Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 7ea39e4954b4aa9e84a3da84060fdf02766c70b1..1f4eebef44611d77d996898ec623a9be75b1e807 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -24,7 +24,7 @@ class Geocode
 
     protected $aExcludePlaceIDs = array();
     protected $bDeDupe = true;
-    protected $bReverseInPlan = false;
+    protected $bReverseInPlan = true;
 
     protected $iLimit = 20;
     protected $iFinalLimit = 10;
@@ -126,6 +126,12 @@ class Geocode
         return $this->aExcludePlaceIDs;
     }
 
+
+    public function getCountryCodes()
+    {
+        return $this->aCountryCodes;
+    }
+
     public function getViewBoxString()
     {
         if (!$this->aViewBox) return null;
@@ -908,6 +914,9 @@ class Geocode
         }
 
         $sQuery = $this->sQuery;
+        if (!preg_match('//u', $sQuery)) {
+            userError("Query string is not UTF-8 encoded.");
+        }
 
         // Conflicts between US state abreviations and various words for 'the' in different languages
         if (isset($this->aLangPrefOrder['name:en'])) {
@@ -1317,6 +1326,7 @@ class Geocode
                             $aOrder[0] .= "     SELECT place_id ";
                             $aOrder[0] .= "     FROM location_property_osmline ";
                             $aOrder[0] .= "     WHERE parent_place_id = search_name.place_id";
+                            $aOrder[0] .= "       AND startnumber is not NULL";
                             $aOrder[0] .= "       AND ".intval($aSearch['sHouseNumber']).">=startnumber ";
                             $aOrder[0] .= "       AND ".intval($aSearch['sHouseNumber'])."<=endnumber ";
                             $aOrder[0] .= "     LIMIT 1";
@@ -1328,19 +1338,20 @@ class Geocode
                         // TODO: filter out the pointless search terms (2 letter name tokens and less)
                         // they might be right - but they are just too darned expensive to run
                         if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'], ",")."]";
-                        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
                             ) {
-                                $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'], $aSearch['aAddressNonSearch']), ",")."]";
+                                //$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'],",")."]";
                             } 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'], ",")."]";
-                                }
+                                }*/
                             }
                         }
                         if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
@@ -1453,7 +1464,7 @@ class Geocode
                                 // do we need to use transliteration and the regex for housenumbers???
                                 //new query for lines, not housenumbers anymore
                                 $sSQL = "SELECT distinct place_id FROM location_property_osmline";
-                                $sSQL .= " WHERE parent_place_id in (".$sPlaceIDs.") and (";
+                                $sSQL .= " WHERE startnumber is not NULL and parent_place_id in (".$sPlaceIDs.") and (";
                                 if ($searchedHousenumber%2 == 0) {
                                     //if housenumber is even, look for housenumber in streets with interpolationtype even or all
                                     $sSQL .= "interpolationtype='even'";
@@ -1672,7 +1683,7 @@ class Geocode
                     $sSQL .= "  SELECT place_id ";
                     $sSQL .= "  FROM location_property_osmline ";
                     $sSQL .= "  WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).")";
-                    $sSQL .= "    AND (30 between $this->iMinAddressRank and $this->iMaxAddressRank)";
+                    $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();