ($contents =~ m,<h\d[^>]*>([^<]+)</h\d>,is) {
58			$new_title = $1;
59			}
60			}
61
62			if ($new_title) {
63			# check if new title is same as collection name
64			my ($a,$b) = ($new_title,$collection);
65			$a =~ s/([^a-zA-Z])+/ /gs;
66			$b =~ s/([^a-zA-Z])+/ /gs;
67			if ($a =~ m/$b/i) {
68			$new_title = $collection;
69			} else {
70			$new_title = "$collection :: $new_title";
71			}
72			} else {
73			# fall-back to collection title
74			$new_title = $collection;
75			}
76
77			$new_title =~ s/\s\s+/ /g;
78
79			if ($contents =~ s,<title>(.*)</title>,$new_title,is) {
80			print STDERR "replace title '$1' with '$new_title'\n";
81			} else {
82			print STDERR "adding new title '$new_title'\n";
83			# try to insert after <head>, <html> or at top
84			$contents =~ s,(<head>),$1<title>$new_title</title>,is \|\|
85			$contents =~ s,(<html>),$1<title>$new_title</title>,is \|\|
86			$contents =~ s,^,<title>$new_title</title>,;
87			}
88
89	dpavlin	46	return $contents;
90			}
91
92			1;