]> git.openstreetmap.org Git - nominatim.git/blobdiff - lib/Geocode.php
projects / nominatim.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
penalize search order where a country comes first
[nominatim.git] / lib / Geocode.php
diff --git a/lib/Geocode.php b/lib/Geocode.php
index eae3baf89aa1079c821c5bf407e40fc5dae3f7d5..bced2a36a555c2166190f9d0c4d0f5caedd93fd3 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -51,10 +51,22 @@ class Geocode
     protected $sQuery = false;
     protected $aStructuredQuery = false;
 
     protected $sQuery = false;
     protected $aStructuredQuery = false;
 
+    protected $oNormalizer = null;
+
 
     public function __construct(&$oDB)
     {
         $this->oDB =& $oDB;
 
     public function __construct(&$oDB)
     {
         $this->oDB =& $oDB;
+        $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
+    }
+
+    private function normTerm($sTerm)
+    {
+        if ($this->oNormalizer === null) {
+            return $sTerm;
+        }
+
+        return $this->oNormalizer->transliterate($sTerm);
     }
 
     public function setReverseInPlan($bReverse)
     }
 
     public function setReverseInPlan($bReverse)
@@ -292,7 +304,7 @@ class Geocode
         $aViewbox = $oParams->getStringList('viewboxlbrt');
         if ($aViewbox) {
             if (count($aViewbox) != 4) {
         $aViewbox = $oParams->getStringList('viewboxlbrt');
         if ($aViewbox) {
             if (count($aViewbox) != 4) {
-                userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
+                userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
             }
             $this->setViewbox($aViewbox);
         } else {
             }
             $this->setViewbox($aViewbox);
         } else {
@@ -301,12 +313,7 @@ class Geocode
                 if (count($aViewbox) != 4) {
                     userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
                 }
                 if (count($aViewbox) != 4) {
                     userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
                 }
-                $this->setViewBox(array(
-                                   $aViewbox[0],
-                                   $aViewbox[3],
-                                   $aViewbox[2],
-                                   $aViewbox[1]
-                                  ));
+                $this->setViewBox($aViewbox);
             } else {
                 $aRoute = $oParams->getStringList('route');
                 $fRouteWidth = $oParams->getFloat('routewidth');
             } else {
                 $aRoute = $oParams->getStringList('route');
                 $fRouteWidth = $oParams->getFloat('routewidth');
@@ -360,7 +367,7 @@ class Geocode
         $this->aAddressRankList = array();
 
         $this->aStructuredQuery = array();
         $this->aAddressRankList = array();
 
         $this->aStructuredQuery = array();
-        $this->sAllowedTypesSQLList = '';
+        $this->sAllowedTypesSQLList = False;
 
         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 
         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
@@ -373,7 +380,7 @@ class Geocode
         if (sizeof($this->aStructuredQuery) > 0) {
             $this->sQuery = join(', ', $this->aStructuredQuery);
             if ($this->iMaxAddressRank < 30) {
         if (sizeof($this->aStructuredQuery) > 0) {
             $this->sQuery = join(', ', $this->aStructuredQuery);
             if ($this->iMaxAddressRank < 30) {
-                $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+                $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
             }
         }
     }
             }
         }
     }
@@ -410,8 +417,15 @@ class Geocode
         $sPlaceIDs = join(',', array_keys($aPlaceIDs));
 
         $sImportanceSQL = '';
         $sPlaceIDs = join(',', array_keys($aPlaceIDs));
 
         $sImportanceSQL = '';
-        if ($this->sViewboxSmallSQL) $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
-        if ($this->sViewboxLargeSQL) $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
+        $sImportanceSQLGeom = '';
+        if ($this->sViewboxSmallSQL) {
+            $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
+            $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, geometry) THEN 1 ELSE 0.75 END * ";
+        }
+        if ($this->sViewboxLargeSQL) {
+            $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
+            $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, geometry) THEN 1 ELSE 0.75 END * ";
+        }
 
         $sSQL  = "SELECT ";
         $sSQL .= "    osm_type,";
 
         $sSQL  = "SELECT ";
         $sSQL .= "    osm_type,";
@@ -480,7 +494,7 @@ class Geocode
         $sSQL .= "UNION ";
         $sSQL .= "SELECT";
         $sSQL .= "  'P' as osm_type,";
         $sSQL .= "UNION ";
         $sSQL .= "SELECT";
         $sSQL .= "  'P' as osm_type,";
-        $sSQL .= "  (SELECT osm_id from placex p WHERE p.place_id = parent_place_id) as osm_id,";
+        $sSQL .= "  (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,";
         $sSQL .= "  'place' as class, 'postcode' as type,";
         $sSQL .= "  null as admin_level, rank_search, rank_address,";
         $sSQL .= "  place_id, parent_place_id, country_code,";
         $sSQL .= "  'place' as class, 'postcode' as type,";
         $sSQL .= "  null as admin_level, rank_search, rank_address,";
         $sSQL .= "  place_id, parent_place_id, country_code,";
@@ -490,19 +504,19 @@ class Geocode
         if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
         if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
         $sSQL .= "  ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
         if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
         if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
         $sSQL .= "  ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
-        $sSQL .=    $sImportanceSQL."(0.75-(rank_search::float/40)) AS importance, ";
+        $sSQL .=    $sImportanceSQLGeom."(0.75-(rank_search::float/40)) AS importance, ";
         $sSQL .= "  (";
         $sSQL .= "     SELECT max(p.importance*(p.rank_address+2))";
         $sSQL .= "     FROM ";
         $sSQL .= "       place_addressline s, ";
         $sSQL .= "       placex p";
         $sSQL .= "  (";
         $sSQL .= "     SELECT max(p.importance*(p.rank_address+2))";
         $sSQL .= "     FROM ";
         $sSQL .= "       place_addressline s, ";
         $sSQL .= "       placex p";
-        $sSQL .= "     WHERE s.place_id = parent_place_id";
+        $sSQL .= "     WHERE s.place_id = lp.parent_place_id";
         $sSQL .= "       AND p.place_id = s.address_place_id ";
         $sSQL .= "       AND s.isaddress";
         $sSQL .= "       AND p.importance is not null";
         $sSQL .= "  ) AS addressimportance, ";
         $sSQL .= "  null AS extra_place ";
         $sSQL .= "       AND p.place_id = s.address_place_id ";
         $sSQL .= "       AND s.isaddress";
         $sSQL .= "       AND p.importance is not null";
         $sSQL .= "  ) AS addressimportance, ";
         $sSQL .= "  null AS extra_place ";
-        $sSQL .= "FROM location_postcode";
+        $sSQL .= "FROM location_postcode lp";
         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 
         if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 
         if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
@@ -695,6 +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];
@@ -718,6 +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']++;
@@ -729,38 +747,36 @@ 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) {
+                                            $iGlobalRank++;
+                                        }
                                     }
                                     }
-                                } elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) {
-                                    if ($aSearch['oNear'] === false) {
-                                        $aSearch['oNear'] = new NearPoint(
-                                            $aSearchTerm['lat'],
-                                            $aSearchTerm['lon'],
-                                            $aSearchTerm['radius']
-                                        );
-                                        if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
-                                    }
-                                } elseif ($sPhraseType == 'postalcode' || ($aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode')) {
+                                } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode') {
                                     // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
                                     // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
-                                    if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
+                                    if ($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.
                                         // If we have structured search or this is the first term,
                                         // make the postcode the primary search element.
-                                        if ($aSearchTerm['operator'] == '' && ($sPhraseType == 'postalcode' || sizeof($aSearch['aName']) == 0)) {
+                                        if (!$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']] = substr($aSearchTerm['word_token'], 1);
+                                            $aNewSearch['aName'] = array($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,
                                         // add the postcode as an addendum.
                                         }
 
                                         // If we have a structured search or this is not the first term,
                                         // add the postcode as an addendum.
-                                        if ($sPhraseType == 'postalcode' || sizeof($aSearch['aName'])) {
-                                            $aSearch['sPostcode'] = $aSearchTerm['word_token'];
+                                        if ($aSearch['sOperator'] !== 'postcode' && ($sPhraseType == 'postalcode' || sizeof($aSearch['aName']))) {
+                                            $aSearch['sPostcode'] = $aSearchTerm['word'];
                                             if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                         }
                                     }
                                 } elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') {
                                             if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                         }
                                     }
                                 } elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') {
-                                    if ($aSearch['sHouseNumber'] === '') {
+                                    if ($aSearch['sHouseNumber'] === '' && $aSearch['sOperator'] !== 'postcode') {
                                         $aSearch['sHouseNumber'] = $sToken;
                                         // sanity check: if the housenumber is not mainly made
                                         // up of numbers, add a penalty
                                         $aSearch['sHouseNumber'] = $sToken;
                                         // sanity check: if the housenumber is not mainly made
                                         // up of numbers, add a penalty
@@ -779,7 +795,7 @@ class Geocode
                                     // require a normalized exact match of the term
                                     // if we have the normalizer version of the query
                                     // available
                                     // require a normalized exact match of the term
                                     // if we have the normalizer version of the query
                                     // available
-                                    if ($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'];
@@ -904,6 +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;
     }
 
@@ -943,13 +973,7 @@ class Geocode
     {
         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 
     {
         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 
-        $oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
-        if ($oNormalizer !== null) {
-            $sNormQuery = $oNormalizer->transliterate($this->sQuery);
-        } else {
-            $sNormQuery = null;
-        }
-
+        $sNormQuery = $this->normTerm($this->sQuery);
         $sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]";
         $sCountryCodesSQL = false;
         if ($this->aCountryCodes) {
         $sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]";
         $sCountryCodesSQL = false;
         if ($this->aCountryCodes) {
@@ -1016,15 +1040,24 @@ class Geocode
 
             // Any 'special' terms in the search?
             $bSpecialTerms = false;
 
             // Any 'special' terms in the search?
             $bSpecialTerms = false;
-            preg_match_all('/\\[(.*)=(.*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
-            $aSpecialTerms = array();
+            preg_match_all('/\\[([\\w_]*)=([\\w_]*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
             foreach ($aSpecialTermsRaw as $aSpecialTerm) {
                 $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
             foreach ($aSpecialTermsRaw as $aSpecialTerm) {
                 $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
-                $aSpecialTerms[strtolower($aSpecialTerm[1])] = $aSpecialTerm[2];
+                if (!$bSpecialTerms) {
+                    $aNewSearches = array();
+                    foreach ($aSearches as $aSearch) {
+                        $aNewSearch = $aSearch;
+                        $aNewSearch['sClass'] = $aSpecialTerm[1];
+                        $aNewSearch['sType'] = $aSpecialTerm[2];
+                        $aNewSearches[] = $aNewSearch;
+                    }
+
+                    $aSearches = $aNewSearches;
+                    $bSpecialTerms = true;
+                }
             }
 
             preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
             }
 
             preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
-            $aSpecialTerms = array();
             if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity']) {
                 $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
                 unset($this->aStructuredQuery['amenity']);
             if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity']) {
                 $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
                 unset($this->aStructuredQuery['amenity']);
@@ -1032,32 +1065,28 @@ class Geocode
 
             foreach ($aSpecialTermsRaw as $aSpecialTerm) {
                 $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 
             foreach ($aSpecialTermsRaw as $aSpecialTerm) {
                 $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
-                $sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".$aSpecialTerm[1]."') AS string"));
+                if ($bSpecialTerms) {
+                    continue;
+                }
+
+                $sToken = chksql($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 .= '   FROM word ';
                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
                 $sSQL .= ') AS x ';
                 $sSQL = 'SELECT * ';
                 $sSQL .= 'FROM ( ';
                 $sSQL .= '   SELECT word_id, word_token, word, class, type, country_code, operator';
                 $sSQL .= '   FROM word ';
                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
                 $sSQL .= ') AS x ';
-                $sSQL .= ' WHERE (class is not null AND class not in (\'place\')) ';
-                $sSQL .= ' OR country_code is not null';
+                $sSQL .= ' WHERE (class is not null AND class not in (\'place\'))';
                 if (CONST_Debug) var_Dump($sSQL);
                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
                 $aNewSearches = array();
                 foreach ($aSearches as $aSearch) {
                     foreach ($aSearchWords as $aSearchTerm) {
                         $aNewSearch = $aSearch;
                 if (CONST_Debug) var_Dump($sSQL);
                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
                 $aNewSearches = array();
                 foreach ($aSearches as $aSearch) {
                     foreach ($aSearchWords as $aSearchTerm) {
                         $aNewSearch = $aSearch;
-                        if ($aSearchTerm['country_code']) {
-                            $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
-                            $aNewSearches[] = $aNewSearch;
-                            $bSpecialTerms = true;
-                        }
-                        if ($aSearchTerm['class']) {
-                            $aNewSearch['sClass'] = $aSearchTerm['class'];
-                            $aNewSearch['sType'] = $aSearchTerm['type'];
-                            $aNewSearches[] = $aNewSearch;
-                            $bSpecialTerms = true;
-                        }
+                        $aNewSearch['sClass'] = $aSearchTerm['class'];
+                        $aNewSearch['sType'] = $aSearchTerm['type'];
+                        $aNewSearches[] = $aNewSearch;
+                        $bSpecialTerms = true;
                     }
                 }
                 $aSearches = $aNewSearches;
                     }
                 }
                 $aSearches = $aNewSearches;
@@ -1106,14 +1135,10 @@ class Geocode
                 if (CONST_Debug) var_Dump($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 $aToken) {
@@ -1134,21 +1159,9 @@ class Geocode
                 }
                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 
                 }
                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 
-                // Try and calculate GB postcodes we might be missing
+                // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
                 foreach ($aTokens as $sToken) {
                 foreach ($aTokens as $sToken) {
-                    // Source of gb postcodes is now definitive - always use
-                    if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) {
-                        if (substr($aData[1], -2, 1) != ' ') {
-                            $aData[0] = substr($aData[0], 0, strlen($aData[1])-1).' '.substr($aData[0], strlen($aData[1])-1);
-                            $aData[1] = substr($aData[1], 0, -1).' '.substr($aData[1], -1, 1);
-                        }
-                        $aGBPostcodeLocation = gbPostcodeCalculate($aData[0], $aData[1], $aData[2], $this->oDB);
-                        if ($aGBPostcodeLocation) {
-                            $aValidTokens[$sToken] = $aGBPostcodeLocation;
-                        }
-                    } elseif (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
-                        // US ZIP+4 codes - if there is no token,
-                        // merge in the 5-digit ZIP code
+                    if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
                         if (isset($aValidTokens[$aData[1]])) {
                             foreach ($aValidTokens[$aData[1]] as $aToken) {
                                 if (!$aToken['class']) {
                         if (isset($aValidTokens[$aData[1]])) {
                             foreach ($aValidTokens[$aData[1]] as $aToken) {
                                 if (!$aToken['class']) {
@@ -1215,7 +1228,6 @@ class Geocode
                 ksort($aGroupedSearches);
             }
 
                 ksort($aGroupedSearches);
             }
 
-            if (CONST_Debug) var_Dump($aGroupedSearches);
             if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0) {
                 $aCopyGroupedSearches = $aGroupedSearches;
                 foreach ($aCopyGroupedSearches as $iGroup => $aSearches) {
             if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0) {
                 $aCopyGroupedSearches = $aGroupedSearches;
                 foreach ($aCopyGroupedSearches as $iGroup => $aSearches) {
@@ -1272,13 +1284,16 @@ class Geocode
                     if (CONST_Debug) echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
                     if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
 
                     if (CONST_Debug) echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
                     if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
 
+                    if ($sCountryCodesSQL && $aSearch['sCountryCode'] && !in_array($aSearch['sCountryCode'], $this->aCountryCodes)) {
+                        continue;
+                    }
+
                     // No location term?
                     // No location term?
-                    if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['oNear']) {
-                        if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) {
+                    if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress'])) {
+                        if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber'] && !$aSearch['oNear']) {
                             // Just looking for a country by code - look it up
                             if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank) {
                                 $sSQL = "SELECT place_id FROM placex WHERE country_code='".$aSearch['sCountryCode']."' AND rank_search = 4";
                             // Just looking for a country by code - look it up
                             if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank) {
                                 $sSQL = "SELECT place_id FROM placex WHERE country_code='".$aSearch['sCountryCode']."' AND rank_search = 4";
-                                if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
                                 if ($bBoundingBoxSearch)
                                     $sSQL .= " AND _st_intersects($this->sViewboxSmallSQL, geometry)";
                                 $sSQL .= " ORDER BY st_area(geometry) DESC LIMIT 1";
                                 if ($bBoundingBoxSearch)
                                     $sSQL .= " AND _st_intersects($this->sViewboxSmallSQL, geometry)";
                                 $sSQL .= " ORDER BY st_area(geometry) DESC LIMIT 1";
@@ -1295,39 +1310,32 @@ class Geocode
                             if (chksql($this->oDB->getOne($sSQL))) {
                                 $sSQL = "SELECT place_id FROM place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
                                 if ($sCountryCodesSQL) $sSQL .= " JOIN placex USING (place_id)";
                             if (chksql($this->oDB->getOne($sSQL))) {
                                 $sSQL = "SELECT place_id FROM place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
                                 if ($sCountryCodesSQL) $sSQL .= " JOIN placex USING (place_id)";
-                                $sSQL .= " WHERE st_contains($this->sViewboxSmallSQL, ct.centroid)";
+                                if ($aSearch['oNear']) {
+                                    $sSQL .= " WHERE ".$aSearch['oNear']->withinSQL('ct.centroid');
+                                } else {
+                                    $sSQL .= " WHERE st_contains($this->sViewboxSmallSQL, ct.centroid)";
+                                }
                                 if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
                                 if (sizeof($this->aExcludePlaceIDs)) {
                                     $sSQL .= " AND place_id not in (".join(',', $this->aExcludePlaceIDs).")";
                                 }
                                 if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
                                 if (sizeof($this->aExcludePlaceIDs)) {
                                     $sSQL .= " AND place_id not in (".join(',', $this->aExcludePlaceIDs).")";
                                 }
-                                if ($this->sViewboxCentreSQL) $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
+                                if ($this->sViewboxCentreSQL) {
+                                    $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
+                                } elseif ($aSearch['oNear']) {
+                                    $sSQL .= " ORDER BY ".$aSearch['oNear']->distanceSQL('ct.centroid').' ASC';
+                                }
                                 $sSQL .= " limit $this->iLimit";
                                 if (CONST_Debug) var_dump($sSQL);
                                 $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
                                 $sSQL .= " limit $this->iLimit";
                                 if (CONST_Debug) var_dump($sSQL);
                                 $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
-
-                                // If excluded place IDs are given, it is fair to assume that
-                                // there have been results in the small box, so no further
-                                // expansion in that case.
-                                // Also don't expand if bounded results were requested.
-                                if (!sizeof($aPlaceIDs) && !sizeof($this->aExcludePlaceIDs) && !$this->bBoundedSearch) {
-                                    $sSQL = "SELECT place_id FROM place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
-                                    if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
-                                    $sSQL .= " WHERE ST_Contains($this->sViewboxLargeSQL, ct.centroid)";
-                                    if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
-                                    if ($this->sViewboxCentreSQL) $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
-                                    $sSQL .= " LIMIT $this->iLimit";
-                                    if (CONST_Debug) var_dump($sSQL);
-                                    $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
-                                }
-                            } else {
+                            } else if ($aSearch['oNear']) {
                                 $sSQL = "SELECT place_id ";
                                 $sSQL .= "FROM placex ";
                                 $sSQL .= "WHERE class='".$aSearch['sClass']."' ";
                                 $sSQL .= "  AND type='".$aSearch['sType']."'";
                                 $sSQL = "SELECT place_id ";
                                 $sSQL .= "FROM placex ";
                                 $sSQL .= "WHERE class='".$aSearch['sClass']."' ";
                                 $sSQL .= "  AND type='".$aSearch['sType']."'";
-                                $sSQL .= "  AND ST_Contains($this->sViewboxSmallSQL, geometry) ";
+                                $sSQL .= "  AND ".$aSearch['oNear']->withinSQL('geometry');
                                 $sSQL .= "  AND linked_place_id is null";
                                 if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
                                 $sSQL .= "  AND linked_place_id is null";
                                 if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
-                                if ($this->sViewboxCentreSQL)   $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, centroid) ASC";
+                                $sSQL .= " ORDER BY ".$aSearch['oNear']->distanceSQL('centroid')." ASC";
                                 $sSQL .= " LIMIT $this->iLimit";
                                 if (CONST_Debug) var_dump($sSQL);
                                 $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
                                 $sSQL .= " LIMIT $this->iLimit";
                                 if (CONST_Debug) var_dump($sSQL);
                                 $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
@@ -1349,8 +1357,7 @@ class Geocode
                         $sSQL .= "p.postcode = '".pg_escape_string(reset($aSearch['aName']))."'";
                         if ($aSearch['sCountryCode']) {
                             $sSQL .= " AND p.country_code = '".$aSearch['sCountryCode']."'";
                         $sSQL .= "p.postcode = '".pg_escape_string(reset($aSearch['aName']))."'";
                         if ($aSearch['sCountryCode']) {
                             $sSQL .= " AND p.country_code = '".$aSearch['sCountryCode']."'";
-                        }
-                        if ($sCountryCodesSQL) {
+                        } elseif ($sCountryCodesSQL) {
                             $sSQL .= " AND p.country_code in ($sCountryCodesSQL)";
                         }
                         $sSQL .= " LIMIT $this->iLimit";
                             $sSQL .= " AND p.country_code in ($sCountryCodesSQL)";
                         }
                         $sSQL .= " LIMIT $this->iLimit";
@@ -1409,7 +1416,7 @@ class Geocode
                         if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
                         if ($aSearch['sHouseNumber']) {
                             $aTerms[] = "address_rank between 16 and 27";
                         if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
                         if ($aSearch['sHouseNumber']) {
                             $aTerms[] = "address_rank between 16 and 27";
-                        } else {
+                        } elseif (!$aSearch['sClass'] || $aSearch['sOperator'] == 'name') {
                             if ($this->iMinAddressRank > 0) {
                                 $aTerms[] = "address_rank >= ".$this->iMinAddressRank;
                             }
                             if ($this->iMinAddressRank > 0) {
                                 $aTerms[] = "address_rank >= ".$this->iMinAddressRank;
                             }
@@ -1421,6 +1428,12 @@ class Geocode
                             $aTerms[] = $aSearch['oNear']->withinSQL('centroid');
 
                             $aOrder[] = $aSearch['oNear']->distanceSQL('centroid');
                             $aTerms[] = $aSearch['oNear']->withinSQL('centroid');
 
                             $aOrder[] = $aSearch['oNear']->distanceSQL('centroid');
+                        } elseif ($aSearch['sPostcode']) {
+                            if (!sizeof($aSearch['aAddress'])) {
+                                $aTerms[] = "EXISTS(SELECT place_id FROM location_postcode p WHERE p.postcode = '".$aSearch['sPostcode']."' AND ST_DWithin(search_name.centroid, p.geometry, 0.1))";
+                            } else {
+                                $aOrder[] = "(SELECT min(ST_Distance(search_name.centroid, p.geometry)) FROM location_postcode p WHERE p.postcode = '".$aSearch['sPostcode']."')";
+                            }
                         }
                         if (sizeof($this->aExcludePlaceIDs)) {
                             $aTerms[] = "place_id not in (".join(',', $this->aExcludePlaceIDs).")";
                         }
                         if (sizeof($this->aExcludePlaceIDs)) {
                             $aTerms[] = "place_id not in (".join(',', $this->aExcludePlaceIDs).")";
@@ -1570,7 +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
                                 $searchedHousenumber = -1;
@@ -1646,7 +1660,7 @@ class Geocode
                                         } elseif ($sPlaceIDs) {
                                             $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
                                         } elseif ($sPlaceGeom) {
                                         } elseif ($sPlaceIDs) {
                                             $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
                                         } elseif ($sPlaceGeom) {
-                                            $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
+                                            $sOrderBySQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
                                         }
 
                                         $sSQL = "select distinct i.place_id".($sOrderBySQL?', i.order_term':'')." from (";
                                         }
 
                                         $sSQL = "select distinct i.place_id".($sOrderBySQL?', i.order_term':'')." from (";
@@ -1682,7 +1696,7 @@ class Geocode
                                             $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
                                         }
 
                                             $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
                                         }
 
-                                        $sSQL = "SELECT distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'');
+                                        $sSQL = "SELECT distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'');
                                         $sSQL .= " FROM placex as l, placex as f ";
                                         $sSQL .= " WHERE f.place_id in ($sPlaceIDs) ";
                                         $sSQL .= "  AND ST_DWithin(l.geometry, f.centroid, $fRange) ";
                                         $sSQL .= " FROM placex as l, placex as f ";
                                         $sSQL .= " WHERE f.place_id in ($sPlaceIDs) ";
                                         $sSQL .= "  AND ST_DWithin(l.geometry, f.centroid, $fRange) ";
@@ -1692,7 +1706,7 @@ class Geocode
                                             $sSQL .= " AND l.place_id not in (".join(',', $this->aExcludePlaceIDs).")";
                                         }
                                         if ($sCountryCodesSQL) $sSQL .= " AND l.country_code in ($sCountryCodesSQL)";
                                             $sSQL .= " AND l.place_id not in (".join(',', $this->aExcludePlaceIDs).")";
                                         }
                                         if ($sCountryCodesSQL) $sSQL .= " AND l.country_code in ($sCountryCodesSQL)";
-                                        if ($sOrderBy) $sSQL .= "ORDER BY ".$OrderBysSQL." ASC";
+                                        if ($sOrderBySQL) $sSQL .= "ORDER BY ".$sOrderBySQL." ASC";
                                         if ($this->iOffset) $sSQL .= " OFFSET $this->iOffset";
                                         $sSQL .= " limit $this->iLimit";
                                         if (CONST_Debug) var_dump($sSQL);
                                         if ($this->iOffset) $sSQL .= " OFFSET $this->iOffset";
                                         $sSQL .= " limit $this->iLimit";
                                         if (CONST_Debug) var_dump($sSQL);
@@ -1709,6 +1723,21 @@ class Geocode
                         var_Dump($aPlaceIDs);
                     }
 
                         var_Dump($aPlaceIDs);
                     }
 
+                    if (sizeof($aPlaceIDs) && $aSearch['sPostcode']) {
+                        $sSQL = 'SELECT place_id FROM placex';
+                        $sSQL .= ' WHERE place_id in ('.join(',', $aPlaceIDs).')';
+                        $sSQL .= " AND postcode = '".pg_escape_string($aSearch['sPostcode'])."'";
+                        if (CONST_Debug) var_dump($sSQL);
+                        $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
+                        if ($aFilteredPlaceIDs) {
+                            $aPlaceIDs = $aFilteredPlaceIDs;
+                            if (CONST_Debug) {
+                                echo "<br><b>Place IDs after postcode filtering:</b> ";
+                                var_Dump($aPlaceIDs);
+                            }
+                        }
+                    }
+
                     foreach ($aPlaceIDs as $iPlaceID) {
                         // array for placeID => -1 | Tiger housenumber
                         $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber;
                     foreach ($aPlaceIDs as $iPlaceID) {
                         // array for placeID => -1 | Tiger housenumber
                         $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber;
@@ -1720,9 +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 <= $this->iMaxAddressRank) {
@@ -1731,20 +1762,23 @@ class Geocode
                     if ($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();
Open Source search based on OpenStreetMap data