avoid dropping tokens completely just because they are expensive. Use ' ' token...

[nominatim.git] / website / search.php

diff --git a/website/search.php b/website/search.php
index 80a05a6442c142a1fcc969c34d046c6d60d9bf70..2c1d2befecce66307a15714e7254c70fd53d378c 100755 (executable)
--- a/website/search.php
+++ b/website/search.php
@@ -416,6 +416,13 @@
                        $aWordFrequencyScores = array();
                        foreach($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;
@@ -631,10 +638,30 @@
                                                                                                $aSearch = $aCurrentSearch;
                                                                                                $aSearch['iSearchRank'] += 1;
                                                                                                if ($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[] = $aSearch;
+                                                                                                       if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+                                                                                               }
                                                                                        }
 
                                                                                        if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
@@ -893,6 +920,7 @@
                                                        // TODO: filter out the pointless search terms (2 letter name tokens and less)
                                                        // they might be right - but they are just too darned expensive to run
                                                        if (sizeof($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'] != $aSearch['aAddress']) 
                                                        {
                                                                // For infrequent name terms disable index usage for address
@@ -900,11 +928,12 @@
                                                                        sizeof($aSearch['aName']) == 1 && 
                                                                        $aWordFrequencyScores[$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'])."'";