PHP Entity::addExtension方法代码示例

 public function processRow($row)
 {
     $arr = str_getcsv($row);
     $ticker = $arr[0];
     $name = $arr[1];
     $name = str_replace(''', "'", $name);
     if ($name == 'Name') {
         return false;
     }
     $cap = $arr[3];
     if ($cap < $this->min_market_cap) {
         $this->too_small_ct++;
         return false;
     } else {
         $corp = Doctrine::getTable('PublicCompany')->findOneByTicker($ticker);
         if ($corp) {
             $this->existing_ct++;
             return false;
         } else {
             $corp = new Entity();
             $corp->addExtension('Org');
             $corp->addExtension('Business');
             $corp->addExtension('PublicCompany');
             $corp->ticker = $ticker;
             $corp->name = $name;
             $corp->save();
             $this->printDebug("New company added: " . $name);
             $this->added_ct++;
         }
     }
 }

 public function import($school)
 {
     if (EntityTable::getByExtensionQuery('Org')->addWhere('LOWER(org.name) LIKE ?', '%' . strtolower($school->instnm) . "%")->fetchOne()) {
         $this->printDebug("School exists in database: " . $school->instnm);
     } else {
         $address = new Address();
         $address->street1 = isset($school->addr) ? $school->addr : null;
         $address->street2 = isset($school->street2) ? $school->street2 : null;
         $address->city = $school->city;
         if ($state = AddressStateTable::retrieveByText($school->stabbr)) {
             $address->State = $state;
         }
         $address->postal = $school->zip;
         $aliases = explode("|", $school->ialias);
         $website = null;
         if (!preg_match('/^http\\:\\/\\//i', trim($school->webaddr))) {
             $website = "http://" . strtolower($school->webaddr);
         }
         $this->printDebug($website);
         $newschool = new Entity();
         $newschool->addExtension('Org');
         $newschool->addExtension('School');
         $newschool->name = $school->instnm;
         $newschool->website = $website;
         $newschool->addAddress($address);
         $newschool->save();
         foreach ($aliases as $alias) {
             try {
                 $newalias = new Alias();
                 $newalias->Entity = $newschool;
                 $newalias->name = $alias;
                 $newalias->save();
             } catch (Exception $e) {
                 $this->printDebug("An alias exception. No biggie. It's most likely that the name already exists. so we ignore it and move on: " . $e);
             }
         }
         $this->printDebug("Adding new school: " . $school->instnm);
     }
 }

//.........这里部分代码省略.........
                     }
                 }
                 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;
 }

//.........这里部分代码省略.........
         } elseif ($this->year > 2004 && preg_match('/<blockquote class="spaced">(.*)<\\/blockquote>/ismU', $text, $match)) {
             $summary = str_replace(array('  ', "\n"), array(' ', ' '), html_entity_decode(trim(strip_tags($match[1]))));
         }
         //get revenue
         if ($this->year > 1995 && $this->year < 2000 && preg_match('/<td class="mainlisttxt">\\$([\\S]+) mil<sup>e?<\\/sup><\\/td>/ismU', $text, $match)) {
             $this->printDebug($match[1]);
             $revenue = str_replace(",", "", $match[1] . ",000,000");
         } elseif ($this->year > 1999 && $this->year < 2005 && preg_match('/<td class="mainlisttxt" nowrap>([^<]+)<sup>e?<\\/sup><\\/td>/ismU', $text, $match)) {
             $this->printDebug($match[1]);
             $revenue = str_replace(",", "", $match[1] . ",000,000");
         } elseif ($this->year > 2004 && preg_match('/<td class="highlight" nowrap="nowrap">\\$([\\S]+) bil.*<\\/td> <td class="highlight" nowrap="nowrap">[^<]+<\\/td> <td class="highlight" nowrap="nowrap">([^<]+)<\\/td>/ismU', $text, $match)) {
             $revenue = 1000000000 * $match[1];
         }
         //get employees
         if ($this->year > 1995 && $this->year < 2005 && preg_match('/mil<\\/td>.+<td class="mainlisttxt"( nowrap)?>(\\d[^<]+)<\\/td>.+<td class="mainlisttxt">[a-zA-Z]+<\\/td>/ismU', $text, $match)) {
             $employees = str_replace(',', '', $match[2]);
         } elseif ($this->year > 1999 && $this->year < 2005 && preg_match('/<sup>e?<\\/sup><\\/td> <td class="mainlisttxt"( nowrap)?>(\\d[^<]+)<sup>e?<\\/sup><\\/td> <td class="mainlisttxt">[a-zA-Z]+<\\/td>/ismU', $text, $match)) {
             $employees = str_replace(',', '', $match[2]);
         } elseif ($this->year > 2004 && preg_match('/<td class="highlight" nowrap="nowrap">([\\d,]+)<\\/td> <td class="highlight" nowrap="nowrap">[A-Z][a-z]{2,}<\\/td>/', $text, $match)) {
             $employees = str_replace(',', '', $match[1]);
         }
         /*$this->printDebug( "URL: ". $url);
           $this->printDebug( "Rank: " . $rank );
           $this->printDebug( "Name: " . $name );
           $this->printDebug( "Industry: " . $industryName );
           $this->printDebug( "Street: " . $street1 );
           $this->printDebug( "Street2: " . $street2 );
           $this->printDebug( "City: " . $city );
           $this->printDebug( "State: " . $state );
           $this->printDebug( "Postal: " . $postal );
           $this->printDebug( "Phone: " . $phone );
           $this->printDebug( "Fax: " . $fax );
           $this->printDebug( "Website: " . $website );
           $this->printDebug( "CEO: " . $ceoName . "  " . $ceoBirthYear);
           $this->printDebug( "Summary: " . $summary );
           $this->printDebug( "Revenue: " . $revenue );
           $this->printDebug( "Employees: " . $employees );*/
         $search_company_name = trim(implode(' ', array_diff(explode(' ', ucwords(strtolower($name))), array_merge(LsLanguage::$business, LsLanguage::$businessAbbreviations))));
         //continue;
         $this->printDebug("{$search_company_name} == {$name}");
         if ($company = EntityTable::getByExtensionQuery(array('Org', 'PrivateCompany'))->addWhere("LOWER(REPLACE( org.name, '-' , '')) = ?", strtolower($name))->fetchOne()) {
             $this->printDebug("Company exists");
             $company->revenue = $revenue;
             $company->save();
         } else {
             $this->printDebug("Creating new company {$name}");
             Doctrine::getTable('ExtensionDefinition')->clear();
             $company = new Entity();
             $company->addExtension('Org');
             $company->addExtension('Business');
             $company->addExtension('PrivateCompany');
             $company->name = LsLanguage::titleize($name);
             $company->employees = strlen($employees) ? $employees : null;
             $company->revenue = strlen($revenue) ? $revenue : null;
             $company->website = strlen($website) ? $website : null;
             $company->summary = strlen($summary) ? trim($summary) : null;
             //add address
             if ($phone) {
                 $company->addPhone($phone);
             }
             if ($fax) {
                 //$company->addPhone($fax);
             }
             if ($city && $state) {
                 $address = new Address();
                 $address->street1 = strlen($street1) ? $street1 : null;
                 $address->street2 = strlen($street2) ? $street2 : null;
                 $address->city = strlen($city) ? $city : null;
                 if ($state = AddressStateTable::retrieveByText($state)) {
                     $address->State = $state;
                 }
                 $address->postal = $postal;
                 $company->addAddress($address);
                 $address->save();
                 $address->addReference($source = $url, $excerpt = null, $fields = array('city', 'country_id', 'postal', 'state_id', 'street1'), $name = 'Forbes.com', $detail = null, $date = null);
             }
         }
         /*$this->printDebug( "URL: ". $url);
           $this->printDebug( "Rank: " . $rank );
           $this->printDebug( "Name: " . $name );
           $this->printDebug( "Industry: " . $industryName );
           $this->printDebug( "Street: " . $street1 );
           $this->printDebug( "Street2: " . $street2 );
           $this->printDebug( "City: " . $city );
           $this->printDebug( "State: " . $state );
           $this->printDebug( "Postal: " . $postal );
           $this->printDebug( "Phone: " . $phone );
           $this->printDebug( "Fax: " . $fax );
           $this->printDebug( "Website: " . $website );
           $this->printDebug( "CEO: " . $ceoName . "  " . $ceoBirthYear);
           $this->printDebug( "Summary: " . $summary );
           $this->printDebug( "Revenue: " . $revenue );
           $this->printDebug( "Employees: " . $employees );*/
         $company->save();
         $company->addReference($source = $url, $excerpt = null, $fields = array('website', 'name', 'website', 'summary', 'revenue', 'employees'), $name = 'Forbes.com', $detail = null, $date = null);
         $this->saveToList($company, $rank);
     } else {
         $this->printDebug("Couldn't get company: " . $url);
     }
 }

//.........这里部分代码省略.........
                     $extensions_arr[] = $ext->name;
                 }
                 $this->matches = array();
                 if (isset($names) && count($names) > 0) {
                     for ($i = 0; $i < count($names); $i++) {
                         if (isset($names[$i]['name']) && trim($names[$i]['name']) != '') {
                             $name = $names[$i]['name'];
                             $name_terms = $name;
                             if ($this->default_type == 'Person') {
                                 $name_parts = preg_split('/\\s+/', $name);
                                 if (count($name_parts) > 1) {
                                     $name_terms = PersonTable::nameSearch($name, true);
                                 }
                                 $terms = $name_terms;
                                 $primary_ext = "Person";
                             } else {
                                 if ($this->default_type == 'Org') {
                                     $name_terms = OrgTable::nameSearch($name);
                                     $terms = $name_terms;
                                     $primary_ext = "Org";
                                 } else {
                                     $terms = $name_terms;
                                     $primary_ext = null;
                                 }
                             }
                             $pager = EntityTable::getSphinxPager($terms, $page = 1, $num = 20, $listIds = null, $aliases = true, $primary_ext);
                             $match = array('name' => $name);
                             $match['search_results'] = $pager->execute();
                             $match['blurb'] = isset($names[$i]['blurb']) ? $names[$i]['blurb'] : null;
                             $match['rank'] = isset($names[$i]['rank']) ? $names[$i]['rank'] : null;
                             $match['types'] = array();
                             if (isset($names[$i]['types'])) {
                                 $types = explode(',', $names[$i]['types']);
                                 $types = array_map('trim', $types);
                                 foreach ($types as $type) {
                                     if (in_array($type, $extensions_arr)) {
                                         $match['types'][] = $type;
                                     }
                                 }
                             }
                             $this->matches[] = $match;
                         }
                     }
                 }
             }
         } else {
             if ($request->hasParameter('ref_id')) {
                 $this->ref_id = $this->getRequestParameter('ref_id');
                 $entity_ids = array();
                 $default_type = $this->getRequestParameter('default_type');
                 for ($i = 0; $i < $this->getRequestParameter('count'); $i++) {
                     if ($entity_id = $request->getParameter('entity_' . $i)) {
                         $selected_entity_id = null;
                         if ($entity_id == 'new') {
                             $name = $request->getParameter('new_name_' . $i);
                             if ($default_type == 'Person') {
                                 $new_entity = PersonTable::parseFlatName($name);
                             } else {
                                 $new_entity = new Entity();
                                 $new_entity->addExtension('Org');
                                 $new_entity->name = trim($name);
                             }
                             if ($types = $request->getParameter('new_extensions_' . $i)) {
                                 foreach ($types as $type) {
                                     $new_entity->addExtension($type);
                                 }
                             }
                             $new_entity->save();
                             $new_entity->blurb = $request->getParameter('new_blurb_' . $i);
                             $ref = Doctrine::getTable('Reference')->find($request->getParameter('ref_id'));
                             $new_entity->addReference($ref->source, null, null, $ref->name);
                             $new_entity->save();
                             $selected_entity_id = $new_entity->id;
                         } else {
                             if ($entity_id > 0) {
                                 $selected_entity_id = $entity_id;
                             }
                         }
                         if ($selected_entity_id) {
                             $q = LsDoctrineQuery::create()->from('LsListEntity le')->where('le.list_id = ? AND le.entity_id = ?', array($this->list['id'], $selected_entity_id));
                             if (!$q->count()) {
                                 $ls_list_entity = new LsListEntity();
                                 $ls_list_entity->list_id = $this->list->id;
                                 $ls_list_entity->entity_id = $selected_entity_id;
                                 $ls_list_entity->rank = $request->getParameter('entity_' . $i . '_rank');
                                 $ls_list_entity->save();
                                 LsCache::clearEntityCacheById($selected_entity_id);
                             }
                         }
                     }
                 }
                 $this->clearCache($this->list);
                 $this->clearRailsCache($this->list->id);
                 $this->redirect($this->list->getInternalUrl());
             } else {
                 $request->setError('name', 'The name you entered is invalid');
             }
         }
     }
 }

 private function addGovernmentBodyEntity($name, $fedspending_name, $parent_id = null)
 {
     $new_gov_body = new Entity();
     $new_gov_body->addExtension('Org');
     $new_gov_body->addExtension('GovernmentBody');
     $new_gov_body->name = $name;
     $new_gov_body->is_federal = 1;
     if ($parent_id) {
         $new_gov_body->parent_id = $parent_id;
     }
     $new_gov_body->save();
     $alias = new Alias();
     $alias->context = 'fedspending_government_body';
     $alias->name = $fedspending_name;
     $alias->entity_id = $new_gov_body->id;
     $alias->save();
     return $new_gov_body;
 }

 protected function import($url)
 {
     $person = null;
     $this->printDebug($url);
     if (!$this->browser->get($url)->responseIsError()) {
         $text = $this->browser->getResponseText();
         $bio = null;
         $name = null;
         $netWorth = null;
         $birthYear = null;
         $schools = null;
         $schools = null;
         $imageUrl = null;
         $rank = null;
         //get name & rank
         if ($this->year > 2005 && preg_match('/<b>#(\\d+) ([^<]+)<\\/b>/', $text, $match)) {
             $name = trim($match[2]);
             $rank = $match[1];
         }
         if ($this->year == 2005 && preg_match('/<h2>#(\\d+) ([^<]+)<\\/h2>/', $text, $match)) {
             $name = trim($match[2]);
             $rank = $match[1];
         }
         //get net worth
         if (preg_match('/Net Worth<\\/span> <span class="red">\\$([\\S]+) billion/', $text, $match)) {
             $netWorth = $match[1] * 1000000000;
         }
         //get birth year
         if (preg_match('/>Age<\\/span> (\\d+)/', $text, $match)) {
             $birthYear = date("Y") - $match[1] . "-00-00";
         }
         //get schools
         if (preg_match('/Education<\\/span>(.*)<\\/td>/isU', $text, $match)) {
             $schools = array();
             $schoolParts = explode('<br>', $match[1]);
             while ($schoolPart = current($schoolParts)) {
                 if (preg_match('/^([^,]+),\\s+<b>([^<]+)<\\/b>/is', trim($schoolPart), $match)) {
                     $schoolOrg = trim($match[1]);
                     if ($schoolOrg == 'High School') {
                         next($schoolParts);
                         continue;
                     }
                     $schoolDegree = trim($match[2]);
                     $schools[] = array('org' => $schoolOrg, 'degree' => $schoolDegree);
                 }
                 next($schoolParts);
             }
         }
         if (preg_match('#<br>[\\n\\s]<br>(.+?)<br>[\\n\\s]<br>[\\n\\s]<img#isU', $text, $match)) {
             $bio = strip_tags(trim($match[1]));
         } else {
             $wikipedia = new LsWikipedia();
             if ($wikipedia->request($name)) {
                 $bio = $wikipedia->getIntroduction();
             }
         }
         //get image
         $regexp = '#([A-Z1-9]{4}).html#';
         if (preg_match($regexp, $url, $match)) {
             $imageFilename = $match[1] . ".jpg";
             $imageUrl = $this->list_urls[$this->year]['img_src'] . $imageFilename;
         }
         //echo "Rank: " . $rank . "\n";
         $this->printDebug("Rank: " . $rank);
         $this->printDebug("Name: " . $name);
         $this->printDebug("Image: " . $imageUrl);
         $this->printDebug("Net worth: " . $netWorth);
         $this->printDebug("Birth year: " . $birthYear);
         $this->printDebug("Bio: " . $bio);
         $person = $this->generatePerson($name, $bio);
         $person_exists = $this->getBusinessPersonQuery()->addWhere("person.name_first = ? AND person.name_last = ?", array($person->name_first, $person->name_last))->fetchOne();
         if ($person_exists != false) {
             $this->printDebug('Person exists');
             $person = $person_exists;
         } else {
             $this->printDebug('Saving new person');
         }
         //parse name and create person object
         $person->addExtension('BusinessPerson');
         $person->start_date = $person->start_date == null ? $birthYear : $person->start_date;
         $person->summary = $person->summary == null ? $bio : $person->summary;
         $person->net_worth = $person->net_worth == null ? $netWorth : $person->net_worth;
         //go through schools person attended
         foreach ($schools as $school) {
             //does the current school exist?
             $current_school = EntityTable::getByExtensionQuery('Org')->addWhere("org.name = ?", $school['org'])->fetchOne();
             if ($current_school) {
                 $this->printDebug("  Found School " . $school['org']);
             } else {
                 //clear cache
                 Doctrine::getTable('ExtensionDefinition')->clear();
                 $current_school = new Entity();
                 $current_school->addExtension('Org');
                 $current_school->addExtension('School');
                 $current_school->name = LsLanguage::titleize($school['org']);
                 $current_school->save();
                 $current_school->addReference($source = $url, $excerpt = null, $fields = array('name'), $name = 'Forbes.com', $detail = null, $date = null);
                 $this->printDebug("  Adding new school: " . $school['org']);
             }
             //if there is no relationship between person and school. connect them!
//.........这里部分代码省略.........

 function import(Entity $person, $possible_persons)
 {
     //loop through the people we found. usually just one.
     foreach ($possible_persons as $possible_person) {
         $this->printDebug('Query returned ' . count($possible_person) . ' person named ' . $possible_person->name);
         //this person does not provide education. we skip
         if (count($possible_person->education)) {
             $this->printDebug('Education found');
         } else {
             $this->printDebug('No education history found');
             continue;
         }
         //get employement info for this possible match
         $possible_person_bio = $possible_person->summary;
         if (count($possible_person->employment_history)) {
             foreach ($possible_person->employment_history as $employment) {
                 $possible_person_bio .= ' ' . $employment->company . " ";
             }
             $this->printDebug('Employment found');
         } else {
             $this->printDebug('No employment history found');
             continue;
         }
         //get employment info for the person in our database
         $relationship_orgs = $person->getRelatedEntitiesQuery('Org', RelationshipTable::POSITION_CATEGORY, null, null, null, false, 1)->execute();
         $person_bio = $person->summary;
         foreach ($relationship_orgs as $org) {
             $person_bio .= ' ' . $org->name;
         }
         //lets see how many matches we get
         $matches = LsLanguage::getCommonPronouns($person_bio, trim($possible_person_bio), LsLanguage::$business);
         if (count($matches)) {
             foreach ($possible_person->education as $school) {
                 $school->institution = mb_convert_encoding($school->institution, 'UTF-8');
                 $school->institution = preg_replace('/–/isu', ' ', $school->institution);
                 $this->printDebug('Looking for the school: ' . $school->institution);
                 $current_school = EntityTable::findByAlias($school->institution, $context = 'bw_school');
                 //find school
                 if ($current_school) {
                     $this->printDebug('Found school');
                 } else {
                     $current_school = EntityTable::getByExtensionQuery(array('Org', 'School'))->addWhere('LOWER(org.name) LIKE ?', '%' . strtolower($school->institution) . "%")->fetchOne();
                     if (!$current_school) {
                         $new_school = new Entity();
                         $new_school->addExtension('Org');
                         $new_school->addExtension('School');
                         $new_school->name = $school->institution;
                         $wikipedia = new LsWikipedia();
                         $wikipedia->request($school->institution);
                         if ($wikipedia->execute() && !$wikipedia->isDisambiguation()) {
                             $info_box = $wikipedia->getInfoBox();
                             if (isset($info_box['students']) && preg_match('/([\\d\\,]{2,})/isu', $info_box['students']['clean'], $match)) {
                                 $new_school->students = LsNumber::clean($match[1]);
                             } else {
                                 $student_types = array('undergrad', 'postgrad', 'grad', 'doctoral');
                                 $num_students = 0;
                                 foreach ($student_types as $st) {
                                     if (isset($info_box[$st]) && preg_match('/([\\d\\,]{2,})/isu', $info_box[$st]['clean'], $match)) {
                                         $num_students += LsNumber::clean($match[1]);
                                     }
                                 }
                                 if ($num_students > 0) {
                                     $new_school->students = $num_students;
                                 }
                             }
                             if (isset($info_box['faculty']) && preg_match('/([\\d\\,]{2,})/isu', $info_box['faculty']['clean'], $match)) {
                                 $new_school->faculty = LsNumber::clean($match[1]);
                             }
                             if (isset($info_box['type'])) {
                                 if (stristr($info_box['type']['clean'], 'public')) {
                                     $new_school->is_private = 0;
                                 } else {
                                     if (stristr($info_box['type']['clean'], 'private')) {
                                         $new_school->is_private = 1;
                                     }
                                 }
                             }
                             if (isset($info_box['endowment'])) {
                                 if (preg_match('/(\\$[\\d\\,\\.\\s]+)(million|billion)/isu', $info_box['endowment']['clean'], $match)) {
                                     if (strtolower($match[2]) == 'billion') {
                                         $factor = 1000000000;
                                     } else {
                                         $factor = 1000000;
                                     }
                                     $new_school->endowment = LsNumber::formatDollarAmountAsNumber($match[1], $factor);
                                 }
                             }
                             if (isset($info_box['established'])) {
                                 $year = null;
                                 if ($date = LsDate::convertDate($info_box['established']['clean'])) {
                                     $new_school->start_date = $date;
                                 } else {
                                     if (preg_match('/\\b(\\d\\d\\d\\d)\\b/isu', $info_box['established']['clean'], $match)) {
                                         $new_school->start_date = $match[1];
                                     }
                                 }
                             }
                             $summary = trim($wikipedia->getIntroduction());
                             $summary = preg_replace('/\\n\\s*\\n/isu', '', $summary);
                             if (strlen($summary) > 10) {
//.........这里部分代码省略.........

 static function parseBioguideName($str)
 {
     $entity = new Entity();
     $entity->addExtension('Person');
     //extract nickname
     if (preg_match('/\\(([^(]+)\\)/', $str, $nick)) {
         $entity->name_nick = $nick[1];
         $str = preg_replace('/\\(.*\\)/U', '', $str);
     }
     $str = preg_replace('/\\s{2,}/', ' ', $str);
     $str = str_replace('.', '', $str);
     $parts = explode(',', trim($str));
     if (count($parts) > 1) {
         $entity->name_last = LsLanguage::nameize(mb_strtolower(trim($parts[0]), mb_detect_encoding(trim($parts[0]))));
         $other = explode(' ', trim($parts[1]));
         $entity->name_first = trim($other[0]);
         if (count($other) > 1) {
             $middles = array_slice($other, 1);
             $middle = trim(implode($middles, ' '));
             $entity->name_middle = $middle;
         }
         if (count($parts) > 2) {
             $suffix = trim($parts[2]);
             $entity->name_suffix = $suffix;
         }
     } else {
         return null;
     }
     return $entity;
 }

//.........这里部分代码省略.........
     $prefixes = self::$nameParsePrefixes;
     while ($prefix = current($prefixes)) {
         if ($str != ($new = preg_replace('/^' . $prefix . ' /i', '', $str))) {
             if (!LsArray::inArrayNoCase($prefix, LsLanguage::$commonPrefixes)) {
                 $namePrefix .= $prefix . ' ';
             }
             $str = trim($new);
             reset($prefixes);
             continue;
         }
         next($prefixes);
     }
     $namePrefix = $namePrefix ? trim($namePrefix) : null;
     //get suffixes
     $suffixes = self::$nameParseSuffixes;
     while ($suffix = current($suffixes)) {
         if ($str != ($new = preg_replace('/ ' . $suffix . '$/i', '', $str))) {
             $nameSuffix = $suffix . ' ' . $nameSuffix;
             $str = trim($new);
             reset($suffixes);
             continue;
         }
         next($suffixes);
     }
     $nameSuffix = $nameSuffix ? trim($nameSuffix) : null;
     //remove commas left over from suffixes
     $str = trim(str_replace(',', '', $str));
     //find nickname in quotes
     if (preg_match('/["\']([\\S]+)[\'"]/', $str, $nickFound)) {
         $nameNick = $nickFound[1] ? $nickFound[1] : $nickFound[2];
         $str = trim(preg_replace('/["\']([\\S]+)[\'"]/', '', $str));
     }
     //condense multiple spaces
     $str = preg_replace('/\\s{2,}/', ' ', $str);
     //split into parts
     $parts = explode(' ', $str);
     switch (count($parts)) {
         case 1:
             if ($namePrefix) {
                 $nameFirst = $namePrefix;
                 $nameLast = $parts[0];
                 $namePrefix = null;
             } else {
                 if ($nameSuffix) {
                     $nameFirst = $parts[0];
                     $nameLast = $nameSuffix;
                     $nameSuffix = null;
                 } else {
                     if (strtolower($sub) == strtolower($parts[0])) {
                         $nameLast = $parts[0];
                     } else {
                         $nameFirst = $parts[0];
                     }
                 }
             }
             break;
         case 2:
             $nameFirst = $parts[0];
             $nameLast = $parts[1];
             break;
         case 3:
             $nameFirst = $parts[0];
             $nameMiddle = $parts[1];
             $nameLast = $parts[2];
             break;
         default:
             $nameFirst = $parts[0];
             $nameLast = $parts[count($parts) - 1];
             for ($n = 1; $n < count($parts) - 1; $n++) {
                 $nameMiddle .= $parts[$n] . ' ';
             }
             $nameMiddle = trim($nameMiddle);
             break;
     }
     $nameLast = str_replace('_', ' ', $nameLast);
     $name = array('name_first' => $nameFirst, 'name_last' => $nameLast, 'name_middle' => $nameMiddle, 'name_prefix' => $namePrefix, 'name_suffix' => $nameSuffix, 'name_nick' => $nameNick);
     foreach ($name as $nk => &$nv) {
         if ($nv && $nk != 'name_suffix' && $nk != 'name_prefix') {
             $nv = preg_replace('/^(\\P{L})+|(\\P{L})+$/u', '', $nv);
             $case = LsString::checkCase($nv);
             $nv = $case == 'upper' || $case == 'lower' ? LsLanguage::nameize($nv) : $nv;
             if ($nk != 'name_last') {
                 $nv = LsLanguage::hgCaser($nv, false);
             }
         }
     }
     unset($nv);
     if ($returnArray) {
         return $name;
     }
     $person = new Entity();
     $person->addExtension('Person');
     $person->name_first = $name['name_first'];
     $person->name_middle = $name['name_middle'];
     $person->name_last = $name['name_last'];
     $person->name_nick = $name['name_nick'];
     $person->name_prefix = $name['name_prefix'];
     $person->name_suffix = $name['name_suffix'];
     return $person;
 }

 private function mergePeople($proxyPerson, $formPerson)
 {
     $formPerson->name_first;
     $formPerson->name_last;
     $person = new Entity();
     $person->addExtension('Person');
     $person->name_first = $formPerson->name_first;
     $person->name_middle = $formPerson->name_middle;
     $person->name_last = $formPerson->name_last;
     $person->name_nick = $formPerson->name_nick;
     $person->name_prefix = $formPerson->name_prefix;
     $person->name_suffix = $formPerson->name_suffix;
     $compatible = false;
     if (!$proxyPerson->name_first or $proxyPerson->name_first == '') {
         return $person;
     }
     //check first name compatibility before doing anything
     if (stripos($proxyPerson->name_first, $formPerson->name_first) === 0 || stripos($formPerson->name_first, $proxyPerson->name_first) === 0) {
         if ($formPerson->name_middle == null || $formPerson->name_middle == '') {
             $compatible = true;
             $person->name_middle = $proxyPerson->name_middle;
             if (strlen($proxyPerson->name_first) > strlen($formPerson->name_first)) {
                 $person->name_first = $proxyPerson->name_first;
             }
         } else {
             if (!$proxyPerson->name_middle || $proxyPerson->name_middle == '') {
                 $compatible = true;
                 if (strlen($proxyPerson->name_first) > strlen($formPerson->name_first)) {
                     $person->name_first = $proxyPerson->name_first;
                 }
             } else {
                 if (stripos($proxyPerson->name_middle, $formPerson->name_middle) === 0 || stripos($formPerson->name_middle, $proxyPerson->name_middle) === 0) {
                     $compatible = true;
                     if (strlen($proxyPerson->name_middle) > strlen($formPerson->name_middle)) {
                         $person->name_middle = $proxyPerson->name_middle;
                     }
                     if (strlen($proxyPerson->name_first) > strlen($formPerson->name_first)) {
                         $person->name_first = $proxyPerson->name_first;
                     }
                 }
             }
         }
     }
     //if names have proven compatible, then check generational suffixes (Jr, Sr etc)
     if ($compatible == true) {
         $form_suffixes = explode(' ', $formPerson->name_suffix);
         $form_generationals = array_intersect($form_suffixes, LsLanguage::$generationalSuffixes);
         $proxy_suffixes = explode(' ', $proxyPerson->name_suffix);
         $common_generationals = array_intersect($form_generationals, $proxy_suffixes);
         //if there are no generationals in the form 4 name, go ahead and grab prefixes, suffixes, etc from proxy name
         if (count($form_generationals) == 0 || ($form_generationals = $common_generationals)) {
             if (strlen($proxyPerson->name_nick) > strlen($formPerson->name_nick)) {
                 $person->name_nick = $proxyPerson->name_nick;
             }
             if (strlen($proxyPerson->name_prefix) > strlen($formPerson->name_prefix)) {
                 $person->name_prefix = $proxyPerson->name_prefix;
             }
             if (strlen($proxyPerson->name_suffix) > strlen($formPerson->name_suffix)) {
                 $person->name_suffix = $proxyPerson->name_suffix;
             }
         }
     }
     return $person;
 }

 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
//.........这里部分代码省略.........

 protected function generatePerson($name_str, $summary = null, $orgs = null)
 {
     $name_arr = LsLanguage::parseFlatName($name_str);
     extract($name_arr);
     $person = new Entity();
     $person->addExtension('Person');
     $person->name_prefix = $name_prefix;
     $person->name_first = $name_first;
     $person->name_middle = $name_middle;
     $person->name_last = $name_last;
     $person->name_suffix = $name_suffix;
     $person->name_nick = $name_nick;
     return $person;
 }

 protected function createCampaignCommittee($entityId, $recipientId)
 {
     $sql = 'SELECT cycle, name, committee_id FROM os_committee WHERE recipient_id = ? AND committee_id <> recipient_id AND name IS NOT NULL';
     $stmt = $this->rawDb->execute($sql, array($recipientId));
     /*
     if (!count($committees = $stmt->fetchAll(PDO::FETCH_ASSOC)))
     {
       $sql = 'SELECT fec_id FROM os_candidate WHERE candidate_id = ?';
       $stmt = $this->rawDb->execute($sql, array($recipientId));
       
       if (count($fecIds = $stmt->fetchAll(PDO::FETCH_COLUMN)))
       {
         $sql = 'SELECT name, committee_id FROM os_committee WHERE candidate_id IN (\'' . implode('\',\'', $fecIds) . '\')';
         $stmt = $this->rawDb->execute($sql);
         $committees = $stmt->fetchAll(PDO::FETCH_ASSOC);
       }
     }
     */
     $committees = array();
     //group committees by committee_id, and get most recent name
     foreach ($stmt->fetchAll(PDO::FETCH_NUM) as $row) {
         list($cycle, $name, $committeeId) = $row;
         if (isset($committees[$committeeId])) {
             if ($cycle > $committees[$committeeId]['cycle']) {
                 $committees[$committeeId]['cycle'] = $cycle;
                 $committees[$committeeId]['name'] = $name;
             }
             $committees[$committeeId]['aliases'][] = $name;
         } else {
             $committees[$committeeId] = array('cycle' => $cycle, 'name' => $name, 'aliases' => array());
         }
     }
     foreach ($committees as $committeeId => $ary) {
         $name = $ary['name'];
         $aliases = array_unique($ary['aliases']);
         $sql = 'SELECT e.* FROM entity e LEFT JOIN political_fundraising p ON (p.entity_id = e.id) ' . 'WHERE p.fec_id = ? AND e.is_deleted = 0';
         $stmt = $this->db->execute($sql, array($committeeId));
         if ($entity = $stmt->fetch(PDO::FETCH_ASSOC)) {
             //check for an existing relationship
             $sql = 'SELECT COUNT(*) FROM relationship r ' . 'WHERE r.entity1_id = ? AND r.entity2_id = ? AND r.category_id = ? AND r.is_deleted = 0';
             $stmt = $this->db->execute($sql, array($entityId, $entity['id'], RelationshipTable::POSITION_CATEGORY));
             $createRel = $stmt->fetch(PDO::FETCH_COLUMN) ? false : true;
         } else {
             Doctrine_Manager::getInstance()->setCurrentConnection('main');
             //create new entity and relationship
             $entity = new Entity();
             $entity->addExtension('Org');
             $entity->addExtension('PoliticalFundraising');
             $entity->name = $name;
             $entity->fec_id = $committeeId;
             $entity->save();
             if ($this->debugMode) {
                 print "+ Created new entity for " . $name . " (" . $committeeId . ")\n";
             }
             $createRel = true;
         }
         if ($createRel) {
             Doctrine_Manager::getInstance()->setCurrentConnection('main');
             //create relationship
             $rel = new Relationship();
             $rel->setCategory('Position');
             $rel->entity1_id = $entityId;
             $rel->entity2_id = $entity['id'];
             $rel->description1 = 'Candidate';
             $rel->description2 = 'Political Fundraising Committee';
             $rel->is_executive = true;
             $rel->save();
             //create reference for the relationship
             $refName = 'FEC.gov - ' . $name;
             $refSource = $this->fecCommitteeBaseUrl . $committeeId;
             $sql = 'INSERT INTO reference (object_model, object_id, name, source) VALUES (?, ?, ?, ?)';
             $params = array('Relationship', $rel->id, $refName, $refSource);
             $stmt = $this->db->execute($sql, $params);
             if (!$stmt->rowCount()) {
                 throw new Exception("Couldn't insert Reference (" . implode(', ', $params) . ")");
             }
             if ($this->debugMode) {
                 print "+ Created position relationship between candidate (entity " . $entityId . ") and committee (entity " . $entity['id'] . ")\n";
             }
         }
         //create aliases if necessary
         $this->addAliasesToEntityById($entity['id'], $aliases);
     }
 }

 public function executeToolbarCreate($request)
 {
     $this->checkToolbarCredentials();
     if (!$request->isMethod('post')) {
         $this->forward404();
     }
     $name = $request->getParameter('name');
     $ext = $request->getParameter('ext');
     $blurb = $request->getParameter('blurb');
     $position = $request->getParameter('position');
     $listId = $request->getParameter("list_id");
     //save list_id to session for further use
     //$this->getUser()->setAttribute('list' . $position . '_id', $listId);
     if (!$name || !$ext) {
         $this->forward404();
     }
     if ($ext == 'Person') {
         $entity = PersonTable::parseFlatName($name);
         $entity->name = $name;
         if (!$entity->name_last) {
             $this->forward404();
         }
     } else {
         $entity = new Entity();
         $entity->addExtension('Org');
         $entity->name = $name;
     }
     $entity->blurb = $blurb;
     $entity->save();
     if ($listId) {
         $db = Doctrine_Manager::connection();
         $sql = "SELECT COUNT(*) FROM ls_list WHERE id = ? AND is_network = 0 AND is_deleted = 0";
         if (!$this->getUser()->hasCredential('admin')) {
             $sql .= " AND is_admin = 0";
         }
         $stmt = $db->execute($sql, array($listId));
         $count = $stmt->fetch(PDO::FETCH_COLUMN);
         if ($count == "1") {
             $sql = "SELECT COUNT(*) FROM ls_list_entity WHERE list_id = ? AND entity_id = ? AND is_deleted = 0";
             $stmt = $db->execute($sql, array($listId, $entity->id));
             $count = $stmt->fetch(PDO::FETCH_COLUMN);
             if ($count == "0") {
                 $le = new LsListEntity();
                 $le->list_id = $listId;
                 $le->entity_id = $entity->id;
                 $le->save();
             }
         } else {
             //save list_id to session for further use
             //$this->getUser()->setAttribute('list' . $position . '_id', null);
         }
     }
     $this->entity = $entity;
     $this->getResponse()->setHttpHeader('Content-Type', 'application/json; charset=utf-8');
 }