/**
* @copydoc Filter::process()
* @param $input string
* @return MetadataDescription
*/
function &process(&$input)
{
$citationString =& $input;
$nullVar = null;
// Freecite requires a post request
$postData = array('citation' => $citationString);
if (is_null($resultDOM = $this->callWebService(FREECITE_WEBSERVICE, $postData, XSL_TRANSFORMER_DOCTYPE_DOM, 'POST'))) {
return $nullVar;
}
// Transform the result into an array of meta-data
if (is_null($metadata =& $this->transformWebServiceResults($resultDOM, dirname(__FILE__) . DIRECTORY_SEPARATOR . 'freecite.xsl'))) {
return $nullVar;
}
// Extract a publisher from the place string if possible
$metadata =& $this->fixPublisherNameAndLocation($metadata);
// Convert the genre
if (isset($metadata['genre'])) {
$genre = $metadata['genre'];
import('lib.pkp.plugins.metadata.nlm30.filter.Openurl10Nlm30CitationSchemaCrosswalkFilter');
$genreMap = Openurl10Nlm30CitationSchemaCrosswalkFilter::_getOpenurl10GenreTranslationMapping();
$metadata['[@publication-type]'] = isset($genreMap[$genre]) ? $genreMap[$genre] : $genre;
unset($metadata['genre']);
}
// Convert article title to source for dissertations
if (isset($metadata['[@publication-type]']) && $metadata['[@publication-type]'] == NLM30_PUBLICATION_TYPE_THESIS && isset($metadata['article-title'])) {
$metadata['source'] = $metadata['article-title'];
unset($metadata['article-title']);
}
unset($metadata['raw_string']);
return $this->getNlm30CitationDescriptionFromMetadataArray($metadata);
}
/**
* @covers Openurl10Nlm30CitationSchemaCrosswalkFilter
* @covers Nlm30Openurl10CrosswalkFilter
*/
public function testExecute()
{
$openurl10Description = $this->getTestOpenurl10Description();
$nlm30Description = $this->getTestNlm30Description();
// Properties that are not part of the OpenURL
// description must be removed from the NLM description
// before we compare the two.
self::assertTrue($nlm30Description->removeStatement('person-group[@person-group-type="editor"]'));
self::assertTrue($nlm30Description->removeStatement('source', 'de_DE'));
self::assertTrue($nlm30Description->removeStatement('article-title', 'de_DE'));
self::assertTrue($nlm30Description->removeStatement('publisher-loc'));
self::assertTrue($nlm30Description->removeStatement('publisher-name'));
self::assertTrue($nlm30Description->removeStatement('pub-id[@pub-id-type="doi"]'));
self::assertTrue($nlm30Description->removeStatement('pub-id[@pub-id-type="pmid"]'));
self::assertTrue($nlm30Description->removeStatement('uri'));
self::assertTrue($nlm30Description->removeStatement('comment'));
self::assertTrue($nlm30Description->removeStatement('annotation'));
$filter = new Openurl10Nlm30CitationSchemaCrosswalkFilter();
self::assertEquals($nlm30Description, $filter->execute($openurl10Description));
}
/**
* @copydoc Filter::process()
* @param $input string
* @return MetadataDescription
*/
function &process(&$input)
{
$citationString =& $input;
$nullVar = null;
// Check the availability of perl
$perlCommand = Config::getVar('cli', 'perl');
if (empty($perlCommand) || !file_exists($perlCommand)) {
return $nullVar;
}
// Convert to ASCII - Paracite doesn't handle UTF-8 well
$citationString = String::utf8_to_ascii($citationString);
// Call the paracite parser
$wrapperScript = dirname(__FILE__) . DIRECTORY_SEPARATOR . 'paracite.pl';
$paraciteCommand = $perlCommand . ' ' . escapeshellarg($wrapperScript) . ' ' . $this->getCitationModule() . ' ' . escapeshellarg($citationString);
$xmlResult = shell_exec($paraciteCommand);
if (empty($xmlResult)) {
return $nullVar;
}
if (Config::getVar('i18n', 'charset_normalization') == 'On' && !String::utf8_compliant($xmlResult)) {
$xmlResult = String::utf8_normalize($xmlResult);
}
// Create a temporary DOM document
$resultDOM = new DOMDocument();
$resultDOM->recover = true;
$resultDOM->loadXML($xmlResult);
// Extract the parser results as an array
$xmlHelper = new XMLHelper();
$metadata = $xmlHelper->xmlToArray($resultDOM->documentElement);
// We have to merge subtitle and title as neither OpenURL
// nor NLM can handle subtitles.
if (isset($metadata['subtitle'])) {
$metadata['title'] .= '. ' . $metadata['subtitle'];
unset($metadata['subtitle']);
}
// Break up the authors field
if (isset($metadata['authors'])) {
$metadata['authors'] = String::trimPunctuation($metadata['authors']);
$metadata['authors'] = String::iterativeExplode(array(':', ';'), $metadata['authors']);
}
// Convert pages to integers
foreach (array('spage', 'epage') as $pageProperty) {
if (isset($metadata[$pageProperty])) {
$metadata[$pageProperty] = (int) $metadata[$pageProperty];
}
}
// Convert titles to title case
foreach (array('title', 'chapter', 'publication') as $titleProperty) {
if (isset($metadata[$titleProperty])) {
$metadata[$titleProperty] = String::titleCase($metadata[$titleProperty]);
}
}
// Map ParaCite results to OpenURL - null means
// throw the value away.
$metadataMapping = array('genre' => 'genre', '_class' => null, 'any' => null, 'authors' => 'au', 'aufirst' => 'aufirst', 'aufull' => null, 'auinit' => 'auinit', 'aulast' => 'aulast', 'atitle' => 'atitle', 'cappublication' => null, 'captitle' => null, 'date' => 'date', 'epage' => 'epage', 'featureID' => null, 'id' => null, 'issue' => 'issue', 'jnl_epos' => null, 'jnl_spos' => null, 'match' => null, 'marked' => null, 'num_of_fig' => null, 'pages' => 'pages', 'publisher' => 'pub', 'publoc' => 'place', 'ref' => null, 'rest_text' => null, 'spage' => 'spage', 'targetURL' => 'url', 'text' => null, 'ucpublication' => null, 'uctitle' => null, 'volume' => 'volume', 'year' => 'date');
// Ignore 'year' if 'date' is set
if (isset($metadata['date'])) {
$metadataMapping['year'] = null;
}
// Set default genre
if (empty($metadata['genre'])) {
$metadata['genre'] = OPENURL10_GENRE_ARTICLE;
}
// Handle title, chapter and publication depending on
// the (inferred) genre. Also instantiate the target schema.
switch ($metadata['genre']) {
case OPENURL10_GENRE_BOOK:
case OPENURL10_GENRE_BOOKITEM:
case OPENURL10_GENRE_REPORT:
case OPENURL10_GENRE_DOCUMENT:
$metadataMapping += array('publication' => 'btitle', 'chapter' => 'atitle');
if (isset($metadata['title'])) {
if (!isset($metadata['publication'])) {
$metadata['publication'] = $metadata['title'];
} elseif (!isset($metadata['chapter'])) {
$metadata['chapter'] = $metadata['title'];
}
unset($metadata['title']);
}
$openurl10SchemaName = 'lib.pkp.plugins.metadata.openurl10.schema.Openurl10BookSchema';
$openurl10SchemaClass = 'Openurl10BookSchema';
break;
case OPENURL10_GENRE_ARTICLE:
case OPENURL10_GENRE_JOURNAL:
case OPENURL10_GENRE_ISSUE:
case OPENURL10_GENRE_CONFERENCE:
case OPENURL10_GENRE_PROCEEDING:
case OPENURL10_GENRE_PREPRINT:
default:
$metadataMapping += array('publication' => 'jtitle');
if (isset($metadata['title'])) {
if (!isset($metadata['publication'])) {
$metadata['publication'] = $metadata['title'];
} elseif (!isset($metadata['atitle'])) {
$metadata['atitle'] = $metadata['title'];
}
unset($metadata['title']);
}
$openurl10SchemaName = 'lib.pkp.plugins.metadata.openurl10.schema.Openurl10JournalSchema';
$openurl10SchemaClass = 'Openurl10JournalSchema';
break;
}
// Instantiate an OpenURL description
$openurl10Description = new MetadataDescription($openurl10SchemaName, ASSOC_TYPE_CITATION);
$openurl10Schema = new $openurl10SchemaClass();
// Map the ParaCite result to OpenURL
foreach ($metadata as $paraciteElementName => $paraciteValue) {
if (!empty($paraciteValue)) {
// Trim punctuation
if (is_string($paraciteValue)) {
$paraciteValue = String::trimPunctuation($paraciteValue);
}
// Transfer the value to the OpenURL result array
assert(array_key_exists($paraciteElementName, $metadataMapping));
$openurl10PropertyName = $metadataMapping[$paraciteElementName];
if (!is_null($openurl10PropertyName) && $openurl10Schema->hasProperty($openurl10PropertyName)) {
if (is_array($paraciteValue)) {
foreach ($paraciteValue as $singleValue) {
$success = $openurl10Description->addStatement($openurl10PropertyName, $singleValue);
assert($success);
}
} else {
$success = $openurl10Description->addStatement($openurl10PropertyName, $paraciteValue);
assert($success);
}
}
}
}
// Crosswalk to NLM
$crosswalkFilter = new Openurl10Nlm30CitationSchemaCrosswalkFilter();
$nlm30Description =& $crosswalkFilter->execute($openurl10Description);
assert(is_a($nlm30Description, 'MetadataDescription'));
// Add 'rest_text' as NLM comment (if given)
if (isset($metadata['rest_text'])) {
$nlm30Description->addStatement('comment', String::trimPunctuation($metadata['rest_text']));
}
// Set display name and sequence id in the meta-data description
// to the corresponding values from the filter. This is important
// so that we later know which result came from which filter.
$nlm30Description->setDisplayName($this->getDisplayName());
$nlm30Description->setSeq($this->getSeq());
return $nlm30Description;
}
