[29673] | 1 | package XML::Tidy;
|
---|
| 2 |
|
---|
| 3 | use strict;
|
---|
| 4 | use warnings;
|
---|
| 5 | use utf8;
|
---|
| 6 | use XML::Parser;
|
---|
| 7 | use base 'Exporter';
|
---|
| 8 |
|
---|
| 9 | our $VERSION = 1.00;
|
---|
| 10 | our @EXPORT = ( 'read_xml', 'write_xml' );
|
---|
| 11 |
|
---|
| 12 | # Reads the servlets partial XML file using XML::Parser,
|
---|
| 13 | # Then converts the obtained data structure to a more readable one
|
---|
| 14 | # Example conversion:
|
---|
| 15 | =xml
|
---|
| 16 | <servlet>
|
---|
| 17 | <servlet-name>library</servlet-name>
|
---|
| 18 | <description attribute="example">The standard gsdl3 library program</description>
|
---|
| 19 | <servlet-class>org.greenstone.gsdl3.LibraryServlet</servlet-class>
|
---|
| 20 | <init-param>
|
---|
| 21 | <param-name>library_name</param-name>
|
---|
| 22 | <param-value>library</param-value>
|
---|
| 23 | </init-param>
|
---|
| 24 | <init-param>
|
---|
| 25 | <param-name>site_name</param-name>
|
---|
| 26 | <param-value>localsite</param-value>
|
---|
| 27 | </init-param>
|
---|
| 28 | </servlet>
|
---|
| 29 | =cut
|
---|
| 30 | =data_structure
|
---|
| 31 | {
|
---|
| 32 | 'servlet' => {
|
---|
| 33 | 'servlet-class' => 'org.greenstone.gsdl3.LibraryServlet',
|
---|
| 34 | 'servlet-name' => 'library',
|
---|
| 35 | 'description' => {
|
---|
| 36 | '.value' => 'The standard gsdl3 library program',
|
---|
| 37 | '.attr' => {
|
---|
| 38 | 'attribute' => 'example'
|
---|
| 39 | }
|
---|
| 40 | },
|
---|
| 41 | 'init-param' => [
|
---|
| 42 | {
|
---|
| 43 | 'param-value' => 'library',
|
---|
| 44 | 'param-name' => 'library_name'
|
---|
| 45 | },
|
---|
| 46 | {
|
---|
| 47 | 'param-name' => 'site_name',
|
---|
| 48 | 'param-value' => 'localsite'
|
---|
| 49 | }
|
---|
| 50 | ]
|
---|
| 51 | }
|
---|
| 52 | }
|
---|
| 53 | =cut
|
---|
| 54 | sub read_xml {
|
---|
| 55 | my $FH = shift;
|
---|
| 56 | local $/ = undef;
|
---|
| 57 | my $xml = <$FH>;
|
---|
| 58 | # The xml needs a root element, so we wrap it in one
|
---|
| 59 | $xml='<__root__>' . $xml . '</__root__>';
|
---|
| 60 | # Parse the data using XML::Parser
|
---|
| 61 | my $data = new XML::Parser (Style => 'Tree')->parse ($xml);
|
---|
| 62 |
|
---|
| 63 | sub tidy {
|
---|
| 64 | my %hash;
|
---|
| 65 | # If the element has attributes, add them to the hash
|
---|
| 66 | my $attr = shift;
|
---|
| 67 | if (scalar keys %$attr > 0) {
|
---|
| 68 | $hash{'.attr'} = $attr;
|
---|
| 69 | }
|
---|
| 70 | # Read any child elements or text value
|
---|
| 71 | while (@_) {
|
---|
| 72 | my $element = shift;
|
---|
| 73 | my $value = shift;
|
---|
| 74 | if ($element eq 0) {
|
---|
| 75 | $value =~ /^\s*$/ || ($hash{'.value'} .= $value);
|
---|
| 76 | } else {
|
---|
| 77 | # If there is more than one of a single tupe of child element,
|
---|
| 78 | # That child element must become an array
|
---|
| 79 | if (defined $hash{$element}) {
|
---|
| 80 | unless (ref $hash{$element} eq 'ARRAY') {
|
---|
| 81 | $hash{$element} = [ $hash{$element} ];
|
---|
| 82 | }
|
---|
| 83 | push @{$hash{$element}}, tidy (@{$value});
|
---|
| 84 | } else {
|
---|
| 85 | $hash{$element} = tidy (@{$value});
|
---|
| 86 | }
|
---|
| 87 | }
|
---|
| 88 | }
|
---|
| 89 | # If the element only has a value, it can become a scalar
|
---|
| 90 | if (scalar keys %hash == 1 and defined $hash{'.value'}) {
|
---|
| 91 | return $hash{'.value'};
|
---|
| 92 | }
|
---|
| 93 | return \%hash;
|
---|
| 94 | }
|
---|
| 95 | return tidy (@{@$data[1]});
|
---|
| 96 | }
|
---|
| 97 |
|
---|
| 98 | # Writes a data structure to an XML file
|
---|
| 99 | sub write_xml {
|
---|
| 100 | my ($hash, $FH) = @_;
|
---|
| 101 | select $FH;
|
---|
| 102 | sub open_tag {
|
---|
| 103 | my ($indent, $tag, $attr) = @_;
|
---|
| 104 | print $indent, "<", $tag;
|
---|
| 105 | if (defined $attr) {
|
---|
| 106 | for my $key (sort keys %$attr) {
|
---|
| 107 | print " ", $key, '="', $attr->{$key}, '"';
|
---|
| 108 | }
|
---|
| 109 | }
|
---|
| 110 | print ">";
|
---|
| 111 | }
|
---|
| 112 | sub write_element {
|
---|
| 113 | my ($indent, $hash) = @_;
|
---|
| 114 | for my $key (sort keys %$hash) {
|
---|
| 115 | $key eq '.attr' && next;
|
---|
| 116 | my $val = $hash->{$key};
|
---|
| 117 | $key eq '.value' && do {
|
---|
| 118 | print $val, "\n";
|
---|
| 119 | next;
|
---|
| 120 | };
|
---|
| 121 | for (ref $val) {
|
---|
| 122 | /^ARRAY$/ && do {
|
---|
| 123 | for my $element (@$val) {
|
---|
| 124 | open_tag ($indent, $key, $element->{'.attr'});
|
---|
| 125 | for (ref $element) {
|
---|
| 126 | /^HASH$/ && do {
|
---|
| 127 | print "\n";
|
---|
| 128 | write_element (" $indent", $element);
|
---|
| 129 | print $indent, "</", $key, ">\n";
|
---|
| 130 | last;
|
---|
| 131 | };
|
---|
| 132 | print $element, "</", $key, ">\n";
|
---|
| 133 | }
|
---|
| 134 | }
|
---|
| 135 | last;
|
---|
| 136 | };
|
---|
| 137 | /^HASH$/ && do {
|
---|
| 138 | open_tag ($indent, $key, $val->{'.attr'});
|
---|
| 139 | # If the element only has a value with attributes,
|
---|
| 140 | # it can be formatted on one line
|
---|
| 141 | if (scalar keys %$val == 2 and defined $val->{'.value'}) {
|
---|
| 142 | print $val->{'.value'}, "</", $key, ">\n";
|
---|
| 143 | } else {
|
---|
| 144 | print "\n";
|
---|
| 145 | write_element (" $indent", $val);
|
---|
| 146 | print $indent, "</", $key, ">\n";
|
---|
| 147 | }
|
---|
| 148 | last;
|
---|
| 149 | };
|
---|
| 150 | open_tag ($indent, $key);
|
---|
| 151 | print $val;
|
---|
| 152 | print "</", $key, ">\n";
|
---|
| 153 | }
|
---|
| 154 | }
|
---|
| 155 | }
|
---|
| 156 | write_element ("", $hash);
|
---|
| 157 | select STDOUT;
|
---|
| 158 | }
|
---|