public function getSummary($str, Entity $e)
{
$str = LsHtml::replaceEntities($str);
$name_re = array();
$name_re[] = $e->getNameRegex();
if ($e->name_nick && $e->name_nick != '') {
$name_re[] = LsString::escapeStringForRegex($e->name_nick);
}
$name_re = implode('|', $name_re);
$style_tags = implode('|', LsHtml::$fontStyleTags);
$layout_tags = implode('|', LsHtml::$layoutTags);
$re = '/((' . $name_re . ')(.*?))<\\/?(' . $layout_tags . ')/isu';
$this->printDebug($re);
$results = null;
if (preg_match_all($re, $str, $matches)) {
$results = $matches[1];
foreach ($results as $result) {
$result = LsString::spacesToSpace(LsHtml::stripTags($result));
$this->printDebug($result);
}
}
return $results;
}
private function findPersonBio($page, $person, $org)
{
//$this->printDebug('');
$name_re = LsString::escapeStringForRegex($person->name_last);
if (preg_match('/<title>([^<]*)<\\/title>/is', $page, $match)) {
if (stristr($match[1], $person->name_last) && stristr($match[1], $person->name_first) && strlen($person->name_first) > 2) {
$name_re .= '|' . LsString::escapeStringForRegex($person->name_first);
}
}
$layout_tags = implode('|', LsHtml::$layoutTags);
$re2 = '/>([^<]*?(' . $name_re . ')(\\s|,|<)(.*?))<\\/?(' . $layout_tags . ')/is';
$re = $re2 . 'u';
//$this->printDebug($re);
$bio_match = null;
if (preg_match_all($re, $page, $matches) || preg_match_all($re2, $page, $matches)) {
//$this->printDebug('matches found');
$arr = array();
$most_reqs = 0;
$qual = false;
$news = false;
foreach ($matches[1] as $match) {
if (stristr($match, '}') || stristr($match, '{') || preg_match('/\\svar\\s/is', $match)) {
//$this->printDebug('FAILED - curly brackets');
continue;
}
$str = LsHtml::replaceEntities($match);
$str = LsHtml::stripTags($str, '');
$str = trim(LsString::spacesToSpace($str));
$this->printDebug(strlen($str));
if (strlen($str) > 3000) {
$this->printDebug('FAILED - str too long');
continue;
}
if (preg_match('/(^|\\b)(' . $name_re . ')\\b/is', $str) == 0) {
$this->printDebug($match . 'FAILED - no name match');
continue;
}
$word_count = count(explode(' ', $str));
if ($word_count < 12) {
$this->printDebug('FAILED - str not long enough');
continue;
} else {
if (stristr($str, 'announce') || stristr($str, 'today') || stristr($str, '—') || stristr($str, '–') || preg_match('/^[^\\-]{0,100}\\-(\\-|\\s)/is', $str)) {
$news = true;
$this->printDebug('FAILED: dash / announced / today');
} else {
if (preg_match('/(^|\\s)([\'"”])([^\\1]+)\\1/is', $str, $qm) && count(explode(' ', $qm[0])) > 6) {
$news = true;
$this->printDebug('FAILED: quote');
} else {
if (preg_match_all('/\\s(\\p{Ll})+\\b/su', $str, $lcm) < 5) {
$this->printDebug('FAILED: not enough lowercase');
} else {
$bio_words = PersonTable::$commonBioWords;
if (in_array('Lobbyist', $person->getExtensions())) {
$bio_words = array_merge($bio_words, LobbyistTable::$commonBioWords);
}
$bio_words = implode('|', $bio_words);
$bio_word_ct = preg_match_all('/\\s(' . $bio_words . ')\\s/is', $str, $matches);
$str = trim($str);
if (preg_match('/\\.$/is', $str) == 0) {
$this->printDebug('no period at end of string');
} else {
if ($bio_word_ct > 1) {
$news = false;
$qual = true;
$arr[] = $str;
} else {
$this->printDebug('less than 2 bio words');
if ($news == false) {
$str = preg_replace('/^[\\,\\.\\:\\;]\\s*/su', '', $str);
$arr[] = $str;
//array('str' => $str, 'bio_words' => $bio_word_ct);
}
}
}
}
}
}
//$this->printDebug('');
}
}
if ($qual) {
$arr = array_unique($arr);
$ret = false;
$bio = implode("\n\n", $arr);
//$this->printDebug($name_re);
if (strlen($bio) < 3000 && LsString::withinN($bio, '(' . $name_re . ')', '(is|was|holds|led|has|had|provides|practices|served|leads)', 2)) {
if (preg_match('/^.*?\\b(' . $name_re . ')\\b/is', $bio, $m) && count(explode(' ', $m[0])) < 20) {
$ret = true;
$this->printDebug('SUCCESS');
}
} else {
$this->printDebug('within N failed !!!!');
}
$org_test = true;
if ($ret && stristr($org->name, $person->name_last)) {
$org_test = false;
if (strlen($person->name_first) > 1) {
if (preg_match('/([^\\s]+\\s+){0,14}/is', $arr[0], $beg_match)) {
$nf_re = LsString::escapeStringForRegex($person->name_first);
if (preg_match('/\\b' . $nf_re . '\\b/is', $beg_match[0]) || preg_match('/\\b(Mr|Mrs|Ms)\\b/su', $arr[0])) {
$org_test = true;
//$this->printDebug('PASSED FIRST NAME TEST');
}
}
} else {
if (preg_match('/\\b(he|she|him|her|his|mr|ms|mrs)\\b/is', $arr[0])) {
$org_test = true;
//$this->printDebug('PASSED POSSESSIVE TEST');
}
}
}
if ($ret && $org_test) {
return $bio;
}
}
} else {
$this->printDebug('no matches found');
}
return false;
}
static function parseNyDonations($str)
{
$re2 = '/(<td.*?>(.*?\\s).*?<.td>\\s*)*?<.tr>/is';
preg_match_all($re2, $str, $matches);
$results = array();
foreach ($matches[0] as $match) {
$result = array("name" => "", "street" => "", "city" => "");
$arr = preg_split('/<.td>\\s*<td.*?>/is', $match);
$name_parts = preg_split('/<br>/is', $arr[0]);
if (count($name_parts) > 1 && count($arr) > 5) {
if (preg_match('/(inc|llp|llc|p\\.c\\.|pc)\\.?$/is', $name_parts[0], $match)) {
$result['name'] = $name_parts[0];
} else {
$np = preg_split('/\\,\\s*/is', $name_parts[0]);
if (count($np) > 1) {
if (count($np) == 3 && stripos($np[2], "jr") !== 0) {
$result['name'] = $np[2] . " " . $np[0] . ", " . $np[1];
} else {
$result['name'] = $np[1] . " " . $np[0];
}
} else {
$result['name'] = $np[0];
}
}
if (count($name_parts) > 1) {
$result['street'] = $name_parts[1];
if (count($name_parts) > 2) {
$result['city'] = $name_parts[2];
}
}
$result['amount'] = $arr[1];
$result['date'] = $arr[2];
$result['committee'] = $arr[3];
foreach ($result as &$r) {
$r = preg_replace('/(\\n|(<.*?>))/is', "", $r);
$r = trim($r);
$r = LsString::spacesToSpace($r);
}
unset($r);
$results[] = $result;
}
}
$str = implode("\t", array_keys($results[0])) . "\n";
if (count($results)) {
foreach ($results as $r) {
$str .= implode("\t", $r);
$str .= "\n";
}
}
$str = trim($str);
return $str;
}
private function importLdaData($lobby_import)
{
$path = $this->_dir . $lobby_import->filename;
$raw = file_get_contents($path);
$xml = new SimpleXMLElement($raw);
$filings = $xml->Filing;
$limit = count($filings);
$this->printDebug('importing data from ' . $lobby_import->filename . ' (record ' . $lobby_import->offset . ' of ' . $limit . ')');
for ($n = (int) $lobby_import->offset; $n < $limit; $n++) {
$this->_count = $this->_count + 1;
if ($this->_count > $this->_limit) {
die;
}
try {
$this->db->beginTransaction();
$lobby_import->offset = $n;
if ($n == $limit - 1) {
$lobby_import->done = 1;
}
$lobby_import->save();
if (!isset($filings[$n])) {
echo 'ok';
var_dump($filings[$n - 1]);
var_dump($filings[$n + 1]);
$this->printDebug('not set' . $n);
$this->db->commit();
continue;
}
$filing = $filings[$n];
if (!isset($filing->Registrant)) {
$this->db->commit();
continue;
}
//var_dump($filing);
$f = new LdaFiling();
$f->federal_filing_id = $filing['ID'];
$f->year = $filing['Year'];
$f->amount = $filing['Amount'];
$f->received = $filing['Received'];
$f->import_id = $lobby_import->id;
$f->offset = $n;
//check for duplicate
if (Doctrine::getTable('LdaFiling')->findOneByFederalFilingId($f->federal_filing_id)) {
$this->db->commit();
continue;
}
//set registrant
if (!($r = Doctrine::getTable('LdaRegistrant')->findOneByFederalRegistrantId($filing->Registrant['RegistrantID']))) {
$r = new LdaRegistrant();
$r->name = LsString::spacesToSpace($filing->Registrant['RegistrantName']);
$r->federal_registrant_id = $filing->Registrant['RegistrantID'];
$r->address = $filing->Registrant['Address'];
$r->description = LsString::spacesToSpace($filing->Registrant['GeneralDescription']);
$r->country = $filing->Registrant['RegistrantCountry'];
$r->save();
}
$f->registrant_id = $r->id;
//set client
if ($filing->Client) {
if (!($c = LsQuery::getByModelAndFieldsQuery('LdaClient', array('registrant_id' => $r->id, 'federal_client_id' => $filing->Client['ClientID']))->execute()->getFirst())) {
$c = new LdaClient();
$c->name = LsString::spacesToSpace($filing->Client['ClientName']);
$c->federal_client_id = $filing->Client['ClientID'];
$c->registrant_id = $r->id;
$c->contact_name = LsString::spacesToSpace($filing->Client['ContactFullname']);
$c->description = LsString::spacesToSpace($filing->Client['GeneralDescription']);
$c->country = $filing->Client['ClientCountry'];
$c->state = $filing->Client['ClientState'];
$c->save();
}
$f->client_id = $c->id;
}
//set filing type
if ($type = (string) $filing['Type']) {
//look for existing type
if (!($t = Doctrine::getTable('LdaType')->findOneByDescription($type))) {
$t = new LdaType();
$t->description = $type;
$t->save();
}
$f->type_id = $t->id;
unset($t);
}
if ($period = (string) $filing['Period']) {
//look for existing period
if (!($p = Doctrine::getTable('LdaPeriod')->findOneByDescription($period))) {
$p = new LdaPeriod();
$p->description = $period;
$p->save();
}
$f->period_id = $p->id;
}
$f->save();
//add lobbyists
if ($filing->Lobbyists) {
foreach ($filing->Lobbyists->Lobbyist as $lobbyist) {
$name = (string) $lobbyist['LobbyistName'];
if (!($l = LsQuery::getByModelAndFieldsQuery('LdaLobbyist', array('registrant_id' => $r->id, 'name' => $name))->execute()->getFirst())) {
$l = new LdaLobbyist();
$l->name = $name;
$l->registrant_id = $r->id;
$l->status = $lobbyist['LobbyistStatus'];
$l->indicator = $lobbyist['LobbyisteIndicator'];
$l->official_position = $lobbyist['OfficialPosition'];
$l->save();
}
$fl = new LdaFilingLobbyist();
$fl->filing_id = $f->id;
$fl->lobbyist_id = $l->id;
$fl->save();
unset($fl);
unset($l);
}
}
//add govt entities
if ($filing->GovernmentEntities) {
foreach ($filing->GovernmentEntities->GovernmentEntity as $govt) {
$govt = trim($govt['GovEntityName']);
if (!($g = Doctrine::getTable('LdaGovt')->findOneByName($govt))) {
$g = new LdaGovt();
$g->name = $govt;
$g->save();
}
$fg = new LdaFilingGovt();
$fg->filing_id = $f->id;
$fg->govt_id = $g->id;
$fg->save();
unset($fg);
unset($g);
}
}
//add issues
if ($filing->Issues) {
foreach ($filing->Issues->Issue as $issue) {
$code = (string) $issue['Code'];
if (!($i = Doctrine::getTable('LdaIssue')->findOneByName($code))) {
$i = new LdaIssue();
$i->name = $code;
$i->save();
}
$fi = new LdaFilingIssue();
$fi->filing_id = $f->id;
$fi->issue_id = $i->id;
$fi->specific_issue = $issue['SpecificIssue'];
$fi->save();
unset($fi);
unset($i);
}
}
$this->printDebug($f->federal_filing_id);
//check for duplicate again
if (Doctrine::getTable('LdaFiling')->findOneByFederalFilingId($f->federal_filing_id)) {
$this->db->rollback();
continue;
}
$this->db->commit();
} catch (Exception $e) {
$this->db->rollback();
throw $e;
}
unset($f);
unset($r);
unset($c);
unset($filing);
}
unset($xml);
unset($raw);
unset($filings);
}
public function parseResults($match)
{
if (isset($match['bio'])) {
$bio_dirty = LsHtml::replaceEntities(LsString::spacesToSpace(LsHtml::stripTags($match['bio'], "; ")));
$bio_dirty = preg_replace('/(\\;\\s)+/is', '; ', $bio_dirty);
}
foreach ($match as $k => &$m) {
$m = LsHtml::replaceEntities(LsString::spacesToSpace(LsHtml::stripTags($m, " ")));
}
if (isset($match['name'])) {
$name = $match['name'];
$bio = '';
if (isset($match['bio'])) {
$bio = $match['bio'];
}
} else {
return;
}
$this->printDebug("_________________________\n\nname: " . $name . "\n");
$this->printDebug("bio: " . $bio . "\n");
$accept = strtolower($this->readline('Process this entity? (n to skip) '));
if ($accept == 'n' || $accept == 'no') {
return false;
}
if (!$this->org_org) {
if ($this->last_first) {
$entity = PersonTable::parseCommaName($name);
} else {
$entity = PersonTable::parseFlatName($name);
}
$similar_entities = PersonTable::getSimilarQuery2($entity)->execute();
} else {
$entity = new Entity();
$entity->addExtension('Org');
foreach ($this->org_extensions as $ext) {
$entity->addExtension($ext);
}
$entity->setEntityField('name', $name);
$name = trim($name);
$name = str_replace('.', '', $name);
$similar_entities = OrgTable::getSimilarQuery($entity)->execute();
}
$matched = false;
foreach ($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)');
$attempts = 1;
while ($accept != 'y' && $accept != 'n' && $attempts < 5) {
$accept = $this->readline(' Is this the same entity? (y or n) ');
$attempts++;
}
if ($accept == 'y') {
$entity = $similar_entity;
$matched = true;
$this->printDebug(' [accepted]');
//sleep(1);
break;
} else {
if ($accept == 'break') {
break;
}
}
}
$created = false;
if (!$matched) {
if ($entity->getPrimaryExtension() == 'Person') {
$this->printDebug(' New person: ' . $entity->name_first . ' ' . $entity->name_last);
} else {
$this->printDebug(' New org: ' . $entity->name);
}
$accept = $this->readline(' create this new entity? (y or n) ');
$attempts = 1;
while ($accept != 'y' && $accept != 'n' && $attempts < 5) {
$accept = $this->readline(' create this new entity? (y or n) ');
$attempts++;
}
if ($accept == 'y') {
if ($entity->getPrimaryExtension() == 'Person') {
$this->printDebug("\n Bio: {$bio} \n");
$accept = $this->readline(' Add this bio? (y or n) ');
$attempts = 1;
while ($accept != 'y' && $accept != 'n' && $attempts < 5) {
$accept = $this->readline(' add this bio? (y or n) ');
$attempts++;
}
if ($accept == 'y') {
$entity->summary = $bio;
}
}
$entity->save();
$entity->addReference($this->url, null, null, $this->url_name);
$created = true;
$this->printDebug(' ' . $entity->name . ' saved');
//sleep(1);
}
}
if (($matched || $created) && $entity->getPrimaryExtension() == 'Person') {
$accept = $this->readline("Parse above bio for possible relationships? (y or n) ");
$attempts = 1;
while ($accept != 'y' && $accept != 'n' && $attempts < 5) {
$accept = $this->readline("Parse above bio for possible relationships? (y or n) ");
$attempts++;
}
if ($accept == 'y') {
$names = $entity->parseBio($bio_dirty);
$this->printDebug(" Orgs that {$entity} has a position at?");
foreach ($names as $name) {
$exists = false;
$name = trim($name);
$accept = $this->readline(" > {$name} :: an org? (y or n or b to break) ");
$attempts = 1;
$accept = strtolower($accept);
while ($accept != 'y' && $accept != 'n' && $accept != 'b' && $attempts < 5) {
$accept = $this->readline(" {$name} :: an org? (y or n or b to break) ");
$accept = strtolower($accept);
$attempts++;
}
if ($accept == 'b') {
break;
} else {
if ($accept == 'y') {
$this->printDebug(' .....looking for names.....');
$orgs = EntityTable::getByExtensionAndNameQuery('Org', $name)->limit(10)->execute();
$related_org = null;
foreach ($orgs as $org) {
$q = LsDoctrineQuery::create()->from('Relationship r')->where('entity1_id = ? and entity2_id = ?', array($entity->id, $org->id))->fetchOne();
if ($q) {
$this->printDebug(' Position already exists, skipping...');
$exists = true;
break;
}
$accept = $this->readline(" Create a position relationship between {$entity->name} and {$org->name}? (y or n) ");
$attempts = 1;
while ($accept != 'y' && $accept != 'n' && $attempts < 5) {
$accept = $this->readline(" Create a position relationship between {$entity->name} and {$org->name}? (y or n) ");
$attempts++;
}
if ($accept == 'y') {
$related_org = $org;
break;
}
}
if (!$related_org && !$exists) {
$accept = $this->readline(" couldn't find org, should this one be created: {$name} (y or n) ");
while ($accept != 'y' && $accept != 'n' && $attempts < 5) {
$accept = $this->readline(" couldn't find org, should this one be created: {$name} (y or n) ");
$attempts++;
}
if ($accept == 'y') {
$related_org = new Entity();
$related_org->addExtension('Org');
$related_org->name = preg_replace('/\\.(?!com)/i', '', $name);
$extensions = $this->readline(" what extensions should this org get? (eg 'Business, LobbyingFirm, LawFirm') ");
$extensions = preg_split('/\\,\\s*/isu', $extensions, -1, PREG_SPLIT_NO_EMPTY);
try {
foreach ($extensions as $extension) {
$related_org->addExtension($extension);
}
$related_org->save();
$related_org->addReference($this->url, null, null, $this->url_name);
} catch (Exception $e) {
$this->printDebug(' !!! problems with org creation, skipping');
$related_org = null;
}
}
}
if ($related_org) {
$q = LsDoctrineQuery::create()->from('Relationship r')->where('r.entity1_id = ? and r.entity2_id = ? and r.category_id = ?', array($entity->id, $related_org->id, 1))->fetchOne();
if ($q) {
$this->printDebug(' (relationship already found, skipping...)');
continue;
}
$relationship = new Relationship();
$relationship->Entity1 = $entity;
$relationship->Entity2 = $related_org;
$relationship->setCategory('Position');
$title = $this->readline(" Title for this position relationship? (<enter> to skip) ");
if (strlen($title) > 2) {
$relationship->description1 = $title;
}
$current = strtolower($this->readline(" Is the relationship current? (y or n or <enter> to skip) "));
if (in_array($current, array('y', 'yes'))) {
$relationship->is_current = 1;
} else {
if (in_array($current, array('n', 'no'))) {
$relationship->is_current = 0;
}
}
$board = strtolower($this->readline(" Is the relationship a board position? (y or n or <enter> to skip) "));
if (in_array($board, array('y', 'yes'))) {
$relationship->is_board = 1;
} else {
if (in_array($board, array('n', 'no'))) {
$relationship->is_board = 0;
}
}
$relationship->save();
$relationship->addReference($this->url, null, null, $this->url_name);
$this->printDebug(" Relationship saved: {$relationship}");
}
}
}
}
}
}
if ($matched || $created) {
if ($this->list) {
$q = LsDoctrineQuery::create()->from('LsListEntity l')->where('l.entity_id = ? and l.list_id = ?', array($entity->id, $this->list->id))->fetchOne();
if (!$q) {
$le = new LsListEntity();
$le->Entity = $entity;
$le->LsList = $this->list;
if (isset($match['rank'])) {
if (preg_match('/(\\d+)/isu', $match['rank'], $m)) {
$le->rank = $m[1];
}
}
$le->save();
$this->printDebug('List membership saved');
}
}
if ($this->org) {
$q = LsDoctrineQuery::create()->from('Relationship r')->where('r.entity1_id = ? and r.entity2_id = ? and r.category_id = ?', array($entity->id, $this->org->id, 1))->fetchOne();
if ($q) {
$this->printDebug(' (relationship already found, skipping...)');
return;
}
$relationship = new Relationship();
$relationship->Entity1 = $entity;
$relationship->Entity2 = $this->org;
$relationship->setCategory($this->relationship_category);
if ($this->description1) {
$relationship->description1 = $this->description1;
} else {
$description = $this->readline(" what description to give this relationship ({$relationship}) ? (less than 3 chars will skip)");
if (strlen($description) > 2) {
$relationship->description1 = $description;
}
}
if ($this->relationship_category == 'Position') {
$relationship->is_board = $this->is_board;
} else {
if ($this->relationship_category == 'Donation') {
if ($this->amount) {
$relationship->amount = $this->amount;
} else {
$amount = $this->readline(" what amount ({$relationship}) ? (less than 3 chars will skip)");
if (strlen($amount) > 1) {
$relationship->amount = $amount;
}
}
}
}
$relationship->save();
$relationship->addReference($this->url, null, null, $this->url_name);
$this->printDebug(" Relationship saved: {$relationship}");
}
}
//dump history
if (isset($match['affiliation1'])) {
$affiliation = $match['affiliation'];
//$this->printDebug($affiliation);
}
}
public function getCleanFirstParagraph()
{
if ($this->_paragraphs) {
return null;
}
$first = $this->_paragraphs[0];
$first = LsString::spacesToSpace(LsHtml::replaceEntities(LsHtml::stripTags($first)));
return $first;
}
private function getProxyData($roster, $url, $proxy_year)
{
echo "fetching data from proxy at {$url} \n\n";
$people_count = 0;
if (!$this->browser->get($url)->responseIsError()) {
$this->proxyText = $this->browser->getResponseText();
$this->proxyText = LsHtml::replaceEntities($this->proxyText, ENT_QUOTES, 'UTF-8');
$this->proxyText = LsString::utf8TransUnaccent($this->proxyText);
foreach ($roster as &$r) {
//make sure this is not form 4 data for a corporation, continue to the next if it is
if ($r['officerTitle'] == '' && $r['isDirector'] != 1 && strtoupper($r['isDirector']) != strtoupper('true')) {
continue;
}
//echo $re;
$parts = preg_split("/[\\s|\\.]+/", $r['personName'], -1, PREG_SPLIT_NO_EMPTY);
//first word, but has to be part of last name because form4 names are in format RUBIN ROBERT E
$last = trim($parts[0]);
//sometimes O'LEARY can appear as O LEARY in the form 4
if (strlen($last) == 1) {
$r['personName'] = $last . substr($r['personName'], 2);
$parts = preg_split("/[\\s|\\.]+/", $r['personName'], -1, PREG_SPLIT_NO_EMPTY);
$last = trim($parts[0]);
}
//prepare regex to match occurrences of full name
//case insensitive to accommodate for various irregularities in names
$re = LsLanguage::buildLooseNameRegex($r['personName']);
$offset = 0;
$found = true;
//use stripos (much faster than regex) to find occurrences of the first word in the form 4 name (assumed to be part of the last name)
//needs to be case insensitive
//continue searching for last name in proxy until a matching full name (proxyName) is found
while (!isset($r['proxyName']) && $found !== false) {
$found = stripos($this->proxyText, $last, $offset);
//$this->printDebug('found at pos:' . $found);
$offset = $found + 1;
if ($found !== false) {
$str = substr($this->proxyText, $found - 70, 120);
//$this->printDebug('found string: ' . $str);
//$this->printDebug($re);
preg_match_all($re, $str, $matches, PREG_OFFSET_CAPTURE | PREG_SET_ORDER);
//$this->printDebug('matchcount is ' . count($matches));
foreach ($matches as $match) {
if (stristr($match[1][0], '=')) {
continue;
}
//since we may or may not be working with the full last name, use getLastName to return full last name
$new_last = $this->getLastName($r['personName'], $match[1][0]);
if ($new_last) {
//if last name produced by case insensitive search has no capital letters, not a match
if (preg_match('/\\p{Lu}/su', $new_last) == 0) {
continue;
}
//now that we have a last name, pull the full name from the string
$name = LsLanguage::getNameWithLast($match[0][0], $new_last);
if ($name) {
$parts = preg_split('/\\s+/isu', $name['nameStart'], -1, PREG_SPLIT_NO_EMPTY);
$non_prefixes = array_diff($parts, PersonTable::$nameParsePrefixes);
//if all we've found are matching prefixes, not a match
if (count($non_prefixes) == 0) {
continue;
} else {
$name1_parts = preg_split('/\\s+/', $r['personName'], -1, PREG_SPLIT_NO_EMPTY);
$ct = 0;
//compatibility check to correct for vagueness of regex
foreach ($non_prefixes as $n) {
foreach ($name1_parts as $p) {
if (stripos($n, $p) === 0 || stripos($p, $n) === 0) {
$ct++;
}
}
}
//phew -- if name is (somewhat) compatible, assume we've found it
if ($ct > 0) {
$r['proxyUrl'] = $url;
$r['proxyYear'] = $proxy_year;
$r['nameLast'] = trim(LsString::spacesToSpace($name['nameLast']));
$r['proxyName'] = trim(LsString::spacesToSpace($name['nameFull']));
}
}
}
}
}
}
}
}
unset($r);
} else {
//Error response (eg. 404, 500, etc)
$log = fopen($this->logFile, 'a');
fwrite($log, "Couldn't get " . $url . "\n");
fclose($log);
}
return $roster;
}
