revert customizations prior to merge of refactoring

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index f142731dac0f3233d96d59c12b86b967081a898c..15915f940133273991916093a6d145e75cbe8ece 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -709,6 +709,8 @@ class Geocode
 
              Score how good the search is so they can be ordered
          */
 
              Score how good the search is so they can be ordered
          */

+        $iGlobalRank = 0;
+

         foreach ($aPhrases as $iPhrase => $aPhrase) {
             $aNewPhraseSearches = array();
             if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
         foreach ($aPhrases as $iPhrase => $aPhrase) {
             $aNewPhraseSearches = array();
             if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];


@@ -732,6 +734,8 @@ class Geocode
 
                         // If the token is valid
                         if (isset($aValidTokens[' '.$sToken])) {
 
                         // If the token is valid
                         if (isset($aValidTokens[' '.$sToken])) {

+                            // TODO variable should go into aCurrentSearch
+                            $bHavePostcode = false;

                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
                                 $aSearch = $aCurrentSearch;
                                 $aSearch['iSearchRank']++;
                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
                                 $aSearch = $aCurrentSearch;
                                 $aSearch['iSearchRank']++;


@@ -743,19 +747,25 @@ class Geocode
                                             $aSearch['iSearchRank'] += 5;
                                         }
                                         if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                             $aSearch['iSearchRank'] += 5;
                                         }
                                         if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;

+                                        // If it is at the beginning, we can be almost sure that this is the wrong order
+                                        // Increase score for all searches.
+                                        if ($iToken == 0 && $iPhrase == 0) {
+                                            $iGlobalRank++;
+                                        }

                                     }
                                     }

-                                } elseif ($sPhraseType == 'postalcode' || ($aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode')) {
+                                } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode') {

                                     // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
                                     // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both

-                                    if ($aSearch['sPostcode'] === '' && $aSearch['sHouseNumber'] === '' &&
+                                    if ($aSearch['sPostcode'] === '' &&

                                         isset($aSearchTerm['word']) && $aSearchTerm['word'] && strpos($sNormQuery, $this->normTerm($aSearchTerm['word'])) !== false) {
                                         // If we have structured search or this is the first term,
                                         // make the postcode the primary search element.
                                         isset($aSearchTerm['word']) && $aSearchTerm['word'] && strpos($sNormQuery, $this->normTerm($aSearchTerm['word'])) !== false) {
                                         // If we have structured search or this is the first term,
                                         // make the postcode the primary search element.

-                                        if ($aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) {
+                                        if (!$bHavePostcode && $aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) {

                                             $aNewSearch = $aSearch;
                                             $aNewSearch['sOperator'] = 'postcode';
                                             $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);
                                             $aNewSearch = $aSearch;
                                             $aNewSearch['sOperator'] = 'postcode';
                                             $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);

-                                            $aNewSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word'];
+                                            $aNewSearch['aName'] = array($aSearchTerm['word_id'] => $aSearchTerm['word']);

                                             if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch;
                                             if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch;

+                                            $bHavePostcode = true;

                                         }
 
                                         // If we have a structured search or this is not the first term,
                                         }
 
                                         // If we have a structured search or this is not the first term,


@@ -771,8 +781,8 @@ class Geocode
                                         // sanity check: if the housenumber is not mainly made
                                         // up of numbers, add a penalty
                                         if (preg_match_all("/[^0-9]/", $sToken, $aMatches) > 2) $aSearch['iSearchRank']++;
                                         // sanity check: if the housenumber is not mainly made
                                         // up of numbers, add a penalty
                                         if (preg_match_all("/[^0-9]/", $sToken, $aMatches) > 2) $aSearch['iSearchRank']++;

-                                        // also housenumbers should appear in the first or second phrase
-                                        if ($iPhrase > 1) $aSearch['iSearchRank'] += 1;
+                                        // also must not appear in the middle of the address
+                                        if ($aSearch['aAddress'] || $aSearch['aAddressNonSearch']) $aSearch['iSearchRank'] += 1;

                                         if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                         /*
                                         // Fall back to not searching for this item (better than nothing)
                                         if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                         /*
                                         // Fall back to not searching for this item (better than nothing)


@@ -785,7 +795,7 @@ class Geocode
                                     // require a normalized exact match of the term
                                     // if we have the normalizer version of the query
                                     // available
                                     // require a normalized exact match of the term
                                     // if we have the normalizer version of the query
                                     // available

-                                    if ($aSearch['sClass'] === ''
+                                    if ($aSearch['sOperator'] === ''

                                         && ($sNormQuery === null || !($aSearchTerm['word'] && strpos($sNormQuery, $aSearchTerm['word']) === false))) {
                                         $aSearch['sClass'] = $aSearchTerm['class'];
                                         $aSearch['sType'] = $aSearchTerm['type'];
                                         && ($sNormQuery === null || !($aSearchTerm['word'] && strpos($sNormQuery, $aSearchTerm['word']) === false))) {
                                         $aSearch['sClass'] = $aSearchTerm['class'];
                                         $aSearch['sType'] = $aSearchTerm['type'];


@@ -849,7 +859,8 @@ class Geocode
                                         }
                                     }
 
                                         }
                                     }
 

-                                    if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) {
+                                    if ((!$aCurrentSearch['sPostcode'] && !$aCurrentSearch['aAddress'] && !$aCurrentSearch['aAddressNonSearch'])
+                                        && (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)) {

                                         $aSearch = $aCurrentSearch;
                                         $aSearch['iSearchRank'] += 1;
                                         if (!sizeof($aCurrentSearch['aName'])) $aSearch['iSearchRank'] += 1;
                                         $aSearch = $aCurrentSearch;
                                         $aSearch['iSearchRank'] += 1;
                                         if (!sizeof($aCurrentSearch['aName'])) $aSearch['iSearchRank'] += 1;


@@ -919,6 +930,7 @@ class Geocode
                     continue;
                 }
             }
                     continue;
                 }
             }

+            $aSearch['iSearchRank'] += $iGlobalRank;

             $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
         }
         ksort($aGroupedSearches);
             $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
         }
         ksort($aGroupedSearches);


@@ -1058,7 +1070,7 @@ class Geocode
                     continue;
                 }
 
                     continue;
                 }
 

-                $sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".$aSpecialTerm[1]."') AS string"));
+                $sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".pg_escape_string($aSpecialTerm[1])."') AS string"));

                 $sSQL = 'SELECT * ';
                 $sSQL .= 'FROM ( ';
                 $sSQL .= '   SELECT word_id, word_token, word, class, type, country_code, operator';
                 $sSQL = 'SELECT * ';
                 $sSQL .= 'FROM ( ';
                 $sSQL .= '   SELECT word_id, word_token, word, class, type, country_code, operator';


@@ -1387,20 +1399,19 @@ class Geocode
                         // 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'], ",")."]";
                         // 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['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
                                 && $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold
                             ) {
                         if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) {
                             // For infrequent name terms disable index usage for address
                             if (CONST_Search_NameOnlySearchFrequencyThreshold
                                 && sizeof($aSearch['aName']) == 1
                                 && $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold
                             ) {

-                                //$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'], $aSearch['aAddressNonSearch']), ",")."]";
-                                $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($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'], ",")."]";
                             } else {
                                 $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'], ",")."]";

-                                /*if (sizeof($aSearch['aAddressNonSearch'])) {
+                                if (sizeof($aSearch['aAddressNonSearch'])) {

                                     $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'], ",")."]";
                                     $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'], ",")."]";

-                                }*/
+                                }

                             }
                         }
                         if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
                             }
                         }
                         if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";