/**
* @covers PersonStringNlm30NameSchemaFilter
* @covers Nlm30PersonStringFilter
* @depends testExecuteWithSinglePersonString
*/
public function testExecuteWithMultiplePersonsStrings()
{
$personsString = 'MULLER:IFC Peterberg:Peters HC:Yu QK:Hans Peter B. Sperling:et al';
$expectedResults = array(array(null, null, null, 'Muller'), array(null, array('I', 'F', 'C'), null, 'Peterberg'), array(null, array('H', 'C'), null, 'Peters'), array(null, array('Q', 'K'), null, 'Yu'), array(null, array('Hans', 'Peter', 'B'), null, 'Sperling'));
$personStringNlm30NameSchemaFilter = new PersonStringNlm30NameSchemaFilter(ASSOC_TYPE_AUTHOR, PERSON_STRING_FILTER_MULTIPLE);
$personDescriptions =& $personStringNlm30NameSchemaFilter->execute($personsString);
// The last description should be an 'et-al' string
self::assertEquals(PERSON_STRING_FILTER_ETAL, array_pop($personDescriptions));
foreach ($personDescriptions as $testNumber => $personDescription) {
$this->assertPerson($expectedResults[$testNumber], $personDescription, $testNumber);
}
// Test again, this time with title and degrees
$personsString = 'Dr. MULLER; IFC Peterberg; Prof. Peters HC, MSc.; Yu QK;Hans Peter B. Sperling; etal';
$expectedResults = array(array('Dr.', null, null, 'Muller'), array(null, array('I', 'F', 'C'), null, 'Peterberg'), array('Prof. - MSc', array('H', 'C'), null, 'Peters'), array(null, array('Q', 'K'), null, 'Yu'), array(null, array('Hans', 'Peter', 'B'), null, 'Sperling'));
$personStringNlm30NameSchemaFilter->setFilterTitle(true);
$personStringNlm30NameSchemaFilter->setFilterDegrees(true);
$personDescriptions =& $personStringNlm30NameSchemaFilter->execute($personsString);
// The last description should be an 'et-al' string
self::assertEquals(PERSON_STRING_FILTER_ETAL, array_pop($personDescriptions));
foreach ($personDescriptions as $testNumber => $personDescription) {
$this->assertPerson($expectedResults[$testNumber], $personDescription, $testNumber);
}
// Test whether Vancouver style comma separation works correctly
$personsString = 'Peterberg IFC, Peters HC, Sperling HP';
$expectedResults = array(array(null, array('I', 'F', 'C'), null, 'Peterberg'), array(null, array('H', 'C'), null, 'Peters'), array(null, array('H', 'P'), null, 'Sperling'));
$personStringNlm30NameSchemaFilter->setFilterTitle(false);
$personStringNlm30NameSchemaFilter->setFilterDegrees(false);
$personDescriptions =& $personStringNlm30NameSchemaFilter->execute($personsString);
foreach ($personDescriptions as $testNumber => $personDescription) {
$this->assertPerson($expectedResults[$testNumber], $personDescription, $testNumber);
}
// Single name strings should not be cut when separated by comma.
$personsString = 'Willinsky, John';
$expectedResult = array(null, array('John'), null, 'Willinsky');
$personDescriptions =& $personStringNlm30NameSchemaFilter->execute($personsString);
$this->assertEquals(1, count($personDescriptions));
$this->assertPerson($expectedResult, $personDescriptions[0], $testNumber);
// Test APA style author tokenization.
$singleAuthor = array(1 => 'Berndt, T. J.');
$twoAuthors = array(2 => 'Wegener-Prent, D. T., & Petty, R. E.');
$threeToSevenAuthors = array(6 => 'Kernis Wettelberger, M. H., Cornell, D. P., Sun, C. R., Berry, A., Harlow, T., & Bach, J. S.');
$moreThanSevenAuthors = array(7 => 'Miller, F. H., Choi, M.J., Angeli, L. L., Harland, A. A., Stamos, J. A., Thomas, S. T., . . . Rubin, L. H.');
$singleEditor = array(1 => 'A. Editor');
$twoEditors = array(2 => 'A. Editor-Double & B. Editor');
$threeToSevenEditors = array(6 => 'M.H. Kernis Wettelberger, D. P. Cornell, C.R. Sun, A. Berry, T. Harlow & J.S. Bach');
$moreThanSevenEditors = array(7 => 'F. H. Miller, M. J. Choi, L. L. Angeli, A. A. Harland, J. A. Stamos, S. T. Thomas . . . L. H. Rubin');
foreach (array($singleAuthor, $twoAuthors, $threeToSevenAuthors, $moreThanSevenAuthors, $singleEditor, $twoEditors, $threeToSevenEditors, $moreThanSevenEditors) as $test) {
$expectedNumber = key($test);
$testString = current($test);
$personDescriptions =& $personStringNlm30NameSchemaFilter->execute($testString);
$this->assertEquals($expectedNumber, count($personDescriptions), 'Offending string: ' . $testString);
}
}
/**
* Map OpenURL properties to NLM 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;
// Instantiate the target description.
$output = new MetadataDescription('lib.pkp.plugins.metadata.nlm30.schema.Nlm30CitationSchema', $input->getAssocType());
// Parse au statements into name descriptions
import('lib.pkp.plugins.metadata.nlm30.filter.PersonStringNlm30NameSchemaFilter');
$personStringFilter = new PersonStringNlm30NameSchemaFilter(ASSOC_TYPE_AUTHOR);
$authors =& $input->getStatement('au');
if (is_array($authors) && count($authors)) {
// TODO: We might improve results here by constructing the
// first author from aufirst, aulast fields.
foreach ($authors as $author) {
$authorDescription =& $personStringFilter->execute($author);
$success = $output->addStatement('person-group[@person-group-type="author"]', $authorDescription);
assert($success);
unset($authorDescription);
}
}
// Publication type
$publicationType = null;
if ($input->hasStatement('genre')) {
$genre = $input->getStatement('genre');
$genreMap = $this->_getOpenurl10GenreTranslationMapping();
$publicationType = isset($genreMap[$genre]) ? $genreMap[$genre] : $genre;
$success = $output->addStatement('[@publication-type]', $publicationType);
assert($success);
}
// Get NLM => OpenURL property mapping.
$propertyMap =& $this->nlmOpenurl10Mapping($publicationType, $input->getMetadataSchema());
// Transfer mapped properties with default locale
foreach ($propertyMap as $nlm30Property => $openurl10Property) {
if ($input->hasStatement($openurl10Property)) {
$success = $output->addStatement($nlm30Property, $input->getStatement($openurl10Property));
assert($success);
}
}
return $output;
}
/**
* Custom implementation of Form::validate() that validates
* meta-data form data and injects it into the internal citation
* object.
*
* NB: The configuration of the internal citation object
* would normally be done in readInputData(). Validation and
* injection can easily be done in one step. It therefore avoids
* code duplication and improves performance to do both here.
*/
function validate()
{
// Make sure that this method is not called twice which
// would corrupt internal state.
assert(empty($this->_metadataDescriptions));
parent::validate();
// Validate form data and inject it into
// the associated citation object.
$citation =& $this->getCitation();
$citation->setRawCitation($this->getData('rawCitation'));
if ($this->getData('citationApproved') == 'citationApproved') {
// Editor's shortcut to the approved state, e.g. for manually edited citations.
$citation->setCitationState(CITATION_APPROVED);
} elseif (in_array($this->getData('citationState'), Citation::_getSupportedCitationStates())) {
// Reset citation state if necessary
if ($this->getData('citationState') == CITATION_APPROVED) {
$this->setData('citationState', CITATION_LOOKED_UP);
}
$citation->setCitationState($this->getData('citationState'));
}
// Extract data from citation form fields and inject it into the citation
import('lib.pkp.classes.metadata.MetadataDescription');
$metadataSchemas = $citation->getSupportedMetadataSchemas();
foreach ($metadataSchemas as $metadataSchema) {
/* @var $metadataSchema MetadataSchema */
// Instantiate a meta-data description for the given schema
$metadataDescription = new MetadataDescription($metadataSchema->getClassName(), ASSOC_TYPE_CITATION);
// Set the meta-data statements
foreach ($metadataSchema->getProperties() as $propertyName => $property) {
$fieldName = $metadataSchema->getNamespacedPropertyId($propertyName);
$fieldValue = trim($this->getData($fieldName));
if (empty($fieldValue)) {
// Delete empty statements so that previously set
// statements (if any) will be deleted.
$metadataDescription->removeStatement($propertyName);
if ($property->getMandatory()) {
// A mandatory field is missing - add a validation error.
$this->addError($fieldName, __($property->getValidationMessage()));
$this->addErrorField($fieldName);
}
} else {
// Try to convert the field value to (a) strongly
// typed object(s) if applicable. Start with the most
// specific allowed type so that we always get the
// most strongly typed result possible.
$allowedTypes = $property->getAllowedTypes();
switch (true) {
case isset($allowedTypes[METADATA_PROPERTY_TYPE_VOCABULARY]) && is_numeric($fieldValue):
case isset($allowedTypes[METADATA_PROPERTY_TYPE_INTEGER]) && is_numeric($fieldValue):
$typedFieldValues = array((int) $fieldValue);
break;
case isset($allowedTypes[METADATA_PROPERTY_TYPE_DATE]):
import('lib.pkp.classes.metadata.DateStringNormalizerFilter');
$dateStringFilter = new DateStringNormalizerFilter();
assert($dateStringFilter->supportsAsInput($fieldValue));
$typedFieldValues = array($dateStringFilter->execute($fieldValue));
break;
case isset($allowedTypes[METADATA_PROPERTY_TYPE_COMPOSITE]):
// We currently only support name composites
$allowedAssocIds = $allowedTypes[METADATA_PROPERTY_TYPE_COMPOSITE];
if (in_array(ASSOC_TYPE_AUTHOR, $allowedAssocIds)) {
$assocType = ASSOC_TYPE_AUTHOR;
} elseif (in_array(ASSOC_TYPE_EDITOR, $allowedAssocIds)) {
$assocType = ASSOC_TYPE_EDITOR;
} else {
assert(false);
}
// Try to transform the field to a name composite.
import('lib.pkp.plugins.metadata.nlm30.filter.PersonStringNlm30NameSchemaFilter');
$personStringFilter = new PersonStringNlm30NameSchemaFilter($assocType, PERSON_STRING_FILTER_MULTIPLE);
assert($personStringFilter->supportsAsInput($fieldValue));
$typedFieldValues =& $personStringFilter->execute($fieldValue);
break;
default:
$typedFieldValues = array($fieldValue);
}
// Inject data into the meta-data description and thereby
// implicitly validate the field value.
foreach ($typedFieldValues as $typedFieldValue) {
if (!$metadataDescription->addStatement($propertyName, $typedFieldValue)) {
// Add form field error
$this->addError($fieldName, __($property->getValidationMessage()));
$this->addErrorField($fieldName);
}
unset($typedFieldValue);
}
unset($typedFieldValues);
}
}
// Inject the meta-data into the citation.
$citation->injectMetadata($metadataDescription);
// Save the meta-data description for later usage.
$this->_metadataDescriptions[] =& $metadataDescription;
unset($metadataDescription);
}
return $this->isValid();
}
/**
* Post processes an NLM meta-data array
* @param $preliminaryNlm30Array array
* @return array
*/
function &postProcessMetadataArray(&$preliminaryNlm30Array)
{
// Clean array
$preliminaryNlm30Array = arrayClean($preliminaryNlm30Array);
// Trim punctuation
$preliminaryNlm30Array =& $this->_recursivelyTrimPunctuation($preliminaryNlm30Array);
// Parse (=filter) author/editor strings into NLM name descriptions
foreach (array('author' => ASSOC_TYPE_AUTHOR, 'editor' => ASSOC_TYPE_EDITOR) as $personType => $personAssocType) {
if (isset($preliminaryNlm30Array[$personType])) {
// Get the author/editor strings from the result
$personStrings = $preliminaryNlm30Array[$personType];
unset($preliminaryNlm30Array[$personType]);
// Parse the author/editor strings into NLM name descriptions
// Interpret a scalar as a textual authors list
if (is_scalar($personStrings)) {
$personStringFilter = new PersonStringNlm30NameSchemaFilter($personAssocType, PERSON_STRING_FILTER_MULTIPLE);
$persons =& $personStringFilter->execute($personStrings);
} else {
$personStringFilter = new PersonStringNlm30NameSchemaFilter($personAssocType, PERSON_STRING_FILTER_SINGLE);
$persons =& array_map(array($personStringFilter, 'execute'), $personStrings);
}
$preliminaryNlm30Array['person-group[@person-group-type="' . $personType . '"]'] = $persons;
unset($persons);
}
}
// Join comments
if (isset($preliminaryNlm30Array['comment']) && is_array($preliminaryNlm30Array['comment'])) {
// Implode comments from the result into a single string
// as required by the NLM citation schema.
$preliminaryNlm30Array['comment'] = implode("\n", $preliminaryNlm30Array['comment']);
}
// Normalize date strings
foreach (array('date', 'conf-date', 'access-date') as $dateProperty) {
if (isset($preliminaryNlm30Array[$dateProperty])) {
$dateFilter = new DateStringNormalizerFilter();
$preliminaryNlm30Array[$dateProperty] = $dateFilter->execute($preliminaryNlm30Array[$dateProperty]);
}
}
// Cast strings to integers where necessary
foreach (array('fpage', 'lpage', 'size') as $integerProperty) {
if (isset($preliminaryNlm30Array[$integerProperty]) && is_numeric($preliminaryNlm30Array[$integerProperty])) {
$preliminaryNlm30Array[$integerProperty] = (int) $preliminaryNlm30Array[$integerProperty];
}
}
// Rename elements that are stored in attributes in NLM citation
$elementToAttributeMap = array('access-date' => 'date-in-citation[@content-type="access-date"]', 'issn-ppub' => 'issn[@pub-type="ppub"]', 'issn-epub' => 'issn[@pub-type="epub"]', 'pub-id-doi' => 'pub-id[@pub-id-type="doi"]', 'pub-id-publisher-id' => 'pub-id[@pub-id-type="publisher-id"]', 'pub-id-coden' => 'pub-id[@pub-id-type="coden"]', 'pub-id-sici' => 'pub-id[@pub-id-type="sici"]', 'pub-id-pmid' => 'pub-id[@pub-id-type="pmid"]', 'publication-type' => '[@publication-type]');
foreach ($elementToAttributeMap as $elementName => $nlm30PropertyName) {
if (isset($preliminaryNlm30Array[$elementName])) {
$preliminaryNlm30Array[$nlm30PropertyName] = $preliminaryNlm30Array[$elementName];
unset($preliminaryNlm30Array[$elementName]);
}
}
// Guess a publication type if none has been set by the
// citation service.
$this->_guessPublicationType($preliminaryNlm30Array);
// Some services return the title as article-title although
// the publication type is a book.
if (isset($preliminaryNlm30Array['[@publication-type]']) && $preliminaryNlm30Array['[@publication-type]'] == 'book') {
if (isset($preliminaryNlm30Array['article-title']) && !isset($preliminaryNlm30Array['source'])) {
$preliminaryNlm30Array['source'] = $preliminaryNlm30Array['article-title'];
unset($preliminaryNlm30Array['article-title']);
}
}
return $preliminaryNlm30Array;
}
