Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index fc2d6c65190ee17ecd0949c0f8c837336e88604f..f142731dac0f3233d96d59c12b86b967081a898c 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -25,7 +25,7 @@ class Geocode
 
     protected $aExcludePlaceIDs = array();
     protected $bDeDupe = true;
-    protected $bReverseInPlan = false;
+    protected $bReverseInPlan = true;
 
     protected $iLimit = 20;
     protected $iFinalLimit = 10;
@@ -910,6 +910,19 @@ class Geocode
 
             //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;
+                }
+            }
+            $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
+        }
+        ksort($aGroupedSearches);
+
         return $aGroupedSearches;
     }
 
@@ -1111,14 +1124,10 @@ class Geocode
                 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 $aToken) {
@@ -1378,19 +1387,20 @@ class Geocode
                         // TODO: filter out the pointless search terms (2 letter name tokens and less)
                         // they might be right - but they are just too darned expensive to run
                         if (sizeof($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'])."'";
@@ -1563,7 +1573,8 @@ class Geocode
                             }
 
                             // 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;
@@ -1728,9 +1739,11 @@ 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
+                    $sWherePlaceId = 'WHERE place_id in (';
+                    $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') ';
+
                     $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 <= $this->iMaxAddressRank) {
@@ -1739,20 +1752,23 @@ class Geocode
                     if ($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 .= "  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();