X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/f9a7f032391965df507318691897ad80eca4f926..adcafde48149001fa263f9c46a503233e2252d7f:/website/search.php

diff --git a/website/search.php b/website/search.php
index a1860f25..00ac3b95 100755
--- a/website/search.php
+++ b/website/search.php
@@ -286,6 +286,7 @@
 			// Start with a blank search
 			$aSearches = array(
 				array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), 
+					'aNameNonSearch'=>array(), 'aAddressNonSearch'=>array(),
 					'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'')
 			);
 
@@ -397,7 +398,7 @@
 			// Check which tokens we have, get the ID numbers			
 			$sSQL = 'select word_id,word_token, word, class, type, location, country_code, operator, search_name_count';
 			$sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$aTokens)).')';
-			$sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency;
+//			$sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency;
 //			$sSQL .= ' group by word_token, word, class, type, location, country_code';
 
 			if (CONST_Debug) var_Dump($sSQL);
@@ -412,8 +413,16 @@
 				failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords);
 			}
 			$aPossibleMainWordIDs = array();
+			$aWordFrequencyScores = array();
 			foreach($aDatabaseWords as $aToken)
 			{
+				// Very special case - require 2 letter country param to match the country code found
+				if ($bStructuredPhrases && $aToken['country_code'] && !empty($aStructuredQuery['country']) 
+                                  && strlen($aStructuredQuery['country']) == 2 && strtolower($aStructuredQuery['country']) != $aToken['country_code'])
+				{
+					continue;
+				}
+
 				if (isset($aValidTokens[$aToken['word_token']]))
 				{
 					$aValidTokens[$aToken['word_token']][] = $aToken;
@@ -422,7 +431,8 @@
 				{
 					$aValidTokens[$aToken['word_token']] = array($aToken);
 				}
-				if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1 + $aToken['search_name_count'];
+				if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
+				$aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
 			}
 			if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 
@@ -598,7 +608,7 @@
 										{
 											if (sizeof($aSearch['aName']))
 											{
-												if (($sPhraseType != 'street' && $sPhraseType != 'country') && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
+												if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
 												{
 													$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
 												}
@@ -623,12 +633,35 @@
 									{
 										if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
 										{
-											if (($sPhraseType != 'street') && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4)
+											if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4)
 											{
 	  											$aSearch = $aCurrentSearch;
 												$aSearch['iSearchRank'] += 1;
-												$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
-	  											if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+												if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
+												{
+													$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+		  											if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+												}
+												elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version?
+												{
+													foreach($aValidTokens[' '.$sToken] as $aSearchTermToken)
+													{
+														if (empty($aSearchTermToken['country_code']) 
+															&& empty($aSearchTermToken['lat'])
+															&& empty($aSearchTermToken['class']))
+														{
+	  														$aSearch = $aCurrentSearch;
+															$aSearch['iSearchRank'] += 1;
+															$aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
+				  											if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+														}
+													}
+												}
+												else
+												{
+													$aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+		  											if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+												}
 											}
 
 											if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
@@ -636,7 +669,10 @@
 	  											$aSearch = $aCurrentSearch;
 												$aSearch['iSearchRank'] += 2;
 												if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
-												$aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+												if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
+													$aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+												else
+													$aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
 												$aSearch['iNamePhrase'] = $iPhrase;
   												if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
 											}
@@ -884,18 +920,20 @@
 							// TODO: filter out the pointless search terms (2 letter name tokens and less)
 							// they might be right - but they are just too darned expensive to run
 							if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]";
+							if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]";
 							if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) 
 							{
 								// For infrequent name terms disable index usage for address
 								if (CONST_Search_NameOnlySearchFrequencyThreshold && 
 									sizeof($aSearch['aName']) == 1 && 
-									$aPossibleMainWordIDs[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
+									$aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
 								{
-									$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]";
+									$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]";
 								}
 								else
 								{
 									$aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";
+									if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]";
 								}
 							}
 							if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
@@ -1492,7 +1530,7 @@
 		if (sizeof($aSearchResults) >= $iFinalLimit) break;
 	}
 
-	$sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '1 day'::interval,'YYYY/MM/DD') from import_status limit 1");
+	$sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '2 minutes'::interval,'YYYY/MM/DD HH24:MI')||' GMT' from import_status limit 1");
 
 	if (isset($_GET['nearlat']) && isset($_GET['nearlon']))
 	{