Code
/
islandora
7
0
Fork 0
Code Issues Pull Requests 1 Projects Releases Wiki Activity
Browse Source

Add Solr option for Orphaned Objects list

pull/807/head
Brandon Weigel 4 years ago
parent
5e88176861
commit
064b115024
2 changed files with 122 additions and 35 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 10
      includes/admin.form.inc
  2. 147
      includes/orphaned_objects.inc

10
includes/admin.form.inc
Unescape View File

@ -127,6 +127,16 @@ function islandora_repository_admin(array $form, array &$form_state) {
),
),
),
'islandora_orphaned_objects_backend' => array(
'#type' => 'radios',
'#title' => t('Orphaned Objects query'),
'#description' => t('How the Orphaned Islandora Objects list is generated.'),
'#default_value' => variable_get('islandora_orphaned_objects_backend', 'Solr'),
'#options' => array(
'Solr' => t('Solr'),
'SPARQL' => t('SPARQL'),
),
),
'islandora_risearch_use_itql_when_necessary' => array(
'#type' => 'checkbox',
'#title' => t('Use iTQL for particular queries'),

147
includes/orphaned_objects.inc
Unescape View File

@ -54,11 +54,23 @@ function islandora_manage_orphaned_objects_form(array $form, array $form_state)
or a variety of other reasons. Some of these orphans may exist intentionally.
Please be cautious when deleting, as this action is irreversible.'), 'warning');
$orphaned_objects = islandora_get_orphaned_objects();
$query_method = variable_get('islandora_orphaned_objects_backend', 'Solr');
$rows = array();
foreach ($orphaned_objects as $orphaned_object) {
$pid = $orphaned_object['object']['value'];
if ($query_method == 'SPARQL') {
$pid = $orphaned_object['object']['value'];
}
elseif ($query_method == 'Solr') {
$pid = $orphaned_object['PID'];
}
if (islandora_namespace_accessible($pid)) {
$rows[$pid] = array(l($orphaned_object['title']['value'] . " (" . $pid . ")", "islandora/object/$pid"));
if ($query_method == 'SPARQL') {
$rows[$pid] = array(l($orphaned_object['title']['value'] . " (" . $pid . ")", "islandora/object/$pid"));
}
elseif ($query_method == 'Solr') {
$rows[$pid] = array(l($orphaned_object['object_label'] . " (" . $pid . ")", "islandora/object/$pid"));
}
}
}
ksort($rows);
@ -144,9 +156,73 @@ function islandora_manage_orphaned_objects_confirm_submit(array $form, array &$f
* An array containing the results of the orphaned objects queries.
*/
function islandora_get_orphaned_objects() {
$connection = islandora_get_tuque_connection();
// SPARQL: get orphaned objects, exclude any with a living parent.
$object_query = <<<EOQ
$query_method = variable_get('islandora_orphaned_objects_backend', 'Solr');
if ($query_method == 'Solr') {
// Step 1: Solr query for the RELS_EXTs
$collection_field = variable_get('islandora_solr_member_of_collection_field', 'RELS_EXT_isMemberOfCollection_uri_ms');
$label_field = variable_get('islandora_solr_object_label_field', 'fgs_label_s');
$member_field = variable_get('islandora_solr_member_of_field', 'RELS_EXT_isMemberOf_uri_ms');
$params = "PID, " . $label_field . ", " . $collection_field . ", " . $member_field;
$query = "PID:*";
$qp = new islandoraSolrQueryProcessor();
$qp->buildQuery($query);
$qp->solrParams['fl'] = $params;
$qp->solrLimit = 1000000000;
// Check islandora_compound_object settings and changes query filters to include compound children if necessary.
if (variable_get('islandora_compound_object_hide_child_objects_solr', TRUE)) {
$fq = variable_get('islandora_compound_object_solr_fq', '-RELS_EXT_isConstituentOf_uri_mt:[* TO *]');
if (!empty($fq)) {
// delete islandora_compound_object_solr_fq from the list of filters
$filters = $qp->solrParams['fq'];
if (($key = array_search($fq, $filters)) !== FALSE) {
unset($filters[$key]);
$qp->solrParams['fq'] = $filters;
}
}
}
$qp->executeQuery(FALSE);
try {
$results = $qp->islandoraSolrResult['response']['objects'];
}
catch (Exception $e) {
watchdog_exception('Islandora', $e, 'Got an exception searching for parent objects .', array(), WATCHDOG_ERROR);
$results = array();
}
$orphaned_objects = array();
// Now we have $results[key]['PID'] for the pid. sparql: [key][object][value] and [key][title][value]
// Step 2: Check all results for PIDs that don't exist
foreach ($results AS $result) {
if (array_key_exists($collection_field, $result['solr_doc'])) {
foreach ($result['solr_doc'][$collection_field] AS $collection) {
$test = islandora_object_load($collection);
if (!$test) {
$orphaned_objects[] = $result;
}
}
}
if (array_key_exists($member_field, $result['solr_doc'])) {
foreach ($result['solr_doc'][$member_field] AS $membership) {
$test = islandora_object_load($membership);
if (!$test) {
$orphaned_objects[] = $result;
}
}
}
}
$results = $orphaned_objects;
}
elseif($query_method == "SPARQL") {
$connection = islandora_get_tuque_connection();
// SPARQL: get orphaned objects, exclude any with a living parent.
$object_query = <<<EOQ
!prefix
SELECT DISTINCT ?object ?title
WHERE {
@ -171,37 +247,38 @@ WHERE {
FILTER (!bound(?liveparent))
} ORDER BY ?object
EOQ;
$parent_relationships = module_invoke_all('islandora_solution_pack_child_relationships', 'all');
$parent_relationships['prefix'] = array_unique($parent_relationships['prefix']);
$parent_relationships['predicate'] = array_unique($parent_relationships['predicate']);
if (count($parent_relationships['predicate']) == 0) {
// No predicates to search for. Exit early.
return array();
}
$optionals = (array) module_invoke('islandora_xacml_api', 'islandora_basic_collection_get_query_optionals', 'view');
$filter_modules = array(
'islandora_xacml_api',
'islandora',
);
$filters = array();
foreach ($filter_modules as $module) {
$filters = array_merge_recursive($filters, (array) module_invoke($module, 'islandora_basic_collection_get_query_filters', 'view'));
$parent_relationships = module_invoke_all('islandora_solution_pack_child_relationships', 'all');
$parent_relationships['prefix'] = array_unique($parent_relationships['prefix']);
$parent_relationships['predicate'] = array_unique($parent_relationships['predicate']);
if (count($parent_relationships['predicate']) == 0) {
// No predicates to search for. Exit early.
return array();
}
$optionals = (array) module_invoke('islandora_xacml_api', 'islandora_basic_collection_get_query_optionals', 'view');
$filter_modules = array(
'islandora_xacml_api',
'islandora',
);
$filters = array();
foreach ($filter_modules as $module) {
$filters = array_merge_recursive($filters, (array) module_invoke($module, 'islandora_basic_collection_get_query_filters', 'view'));
}
$filter_map = function ($filter) {
return "FILTER($filter)";
};
$parent_map = function ($parent) {
return "?object $parent ?liveparent";
};
// Use separate queries for different object types.
$sparql_query_objects = format_string($object_query, array(
'!optionals' => !empty($optionals) ? ('OPTIONAL {{' . implode('} UNION {', $optionals) . '}}') : '',
'!filters' => !empty($filters) ? implode(' ', array_map($filter_map, $filters)) : '',
'!dead_parent_relationships' => '?p = ' . implode(' || ?p = ', $parent_relationships['predicate']),
'!live_parent_relationships' => '{' . implode(' } UNION { ', array_map($parent_map, $parent_relationships['predicate'])) . '}',
'!prefix' => implode("\n", $parent_relationships['prefix']),
));
$results = $connection->repository->ri->sparqlQuery($sparql_query_objects);
}
$filter_map = function ($filter) {
return "FILTER($filter)";
};
$parent_map = function ($parent) {
return "?object $parent ?liveparent";
};
// Use separate queries for different object types.
$sparql_query_objects = format_string($object_query, array(
'!optionals' => !empty($optionals) ? ('OPTIONAL {{' . implode('} UNION {', $optionals) . '}}') : '',
'!filters' => !empty($filters) ? implode(' ', array_map($filter_map, $filters)) : '',
'!dead_parent_relationships' => '?p = ' . implode(' || ?p = ', $parent_relationships['predicate']),
'!live_parent_relationships' => '{' . implode(' } UNION { ', array_map($parent_map, $parent_relationships['predicate'])) . '}',
'!prefix' => implode("\n", $parent_relationships['prefix']),
));
$results = $connection->repository->ri->sparqlQuery($sparql_query_objects);
return $results;
}

Write Preview
Loading…
Cancel
Save