order search results by existance of housenumbers

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 85aa31f08202d99ac1e26ba80ee3444ee4c785e4..2fa170a6d5f8bfc84f47d8946b97357d825e0a9c 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -443,7 +443,7 @@
                        return $aSearchResults;
                }
 
-               function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases)
+               function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases)
                {
                        /*
                           Calculate all searches using aValidTokens i.e.
@@ -555,6 +555,8 @@
                                                                                        // 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) $aSearch['iSearchRank'] += 1;
                                                                                        if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        /*
                                                                                        // Fall back to not searching for this item (better than nothing)
@@ -613,7 +615,10 @@
                                                                        }
                                                                }
                                                        }
-                                                       if (isset($aValidTokens[$sToken]))
+                                                       // Look for partial matches.
+                                                       // Note that there is no point in adding country terms here
+                                                       // because country are omitted in the address.
+                                                       if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country')
                                                        {
                                                                // Allow searching for a word - but at extra cost
                                                                foreach($aValidTokens[$sToken] as $aSearchTerm)
@@ -1041,7 +1046,7 @@
                                        // Start the search process
                                        $aResultPlaceIDs = array();
 
-                                       $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
+                                       $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
 
                                        if ($this->bReverseInPlan)
                                        {
@@ -1055,7 +1060,7 @@
                                                        $aFinalPhrase = end($aPhrases);
                                                        $aFinalPhrase['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
                                                }
-                                               $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $aWordFrequencyScores, false);
+                                               $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false);
 
                                                foreach($aGroupedSearches as $aSearches)
                                                {
@@ -1171,7 +1176,7 @@
                                                                        $sSQL = "select place_id from placex where calculated_country_code='".$aSearch['sCountryCode']."' and rank_search = 4";
                                                                        if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
                                     if ($bBoundingBoxSearch)
-                                        $sSQL .= " and st_overlaps($this->sViewboxSmallSQL, geometry)";
+                                        $sSQL .= " and _st_intersects($this->sViewboxSmallSQL, geometry)";
                                                                        $sSQL .= " order by st_area(geometry) desc limit 1";
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
@@ -1237,6 +1242,12 @@
                                                        $aTerms = array();
                                                        $aOrder = array();
 
+                                                       if ($aSearch['sHouseNumber'])
+                                                       {
+                                                               $sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M';
+                                                               $aOrder[] = "exists(select place_id from placex where parent_place_id = search_name.place_id and transliteration(housenumber) ~* E'".$sHouseNumberRegex."' limit 1) desc";
+                                                       }
+
                                                        // 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'],",")."]";
@@ -1318,7 +1329,7 @@
                                                                $sSQL .= " where ".join(' and ',$aTerms);
                                                                $sSQL .= " order by ".join(', ',$aOrder);
                                                                if ($aSearch['sHouseNumber'] || $aSearch['sClass'])
-                                                                       $sSQL .= " limit 50";
+                                                                       $sSQL .= " limit 20";
                                                                elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass'])
                                                                        $sSQL .= " limit 1";
                                                                else
@@ -1785,7 +1796,6 @@
                        $bFirst = true;
                        foreach($aToFilter as $iResNum => $aResult)
                        {
-                               if ($aResult['type'] == 'adminitrative') $aResult['type'] = 'administrative';
                                $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
                                if ($bFirst)
                                {