/**
* Returns id for the row with the specified name (and type) or idno (regardless of specified type.) If the row does not already
* exist then it will be created with the specified name, type and locale, as well as with any specified values in the $pa_values array.
* $pa_values keys should be either valid entity fields or attributes.
*
* @param string $ps_table The table to match and/or create rows in
* @param array $pa_label Array with values for row label
* @param int $pn_parent_id
* @param int $pn_type_id The type_id or code of the type to use if the row needs to be created
* @param int $pn_locale_id The locale_id to use if the row needs to be created (will be used for both the row locale as well as the label locale)
* @param array $pa_values An optional array of additional values to populate newly created rows with. These values are *only* used for newly created rows; they will not be applied if the row named already exists unless the forceUpdate option is set, in which case attributes (but not intrinsics) will be updated. The array keys should be names of fields or valid attributes. Values should be either a scalar (for single-value attributes) or an array of values for (multi-valued attributes)
* @param array $pa_options An optional array of options, which include:
* outputErrors - if true, errors will be printed to console [default=false]
* dontCreate - if true then new entities will not be created [default=false]
* matchOn = optional list indicating sequence of checks for an existing record; values of array can be "label" and "idno". Ex. array("idno", "label") will first try to match on idno and then label if the first match fails. For entities only you may also specifiy "displayname", "surname" and "forename" to match on the text of the those label fields exclusively.
* matchOnDisplayName if true then entities are looked up exclusively using displayname, otherwise forename and surname fields are used [default=false]
* transaction - if Transaction instance is passed, use it for all Db-related tasks [default=null]
* returnInstance = return ca_entities instance rather than entity_id. Default is false.
* generateIdnoWithTemplate = A template to use when setting the idno. The template is a value with automatically-set SERIAL values replaced with % characters. Eg. 2012.% will set the created row's idno value to 2012.121 (assuming that 121 is the next number in the serial sequence.) The template is NOT used if idno is passed explicitly as a value in $pa_values.
* importEvent = if ca_data_import_events instance is passed then the insert/update of the entity will be logged as part of the import
* importEventSource = if importEvent is passed, then the value set for importEventSource is used in the import event log as the data source. If omitted a default value of "?" is used
* nonPreferredLabels = an optional array of nonpreferred labels to add to any newly created entities. Each label in the array is an array with required entity label values.
* forceUpdate = update attributes set in $pa_values even if row already exists. [Default=false; no values are updated in existing rows]
* matchMediaFilesWithoutExtension = For ca_object_representations, if media path is invalid, attempt to find media in referenced directory and sub-directories that has a matching name, regardless of file extension. [default=false]
* log = if KLogger instance is passed then actions will be logged
* ignoreParent = Don't take into account parent_id value when looking for matching rows [Default is false]
* @return bool|BaseModel|mixed|null
*/
private static function _getID($ps_table, $pa_label, $pn_parent_id, $pn_type_id, $pn_locale_id, $pa_values = null, $pa_options = null)
{
if (!is_array($pa_options)) {
$pa_options = array();
}
$o_dm = Datamodel::load();
/** @var KLogger $o_log */
$o_log = isset($pa_options['log']) && $pa_options['log'] instanceof KLogger ? $pa_options['log'] : null;
if (!($t_instance = $o_dm->getInstanceByTableName($ps_table, true))) {
return null;
}
$vs_table_display_name = $t_instance->getProperty('NAME_SINGULAR');
$vs_table_class = $t_instance->tableName();
$vs_label_display_fld = $t_instance->getLabelDisplayField();
$vs_label = $pa_label[$vs_label_display_fld];
$pb_output_errors = caGetOption('outputErrors', $pa_options, false);
$pb_match_on_displayname = caGetOption('matchOnDisplayName', $pa_options, false);
$pa_match_on = caGetOption('matchOn', $pa_options, array('label', 'idno', 'displayname'), array('castTo' => "array"));
$ps_event_source = caGetOption('importEventSource', $pa_options, '?');
$pb_match_media_without_ext = caGetOption('matchMediaFilesWithoutExtension', $pa_options, false);
$pb_ignore_parent = caGetOption('ignoreParent', $pa_options, false);
$vn_parent_id = $pn_parent_id ? $pn_parent_id : caGetOption('parent_id', $pa_values, null);
if (!$vn_parent_id) {
$vn_parent_id = null;
}
$vs_idno_fld = $t_instance->getProperty('ID_NUMBERING_ID_FIELD');
$vs_idno = caGetOption($vs_idno_fld, $pa_values, null);
/** @var ca_data_import_events $o_event */
$o_event = isset($pa_options['importEvent']) && $pa_options['importEvent'] instanceof ca_data_import_events ? $pa_options['importEvent'] : null;
if (isset($pa_options['transaction']) && $pa_options['transaction'] instanceof Transaction) {
$t_instance->setTransaction($pa_options['transaction']);
if ($o_event) {
$o_event->setTransaction($pa_options['transaction']);
}
}
if (preg_match('!\\%!', $vs_idno)) {
$pa_options['generateIdnoWithTemplate'] = $vs_idno;
$vs_idno = null;
}
if (!$vs_idno) {
if (isset($pa_options['generateIdnoWithTemplate']) && $pa_options['generateIdnoWithTemplate']) {
$pa_values[$vs_idno_fld] = $vs_idno = $t_instance->setIdnoWithTemplate($pa_options['generateIdnoWithTemplate'], array('dontSetValue' => true));
}
}
$va_regex_list = $va_replacements_list = null;
if ($vs_table_class == 'ca_object_representations') {
// Get list of regular expressions that user can use to transform file names to match object idnos
$va_regex_list = caBatchGetMediaFilenameToIdnoRegexList(array('log' => $o_log));
// Get list of replacements that user can use to transform file names to match object idnos
$va_replacements_list = caBatchGetMediaFilenameReplacementRegexList(array('log' => $o_log));
}
$vn_id = null;
foreach ($pa_match_on as $vs_match_on) {
switch (strtolower($vs_match_on)) {
case 'idno':
if ($vs_idno == '%') {
break;
}
// don't try to match on an unreplaced idno placeholder
switch ($vs_table_class) {
case 'ca_object_representations':
//
// idno lookups for representations use media batch importer rules
//
$va_idnos_to_match = array($vs_idno);
if (is_array($va_replacements_list)) {
foreach ($va_replacements_list as $vs_replacement_code => $va_replacement) {
if (isset($va_replacement['search']) && is_array($va_replacement['search'])) {
$va_replace = caGetOption('replace', $va_replacement);
$va_search = array();
foreach ($va_replacement['search'] as $vs_search) {
$va_search[] = '!' . $vs_search . '!';
}
if ($vs_idno_proc = @preg_replace($va_search, $va_replace, $vs_idno)) {
$va_idnos_to_match[] = $vs_idno_proc;
}
}
}
}
if (is_array($va_regex_list) && sizeof($va_regex_list)) {
foreach ($va_regex_list as $vs_regex_name => $va_regex_info) {
foreach ($va_regex_info['regexes'] as $vs_regex) {
foreach ($va_idnos_to_match as $vs_idno_match) {
if (!$vs_idno_match) {
continue;
}
if (preg_match('!' . $vs_regex . '!', $vs_idno_match, $va_matches)) {
if ($vn_id = ca_object_representations::find(array('idno' => $va_matches[1]), array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction']))) {
break 5;
}
}
}
}
}
} else {
foreach ($va_idnos_to_match as $vs_idno_match) {
if (!$vs_idno_match) {
continue;
}
if ($vn_id = ca_object_representations::find(array('idno' => $vs_idno_match), array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction']))) {
break 3;
}
}
}
break;
default:
//
// Standard idno lookup for most tables
//
$va_find_vals = array($vs_idno_fld => $vs_idno ? $vs_idno : ($pa_label['_originalText'] ? $pa_label['_originalText'] : $vs_label), 'type_id' => $pn_type_id);
if (!$pb_ignore_parent && isset($pa_values['parent_id'])) {
$va_find_vals['parent_id'] = $pa_values['parent_id'];
}
if (($vs_idno || trim($pa_label['_originalText'] || $vs_label)) && ($vn_id = $vs_table_class::find($va_find_vals, array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction'])))) {
break 2;
}
break;
}
break;
case 'label':
case 'labels':
if ($pb_match_on_displayname && strlen(trim($pa_label['displayname'])) > 0) {
// entities only
$va_params = array('preferred_labels' => array('displayname' => $pa_label['displayname']), 'type_id' => $pn_type_id);
if (!$pb_ignore_parent && isset($pa_values['parent_id'])) {
$va_params['parent_id'] = $vn_parent_id;
}
$vn_id = $vs_table_class::find($va_params, array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction']));
} elseif ($vs_table_class == 'ca_entities') {
// entities only
$va_params = array('preferred_labels' => array('forename' => $pa_label['forename'], 'middlename' => $pa_label['middlename'], 'surname' => $pa_label['surname']), 'type_id' => $pn_type_id);
if (!$pb_ignore_parent) {
$va_params['parent_id'] = $vn_parent_id;
}
$vn_id = $vs_table_class::find($va_params, array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction']));
} else {
$va_params = array('preferred_labels' => array($vs_label_display_fld => $pa_label[$vs_label_display_fld]), 'type_id' => $pn_type_id);
if (!$pb_ignore_parent && isset($pa_values['parent_id'])) {
$va_params['parent_id'] = $vn_parent_id;
}
$vn_id = $vs_table_class::find($va_params, array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction']));
}
if ($vn_id) {
break 2;
}
break;
//
// For entities only
//
//
// For entities only
//
case 'surname':
$va_params = array('preferred_labels' => array('surname' => $pa_label['surname']), 'type_id' => $pn_type_id);
if (!$pb_ignore_parent && isset($pa_values['parent_id'])) {
$va_params['parent_id'] = $vn_parent_id;
}
$vn_id = $vs_table_class::find($va_params, array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction']));
if ($vn_id) {
break 2;
}
break;
case 'forename':
$va_params = array('preferred_labels' => array('forename' => $pa_label['forename']), 'type_id' => $pn_type_id);
if (!$pb_ignore_parent && isset($pa_values['parent_id'])) {
$va_params['parent_id'] = $vn_parent_id;
}
$vn_id = $vs_table_class::find($va_params, array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction']));
if ($vn_id) {
break 2;
}
break;
case 'displayname':
$va_params = array('preferred_labels' => array('displayname' => $pa_label['displayname']), 'type_id' => $pn_type_id);
if (!$pb_ignore_parent && isset($pa_values['parent_id'])) {
$va_params['parent_id'] = $vn_parent_id;
}
$vn_id = $vs_table_class::find($va_params, array('returnAs' => 'firstId', 'purifyWithFallback' => true, 'transaction' => $pa_options['transaction']));
if ($vn_id) {
break 2;
}
break;
}
}
if (!$vn_id) {
//
// Create new row
//
if (caGetOption('dontCreate', $pa_options, false)) {
return false;
}
if ($o_event) {
$o_event->beginItem($ps_event_source, $vs_table_class, 'I');
}
// If we're creating a new item, it's probably a good idea to *NOT* use a
// BaseModel instance from cache, because those cannot change their type_id
if (!($t_instance = $o_dm->getInstanceByTableName($ps_table, false))) {
return null;
}
if (isset($pa_options['transaction']) && $pa_options['transaction'] instanceof Transaction) {
$t_instance->setTransaction($pa_options['transaction']);
}
$t_instance->setMode(ACCESS_WRITE);
$t_instance->set('locale_id', $pn_locale_id);
$t_instance->set('type_id', $pn_type_id);
$va_intrinsics = array('source_id' => null, 'access' => 0, 'status' => 0, 'lifespan' => null, 'parent_id' => $vn_parent_id, 'lot_status_id' => null, '_interstitial' => null);
if ($vs_hier_id_fld = $t_instance->getProperty('HIERARCHY_ID_FLD')) {
$va_intrinsics[$vs_hier_id_fld] = null;
}
if ($vs_idno_fld) {
$va_intrinsics[$vs_idno_fld] = $vs_idno ? $vs_idno : null;
}
foreach ($va_intrinsics as $vs_fld => $vm_fld_default) {
if ($t_instance->hasField($vs_fld)) {
$t_instance->set($vs_fld, caGetOption($vs_fld, $pa_values, $vm_fld_default));
}
unset($pa_values[$vs_fld]);
}
if ($t_instance->hasField('media') && $t_instance->getFieldInfo('media', 'FIELD_TYPE') == FT_MEDIA && isset($pa_values['media']) && $pa_values['media']) {
if (is_array($pa_values['media'])) {
$pa_values['media'] = array_shift($pa_values['media']);
}
if ($pb_match_media_without_ext && !isURL($pa_values['media']) && !file_exists($pa_values['media'])) {
$vs_dirname = pathinfo($pa_values['media'], PATHINFO_DIRNAME);
$vs_filename = preg_replace('!\\.[A-Za-z0-9]{1,4}$!', '', pathinfo($pa_values['media'], PATHINFO_BASENAME));
$vs_original_path = $pa_values['media'];
$pa_values['media'] = null;
$va_files_in_dir = caGetDirectoryContentsAsList($vs_dirname, true, false, false, false);
foreach ($va_files_in_dir as $vs_filepath) {
if ($o_log) {
$o_log->logDebug(_t("Trying media %1 in place of %2/%3", $vs_filepath, $vs_original_path, $vs_filename));
}
if (pathinfo($vs_filepath, PATHINFO_FILENAME) == $vs_filename) {
if ($o_log) {
$o_log->logNotice(_t("Found media %1 for %2/%3", $vs_filepath, $vs_original_path, $vs_filename));
}
$pa_values['media'] = $vs_filepath;
break;
}
}
}
$t_instance->set('media', $pa_values['media']);
}
$t_instance->insert();
if ($t_instance->numErrors()) {
if ($pb_output_errors) {
print "[Error] " . _t("Could not insert %1 %2: %3", $vs_table_display_name, $pa_label[$vs_label_display_fld], join('; ', $t_instance->getErrors())) . "\n";
}
if ($o_log) {
$o_log->logError(_t("Could not insert %1 %2: %3", $vs_table_display_name, $pa_label[$vs_label_display_fld], join('; ', $t_instance->getErrors())));
}
return null;
}
$vb_label_errors = false;
$t_instance->addLabel($pa_label, $pn_locale_id, null, true);
if ($t_instance->numErrors()) {
if ($pb_output_errors) {
print "[Error] " . _t("Could not set preferred label for %1 %2: %3", $vs_table_display_name, $pa_label[$vs_label_display_fld], join('; ', $t_instance->getErrors())) . "\n";
}
if ($o_log) {
$o_log->logError(_t("Could not set preferred label for %1 %2: %3", $vs_table_display_name, $pa_label[$vs_label_display_fld], join('; ', $t_instance->getErrors())));
}
$vb_label_errors = true;
}
DataMigrationUtils::_setIdno($t_instance, $vs_idno, $pa_options);
$vb_attr_errors = !DataMigrationUtils::_setAttributes($t_instance, $pn_locale_id, $pa_values, $pa_options);
DataMigrationUtils::_setNonPreferredLabels($t_instance, $pn_locale_id, $pa_options);
$vn_id = $t_instance->getPrimaryKey();
if ($o_event) {
if ($vb_attr_errors || $vb_label_errors) {
$o_event->endItem($vn_id, __CA_DATA_IMPORT_ITEM_PARTIAL_SUCCESS__, _t("Errors setting field values: %1", join('; ', $t_instance->getErrors())));
} else {
$o_event->endItem($vn_id, __CA_DATA_IMPORT_ITEM_SUCCESS__, '');
}
}
if ($o_log) {
$o_log->logInfo(_t("Created new %1 %2", $vs_table_display_name, $pa_label[$vs_label_display_fld]));
}
if (isset($pa_options['returnInstance']) && $pa_options['returnInstance']) {
return $t_instance;
}
} else {
if ($o_event) {
$o_event->beginItem($ps_event_source, $vs_table_class, 'U');
}
if ($o_log) {
$o_log->logDebug(_t("Found existing %1 %2 in DataMigrationUtils::_getID()", $vs_table_display_name, $pa_label[$vs_label_display_fld]));
}
$vb_attr_errors = false;
if (($vb_force_update = caGetOption('forceUpdate', $pa_options, false)) || ($vb_return_instance = caGetOption('returnInstance', $pa_options, false))) {
if (!($t_instance = $o_dm->getInstanceByTableName($vs_table_class, false))) {
return null;
}
if (isset($pa_options['transaction']) && $pa_options['transaction'] instanceof Transaction) {
$t_instance->setTransaction($pa_options['transaction']);
}
$vb_has_attr = false;
if ($vb_force_update) {
foreach ($pa_values as $vs_element => $va_values) {
if ($t_instance->hasElement($vs_element)) {
$vb_has_attr = true;
break;
}
}
}
if ($vb_return_instance || $vb_force_update && $vb_has_attr) {
$vn_rc = $t_instance->load($vn_id);
} else {
$vn_rc = true;
}
if (!$vn_rc) {
if ($o_log) {
$o_log->logError(_t("Could not load existing %1 with id %2 (%3) in DataMigrationUtils::_getID() [THIS SHOULD NOT HAPPEN]", $vs_table_display_name, $vn_id, $pa_label[$vs_label_display_fld]));
}
return null;
} else {
if ($vb_force_update && $vb_has_attr) {
if ($vb_attr_errors = !DataMigrationUtils::_setAttributes($t_instance, $pn_locale_id, $pa_values, $pa_options)) {
if ($o_log) {
$o_log->logError(_t("Could not set attributes for %1 with id %2 (%3) in DataMigrationUtils::_getID(): %4", $vs_table_display_name, $vn_id, $pa_label[$vs_label_display_fld], join("; ", $t_instance->getErrors())));
}
}
}
if ($o_event) {
if ($vb_attr_errors) {
$o_event->endItem($vn_id, __CA_DATA_IMPORT_ITEM_PARTIAL_SUCCESS__, _t("Errors setting field values: %1", join('; ', $t_instance->getErrors())));
} else {
$o_event->endItem($vn_id, __CA_DATA_IMPORT_ITEM_SUCCESS__, '');
}
}
if ($vb_return_instance) {
return $t_instance;
}
}
}
if ($o_event) {
$o_event->endItem($vn_id, __CA_DATA_IMPORT_ITEM_SUCCESS__, '');
}
}
return $vn_id;
}
