Open-ILS/src/extras/marc2html

   1 #!/usr/bin/perl
   2
   3 use Error;
   4 use MARC::Batch;
   5 use MARC::File::XML;
   6 use XML::LibXSLT;
   7 use XML::LibXML;
   8 use Unicode::Normalize;
   9 use Getopt::Long;
  10
  11 my ($split,$enc,$marc,$out) = (100);
  12 GetOptions(
  13         'split=i' => \$split,
  14         'marc=s'  => \$marc,
  15         'encoding=s'  => \$enc,
  16         'out_dir=s'  => \$out,
  17 );
  18
  19 if ($enc) {
  20         MARC::Charset->ignore_errors(1);
  21         MARC::Charset->assume_encoding($enc);
  22 }
  23
  24 die "gimme some marc!\n" unless $marc;
  25 die "gimme somewhere to put it!\n" unless $out;
  26
  27 my $xsl = join('',(<DATA>));
  28
  29 my $parser = XML::LibXML->new();
  30 my $xslt = XML::LibXSLT->new();
  31
  32 $stylesheet = $xslt->parse_stylesheet( $parser->parse_string($xsl) );
  33
  34
  35 my $xml = '';
  36 my $current = 1;
  37 my $prev = 0;
  38 my $next = 2;
  39
  40 my $marc = MARC::Batch->new( USMARC => $marc );
  41 $marc->strict_off;
  42 $marc->warnings_off;
  43
  44 while (my $r = $marc->next) {
  45         $xml .= entityize(MARC::File::XML::record($r));
  46
  47         unless ($current % $split) {
  48                 $xml = <<"              XML";
  49                         <collection xmlns="http://www.loc.gov/MARC21/slim">
  50                                 $xml
  51                         </collection>
  52                 XML
  53
  54                 my $doc = $parser->parse_string($xml);
  55                 $xml = '';
  56
  57                 my $results = $stylesheet->transform($doc, prev => "'$prev'", next => "'$next'");
  58                 $prev++;
  59                 $next++;
  60
  61                 open OUT, ">$out/$prev.html";
  62                 print OUT $results->toString;
  63                 close OUT;
  64         }
  65         $current++;
  66 }
  67
  68 my $doc = $parser->parse_string(<<XML);
  69 <marc:collection xmlns:marc="http://www.loc.gov/MARC21/slim">
  70         $xml
  71 </marc:collection>
  72 XML
  73
  74 my $results = $stylesheet->transform($doc, prev => "'$prev'", next => "'0'");
  75 $prev++;
  76
  77 $stylesheet->output_file($results, "$out/$prev.html");
  78
  79
  80 sub entityize {
  81         my $stuff = shift;
  82         my $form = shift;
  83
  84         if ($form eq 'D') {
  85                 $stuff = NFD($stuff);
  86         } else {
  87                 $stuff = NFC($stuff);
  88         }
  89
  90         $stuff =~ s/([\x{0080}-\x{fffd}])/sprintf('&#x%X;',ord($1))/sgoe;
  91         $stuff =~ s/([\x00-\x19])//sgoe;
  92         return $stuff;
  93 }
  94
  95
  96
  97 __DATA__
  98 <?xml version="1.0" encoding="UTF-8"?>
  99 <xsl:stylesheet version="1.0" xmlns:marc="http://www.loc.gov/MARC21/slim" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
 100         <xsl:output method="html"/>
 101
 102         <xsl:template match="/">
 103                 <html>
 104                         <head>
 105
 106                                 <style>
 107
 108                                         .marc_table {}
 109                                         .marc_tag_row {}
 110                                         .marc_tag_data {}
 111                                         .marc_tag_col {}
 112                                         .marc_tag_ind {}
 113                                         .marc_subfields {}
 114                                         .marc_subfield_code {
 115                                                 color: blue;
 116                                                 padding-left: 5px;
 117                                                 padding-right: 5px;
 118                                         }
 119
 120                                 </style>
 121
 122                                 <link href='/css/opac_marc.css' rel='stylesheet' type='text/css'></link>
 123                         </head>
 124                         <body>
 125                                 <xsl:if test="$prev &gt; 0">
 126                                         <a>
 127                                                 <xsl:attribute name="href">
 128                                                         <xsl:value-of select="concat( $prev, '.html')"/>
 129                                                 </xsl:attribute>
 130                                                 <xsl:text>Previous page</xsl:text>
 131                                         </a>
 132                                 </xsl:if>
 133                                 <span> | </span>
 134                                 <xsl:if test="$next &gt; 0">
 135                                         <a>
 136                                                 <xsl:attribute name="href">
 137                                                         <xsl:value-of select="concat( $next, '.html')"/>
 138                                                 </xsl:attribute>
 139                                                 <xsl:text>Next page</xsl:text>
 140                                         </a>
 141                                 </xsl:if>
 142                                 <hr/>
 143                                 <xsl:apply-templates select="//marc:record"/>
 144                                 <xsl:if test="$prev &gt; 0">
 145                                         <a>
 146                                                 <xsl:attribute name="href">
 147                                                         <xsl:value-of select="concat( $prev, '.html')"/>
 148                                                 </xsl:attribute>
 149                                                 <xsl:text>Previous page</xsl:text>
 150                                         </a>
 151                                 </xsl:if>
 152                                 <span> | </span>
 153                                 <xsl:if test="$next &gt; 0">
 154                                         <a>
 155                                                 <xsl:attribute name="href">
 156                                                         <xsl:value-of select="concat( $next, '.html')"/>
 157                                                 </xsl:attribute>
 158                                                 <xsl:text>Next page</xsl:text>
 159                                         </a>
 160                                 </xsl:if>
 161                         </body>
 162                 </html>
 163         </xsl:template>
 164
 165         <xsl:template match="marc:record">
 166                 <table class='marc_table'>
 167                         <tr class='marc_tag_row'>
 168                                 <th class='marc_tag_col' NOWRAP="TRUE" ALIGN="RIGHT" VALIGN="middle">
 169                                         LDR
 170                                 </th>
 171                                 <td class='marc_tag_data' COLSPAN='3'>
 172                                         <xsl:value-of select="marc:leader"/>
 173                                 </td>
 174                         </tr>
 175                         <xsl:apply-templates select="marc:datafield|marc:controlfield"/>
 176                 </table>
 177                 <hr/>
 178         </xsl:template>
 179
 180         <xsl:template match="marc:controlfield">
 181                 <tr class='marc_tag_row'>
 182                         <th class='marc_tag_col' NOWRAP="TRUE" ALIGN="RIGHT" VALIGN="middle">
 183                                 <xsl:value-of select="@tag"/>
 184                         </th>
 185                         <td class='marc_tag_data' COLSPAN='3'>
 186                                 <xsl:value-of select="."/>
 187                         </td>
 188                 </tr>
 189         </xsl:template>
 190
 191         <xsl:template match="marc:datafield">
 192                 <tr class='marc_tag_row'>
 193                         <th class='marc_tag_col' NOWRAP="TRUE" ALIGN="RIGHT" VALIGN="middle">
 194                                 <xsl:value-of select="@tag"/>
 195                         </th>
 196                         <td class='marc_tag_ind'>
 197                                 <xsl:value-of select="@ind1"/>
 198                         </td>
 199
 200                         <td class='marc_tag_ind' style='border-left: 1px solid #A0A0A0; padding-left: 3px;'>
 201                                 <xsl:value-of select="@ind2"/>
 202                                 <span style='color:#FFF'>.</span>
 203                         </td>
 204
 205                         <td class='marc_subfields'>
 206                                 <xsl:apply-templates select="marc:subfield"/>
 207                         </td>
 208                 </tr>
 209         </xsl:template>
 210
 211         <xsl:template match="marc:subfield">
 212                 <span class='marc_subfield_code' >
 213                         &#8225;<xsl:value-of select="@code"/>
 214                 </span><xsl:value-of select="."/>
 215         </xsl:template>
 216
 217 </xsl:stylesheet>
 218