1 |
sub filter { |
sub filter { |
2 |
my $contents = shift || return; |
my $contents = shift || return; |
3 |
|
|
4 |
|
my $verbose = 0; |
5 |
|
|
6 |
# if you don't want content to be indexed, include it in |
# if you don't want content to be indexed, include it in |
7 |
# <noindex> foobar </noindex> tags or surround it with comments |
# <noindex> foobar </noindex> tags or surround it with comments |
8 |
# <!-- noindex --> foobar <!-- /noindex --> |
# <!-- noindex --> foobar <!-- /noindex --> |
48 |
|
|
49 |
if ($contents =~ m,<!--SafTocEntry="([^"]+)"-->,is) { |
if ($contents =~ m,<!--SafTocEntry="([^"]+)"-->,is) { |
50 |
$new_title = $1; |
$new_title = $1; |
51 |
} elsif ($contents =~ m,<h\d\sclass="docPartTitle"[^>]*>([^<]+)</h\d>,is) { |
} elsif ($contents =~ m,<(h\d)\sclass="docPartTitle"[^>]*>(.+?)<\1>,is) { |
52 |
$new_title = $1; |
$new_title = $2; |
53 |
} elsif ($contents =~ m,<h\d\sclass="docChapterTitle"[^>]*>([^<]+)</h\d>,is) { |
} elsif ($contents =~ m,<(h\d)\sclass="docChapterTitle"[^>]*>(.+?)<\1>,is) { |
54 |
$new_title = $1; |
$new_title = $2; |
55 |
} elsif ($contents =~ m,<h\d\sclass="docSection1Title"[^>]*>([^<]+)</h\d>,is) { |
} elsif ($contents =~ m,<(h\d)\sclass="docSection1Title"[^>]*>(.+?)<\1>,is) { |
56 |
$new_title = $1; |
$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 = $2; |
61 |
} else { |
} else { |
62 |
if ($contents =~ m,<title>([^<]+)</title>,is) { |
if ($contents =~ m,<title>([^<]+)</title>,is) { |
63 |
$new_title = $1; |
$new_title = $1; |
67 |
} |
} |
68 |
|
|
69 |
if ($new_title) { |
if ($new_title) { |
70 |
|
# nuke html in title |
71 |
|
$new_title =~ s/<br>\s+/: /gs; |
72 |
|
$new_title =~ s/<\/*[^>]+>//gs; |
73 |
|
|
74 |
# check if new title is same as collection name |
# check if new title is same as collection name |
75 |
my ($a,$b) = ($new_title,$collection); |
my ($a,$b) = ($new_title,$collection); |
76 |
$a =~ s/([^a-zA-Z])+/ /gs; |
$a =~ s/([^a-zA-Z])+/ /gs; |
87 |
|
|
88 |
$new_title =~ s/\s\s+/ /g; |
$new_title =~ s/\s\s+/ /g; |
89 |
|
|
90 |
if ($contents =~ s,<title>(.*)</title>,$new_title,is) { |
if ($contents =~ s,<title>(.*)</title>,<title>$new_title</title>,is) { |
91 |
print STDERR "replace title '$1' with '$new_title'\n"; |
print STDERR "replace title '$1' with '$new_title'\n" if ($verbose); |
92 |
|
} elsif ($contents =~ s,(<head>),$1<title>$new_title</title>,is) { |
93 |
|
print STDERR "adding title '$new_title' after <head>\n" if ($verbose); |
94 |
|
|
95 |
|
} elsif ($contents =~ s,(<html>),$1<head><title>$new_title</title></head>,is) { |
96 |
|
print STDERR "adding title '$new_title' after <html>\n" if ($verbose); |
97 |
|
|
98 |
|
} elsif ($contents =~ s,^,<title>$new_title</title>,) { |
99 |
|
print STDERR "adding new title '$new_title'\n" if ($verbose); |
100 |
} else { |
} else { |
101 |
print STDERR "adding new title '$new_title'\n"; |
print STDERR "WARNING: filter couldn't add new title '$new_title' anywhere!"; |
|
# 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>,; |
|
102 |
} |
} |
103 |
|
|
104 |
return $contents; |
return $contents; |