/**
* @covers Nlm30NameSchemaPersonStringFilter
* @covers Nlm30PersonStringFilter
* @depends testExecuteWithSinglePersonDescription
*/
public function testExecuteWithMultiplePersonDescriptions($personDescription1)
{
$personDescription2 = new MetadataDescription('lib.pkp.plugins.metadata.nlm30.schema.Nlm30NameSchema', ASSOC_TYPE_AUTHOR);
$personDescription2->addStatement('given-names', $givenNames1 = 'Bernardo');
$personDescription2->addStatement('given-names', $givenNames2 = 'Antonio');
$personDescription2->addStatement('surname', $surname = 'Elis');
$personDescriptions = array($personDescription1, $personDescription2, PERSON_STRING_FILTER_ETAL);
$nlm30NameSchemaPersonStringFilter = new Nlm30NameSchemaPersonStringFilter(PERSON_STRING_FILTER_MULTIPLE);
self::assertEquals('Assis Jr, (Machado) de; Elis, A. (Bernardo); et al', $nlm30NameSchemaPersonStringFilter->execute($personDescriptions));
// Test template and delimiter
$nlm30NameSchemaPersonStringFilter->setDelimiter(':');
$nlm30NameSchemaPersonStringFilter->setTemplate('%firstname%%initials%%prefix% %surname%%suffix%');
self::assertEquals('Machado de Assis Jr:Bernardo A. Elis:et al', $nlm30NameSchemaPersonStringFilter->execute($personDescriptions));
}
/**
* Take a structured meta-data statement and transform it into a
* plain text value that can be displayed to the end-user.
*
* @param $property MetadataProperty
* @param $value mixed
* @return string
*/
function _getStringValueFromMetadataStatement(&$property, &$value)
{
if ($property->getCardinality() == METADATA_PROPERTY_CARDINALITY_MANY && !empty($value)) {
$allowedTypes = $property->getAllowedTypes();
if (isset($allowedTypes[METADATA_PROPERTY_TYPE_COMPOSITE])) {
// We currently only can transform composite
// name arrays to strings.
$allowedAssocTypes = $allowedTypes[METADATA_PROPERTY_TYPE_COMPOSITE];
assert(in_array(ASSOC_TYPE_AUTHOR, $allowedAssocTypes) || in_array(ASSOC_TYPE_EDITOR, $allowedAssocTypes));
import('lib.pkp.plugins.metadata.nlm30.filter.Nlm30NameSchemaPersonStringFilter');
$personStringFilter = new Nlm30NameSchemaPersonStringFilter(PERSON_STRING_FILTER_MULTIPLE);
assert($personStringFilter->supportsAsInput($value));
$stringValue = $personStringFilter->execute($value);
} else {
// We currently can't transform properties of
// cardinality "many" to strings.
assert(is_array($value) && count($value) <= 1);
$stringValue = $value[0];
}
} else {
$stringValue = (string) $value;
}
return $stringValue;
}
/**
* Construct an array of search strings from a citation
* description and an array of search templates.
* The templates may contain the placeholders
* %aulast%: the first author's surname
* %au%: the first author full name
* %title%: the article-title (if it exists),
* otherwise the source
* %date%: the publication year
* %isbn%: ISBN
* @param $searchTemplates an array of templates
* @param $citationDescription MetadataDescription
* @return array
*/
function constructSearchStrings(&$searchTemplates, &$citationDescription)
{
// Convert first authors' name description to a string
import('lib.pkp.plugins.metadata.nlm30.filter.Nlm30NameSchemaPersonStringFilter');
$personStringFilter = new Nlm30NameSchemaPersonStringFilter();
// Retrieve the authors
$firstAuthorSurname = $firstAuthor = '';
$authors = $citationDescription->getStatement('person-group[@person-group-type="author"]');
if (is_array($authors) && count($authors)) {
$firstAuthorSurname = (string) $authors[0]->getStatement('surname');
$firstAuthor = $personStringFilter->execute($authors[0]);
}
// Retrieve the editors
$firstEditorSurname = $firstEditor = '';
$editors = $citationDescription->getStatement('person-group[@person-group-type="editor"]');
if (is_array($editors) && count($editors)) {
$firstEditorSurname = (string) $editors[0]->getStatement('surname');
$firstEditor = $personStringFilter->execute($editors[0]);
}
// Retrieve (default language) title
$title = (string) ($citationDescription->hasStatement('article-title') ? $citationDescription->getStatement('article-title') : $citationDescription->getStatement('source'));
// Extract the year from the publication date
$year = (string) $citationDescription->getStatement('date');
$year = String::strlen($year) > 4 ? String::substr($year, 0, 4) : $year;
// Retrieve ISBN
$isbn = (string) $citationDescription->getStatement('isbn');
// Replace the placeholders in the templates
$searchStrings = array();
foreach ($searchTemplates as $searchTemplate) {
// Try editors and authors separately
$searchStrings[] = str_replace(array('%aulast%', '%au%', '%title%', '%date%', '%isbn%'), array($firstAuthorSurname, $firstAuthor, $title, $year, $isbn), $searchTemplate);
$searchStrings[] = str_replace(array('%aulast%', '%au%', '%title%', '%date%', '%isbn%'), array($firstEditorSurname, $firstEditor, $title, $year, $isbn), $searchTemplate);
}
// Remove empty or duplicate searches
$searchStrings = array_map(array('String', 'trimPunctuation'), $searchStrings);
$searchStrings = array_unique($searchStrings);
$searchStrings = arrayClean($searchStrings);
return $searchStrings;
}
/**
* Map NLM properties to OpenURL properties.
* NB: OpenURL has no i18n so we use the default
* locale when mapping.
* @see Filter::process()
* @param $input MetadataDescription
* @return MetadataDescription
*/
function &process(&$input)
{
$nullVar = null;
// Identify the genre of the target record and
// instantiate the target description.
$publicationType = $input->getStatement('[@publication-type]');
switch ($publicationType) {
case NLM30_PUBLICATION_TYPE_JOURNAL:
case NLM30_PUBLICATION_TYPE_CONFPROC:
$outputSchemaName = 'lib.pkp.plugins.metadata.openurl10.schema.Openurl10JournalSchema';
break;
case NLM30_PUBLICATION_TYPE_BOOK:
$outputSchemaName = 'lib.pkp.plugins.metadata.openurl10.schema.Openurl10BookSchema';
break;
case NLM30_PUBLICATION_TYPE_THESIS:
$outputSchemaName = 'lib.pkp.plugins.metadata.openurl10.schema.Openurl10DissertationSchema';
break;
default:
// Unsupported type
return $nullVar;
}
// Create the target description
$output = new MetadataDescription($outputSchemaName, $input->getAssocType());
// Transform authors
import('lib.pkp.plugins.metadata.nlm30.filter.Nlm30NameSchemaPersonStringFilter');
$personStringFilter = new Nlm30NameSchemaPersonStringFilter();
$authors =& $input->getStatement('person-group[@person-group-type="author"]');
if (is_array($authors) && count($authors)) {
$aulast = $authors[0]->hasStatement('prefix') ? $authors[0]->getStatement('prefix') . ' ' : '';
$aulast .= $authors[0]->getStatement('surname');
if (!empty($aulast)) {
$success = $output->addStatement('aulast', $aulast);
assert($success);
}
$givenNames = $authors[0]->getStatement('given-names');
if (is_array($givenNames) && count($givenNames)) {
$aufirst = implode(' ', $givenNames);
if (!empty($aufirst)) {
$success = $output->addStatement('aufirst', $aufirst);
assert($success);
}
$initials = array();
foreach ($givenNames as $givenName) {
$initials[] = substr($givenName, 0, 1);
}
$auinit1 = array_shift($initials);
if (!empty($auinit1)) {
$success = $output->addStatement('auinit1', $auinit1);
assert($success);
}
$auinitm = implode('', $initials);
if (!empty($auinitm)) {
$success = $output->addStatement('auinitm', $auinitm);
assert($success);
}
$auinit = $auinit1 . $auinitm;
if (!empty($auinit)) {
$success = $output->addStatement('auinit', $auinit);
assert($success);
}
}
$ausuffix = $authors[0]->getStatement('suffix');
if (!empty($ausuffix)) {
$success = $output->addStatement('ausuffix', $ausuffix);
assert($success);
}
foreach ($authors as $author) {
if ($author == PERSON_STRING_FILTER_ETAL) {
$au = $author;
} else {
$au = $personStringFilter->execute($author);
}
$success = $output->addStatement('au', $au);
assert($success);
unset($au);
}
}
// Genre: Guesswork
if (is_a($output->getMetadataSchema(), 'Openurl10JournalBookBaseSchema')) {
switch ($publicationType) {
case NLM30_PUBLICATION_TYPE_JOURNAL:
$genre = $input->hasProperty('article-title') ? OPENURL10_GENRE_ARTICLE : OPENURL10_GENRE_JOURNAL;
break;
case NLM30_PUBLICATION_TYPE_CONFPROC:
$genre = $input->hasProperty('article-title') ? OPENURL10_GENRE_PROCEEDING : OPENURL10_GENRE_CONFERENCE;
break;
case NLM30_PUBLICATION_TYPE_BOOK:
$genre = $input->hasProperty('article-title') ? OPENURL10_GENRE_BOOKITEM : OPENURL10_GENRE_BOOK;
break;
}
assert(!empty($genre));
$success = $output->addStatement('genre', $genre);
assert($success);
}
// Map remaining properties (NLM => OpenURL)
$propertyMap =& $this->nlmOpenurl10Mapping($publicationType, $output->getMetadataSchema());
// Transfer mapped properties with default locale
foreach ($propertyMap as $nlm30Property => $openurl10Property) {
if ($input->hasStatement($nlm30Property)) {
$success = $output->addStatement($openurl10Property, $input->getStatement($nlm30Property));
assert($success);
}
}
return $output;
}
/**
* @see Filter::process()
* @param $citationDescription MetadataDescription
* @return MetadataDescription
*/
function &process(&$citationDescription)
{
$pmid = $citationDescription->getStatement('pub-id[@pub-id-type="pmid"]');
// If the citation does not have a PMID, try to get one from eSearch
// otherwise skip directly to eFetch.
if (empty($pmid)) {
// Initialize search result arrays.
$pmidArrayFromAuthorsSearch = $pmidArrayFromTitleSearch = $pmidArrayFromStrictSearch = array();
// 1) Try a "loose" search based on the author list.
// (This works surprisingly well for pubmed.)
$authors =& $citationDescription->getStatement('person-group[@person-group-type="author"]');
if (is_array($authors)) {
import('lib.pkp.plugins.metadata.nlm30.filter.Nlm30NameSchemaPersonStringFilter');
$personNameFilter = new Nlm30NameSchemaPersonStringFilter(PERSON_STRING_FILTER_MULTIPLE, '%firstname%%initials%%prefix% %surname%%suffix%', ', ');
$authorsString = (string) $personNameFilter->execute($authors);
if (!empty($authorsString)) {
$pmidArrayFromAuthorsSearch =& $this->_search($authorsString);
}
}
// 2) Try a "loose" search based on the article title
$articleTitle = (string) $citationDescription->getStatement('article-title');
if (!empty($articleTitle)) {
$pmidArrayFromTitleSearch =& $this->_search($articleTitle);
}
// 3) Try a "strict" search based on as much information as possible
$searchProperties = array('article-title' => '', 'person-group[@person-group-type="author"]' => '[Auth]', 'source' => '[Jour]', 'date' => '[DP]', 'volume' => '[VI]', 'issue' => '[IP]', 'fpage' => '[PG]');
$searchTerms = '';
$statements = $citationDescription->getStatements();
foreach ($searchProperties as $nlm30Property => $pubmedProperty) {
if (isset($statements[$nlm30Property])) {
if (!empty($searchTerms)) {
$searchTerms .= ' AND ';
}
// Special treatment for authors
if ($nlm30Property == 'person-group[@person-group-type="author"]') {
assert(isset($statements['person-group[@person-group-type="author"]'][0]));
$firstAuthor =& $statements['person-group[@person-group-type="author"]'][0];
// Add surname
$searchTerms .= (string) $firstAuthor->getStatement('surname');
// Add initial of the first given name
$givenNames = $firstAuthor->getStatement('given-names');
if (is_array($givenNames)) {
$searchTerms .= ' ' . String::substr($givenNames[0], 0, 1);
}
} else {
$searchTerms .= $citationDescription->getStatement($nlm30Property);
}
$searchTerms .= $pubmedProperty;
}
}
$pmidArrayFromStrictSearch =& $this->_search($searchTerms);
// TODO: add another search like strict, but without article title
// e.g. ...term=Baumgart+Dc[Auth]+AND+Lancet[Jour]+AND+2005[DP]+AND+366[VI]+AND+9492[IP]+AND+1210[PG]
// Compare the arrays to try to narrow it down to one PMID
switch (true) {
// strict search has a single result
case count($pmidArrayFromStrictSearch) == 1:
$pmid = $pmidArrayFromStrictSearch[0];
break;
// 3-way union
// 3-way union
case count($intersect = array_intersect($pmidArrayFromTitleSearch, $pmidArrayFromAuthorsSearch, $pmidArrayFromStrictSearch)) == 1:
$pmid = current($intersect);
break;
// 2-way union: title / strict
// 2-way union: title / strict
case count($pmid_2way1 = array_intersect($pmidArrayFromTitleSearch, $pmidArrayFromStrictSearch)) == 1:
$pmid = current($pmid_2way1);
break;
// 2-way union: authors / strict
// 2-way union: authors / strict
case count($pmid_2way2 = array_intersect($pmidArrayFromAuthorsSearch, $pmidArrayFromStrictSearch)) == 1:
$pmid = current($pmid_2way2);
break;
// 2-way union: authors / title
// 2-way union: authors / title
case count($pmid_2way3 = array_intersect($pmidArrayFromAuthorsSearch, $pmidArrayFromTitleSearch)) == 1:
$pmid = current($pmid_2way3);
break;
// we only have one result for title
// we only have one result for title
case count($pmidArrayFromTitleSearch) == 1:
$pmid = $pmidArrayFromTitleSearch[0];
break;
// we only have one result for authors
// we only have one result for authors
case count($pmidArrayFromAuthorsSearch) == 1:
$pmid = $pmidArrayFromAuthorsSearch[0];
break;
// we were unable to find a PMID
// we were unable to find a PMID
default:
$pmid = '';
}
}
// If we have a PMID, get a metadata array for it
if (!empty($pmid)) {
$citationDescription =& $this->_lookup($pmid);
return $citationDescription;
}
// Nothing found
$nullVar = null;
return $nullVar;
}
