public function testCitationCrud()
{
$nameSchema = new NlmNameSchema();
$nameDescription = new MetadataDescription($nameSchema, ASSOC_TYPE_AUTHOR);
$nameDescription->addStatement('given-names', $value = 'Peter');
$nameDescription->addStatement('given-names', $value = 'B');
$nameDescription->addStatement('surname', $value = 'Bork');
$nameDescription->addStatement('prefix', $value = 'Mr.');
$citationSchema = new NlmCitationSchema();
$citationDescription = new MetadataDescription($citationSchema, ASSOC_TYPE_CITATION);
$citationDescription->addStatement('person-group[@person-group-type="author"]', $nameDescription);
$citationDescription->addStatement('article-title', $value = 'PHPUnit in a nutshell', 'en_US');
$citationDescription->addStatement('article-title', $value = 'PHPUnit in Kürze', 'de_DE');
$citationDescription->addStatement('date', $value = '2009-08-17');
$citationDescription->addStatement('size', $value = 320);
$citationDescription->addStatement('uri', $value = 'http://phpunit.org/nutshell');
$citation = new Citation('raw citation');
$citation->setAssocType(ASSOC_TYPE_ARTICLE);
$citation->setAssocId(5);
$citation->setEditedCitation('edited citation');
$citation->setParseScore(50);
$citation->injectMetadata($citationDescription);
$citationId = $this->citationDAO->insertCitation($citation);
self::assertTrue(is_numeric($citationId));
self::assertTrue($citationId > 0);
}
public function testCitationCrud()
{
$nameSchema = new NlmNameSchema();
$nameDescription = new MetadataDescription($nameSchema, ASSOC_TYPE_AUTHOR);
$nameDescription->addStatement('given-names', $value = 'Peter');
$nameDescription->addStatement('given-names', $value = 'B');
$nameDescription->addStatement('surname', $value = 'Bork');
$nameDescription->addStatement('prefix', $value = 'Mr.');
$citationSchema = new NlmCitationSchema();
$citationDescription = new MetadataDescription($citationSchema, ASSOC_TYPE_CITATION);
$citationDescription->addStatement('person-group[@person-group-type="author"]', $nameDescription);
$citationDescription->addStatement('article-title', $value = 'PHPUnit in a nutshell', 'en_US');
$citationDescription->addStatement('article-title', $value = 'PHPUnit in Kürze', 'de_DE');
$citationDescription->addStatement('date', $value = '2009-08-17');
$citationDescription->addStatement('size', $value = 320);
$citationDescription->addStatement('uri', $value = 'http://phpunit.org/nutshell');
$citation = new Citation('raw citation');
$citation->setAssocType(ASSOC_TYPE_ARTICLE);
$citation->setAssocId(999999);
$citation->setEditedCitation('edited citation');
$citation->setParseScore(50);
$citation->injectMetadata($citationDescription);
// Create citation
$citationId = $this->citationDAO->insertCitation($citation);
self::assertTrue(is_numeric($citationId));
self::assertTrue($citationId > 0);
// Retrieve citation
$citationById = $this->citationDAO->getCitation($citationId);
$citationById->getMetadataFieldNames();
// Initializes internal state for comparison.
self::assertEquals($citation, $citationById);
$citationsByAssocIdDaoFactory = $this->citationDAO->getCitationsByAssocId(ASSOC_TYPE_ARTICLE, 999999);
$citationsByAssocId = $citationsByAssocIdDaoFactory->toArray();
self::assertEquals(1, count($citationsByAssocId));
$citationsByAssocId[0]->getMetadataFieldNames();
// Initializes internal state for comparison.
self::assertEquals($citation, $citationsByAssocId[0]);
// Update citation
$citationDescription->removeStatement('date');
$citationDescription->addStatement('article-title', $value = 'PHPUnit rápido', 'pt_BR');
$updatedCitation = new Citation('another raw citation');
$updatedCitation->setId($citationId);
$updatedCitation->setAssocType(ASSOC_TYPE_ARTICLE);
$updatedCitation->setAssocId(999998);
$updatedCitation->setEditedCitation('another edited citation');
$updatedCitation->setParseScore(50);
$updatedCitation->injectMetadata($citationDescription);
$this->citationDAO->updateCitation($updatedCitation);
$citationAfterUpdate = $this->citationDAO->getCitation($citationId);
$citationAfterUpdate->getMetadataFieldNames();
// Initializes internal state for comparison.
self::assertEquals($updatedCitation, $citationAfterUpdate);
// Delete citation
$this->citationDAO->deleteCitationsByAssocId(ASSOC_TYPE_ARTICLE, 999998);
self::assertNull($this->citationDAO->getCitation($citationId));
}
/**
* Inject the given meta-data into a new citation object.
* @param $metadataDescription MetadataDescription
* @return Citation
*/
protected function &getCitation($citationDescription)
{
// Instantiate the citation and inject the meta-data.
import('lib.pkp.classes.citation.Citation');
$citation = new Citation('raw citation');
$citation->setAssocType($this->assocType);
$citation->setAssocId($this->assocId);
$citation->injectMetadata($citationDescription);
return $citation;
}
/**
* @covers CitationDAO
*/
public function testCitationCrud()
{
$citationDao = DAORegistry::getDAO('CitationDAO');
/* @var $citationDao CitationDAO */
$nameSchemaName = 'lib.pkp.plugins.metadata.nlm30.schema.Nlm30NameSchema';
$nameDescription = new MetadataDescription($nameSchemaName, ASSOC_TYPE_AUTHOR);
$nameDescription->addStatement('given-names', $value = 'Peter');
$nameDescription->addStatement('given-names', $value = 'B');
$nameDescription->addStatement('surname', $value = 'Bork');
$nameDescription->addStatement('prefix', $value = 'Mr.');
$citationSchemaName = 'lib.pkp.plugins.metadata.nlm30.schema.Nlm30CitationSchema';
$citationDescription = new MetadataDescription($citationSchemaName, ASSOC_TYPE_CITATION);
$citationDescription->addStatement('person-group[@person-group-type="author"]', $nameDescription);
$citationDescription->addStatement('article-title', $value = 'PHPUnit in a nutshell', 'en_US');
$citationDescription->addStatement('article-title', $value = 'PHPUnit in Kürze', 'de_DE');
$citationDescription->addStatement('date', $value = '2009-08-17');
$citationDescription->addStatement('size', $value = 320);
$citationDescription->addStatement('uri', $value = 'http://phpunit.org/nutshell');
// Add a simple source description.
$sourceDescription = new MetadataDescription($citationSchemaName, ASSOC_TYPE_CITATION);
$sourceDescription->setDisplayName('test');
$sourceDescription->addStatement('article-title', $value = 'a simple source description', 'en_US');
$sourceDescription->setSeq(0);
$citation = new Citation('raw citation');
$citation->setAssocType(ASSOC_TYPE_ARTICLE);
$citation->setAssocId(999999);
$citation->setSeq(50);
$citation->addSourceDescription($sourceDescription);
$citation->injectMetadata($citationDescription);
// Create citation.
$citationId = $citationDao->insertObject($citation);
self::assertTrue(is_numeric($citationId));
self::assertTrue($citationId > 0);
// Retrieve citation.
$citationById = $citationDao->getObjectById($citationId);
// Fix state differences for comparison.
$citation->removeSupportedMetadataAdapter($citationSchemaName);
$citationById->removeSupportedMetadataAdapter($citationSchemaName);
$citationById->_extractionAdaptersLoaded = true;
$citationById->_injectionAdaptersLoaded = true;
$sourceDescription->setAssocId($citationId);
$sourceDescription->removeSupportedMetadataAdapter($citationSchemaName);
$sourceDescriptions = $citationById->getSourceDescriptions();
$sourceDescriptions['test']->getMetadataSchema();
// this will instantiate the meta-data schema internally.
self::assertEquals($citation, $citationById);
$citationsByAssocIdDaoFactory = $citationDao->getObjectsByAssocId(ASSOC_TYPE_ARTICLE, 999999);
$citationsByAssocId = $citationsByAssocIdDaoFactory->toArray();
self::assertEquals(1, count($citationsByAssocId));
// Fix state differences for comparison.
$citationsByAssocId[0]->_extractionAdaptersLoaded = true;
$citationsByAssocId[0]->_injectionAdaptersLoaded = true;
$citationsByAssocId[0]->removeSupportedMetadataAdapter($citationSchemaName);
$sourceDescriptionsByAssocId = $citationsByAssocId[0]->getSourceDescriptions();
$sourceDescriptionsByAssocId['test']->getMetadataSchema();
// this will instantiate the meta-data schema internally.
self::assertEquals($citation, $citationsByAssocId[0]);
// Update citation.
$citationDescription->removeStatement('date');
$citationDescription->addStatement('article-title', $value = 'PHPUnit rápido', 'pt_BR');
// Update source descriptions.
$sourceDescription->addStatement('article-title', $value = 'edited source description', 'en_US', true);
$updatedCitation = new Citation('another raw citation');
$updatedCitation->setId($citationId);
$updatedCitation->setAssocType(ASSOC_TYPE_ARTICLE);
$updatedCitation->setAssocId(999998);
$updatedCitation->setSeq(50);
$updatedCitation->addSourceDescription($sourceDescription);
$updatedCitation->injectMetadata($citationDescription);
$citationDao->updateObject($updatedCitation);
$citationAfterUpdate = $citationDao->getObjectById($citationId);
// Fix state differences for comparison.
$updatedCitation->removeSupportedMetadataAdapter($citationSchemaName);
$citationAfterUpdate->removeSupportedMetadataAdapter($citationSchemaName);
$citationAfterUpdate->_extractionAdaptersLoaded = true;
$citationAfterUpdate->_injectionAdaptersLoaded = true;
$sourceDescriptionsAfterUpdate = $citationAfterUpdate->getSourceDescriptions();
$sourceDescriptionsAfterUpdate['test']->getMetadataSchema();
// this will instantiate the meta-data schema internally.
$sourceDescription->removeSupportedMetadataAdapter($citationSchemaName);
self::assertEquals($updatedCitation, $citationAfterUpdate);
// Delete citation
$citationDao->deleteObjectsByAssocId(ASSOC_TYPE_ARTICLE, 999998);
self::assertNull($citationDao->getObjectById($citationId));
}
/**
* Take an array of citation parse/lookup results and derive a citation
* with one "best" set of values.
*
* We determine the best values within the citations that have a score above
* the given threshold. Citations with a score below the threshold will be
* ignored.
*
* For these citations we count the frequency of values per meta-data property.
* The most frequent value will be chosen as "best" value.
*
* If two values have the same frequency then decide based on the score. If
* this is still ambivalent then return the first of the remaining values.
*
* @param $scoredCitations
* @param $scoreThreshold integer a number between 0 (=no threshold) and 100
* @return Citation one citation with the "best" values set
*/
function &_guessValues(&$scoredCitations, $scoreThreshold)
{
assert($scoreThreshold >= 0 && $scoreThreshold <= 100);
// Create the target citation description.
$targetDescription = new MetadataDescription('lib.pkp.plugins.metadata.nlm30.schema.Nlm30CitationSchema', ASSOC_TYPE_CITATION);
// Step 1: List all values and max scores that have been identified for a given element
// but only include values from results above a given scoring threshold
// Initialize variables for the first step.
$valuesByPropertyName = array();
$maxScoresByPropertyNameAndValue = array();
// Sort the scored citations by score with the highest score first.
krsort($scoredCitations);
foreach ($scoredCitations as $currentScore => $citationsForCurrentScore) {
// Check whether the current score is below the threshold, if so
// stop the loop. We've sorted our citations by score so the remaining
// citations all have scores below the threshold and we can forget
// about them.
if ($currentScore < $scoreThreshold) {
break;
}
foreach ($citationsForCurrentScore as $citationForCurrentScore) {
$statements = $citationForCurrentScore->getStatements();
// Add the property values and scores of this citation
// to the overall property lists
foreach ($statements as $propertyName => $value) {
// Initialize sub-arrays if necessary
if (!isset($valuesByPropertyName[$propertyName])) {
$valuesByPropertyName[$propertyName] = array();
}
if (!isset($maxScoresByPropertyNameAndValue[$propertyName])) {
$maxScoresByPropertyNameAndValue[$propertyName] = array();
}
// Add the value for the given property, as we want to count
// value frequencies later, we explicitly allow duplicates.
$serializedValue = serialize($value);
$valuesByPropertyName[$propertyName][] = $serializedValue;
// As we have ordered our citations descending by score, the
// first score found for a value is also the maximum score.
if (!isset($maxScoresByPropertyNameAndValue[$propertyName][$serializedValue])) {
$maxScoresByPropertyNameAndValue[$propertyName][$serializedValue] = $currentScore;
}
}
}
}
// Step 2: Find out the values that occur most frequently for each element
// and order these by score.
foreach ($valuesByPropertyName as $propertyName => $values) {
// Count the occurrences of each value within the given element
$valueFrequencies = array_count_values($values);
// Order the most frequent values to the beginning of the array
arsort($valueFrequencies);
// Get the most frequent values (may be several if there are more than one
// with the same frequency).
$scoresOfMostFrequentValues = array();
$previousValueFrequency = 0;
foreach ($valueFrequencies as $value => $valueFrequency) {
// Only extract the most frequent values, jump out of the
// loop when less frequent values start.
if ($previousValueFrequency > $valueFrequency) {
break;
}
$previousValueFrequency = $valueFrequency;
$scoresOfMostFrequentValues[$value] = $maxScoresByPropertyNameAndValue[$propertyName][$value];
}
// Now we can order the most frequent values by score, starting
// with the highest score.
arsort($scoresOfMostFrequentValues);
// Now get the first key which represents the value with the
// highest frequency and the highest score.
reset($scoresOfMostFrequentValues);
$bestValue = unserialize(key($scoresOfMostFrequentValues));
// Set the found "best" element value in the result citation.
$statements = array($propertyName => $bestValue);
$success = $targetDescription->setStatements($statements);
assert($success);
}
// Instantiate the target citation
$targetCitation = new Citation();
$targetCitation->injectMetadata($targetDescription);
return $targetCitation;
}
