trunk/spider/filter.pm

sub filter {
        my $contents = shift || return;

        my $verbose = 0;
        
        # 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"[^>]*>(.+?)<\1>,is) {
                $new_title = $2;
        } elsif ($contents =~ m,<(h\d)\sclass="docChapterTitle"[^>]*>(.+?)<\1>,is) {
                $new_title = $2;
        } elsif ($contents =~ m,<(h\d)\sclass="docSection1Title"[^>]*>(.+?)<\1>,is) {
                $new_title = $2;
        } elsif ($contents =~ m,<(h\d)\sclass="chapter"[^>]*>(.+?)<\1>,is) {
                $new_title = $2;
        } elsif ($contents =~ m,<(h\d)\sclass="sect1"[^>]*>(.+?)<\1>,is) {
                $new_title = $2;
        } else {
                if ($contents =~ m,<title>([^<]+)</title>,is) {
                        $new_title = $1;
                } elsif ($contents =~ m,<h\d[^>]*>([^<]+)</h\d>,is) {
                        $new_title = $1;
                }
        }

        if ($new_title) {
                # nuke html in title
                $new_title =~ s/<br>\s+/: /gs;
                $new_title =~ s/<\/*[^>]+>//gs;
        
                # 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 ($contents =~ s,<title>(.*)</title>,<title>$new_title</title>,is) {
                print STDERR "replace title '$1' with '$new_title'\n" if ($verbose);
        } elsif (
                        # 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>, ) {
                print STDERR "adding new title '$new_title'\n" if ($verbose);
        }

        return $contents;
}

1;
1	sub filter {
2	my $contents = shift \|\| return;
3
4	my $verbose = 0;
5
6	# if you don't want content to be indexed, include it in
7	# <noindex> foobar </noindex> tags or surround it with comments
8	# <!-- noindex --> foobar <!-- /noindex -->
9	# <!-- noindex --> foobar <!-- index --> (also supported by swish)
10	$contents =~ s,<noindex>.+?</noindex>,,isg;
11	$contents =~ s,<!--\snoindex\s-->.+?<!--\s/noindex\s-->,,isg;
12	$contents =~ s,<!--\snoindex\s-->.+?<!--\sindex\s-->,,isg;
13	# this will remove all script from indexing content
14	$contents =~ s,<script[^>]*>.+?</script>,,isg;
15	# remap Windows charset to ISO-8859-2
16	$contents =~ tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2
17	# this will fix badly formatted html in form:
18	# <head><title>some text</title
19	# ></head>
20	# which will confuse indexer (or libxml2?)
21	$contents =~ s/[\n\r]^(>)/$1\n/msg;
22	# remove comments between <html> and <head> texi2html inserts them
23	# there and swish can't find document title then (libxml or swish bug?)
24	while ($contents =~ s/(<html>.)<!--.?-->(.*<head>)/$1$2/msi) { };
25
26	# remote TPJ left column
27	if ($contents =~ s,<!-- BEGIN LEFT SIDE BAR CELL -->.+?<!-- END LEFT SIDE BAR CELL -->,,isg) {
28	my $title;
29	# extract title and add to title
30	if ($contents =~ m,<!-- the article goes here -->\s<h2[^>]>(.+?)</h2>,si) {
31	$title = $1;
32	} elsif ($contents =~ m,<h1[^>]*>(.+?)</h1>,is) {
33	$title = $1;
34	} elsif ($contents =~ m,<h2[^>]*>(.+?)</h2>,is) {
35	$title = $1;
36	} else {
37	$title = "no detail title";
38	}
39	$contents =~ s,(<title>)([^<]+)(</title>),$1$2: $title$3,gsi if ($title);
40
41	}
42
43	# is second argument collection?
44	my $collection = shift \|\| return $contents;
45
46	# construct new title (from various parts of DocBook if available)
47	my $new_title;
48
49	if ($contents =~ m,<!--SafTocEntry="([^"]+)"-->,is) {
50	$new_title = $1;
51	} elsif ($contents =~ m,<(h\d)\sclass="docPartTitle"[^>]*>(.+?)<\1>,is) {
52	$new_title = $2;
53	} elsif ($contents =~ m,<(h\d)\sclass="docChapterTitle"[^>]*>(.+?)<\1>,is) {
54	$new_title = $2;
55	} elsif ($contents =~ m,<(h\d)\sclass="docSection1Title"[^>]*>(.+?)<\1>,is) {
56	$new_title = $2;
57	} elsif ($contents =~ m,<(h\d)\sclass="chapter"[^>]*>(.+?)<\1>,is) {
58	$new_title = $2;
59	} elsif ($contents =~ m,<(h\d)\sclass="sect1"[^>]*>(.+?)<\1>,is) {
60	$new_title = $2;
61	} else {
62	if ($contents =~ m,<title>([^<]+)</title>,is) {
63	$new_title = $1;
64	} elsif ($contents =~ m,<h\d[^>]*>([^<]+)</h\d>,is) {
65	$new_title = $1;
66	}
67	}
68
69	if ($new_title) {
70	# nuke html in title
71	$new_title =~ s/<br>\s+/: /gs;
72	$new_title =~ s/<\/*[^>]+>//gs;
73
74	# check if new title is same as collection name
75	my ($a,$b) = ($new_title,$collection);
76	$a =~ s/([^a-zA-Z])+/ /gs;
77	$b =~ s/([^a-zA-Z])+/ /gs;
78	if ($a =~ m/$b/i) {
79	$new_title = $collection;
80	} else {
81	$new_title = "$collection :: $new_title";
82	}
83	} else {
84	# fall-back to collection title
85	$new_title = $collection;
86	}
87
88	$new_title =~ s/\s\s+/ /g;
89
90	if ($contents =~ s,<title>(.*)</title>,<title>$new_title</title>,is) {
91	print STDERR "replace title '$1' with '$new_title'\n" if ($verbose);
92	} elsif (
93	# try to insert after <head>, <html> or at top
94	$contents =~ s,(<head>),$1<title>$new_title</title>,is \|\|
95	$contents =~ s,(<html>),$1<title>$new_title</title>,is \|\|
96	$contents =~ s,^,<title>$new_title</title>, ) {
97	print STDERR "adding new title '$new_title'\n" if ($verbose);
98	}
99
100	return $contents;
101	}
102
103	1;