| 1 | #!/usr/bin/perl -w
|
|---|
| 2 |
|
|---|
| 3 |
|
|---|
| 4 | BEGIN {
|
|---|
| 5 | die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
|
|---|
| 6 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
|
|---|
| 7 | }
|
|---|
| 8 |
|
|---|
| 9 |
|
|---|
| 10 | use File::Basename;
|
|---|
| 11 | use unicode;
|
|---|
| 12 |
|
|---|
| 13 | my %classifications = (
|
|---|
| 14 | 'Pharmacokinetics' => ['Absorption', 'Distribution', 'Metabolism', 'Elimination',
|
|---|
| 15 | 'Effect of Food', 'Drug Interactions', 'Special Populations'],
|
|---|
| 16 | 'Pharmacodynamics' => ['Trace Metal', 'Excretion', 'Selectivity', 'Copper', 'Zinc',
|
|---|
| 17 | 'Manganese', 'Iron', 'Distribution', 'Tissue', 'Brain',
|
|---|
| 18 | 'Heart', 'Plasma', 'Fetal', 'Liver', 'Kidney'],
|
|---|
| 19 | 'Safety/Side Effects/Toxicity' => ['Acute Toxicity', 'Repeated Dose Toxicity',
|
|---|
| 20 | 'Studies in Rodents', 'Studies in Dogs',
|
|---|
| 21 | 'Overdosage', 'Reproductive Toxicity',
|
|---|
| 22 | 'Teratogenicity', 'Mutagenecity', 'Carcinogenicity'],
|
|---|
| 23 | 'Other' => ['Superoxide Dismutase', 'Ferroxidase I', 'Ferroxidase II',
|
|---|
| 24 | 'Metallothienein', 'Ceruloplasmin'],
|
|---|
| 25 | 'Efficacy' => ['Heart', 'Complications', 'Microvascular', 'Renal', 'Diabetes'],
|
|---|
| 26 | 'Contraindications' => [],
|
|---|
| 27 | 'Precautions' => []
|
|---|
| 28 | );
|
|---|
| 29 |
|
|---|
| 30 | my @meta_files = ();
|
|---|
| 31 |
|
|---|
| 32 | &recursive_rename("import");
|
|---|
| 33 | # process meta files
|
|---|
| 34 | foreach my $mfile (@meta_files) {
|
|---|
| 35 | &process_meta_file($mfile);
|
|---|
| 36 | }
|
|---|
| 37 |
|
|---|
| 38 | sub recursive_rename {
|
|---|
| 39 | my ($dir) = @_;
|
|---|
| 40 |
|
|---|
| 41 | opendir (DIR, "$dir") || die;
|
|---|
| 42 | my @files = readdir DIR;
|
|---|
| 43 | closedir DIR;
|
|---|
| 44 |
|
|---|
| 45 | foreach $file (@files) {
|
|---|
| 46 | next if $file eq "." || $file eq "..";
|
|---|
| 47 | next if $file =~ /processed.htm$/;
|
|---|
| 48 |
|
|---|
| 49 | my $path = "$dir/$file";
|
|---|
| 50 | if (-d $path) {
|
|---|
| 51 | &recursive_rename($path);
|
|---|
| 52 | } else {
|
|---|
| 53 | $newfile = &rename_file($file);
|
|---|
| 54 | if ($newfile ne $file) {
|
|---|
| 55 | print STDERR "renaming $file --> $newfile\n";
|
|---|
| 56 | `mv "$path" "$dir/$newfile"`;
|
|---|
| 57 | }
|
|---|
| 58 |
|
|---|
| 59 | if ($file =~ /^title\.txt$/) {
|
|---|
| 60 | # we'll process all the title.txt files after we've finished
|
|---|
| 61 | # renaming everything
|
|---|
| 62 | push(@meta_files, "$dir/$newfile");
|
|---|
| 63 | }
|
|---|
| 64 |
|
|---|
| 65 | if ($file =~ /\.html?$/i) {
|
|---|
| 66 | &process_html_file("$dir/$newfile");
|
|---|
| 67 | }
|
|---|
| 68 | }
|
|---|
| 69 | }
|
|---|
| 70 | }
|
|---|
| 71 |
|
|---|
| 72 | # clean up the html (currently just use the non-css version)
|
|---|
| 73 | sub process_html_file {
|
|---|
| 74 | my ($filename) = @_;
|
|---|
| 75 |
|
|---|
| 76 | print STDERR "processing $filename\n";
|
|---|
| 77 |
|
|---|
| 78 | open (FILE, $filename) || die;
|
|---|
| 79 | undef $/;
|
|---|
| 80 | my $file = <FILE>;
|
|---|
| 81 | $/ = "\n";
|
|---|
| 82 | close FILE;
|
|---|
| 83 |
|
|---|
| 84 | my $header = "<html>\n<head></head>\n<body bgcolor=\"#FFFFFF\">\n";
|
|---|
| 85 | my $footer = "</body>\n</html>\n";
|
|---|
| 86 |
|
|---|
| 87 | my ($noncss, $css) = $file =~ /document\.write\(\"(.*?[^\\])\"\).*?document\.write\(\"(.*?[^\\])\"\)/si;
|
|---|
| 88 |
|
|---|
| 89 | # remove backslashes added for javascript strings
|
|---|
| 90 | $noncss =~ s/\\\"/\"/sg;
|
|---|
| 91 | $noncss =~ s/\\\n/\n/sg;
|
|---|
| 92 |
|
|---|
| 93 | # alter <img> tags to support renamed files
|
|---|
| 94 | $noncss =~ s/(<img src=\")([^\"]+)/$1 . &rename_file($2)/sige;
|
|---|
| 95 |
|
|---|
| 96 | # remove rules="all" attribute from table tags
|
|---|
| 97 | $noncss =~ s/(<table.*?)rules=\"all\"\s+/$1/sig;
|
|---|
| 98 |
|
|---|
| 99 | open (FILE, ">$filename") || die;
|
|---|
| 100 | print FILE $header . $noncss . $footer;
|
|---|
| 101 | close FILE;
|
|---|
| 102 |
|
|---|
| 103 | }
|
|---|
| 104 |
|
|---|
| 105 | sub rename_file {
|
|---|
| 106 | my ($filename) = @_;
|
|---|
| 107 |
|
|---|
| 108 | $filename =~ s/\s+//g;
|
|---|
| 109 | $filename =~ s/^protemix\(\d+\)-?//;
|
|---|
| 110 |
|
|---|
| 111 | return $filename;
|
|---|
| 112 | }
|
|---|
| 113 |
|
|---|
| 114 | # process a title.txt file and replace it with a meta.xml file
|
|---|
| 115 | sub process_meta_file {
|
|---|
| 116 | my ($filename) = @_;
|
|---|
| 117 |
|
|---|
| 118 | open (FILE, $filename) || die ("couldn't open $filename");
|
|---|
| 119 | undef $/;
|
|---|
| 120 | my $title = <FILE>;
|
|---|
| 121 | $/ = "\n";
|
|---|
| 122 | close FILE;
|
|---|
| 123 |
|
|---|
| 124 | unlink($filename);
|
|---|
| 125 |
|
|---|
| 126 | $title =~ s/\s+/ /gs;
|
|---|
| 127 | $title =~ s/^\s+//;
|
|---|
| 128 | $title =~ s/\s+$//;
|
|---|
| 129 | $title = &unicode::ascii2utf8(\$title); # assumes title is iso-8859-1
|
|---|
| 130 | my $metafile = "<Metafile>\n";
|
|---|
| 131 | $metafile .= " <Metadata name=\"Title\">$title</Metadata>\n";
|
|---|
| 132 |
|
|---|
| 133 | # currently just write some random values for classification metadata
|
|---|
| 134 | my @c1 = ('Animal', 'Human', 'Other');
|
|---|
| 135 | my $rand = int(rand 3);
|
|---|
| 136 | my $classtext = " <Metadata name=\"Class1\">" . $c1[$rand] . "</Metadata>\n";
|
|---|
| 137 | my @c2 = ('Pharmacokinetics', 'Pharmacodynamics', 'Safety/Side Effects/Toxicity',
|
|---|
| 138 | 'Other', 'Efficacy', 'Contraindications', 'Precautions');
|
|---|
| 139 | $rand = int(rand 7);
|
|---|
| 140 | my $c2val = $c2[$rand];
|
|---|
| 141 | $classtext .= " <Metadata name=\"Class2\">$c2val</Metadata>\n";
|
|---|
| 142 | if (defined $classifications{$c2val}) {
|
|---|
| 143 | if (scalar(@{$classifications{$c2val}})) {
|
|---|
| 144 | my $numvals = scalar(@{$classifications{$c2val}});
|
|---|
| 145 | $rand = int(rand $numvals);
|
|---|
| 146 | my $c3val = $classifications{$c2val}->[$rand];
|
|---|
| 147 | $classtext .= " <Metadata name=\"Class3\">$c3val</Metadata>\n";
|
|---|
| 148 | }
|
|---|
| 149 | } else {
|
|---|
| 150 | print STDERR "ERROR: '$c2val' not in clasifications list\n";
|
|---|
| 151 | }
|
|---|
| 152 |
|
|---|
| 153 | my $dir = File::Basename::dirname($filename);
|
|---|
| 154 |
|
|---|
| 155 | opendir(DIR, $dir) || die;
|
|---|
| 156 | my @files = readdir DIR;
|
|---|
| 157 | foreach my $file (@files) {
|
|---|
| 158 | if ($file =~ s/\.html?$//i) {
|
|---|
| 159 | $metafile .= " <Page filename=\"$file\">\n$classtext </Page>\n";
|
|---|
| 160 | }
|
|---|
| 161 | }
|
|---|
| 162 | $metafile .= "</Metafile>\n";
|
|---|
| 163 |
|
|---|
| 164 | $filename = "$dir/meta.xml";
|
|---|
| 165 | open (FILE, ">$filename") || die;
|
|---|
| 166 | print FILE $metafile;
|
|---|
| 167 | close FILE;
|
|---|
| 168 | }
|
|---|
| 169 |
|
|---|
| 170 |
|
|---|
