include GB CodePoint data into location_postcode table

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 82c9b0b68db735285716dd7431f03637ba078133..0546983f19e33beefbdf6db29916eba57d6a1221 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -744,7 +744,7 @@ class Geocode
                                     if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
                                         // If we have structured search or this is the first term,
                                         // make the postcode the primary search element.
-                                        if ($sPhraseType == 'postalcode' || sizeof($aSearch['aName']) == 0) {
+                                        if ($aSearchTerm['operator'] == '' && ($sPhraseType == 'postalcode' || sizeof($aSearch['aName']) == 0)) {
                                             $aNewSearch = $aSearch;
                                             $aNewSearch['sOperator'] = 'postcode';
                                             $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);
@@ -755,7 +755,7 @@ class Geocode
                                         // If we have a structured search or this is not the first term,
                                         // add the postcode as an addendum.
                                         if ($sPhraseType == 'postalcode' || sizeof($aSearch['aName'])) {
-                                            $aSearch['sPostcode'] = $aSearchTerm['word_token'];
+                                            $aSearch['sPostcode'] = substr($aSearchTerm['word_token'], 1);
                                             if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                         }
                                     }
@@ -1134,21 +1134,9 @@ class Geocode
                 }
                 if (CONST_Debug) var_Dump($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 $sToken) {
-                    // Source of gb postcodes is now definitive - always use
-                    if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', 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']) {
@@ -1272,13 +1260,16 @@ 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 ($sCountryCodesSQL && $aSearch['sCountryCode'] && !in_array($aSearch['sCountryCode'], $this->aCountryCodes)) {
+                        continue;
+                    }
+
                     // No location term?
                     if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['oNear']) {
                         if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) {
                             // 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";
@@ -1338,13 +1329,19 @@ class Geocode
                         // be for a name or a special search. Ignore everythin else.
                         $aPlaceIDs = array();
                     } elseif ($aSearch['sOperator'] == 'postcode') {
-                        $sSQL  = "SELECT place_id FROM location_postcode ";
-                        $sSQL .= "WHERE postcode = '".pg_escape_string(reset($aSearch['aName']))."'";
-                        if ($aSearch['sCountryCode']) {
-                            $sSQL .= " AND country_code = '".$aSearch['sCountryCode']."'";
+                        $sSQL  = "SELECT p.place_id FROM location_postcode p ";
+                        if (sizeof($aSearch['aAddress'])) {
+                            $sSQL .= ", search_name s ";
+                            $sSQL .= "WHERE s.place_id = p.parent_place_id ";
+                            $sSQL .= "AND array_cat(s.nameaddress_vector, s.name_vector) @> ARRAY[".join($aSearch['aAddress'], ",")."] AND ";
+                        } else {
+                            $sSQL .= " WHERE ";
                         }
-                        if ($sCountryCodesSQL) {
-                            $sSQL .= " AND country_code in ($sCountryCodesSQL)";
+                        $sSQL .= "p.postcode = '".pg_escape_string(reset($aSearch['aName']))."'";
+                        if ($aSearch['sCountryCode']) {
+                            $sSQL .= " AND p.country_code = '".$aSearch['sCountryCode']."'";
+                        } elseif ($sCountryCodesSQL) {
+                            $sSQL .= " AND p.country_code in ($sCountryCodesSQL)";
                         }
                         $sSQL .= " LIMIT $this->iLimit";
                         if (CONST_Debug) var_dump($sSQL);
@@ -1414,6 +1411,8 @@ class Geocode
                             $aTerms[] = $aSearch['oNear']->withinSQL('centroid');
 
                             $aOrder[] = $aSearch['oNear']->distanceSQL('centroid');
+                        } elseif ($aSearch['sPostcode']) {
+                            $aOrder[] = "(SELECT min(ST_Distance(search_name.centroid, p.geometry)) FROM location_postcode p WHERE p.postcode = '".$aSearch['sPostcode']."')";
                         }
                         if (sizeof($this->aExcludePlaceIDs)) {
                             $aTerms[] = "place_id not in (".join(',', $this->aExcludePlaceIDs).")";
@@ -1702,6 +1701,21 @@ class Geocode
                         var_Dump($aPlaceIDs);
                     }
 
+                    if ($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;