protected function processRow($row)
{
foreach ($row as &$r) {
$r = trim($r);
}
$edit = array('Search Name' => $row['name'], 'Affiliation Name' => $row['affiliation1'], 'Similar Names' => array(), 'New Person' => null, 'Existing Person' => null, 'New Org' => null, 'Existing Org' => null, 'New Relationship' => null);
try {
$this->db->beginTransaction();
$person = null;
$search_person = PersonTable::parseFlatName($row['name']);
$similar = $search_person->getSimilarEntitiesQuery(true)->execute();
$matched_bio = false;
$similar_ids = array();
foreach ($similar as $s) {
$similar_ids[] = $s->id;
$sim_re = LsString::escapeStringForRegex($s->name_first);
$search_re = LsString::escapeStringForRegex($search_person->name_first);
if (preg_match('/^' . $sim_re . '/su', $search_person->name_first) == 0 && preg_match('/^' . $search_re . '/su', $s->name_first) == 0) {
continue;
}
$matched = false;
$affils = array();
$ct = 1;
$matched_affils = array();
$unmatched_affils = array();
while (isset($row['affiliation' . $ct]) && trim($row['affiliation' . $ct]) != '') {
$affil = trim($row['affiliation' . $ct]);
$org = $s->checkAffiliations(array($affil));
if ($org) {
$matched_affils[] = array($org, $affil);
$edit['Existing Org'] = $org->id;
break;
} else {
$unmatched_affils[] = $affil;
}
$ct++;
}
if (count($matched_affils)) {
$person = $s;
break;
//$ret[] = array('person' => $s, $matched_affils, $unmatched_affils);
} else {
/*$str = implode(' ', $unmatched_affils);
if (isset($row['bio']))
{
$str .= ' ' . $row['bio'];
}*/
$bio = $s->getExtendedBio();
foreach ($unmatched_affils as $affil) {
$affil = OrgTable::removeSuffixes($affil);
$this->printDebug($affil);
$this->printDebug($bio);
if (preg_match('/' . OrgTable::getNameRegex($affil) . '/su', $bio)) {
$matched_bio = true;
break;
}
}
if ($matched_bio) {
$person = $s;
break;
} else {
$this->printDebug(' ' . $s->name . ' failed');
}
}
}
$edit['Similar Names'] = array_slice($similar_ids, 0, 5);
$no_match = false;
if (!$person) {
if (isset($row['bio']) && trim($row['bio']) != '') {
$search_person->summary = $row['bio'];
}
$search_person->save();
$this->printDebug(' not found, new person saved: ' . $search_person->name);
$search_person->addReference($this->source_url, null, null, $this->source_name);
$no_match = true;
$edit['New Person'] = $search_person->id;
$person = $search_person;
} else {
if (isset($row['bio']) && trim($row['bio']) != '' && !$person->summary) {
$person->summary = $row['bio'];
$person->save();
}
$this->printDebug(' **person found: ' . $person->name);
$edit['Existing Person'] = $person->id;
}
if ($matched_bio || $no_match) {
$orgs = OrgTable::getOrgsWithSimilarNames($row['affiliation1'], true);
$max = -1;
$affiliated_org = null;
foreach ($orgs as $org) {
$this->printDebug(' found match: ' . $org->name);
$ct = $org->getRelatedEntitiesQuery('Person', RelationshipTable::POSITION_CATEGORY, null, null, null, false, 2)->count();
if ($ct > $max) {
$affiliated_org = $org;
$edit['Existing Org'] = $affiliated_org->id;
$max = $ct;
}
}
if (!$affiliated_org) {
$affiliated_org = new Entity();
$affiliated_org->addExtension('Org');
if (isset($row['affiliation1_extensions']) && $row['affiliation1_extensions'] != '') {
$extensions = explode(',', $row['affiliation1_extensions']);
foreach ($extensions as $ext) {
$ext = trim($ext);
if (in_array($ext, ExtensionDefinitionTable::$extensionNames)) {
$affiliated_org->addExtension($ext);
}
}
} else {
//$affiliated_org->addExtension('Business');
}
$affiliated_org->name = $row['affiliation1'];
$affiliated_org->save();
$affiliated_org->addReference($this->source_url, null, null, $this->source_name);
$edit['New Org'] = $affiliated_org->id;
}
$rel = new Relationship();
$rel->Entity1 = $person;
$rel->Entity2 = $affiliated_org;
$rel->setCategory('Position');
if (isset($row['affiliation1_title']) && $row['affiliation1_title'] != '') {
$description = trim($row['affiliation1_title']);
$rel->description1 = $description;
if ($description == 'Director' || $description == 'Trustee' || preg_match('/^Chair/su', $description)) {
$rel->is_board = 1;
$rel->is_employee = 0;
}
}
$rel->save();
$rel->addReference($this->source_url, null, null, $this->source_name);
$edit['New Relationship'] = $rel->id;
}
if (isset($row['start_date']) && trim($row['start_date']) != '') {
$edit['Relationship']['start_date'] = trim($row['start_date']);
}
if (isset($row['end_date']) && trim($row['end_date']) != '') {
$edit['Relationship']['end_date'] = trim($row['end_date']);
}
if (isset($row['title']) && trim($row['title']) != '') {
$edit['Relationship']['title'] = trim($row['title']);
}
if (isset($row['notes']) && trim($row['notes']) != '') {
$edit['Relationship']['notes'] = trim($row['notes']);
}
if (isset($row['rank']) && $row['rank'] != '') {
$edit['rank'] = $row['rank'];
}
$this->db->commit();
} catch (Exception $e) {
$this->db->rollback();
throw $e;
}
$this->edits[] = $edit;
}
public function processRow($row)
{
if (isset($row['url']) && $row['url'] != '' && isset($row['url_name']) && $row['url_name'] != '') {
$url = $row['url'];
$url_name = $row['url_name'];
} else {
$url = $this->url;
$url_name = $this->url_name;
}
foreach ($row as &$r) {
trim($r);
}
unset($r);
if ($this->entity) {
$required = array('entity_name', 'primary_type', 'relationship_category');
} else {
$required = array('entity_name', 'primary_type');
}
foreach ($required as $req) {
if (!isset($row[$req]) || $row[$req] == '') {
$this->printDebug('!!! > skipping row, ' . $req . ' not set');
return;
}
}
if ($row['primary_type'] != 'Person' && $row['primary_type'] != 'Org') {
$this->printDebug('!!! > primary type not properly set, skipping row...');
return;
}
if ($this->entity) {
$relationship_category = trim($row['relationship_category']);
$relationship_category_id = array_search($relationship_category, RelationshipCategoryTable::$categoryNames);
if (!$relationship_category_id) {
$this->printDebug('!!! > relationship type not properly set, skipping row...');
return;
}
}
$this->printDebug("processing: " . $row['entity_name'] . '......');
if ($row['primary_type'] == 'Person') {
$entity2 = PersonTable::parseFlatName($row['entity_name']);
$similar_entities = PersonTable::getSimilarQuery2($entity2)->execute();
} else {
$entity2 = new Entity();
$entity2->addExtension('Org');
$entity2->setEntityField('name', $row['entity_name']);
$similar_entities = OrgTable::getOrgsWithSimilarNames($entity2->name);
}
$matched = false;
foreach ($similar_entities as $similar_entity) {
if ($similar_entity['primary_ext'] == 'Person') {
$this->printDebug(' POSSIBLE MATCH: ' . $similar_entity->name . ' (Orgs :: ' . $similar_entity->getRelatedOrgsSummary() . " Bio :: {$similar_entity->summary})");
} else {
$this->printDebug(' POSSIBLE MATCH: ' . $similar_entity->name . ' (Summary :: ' . $similar_entity->summary . ')');
}
$accept = $this->readline(' Is this the same entity? (y or n or b to break)');
if ($accept == 'y') {
$entity2 = $similar_entity;
$matched = true;
$this->printDebug(' [accepted]');
break;
} else {
if ($accept == 'b') {
break;
}
}
}
$created = false;
if (!$matched) {
if ($entity2->getPrimaryExtension() == 'Person') {
$this->printDebug(' New person: ' . $entity2->name_first . ' ' . $entity2->name_last);
} else {
$this->printDebug(' New org: ' . $entity2->name);
}
$accept = $this->readline(' create this new entity? (y or n) ');
if ($accept == 'y') {
try {
$extensions = LsString::split($row['entity_extensions'], '\\s*\\,\\s*');
foreach ($extensions as $extension) {
$entity2->addExtension($extension);
}
$entity2->save();
$entity2->addReference($url, null, null, $url_name);
} catch (Exception $e) {
$this->printDebug(' !!! problems with extensions for this row');
}
$fields = array('summary', 'blurb', 'website');
foreach ($fields as $field) {
if (isset($row[$field])) {
$entity2[$field] = $row[$field];
}
}
$entity2->save();
$entity2->addReference($url, null, null, $url_name);
$created = true;
$this->printDebug(' ' . $entity2->name . ' saved');
//sleep(1);
} else {
$entity2 = null;
}
}
// create relationship
if ($entity2) {
if ($this->entity) {
$relationship = new Relationship();
if (isset($row['relationship_order']) && $row['relationship_order'] != '') {
if ($row['relationship_order'] == '1') {
$relationship->Entity1 = $this->entity;
$relationship->Entity2 = $entity2;
} else {
$relationship->Entity2 = $this->entity;
$relationship->Entity1 = $entity2;
}
} else {
if ($relationship_category == 'Position' || $relationship_category == 'Education') {
if ($row['primary_type'] == 'Org') {
$relationship->Entity1 = $this->entity;
$relationship->Entity2 = $entity2;
} else {
$relationship->Entity1 = $entity2;
$relationship->Entity2 = $this->entity;
}
} else {
$relationship->Entity1 = $this->entity;
$relationship->Entity2 = $entity2;
}
}
$relationship->setCategory($relationship_category);
$cols = array('description1', 'description2', 'start_date', 'end_date', 'goods', 'amount', 'is_board', 'is_executive', 'is_employee');
foreach ($cols as $col) {
if (isset($row[$col]) && $row[$col] != '') {
try {
$relationship[$col] = $row[$col];
} catch (Exception $e) {
$this->printDebug(" could not set {$col} for relationship, skipping");
}
}
}
$q = LsDoctrineQuery::create()->from('Relationship r')->where('r.entity1_id = ? and r.entity2_id = ? and r.category_id = ? and r.id <> ?', array($relationship->entity1_id, $relationship->entity2_id, $relationship->category_id, $relationship->id))->fetchOne();
if ($q) {
$this->printDebug(' (relationship already found, skipping...)');
return;
}
$relationship->save();
$relationship->addReference($url, null, null, $url_name);
$this->printDebug(" Relationship saved: {$relationship}\n");
} else {
if ($this->list) {
$q = LsDoctrineQuery::create()->from('LsListEntity le')->where('le.entity_id = ? and le.list_id = ?', array($entity2->id, $this->list->id))->fetchOne();
if ($q) {
$this->printDebug(' (already on list, skipping...)');
return;
}
$le = new LsListEntity();
$le->LsList = $this->list;
$le->Entity = $entity2;
var_dump($row);
if (isset($row['rank'])) {
echo $row['rank'];
$le->rank = $row['rank'];
}
$le->save();
}
}
}
}
