penalize search order where a country comes first

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index fc2d6c65190ee17ecd0949c0f8c837336e88604f..bced2a36a555c2166190f9d0c4d0f5caedd93fd3 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,


@@ -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'];


@@ -910,6 +920,20 @@ class Geocode
 
             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
         }
 
             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
         }

+
+        // Revisit searches, giving penalty to unlikely combinations
+        $aGroupedSearches = array();
+        foreach ($aSearches as $aSearch) {
+            if (!$aSearch['aName']) {
+                if ($aSearch['sHouseNumber']) {
+                    continue;
+                }
+            }
+            $aSearch['iSearchRank'] += $iGlobalRank;
+            $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
+        }
+        ksort($aGroupedSearches);
+

         return $aGroupedSearches;
     }
 
         return $aGroupedSearches;
     }
 


@@ -1045,7 +1069,7 @@ 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';


@@ -1111,14 +1135,10 @@ class Geocode
                 if (CONST_Debug) var_Dump($sSQL);
 
                 $aValidTokens = array();
                 if (CONST_Debug) var_Dump($sSQL);
 
                 $aValidTokens = array();

-                if (sizeof($aTokens)) {
-                    $aDatabaseWords = chksql(
-                        $this->oDB->getAll($sSQL),
-                        "Could not get word tokens."
-                    );
-                } else {
-                    $aDatabaseWords = array();
-                }
+                $aDatabaseWords = chksql(
+                    $this->oDB->getAll($sSQL),
+                    "Could not get word tokens."
+                );

                 $aPossibleMainWordIDs = array();
                 $aWordFrequencyScores = array();
                 foreach ($aDatabaseWords as $aToken) {
                 $aPossibleMainWordIDs = array();
                 $aWordFrequencyScores = array();
                 foreach ($aDatabaseWords as $aToken) {


@@ -1563,7 +1583,8 @@ class Geocode
                             }
 
                             // Fallback to the road (if no housenumber was found)
                             }
 
                             // Fallback to the road (if no housenumber was found)

-                            if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) {
+                            if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])
+                                && ($aSearch['aAddress'] || $aSearch['sCountryCode'])) {

                                 $aPlaceIDs = $aRoadPlaceIDs;
                                 //set to -1, if no housenumbers were found
                                 $searchedHousenumber = -1;
                                 $aPlaceIDs = $aRoadPlaceIDs;
                                 //set to -1, if no housenumbers were found
                                 $searchedHousenumber = -1;


@@ -1728,9 +1749,11 @@ class Geocode
                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
                     // reduces the number of place ids, like a filter
                     // rank_address is 30 for interpolated housenumbers
                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
                     // reduces the number of place ids, like a filter
                     // rank_address is 30 for interpolated housenumbers

+                    $sWherePlaceId = 'WHERE place_id in (';
+                    $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') ';
+

                     $sSQL = "SELECT place_id ";
                     $sSQL = "SELECT place_id ";

-                    $sSQL .= "FROM placex ";
-                    $sSQL .= "WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).") ";
+                    $sSQL .= "FROM placex ".$sWherePlaceId;

                     $sSQL .= "  AND (";
                     $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
                     if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
                     $sSQL .= "  AND (";
                     $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
                     if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {


@@ -1739,20 +1762,23 @@ class Geocode
                     if ($this->aAddressRankList) {
                         $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
                     }
                     if ($this->aAddressRankList) {
                         $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
                     }

-                    if (CONST_Use_US_Tiger_Data) {
-                        $sSQL .= "  ) ";
+                    $sSQL .= "  ) UNION ";
+                    $sSQL .= " SELECT place_id FROM location_postcode lp ".$sWherePlaceId;
+                    $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
+                    if ($this->aAddressRankList) {
+                        $sSQL .= "     OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
+                    }
+                    $sSQL .= ") ";
+                    if (CONST_Use_US_Tiger_Data && $this->iMaxAddressRank == 30) {
+                        $sSQL .= "UNION ";
+                        $sSQL .= "  SELECT place_id ";
+                        $sSQL .= "  FROM location_property_tiger ".$sWherePlaceId;
+                    }
+                    if ($this->iMaxAddressRank == 30) {

                         $sSQL .= "UNION ";
                         $sSQL .= "  SELECT place_id ";
                         $sSQL .= "UNION ";
                         $sSQL .= "  SELECT place_id ";

-                        $sSQL .= "  FROM location_property_tiger ";
-                        $sSQL .= "  WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).") ";
-                        $sSQL .= "    AND (30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
-                        if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',', $this->aAddressRankList).")";
+                        $sSQL .= "  FROM location_property_osmline ".$sWherePlaceId;

                     }
                     }

-                    $sSQL .= ") UNION ";
-                    $sSQL .= "  SELECT place_id ";
-                    $sSQL .= "  FROM location_property_osmline ";
-                    $sSQL .= "  WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).")";
-                    $sSQL .= "    AND startnumber is not NULL AND (30 between $this->iMinAddressRank and $this->iMaxAddressRank)";

                     if (CONST_Debug) var_dump($sSQL);
                     $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
                     $tempIDs = array();
                     if (CONST_Debug) var_dump($sSQL);
                     $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
                     $tempIDs = array();