/**
* Break the MODS topic element text-node metadata on
* the specified character and put into seperate MODS topic elements.
*
* @param string $xmlsnippet The initial MODS topic element.
*
* @param string $breakOnCharacter The charcter break the string.
* The default character is the semicolon ';'.
*
* @return string
* An XML string containing one or more MODS topic elements.
*/
public function breakTopicMetadaOnCharacter($xmlsnippet, $breakOnCharacter = ';')
{
// Break topic metadata on ; into seperate topic elements.
$xml = new \DomDocument();
$xml->loadxml($xmlsnippet, LIBXML_NSCLEAN);
$topicNode = $xml->getElementsByTagName('topic')->item(0);
if (!is_object($topicNode)) {
$xmlstring = $xmlsnippet;
} else {
$topictext = $topicNode->nodeValue;
$topics = explode($breakOnCharacter, $topictext);
// remove old topic node.
$topicNodeParent = $topicNode->parentNode;
$topicNode->parentNode->removeChild($topicNode);
$subjectNode = $xml->getElementsByTagName($this->topLevelNodeName)->item(0);
foreach ($topics as $topic) {
$topic = trim($topic);
$newtopicElement = $xml->createElement('topic');
$topictextNode = $xml->createTextNode($topic);
$newtopicElement->appendChild($topictextNode);
$subjectNode->appendChild($newtopicElement);
unset($topictextNode);
unset($newtopicElement);
}
$xmlstring = $xml->saveXML($subjectNode);
}
return $xmlstring;
}
/**
* General manipulate wrapper method.
*
* @param string $input An XML snippet to be manipulated. We are only interested
* in <abstract> snippets.
*
* @return string
* Manipulated string
*/
public function manipulate($input)
{
$dom = new \DomDocument();
$dom->loadxml($input, LIBXML_NSCLEAN);
$abstracts = $dom->getElementsByTagName('abstract');
if ($abstracts->length == 1) {
$abstract = $abstracts->item(0);
// Use Guzzle to hit the API.
$client = new Client();
try {
$original_text = urlencode($abstract->nodeValue);
$query = "?text={$original_text}&format=json";
$response = $client->get($this->arrpiUrl . $query);
// If there is a Guzzle error, log it and return the original snippet.
} catch (Exception $e) {
$this->log->addWarning("PiratizeAbstract", array('HTTP request error' => $e->getMessage()));
return $input;
}
$body = $response->getBody();
$translation = json_decode($body, true);
$abstract->nodeValue = urldecode($translation['translation']['pirate']);
// Log any instances where the translation differs from the original text.
if (urldecode($original_text) != $abstract->nodeValue) {
$this->log->addInfo("PiratizeAbstract", array('Record key' => $this->record_key, 'Source abstract text' => urldecode($original_text), 'Piratized abstract text' => $abstract->nodeValue));
}
// We're done, so return the modified snippet.
return $dom->saveXML($dom->documentElement);
} else {
return $input;
}
}
/**
* General manipulate wrapper method.
*
* @param string $input The XML fragment to be manipulated. We are only
* interested in the <extension><CSVData> fragment added in the
* MIK mappings file.
*
* @return string
* One of the manipulated XML fragment, the original input XML if the
* input is not the fragment we are interested in, or an empty string,
* which as the effect of removing the empty <extension><CSVData>
* fragement from our MODS (if there was an error, for example, we don't
* want empty extension elements in our MODS documents).
*/
public function manipulate($input)
{
$dom = new \DomDocument();
$dom->loadxml($input, LIBXML_NSCLEAN);
// Test to see if the current fragment is <extension><CSVData>.
$xpath = new \DOMXPath($dom);
$csvdatas = $xpath->query("//extension/CSVData");
// There should only be one <CSVData> fragment in the incoming
// XML. If there is 0 or more than 1, return the original.
if ($csvdatas->length === 1) {
$csvdata = $csvdatas->item(0);
$csvid = $dom->createElement('id_in_csv', $this->record_key);
$csvdata->appendChild($csvid);
$timestamp = date("Y-m-d H:i:s");
// Add the <CSVRecord> element.
$csvrecord = $dom->createElement('CSVRecord');
$now = $dom->createAttribute('timestamp');
$now->value = $timestamp;
$csvrecord->appendChild($now);
$mimetype = $dom->createAttribute('mimetype');
$mimetype->value = 'application/json';
$csvrecord->appendChild($mimetype);
try {
$metadata_path = $this->settings['FETCHER']['temp_directory'] . DIRECTORY_SEPARATOR . $this->record_key . '.metadata';
$metadata_contents = file_get_contents($metadata_path);
$metadata_contents = unserialize($metadata_contents);
$metadata_contents = json_encode($metadata_contents);
} catch (Exception $e) {
$message = "Problem creating <CSVRecord> element for object " . $this->record_key . ":" . $e->getMessage();
$this->log->addInfo("AddCsvData", array('CSV metadata warning' => $message));
return '';
}
// If the metadata contains the CDATA end delimiter, log and return.
if (preg_match('/\\]\\]>/', $metadata_contents)) {
$message = "CSV metadata for object " . $this->record_key . ' contains the CDATA end delimiter ]]>';
$this->log->addInfo("AddCsvData", array('CSV metadata warning' => $message));
return '';
}
// If we've made it this far, add the metadata to <CcvData> as
// CDATA and return the modified XML fragment.
if (strlen($metadata_contents)) {
$cdata = $dom->createCDATASection($metadata_contents);
$csvrecord->appendChild($cdata);
$csvdata->appendChild($csvrecord);
}
return $dom->saveXML($dom->documentElement);
} else {
// If current fragment is not <extension><CSVData>, return it
// unmodified.
return $input;
}
}
/**
* General manipulate wrapper method.
*
* @param string $input An XML snippet to be manipulated.
*
* @return string
* Manipulated string
*/
public function manipulate($input)
{
$dom = new \DomDocument();
$dom->loadxml($input, LIBXML_NSCLEAN);
// Test to see if the current fragment is the one identified in the config file.
$xpath = new \DOMXPath($dom);
$date_elements = $xpath->query('/originInfo/' . $this->destDateElement);
// There should only be one target date element.
if ($date_elements->length === 1) {
// Get the child node, which we will repopulate below if its value
// matches our regex.
$date_element = $date_elements->item(0);
// Get its parent so we can reconstruct it for sending back to the
// metadata parser.
$origin_info_element = $date_element->parentNode;
$this->sourceDateFieldValue = $this->getSourceDateFieldValue();
// See if the value of the date field in the raw metadata matches our
// pattern, and if it does, replace the value of the target MODS element
// with a w3cdtf version of the date value.
// @todo: When 'ca.'' is present, add 'qualifier' attribute with values 'approximate',
// 'inferred', 'questionable'. Set a default (maybe configurable) date in this case?
// Check for dates in \d\d-\d\d-\d\d\d\d.
if (preg_match('/^(\\d\\d)\\-(\\d\\d)\\-(\\d\\d\\d\\d)$/', $this->sourceDateFieldValue, $matches)) {
$date_element->nodeValue = $matches[3] . '-' . $matches[2] . '-' . $matches[1];
// Reassemble the parent and child elements.
$origin_info_element->appendChild($date_element);
// Convert the back to the snippet and return it.
$this->logNormalization($this->sourceDateFieldValue, $origin_info_element, $dom);
return $dom->saveXML($origin_info_element);
} elseif (preg_match('/^(\\d\\d\\d\\d)\\s+(\\d\\d)\\s+(\\d\\d)$/', $this->sourceDateFieldValue, $matches)) {
$date_element->nodeValue = $matches[1] . '-' . $matches[2] . '-' . $matches[3];
$origin_info_element->appendChild($date_element);
$this->logNormalization($this->sourceDateFieldValue, $origin_info_element, $dom);
return $dom->saveXML($origin_info_element);
} elseif (!is_string($this->sourceDateFieldValue) || !strlen($this->sourceDateFieldValue)) {
$this->log->addWarning("NormalizeDate", array('Record key' => $this->record_key, 'Message' => 'Source date value is empty or not a string'));
return $input;
} else {
$this->log->addWarning("NormalizeDate", array('Record key' => $this->record_key, 'Source date value does not match any pattern' => $this->sourceDateFieldValue));
return $input;
}
} else {
// If current fragment does not match our XPath expression, return it.
return $input;
}
}
/**
* General manipulate wrapper method.
*
* @param string $input The XML fragment to be manipulated. We are only
* interested in the <identifier type="uuid"> fragment added in the
* MIK mappings file.
*
* @return string
* One of the manipulated XML fragment, the original input XML if the
* input is not the fragment we are interested in.
*/
public function manipulate($input)
{
$dom = new \DomDocument();
$dom->loadxml($input, LIBXML_NSCLEAN);
// Test to see if the current fragment is <identifier type="uuid">.
$xpath = new \DOMXPath($dom);
$uuid_identifiers = $xpath->query("//identifier[@type='uuid']");
// There should only be one <identifier type="uuid"/> fragment in the
// incoming XML, defined in the mappings file. If there is 0, return
// the original.
if ($uuid_identifiers->length === 1) {
$uuid_identifier = $uuid_identifiers->item(0);
// If our incoming fragment is already a valid UUID v4, return it as
// is. Note that if a identifier with type "uuid" already exists, this
// manipulator will add a new one, since we are processing the MODS
// on an element by element basis, not the entire MODS document.
if (strlen($uuid_identifier->nodeValue) && preg_match('/^[0-9A-F]{8}-[0-9A-F]{4}-4[0-9A-F]{3}-[89AB][0-9A-F]{3}-[0-9A-F]{12}$/i', $uuid_identifier->nodeValue)) {
$this->log->addError("AddUuidToMods", array('UUID already present' => $uuid_identifier->nodeValue));
return $input;
} else {
try {
$uuid4 = Uuid::uuid4();
$uuid4_string = $uuid4->toString();
} catch (UnsatisfiedDependencyException $e) {
// Log error and return $input.
$this->log->addError("AddUuidToMods", array('UUID generation error' => $e->getMessage()));
}
$uuid_identifier->nodeValue = $uuid4_string;
return $dom->saveXML($dom->documentElement);
}
} else {
// If current fragment is not <identifier type="uuid">,
// with or without a valid UUID v4 as a value, return it unmodified.
return $input;
}
}
</item>
<item date="3247283732">
<caption>Testing...</caption>
<content>
<![CDATA[
dies ist ein zweiter test...
]]>
</content>
</item>
</news>';
$xsl2str = '<xsl:stylesheet version = "1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
<xsl:template match="/">
<xsl:apply-templates />
</xsl:template>
<xsl:template match="content">
<b><xsl:value-of select="." /></b>
</xsl:template>
';
$xml = new DomDocument();
$xml->loadxml($xmlstr);
$xsl = new DomDocument();
$xsl->load("test1.xsl");
$xsl2 = new DomDocument();
$xsl2->load("test1-2.xsl");
$proc = new xsltprocessor();
$proc->importStyleSheet($xsl2);
$proc->importStyleSheet($xsl);
echo $proc->transformToXML($xml);
echo '<br><br>';
printf("%s<br>Count: %d", md5("29o3_DEFAULT_THEME"), strlen(md5("test")));
/**
* General manipulate wrapper method.
*
* @param string $input The XML fragment to be manipulated. We are only
* interested in the <extension><CONTENTdmData> fragment added in the
* MIK mappings file.
*
* @return string
* One of the manipulated XML fragment, the original input XML if the
* input is not the fragment we are interested in, or an empty string,
* which as the effect of removing the empty <extension><CONTENTdmData>
* fragement from our MODS (if there was an error, for example, we don't
* want empty extension elements in our MODS documents).
*/
public function manipulate($input)
{
$dom = new \DomDocument();
$dom->loadxml($input, LIBXML_NSCLEAN);
// Test to see if the current fragment is <extension><CONTENTdmData>.
$xpath = new \DOMXPath($dom);
$cdmdatas = $xpath->query("//extension/CONTENTdmData");
// There should only be one <CONTENTdmData> fragment in the incoming
// XML. If there is 0 or more than 1, return the original.
if ($cdmdatas->length === 1) {
$contentdmdata = $cdmdatas->item(0);
$alias = $dom->createElement('alias', $this->alias);
$contentdmdata->appendChild($alias);
$pointer = $dom->createElement('pointer', $this->record_key);
$contentdmdata->appendChild($pointer);
$timestamp = date("Y-m-d H:i:s");
// Add the <dmGetItemInfo> element.
$dmGetItemInfo = $dom->createElement('dmGetItemInfo');
$now = $dom->createAttribute('timestamp');
$now->value = $timestamp;
$dmGetItemInfo->appendChild($now);
$mimetype = $dom->createAttribute('mimetype');
$mimetype->value = 'application/json';
$dmGetItemInfo->appendChild($mimetype);
$source_url = $this->settings['METADATA_PARSER']['ws_url'] . 'dmGetItemInfo/' . $this->alias . '/' . $this->record_key . '/json';
$source = $dom->createAttribute('source');
$source->value = $source_url;
$dmGetItemInfo->appendChild($source);
$item_info = $this->getCdmData($this->alias, $this->record_key, 'dmGetItemInfo', 'json');
// CONTENTdm returns a 200 OK with its error messages, so we can't rely
// on catching all 'errors' with the above try/catch block. Instead, we
// check to see if the string 'dmcreated' (one of the metadata fields
// returned for every object) is in the response body. If it's not,
// assume CONTENTdm has returned an error of some sort, log it, and
// return.
if (!preg_match('/dmcreated/', $item_info)) {
$this->log->addInfo("AddContentdmData", array('CONTENTdm internal error' => $item_info));
return '';
}
// If the CONTENTdm metadata contains the CDATA end delimiter, log and return.
if (preg_match('/\\]\\]>/', $item_info)) {
$message = "CONTENTdm metadata for object " . $this->settings['METADATA_PARSER']['alias'] . '/' . $this->record_key . ' contains the CDATA end delimiter ]]>';
$this->log->addInfo("AddContentdmData", array('CONTENTdm metadata warning' => $message));
return '';
}
// If we've made it this far, add the output of dmGetItemInfo to <CONTENTdmData> as
// CDATA and return the modified XML fragment.
if (strlen($item_info)) {
$cdata = $dom->createCDATASection($item_info);
$dmGetItemInfo->appendChild($cdata);
$contentdmdata->appendChild($dmGetItemInfo);
}
// Add the <dmCompoundObjectInfo> element.
$dmGetCompoundObjectInfo = $dom->createElement('dmGetCompoundObjectInfo');
$now = $dom->createAttribute('timestamp');
$now->value = $timestamp;
$dmGetCompoundObjectInfo->appendChild($now);
$mimetype = $dom->createAttribute('mimetype');
$mimetype->value = 'text/xml';
$dmGetCompoundObjectInfo->appendChild($mimetype);
$source = $dom->createAttribute('source');
$source_url = $this->settings['METADATA_PARSER']['ws_url'] . 'dmGetCompoundObjectInfo/' . $this->alias . '/' . $this->record_key . '/xml';
$source->value = $source_url;
$dmGetCompoundObjectInfo->appendChild($source);
$compound_object_info = $this->getCdmData($this->alias, $this->record_key, 'dmGetCompoundObjectInfo', 'xml');
// Only add the <dmGetCompoundObjectInfo> element if the object is compound.
if (strlen($compound_object_info) && preg_match('/<cpd>/', $compound_object_info)) {
$cdata = $dom->createCDATASection($compound_object_info);
$dmGetCompoundObjectInfo->appendChild($cdata);
$contentdmdata->appendChild($dmGetCompoundObjectInfo);
}
// Add the <GetParent> element.
$GetParent = $dom->createElement('GetParent');
$now = $dom->createAttribute('timestamp');
$now->value = $timestamp;
$GetParent->appendChild($now);
$mimetype = $dom->createAttribute('mimetype');
$mimetype->value = 'text/xml';
$GetParent->appendChild($mimetype);
$source = $dom->createAttribute('source');
$source_url = $this->settings['METADATA_PARSER']['ws_url'] . 'GetParent/' . $this->alias . '/' . $this->record_key . '/xml';
$source->value = $source_url;
$GetParent->appendChild($source);
$parent_info = $this->getCdmData($this->alias, $this->record_key, 'GetParent', 'xml');
// Only add the <GetParent> element if the object has a parent
// pointer of not -1.
if (strlen($parent_info) && !preg_match('/\\-1/', $parent_info)) {
$cdata = $dom->createCDATASection($parent_info);
$GetParent->appendChild($cdata);
$contentdmdata->appendChild($GetParent);
}
return $dom->saveXML($dom->documentElement);
} else {
// If current fragment is not <extension><CONTENTdmData>, return it
// unmodified.
return $input;
}
}