22 |
# remove comments between <html> and <head> texi2html inserts them |
# remove comments between <html> and <head> texi2html inserts them |
23 |
# there and swish can't find document title then (libxml or swish bug?) |
# there and swish can't find document title then (libxml or swish bug?) |
24 |
while ($contents =~ s/(<html>.*)<!--.*?-->(.*<head>)/$1$2/msi) { }; |
while ($contents =~ s/(<html>.*)<!--.*?-->(.*<head>)/$1$2/msi) { }; |
25 |
|
|
26 |
|
# remove empty lines before/after <html> |
27 |
|
$contents =~ s/^\s+(<html>)/$1/is; |
28 |
|
$contents =~ s/(<\/html>)\s+$/$1/is; |
29 |
|
# remove cr |
30 |
|
$contents =~ s/\r//gs; |
31 |
|
|
32 |
|
# remove SQL Magazine header and footer |
33 |
|
$contents =~ s/<!-- begin topnav area -->.+?<!-- end topnav area -->/<\/table>/is; |
34 |
|
$contents =~ s/<!--Begin Footer-->.+?<\/table>/<\/table>/is; |
35 |
|
|
36 |
# remote TPJ left column |
# remote TPJ left column |
37 |
if ($contents =~ s,<!-- BEGIN LEFT SIDE BAR CELL -->.+?<!-- END LEFT SIDE BAR CELL -->,,isg) { |
if ($contents =~ s,<!-- BEGIN LEFT SIDE BAR CELL -->.+?<!-- END LEFT SIDE BAR CELL -->,,isg) { |
58 |
|
|
59 |
if ($contents =~ m,<!--SafTocEntry="([^"]+)"-->,is) { |
if ($contents =~ m,<!--SafTocEntry="([^"]+)"-->,is) { |
60 |
$new_title = $1; |
$new_title = $1; |
61 |
} elsif ($contents =~ m,<(h\d)\sclass="docPartTitle"[^>]*>(.+?)<\1>,is) { |
print STDERR "using title '$new_title' from <!--SafTocEntry-->\n" if ($verbose); |
62 |
|
} elsif ($contents =~ m,<(h\d)\s+class="docPartTitle"[^>]*>(.+?)</\1>,is) { |
63 |
|
$new_title = $2; |
64 |
|
print STDERR "using title '$new_title' from docPartTitle\n" if ($verbose); |
65 |
|
} elsif ($contents =~ m,<(h\d)\s+class="docChapterTitle"[^>]*>(.+?)</\1>,is) { |
66 |
$new_title = $2; |
$new_title = $2; |
67 |
} elsif ($contents =~ m,<(h\d)\sclass="docChapterTitle"[^>]*>(.+?)<\1>,is) { |
print STDERR "using title '$new_title' from docChapterTitle\n" if ($verbose); |
68 |
|
} elsif ($contents =~ m,<(h\d)\s+class="docSection1Title"[^>]*>(.+?)</\1>,is) { |
69 |
$new_title = $2; |
$new_title = $2; |
70 |
} elsif ($contents =~ m,<(h\d)\sclass="docSection1Title"[^>]*>(.+?)<\1>,is) { |
print STDERR "using title '$new_title' from docSection1Title\n" if ($verbose); |
71 |
|
} elsif ($contents =~ m,<(h\d)\s+class="doc[^"]*Title"[^>]*>(.+?)</\1>,is) { |
72 |
$new_title = $2; |
$new_title = $2; |
73 |
} elsif ($contents =~ m,<(h\d)\sclass="chapter"[^>]*>(.+?)<\1>,is) { |
print STDERR "using title '$new_title' from doc.+Title\n" if ($verbose); |
74 |
|
} elsif ($contents =~ m,<(h\d)\s+class="chapter"[^>]*>(.+?)</\1>,is) { |
75 |
$new_title = $2; |
$new_title = $2; |
76 |
} elsif ($contents =~ m,<(h\d)\sclass="sect1"[^>]*>(.+?)<\1>,is) { |
print STDERR "using title '$new_title' from chapter\n" if ($verbose); |
77 |
|
} elsif ($contents =~ m,<(h\d)\s+class="sect1"[^>]*>(.+?)</\1>,is) { |
78 |
$new_title = $2; |
$new_title = $2; |
79 |
|
print STDERR "using title '$new_title' from sect1\n" if ($verbose); |
80 |
} else { |
} else { |
81 |
if ($contents =~ m,<title>([^<]+)</title>,is) { |
if ($contents =~ m,<title>([^<]+)</title>,is) { |
82 |
$new_title = $1; |
$new_title = $1; |
83 |
|
print STDERR "using title '$new_title' from <title>\n" if ($verbose); |
84 |
} elsif ($contents =~ m,<h\d[^>]*>([^<]+)</h\d>,is) { |
} elsif ($contents =~ m,<h\d[^>]*>([^<]+)</h\d>,is) { |
85 |
$new_title = $1; |
$new_title = $1; |
86 |
|
print STDERR "using title '$new_title' from <h_>\n" if ($verbose); |
87 |
} |
} |
88 |
} |
} |
89 |
|
|
98 |
$b =~ s/([^a-zA-Z])+/ /gs; |
$b =~ s/([^a-zA-Z])+/ /gs; |
99 |
if ($a =~ m/$b/i) { |
if ($a =~ m/$b/i) { |
100 |
$new_title = $collection; |
$new_title = $collection; |
101 |
|
print STDERR "new_title and collection are same! [$new_title]\n" if ($verbose); |
102 |
} else { |
} else { |
103 |
$new_title = "$collection :: $new_title"; |
$new_title = "$collection :: $new_title"; |
104 |
} |
} |