Convert Harvard Library Bibliographic Dataset (MARC21) to MODS XML

marc-to-mods.php

<?php
require 'File/MARC.php';

$xsl = new DOMDocument;
$xsl->load('http://www.loc.gov/standards/mods/v3/MARC21slim2MODS3-4.xsl');

$xsltproc = new XSLTProcessor;
$xsltproc->importStylesheet($xsl);

$marcxml = new DOMDocument;
$marcxml->preserveWhiteSpace = false;

$i = 0;

foreach (glob('/marc/*.mrc') as $file) {
	$items = new File_MARC($file);

	while ($record = $items->next()) {
                $marcxml->loadXML($record->toXML(), LIBXML_NOCDATA);
                $mods = $xsltproc->transformToDoc($marcxml);

                $xpath = new DOMXPath($mods);
                $xpath->registerNamespace('mods', 'http://www.loc.gov/mods/v3');

                $nodes = $xpath->query('mods:mods/mods:recordInfo/mods:recordIdentifier');
                if (!$nodes->length) continue; // no identifier

                $id = $nodes->item(0)->textContent;

                $file = sprintf('/mods/%s/%s/%s.xml', substr($id, 0, 3), substr($id, 3, 3), $id);
                if (($i++ % 1000) === 0) print "$file\n";

                $dir = dirname($file);
                if (!file_exists($dir)) mkdir($dir, 0777, true);

                $mods->formatOutput = true;
                file_put_contents($file, $mods->saveXML($mods->documentElement->firstChild));
	}
}

mods-to-cloudsearch.php

<?php

$xsl = new DOMDocument;
$xsl->load('mods-to-cloudsearch.xsl');

$xsltproc = new XSLTProcessor;
$xsltproc->importStylesheet($xsl);

$mods = new DOMDocument;
$mods->preserveWhiteSpace = false;

$i = 0;

$files = glob('/mods/000/000/*.xml');

$output = fopen('/mods/cloud-000-000.xml', 'w');
fwrite($output, "<batch>\n");

foreach ($files as $file) {
    $mods->load($file, LIBXML_NOCDATA);
    fwrite($output, $xsltproc->transformToXML($mods));
}

fwrite($output, "</batch>\n");

mods-to-cloudsearch.xsl

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
  <xsl:output nethod="xml" encoding="utf-8" omit-xml-declaration="yes" standalone="no" indent="yes"/>

  <xsl:template match="/mods">
        <add  id="{recordInfo/recordIdentifier}" version="1" lang="en">
                <xsl:for-each select="titleInfo/title">
                        <field name="title"><xsl:value-of select="."/></field>
                </xsl:for-each>
                <xsl:for-each select="subject[@authority='lcsh']/topic">
                        <field name="subject"><xsl:value-of select="."/></field>
                </xsl:for-each>
        </add>
  </xsl:template>
</xsl:stylesheet>