indexing: precompute row counts

[nominatim.git] / lib-php / ReverseGeocode.php

diff --git a/lib-php/ReverseGeocode.php b/lib-php/ReverseGeocode.php
index b420e5dd99f8ba69fadc0488684784a7ffcb3794..f6ea590fb7c6cfa451f75952fde894a39f34dceb 100644 (file)
--- a/lib-php/ReverseGeocode.php
+++ b/lib-php/ReverseGeocode.php
@@ -1,4 +1,12 @@
 <?php
 <?php

+/**
+ * SPDX-License-Identifier: GPL-2.0-only
+ *
+ * This file is part of Nominatim. (https://nominatim.org)
+ *
+ * Copyright (C) 2022 by the Nominatim developer community.
+ * For a full list of authors see the git log.
+ */

 
 namespace Nominatim;
 
 
 namespace Nominatim;
 


@@ -32,10 +40,10 @@ class ReverseGeocode
                       9 => 12,
                       10 => 17, // City
                       11 => 17,
                       9 => 12,
                       10 => 17, // City
                       11 => 17,

-                      12 => 18, // Town / Village
-                      13 => 18,
-                      14 => 22, // Suburb
-                      15 => 22,
+                      12 => 18, // Town
+                      13 => 19, // Village
+                      14 => 22, // Neighbourhood
+                      15 => 25, // Locality

                       16 => 26, // major street
                       17 => 27, // minor street
                       18 => 30, // or >, Building
                       16 => 26, // major street
                       17 => 27, // minor street
                       18 => 30, // or >, Building


@@ -56,12 +64,15 @@ class ReverseGeocode
     {
         Debug::newFunction('lookupInterpolation');
         $sSQL = 'SELECT place_id, parent_place_id, 30 as rank_search,';
     {
         Debug::newFunction('lookupInterpolation');
         $sSQL = 'SELECT place_id, parent_place_id, 30 as rank_search,';

-        $sSQL .= '  ST_LineLocatePoint(linegeo,'.$sPointSQL.') as fraction,';
-        $sSQL .= '  startnumber, endnumber, interpolationtype,';
+        $sSQL .= '  (CASE WHEN endnumber != startnumber';
+        $sSQL .= '        THEN (endnumber - startnumber) * ST_LineLocatePoint(linegeo,'.$sPointSQL.')';
+        $sSQL .= '        ELSE startnumber END) as fhnr,';
+        $sSQL .= '  startnumber, endnumber, step,';

         $sSQL .= '  ST_Distance(linegeo,'.$sPointSQL.') as distance';
         $sSQL .= ' FROM location_property_osmline';
         $sSQL .= ' WHERE ST_DWithin('.$sPointSQL.', linegeo, '.$fSearchDiam.')';
         $sSQL .= '  ST_Distance(linegeo,'.$sPointSQL.') as distance';
         $sSQL .= ' FROM location_property_osmline';
         $sSQL .= ' WHERE ST_DWithin('.$sPointSQL.', linegeo, '.$fSearchDiam.')';

-        $sSQL .= ' and indexed_status = 0 and startnumber is not NULL ';
+        $sSQL .= '       and indexed_status = 0 and startnumber is not NULL ';
+        $sSQL .= '       and parent_place_id != 0';

         $sSQL .= ' ORDER BY distance ASC limit 1';
         Debug::printSQL($sSQL);
 
         $sSQL .= ' ORDER BY distance ASC limit 1';
         Debug::printSQL($sSQL);
 


@@ -74,7 +85,10 @@ class ReverseGeocode
 
     protected function lookupLargeArea($sPointSQL, $iMaxRank)
     {
 
     protected function lookupLargeArea($sPointSQL, $iMaxRank)
     {

-        $oResult = null;
+        $sCountryCode = $this->getCountryCode($sPointSQL);
+        if (CONST_Search_WithinCountries and $sCountryCode == null) {
+            return  null;
+        }

 
         if ($iMaxRank > 4) {
             $aPlace = $this->lookupPolygon($sPointSQL, $iMaxRank);
 
         if ($iMaxRank > 4) {
             $aPlace = $this->lookupPolygon($sPointSQL, $iMaxRank);


@@ -85,12 +99,12 @@ class ReverseGeocode
 
         // If no polygon which contains the searchpoint is found,
         // searches in the country_osm_grid table for a polygon.
 
         // If no polygon which contains the searchpoint is found,
         // searches in the country_osm_grid table for a polygon.

-        return  $this->lookupInCountry($sPointSQL, $iMaxRank);
+        return  $this->lookupInCountry($sPointSQL, $iMaxRank, $sCountryCode);

     }
 
     }
 

-    protected function lookupInCountry($sPointSQL, $iMaxRank)
+    protected function getCountryCode($sPointSQL)

     {
     {

-        Debug::newFunction('lookupInCountry');
+        Debug::newFunction('getCountryCode');

         // searches for polygon in table country_osm_grid which contains the searchpoint
         // and searches for the nearest place node to the searchpoint in this polygon
         $sSQL = 'SELECT country_code FROM country_osm_grid';
         // searches for polygon in table country_osm_grid which contains the searchpoint
         // and searches for the nearest place node to the searchpoint in this polygon
         $sSQL = 'SELECT country_code FROM country_osm_grid';


@@ -102,8 +116,12 @@ class ReverseGeocode
             null,
             'Could not determine country polygon containing the point.'
         );
             null,
             'Could not determine country polygon containing the point.'
         );

-        Debug::printVar('Country code', $sCountryCode);
+        return $sCountryCode;
+    }

 
 

+    protected function lookupInCountry($sPointSQL, $iMaxRank, $sCountryCode)
+    {
+        Debug::newFunction('lookupInCountry');

         if ($sCountryCode) {
             if ($iMaxRank > 4) {
                 // look for place nodes with the given country code
         if ($sCountryCode) {
             if ($iMaxRank > 4) {
                 // look for place nodes with the given country code


@@ -113,11 +131,13 @@ class ReverseGeocode
                 $sSQL .= ' FROM placex';
                 $sSQL .= ' WHERE osm_type = \'N\'';
                 $sSQL .= ' AND country_code = \''.$sCountryCode.'\'';
                 $sSQL .= ' FROM placex';
                 $sSQL .= ' WHERE osm_type = \'N\'';
                 $sSQL .= ' AND country_code = \''.$sCountryCode.'\'';

+                $sSQL .= ' AND rank_address between 4 and 25'; // needed to select right index

                 $sSQL .= ' AND rank_search between 5 and ' .min(25, $iMaxRank);
                 $sSQL .= ' AND rank_search between 5 and ' .min(25, $iMaxRank);

-                $sSQL .= ' AND class = \'place\' AND type != \'postcode\'';
+                $sSQL .= ' AND type != \'postcode\'';

                 $sSQL .= ' AND name IS NOT NULL ';
                 $sSQL .= ' and indexed_status = 0 and linked_place_id is null';
                 $sSQL .= ' AND name IS NOT NULL ';
                 $sSQL .= ' and indexed_status = 0 and linked_place_id is null';

-                $sSQL .= ' AND ST_DWithin('.$sPointSQL.', geometry, 1.8)) p ';
+                $sSQL .= ' AND ST_Buffer(geometry, reverse_place_diameter(rank_search)) && '.$sPointSQL;
+                $sSQL .= ') as a ';

                 $sSQL .= 'WHERE distance <= reverse_place_diameter(rank_search)';
                 $sSQL .= ' ORDER BY rank_search DESC, distance ASC';
                 $sSQL .= ' LIMIT 1';
                 $sSQL .= 'WHERE distance <= reverse_place_diameter(rank_search)';
                 $sSQL .= ' ORDER BY rank_search DESC, distance ASC';
                 $sSQL .= ' LIMIT 1';


@@ -167,22 +187,28 @@ class ReverseGeocode
     {
         Debug::newFunction('lookupPolygon');
         // polygon search begins at suburb-level
     {
         Debug::newFunction('lookupPolygon');
         // polygon search begins at suburb-level

-        if ($iMaxRank > 25) $iMaxRank = 25;
+        if ($iMaxRank > 25) {
+            $iMaxRank = 25;
+        }

         // no polygon search over country-level
         // no polygon search over country-level

-        if ($iMaxRank < 5) $iMaxRank = 5;
+        if ($iMaxRank < 5) {
+            $iMaxRank = 5;
+        }

         // search for polygon
         $sSQL = 'SELECT place_id, parent_place_id, rank_address, rank_search FROM';
         $sSQL .= '(select place_id, parent_place_id, rank_address, rank_search, country_code, geometry';
         $sSQL .= ' FROM placex';
         $sSQL .= ' WHERE ST_GeometryType(geometry) in (\'ST_Polygon\', \'ST_MultiPolygon\')';
         // search for polygon
         $sSQL = 'SELECT place_id, parent_place_id, rank_address, rank_search FROM';
         $sSQL .= '(select place_id, parent_place_id, rank_address, rank_search, country_code, geometry';
         $sSQL .= ' FROM placex';
         $sSQL .= ' WHERE ST_GeometryType(geometry) in (\'ST_Polygon\', \'ST_MultiPolygon\')';

-        $sSQL .= ' AND rank_address Between 5 AND ' .$iMaxRank;
+        // Ensure that query planner doesn't use the index on rank_search.
+        $sSQL .= ' AND coalesce(rank_search, 0) between 5 and ' .$iMaxRank;
+        $sSQL .= ' AND rank_address between 4 and 25'; // needed for index selection

         $sSQL .= ' AND geometry && '.$sPointSQL;
         $sSQL .= ' AND type != \'postcode\' ';
         $sSQL .= ' AND name is not null';
         $sSQL .= ' AND indexed_status = 0 and linked_place_id is null';
         $sSQL .= ' AND geometry && '.$sPointSQL;
         $sSQL .= ' AND type != \'postcode\' ';
         $sSQL .= ' AND name is not null';
         $sSQL .= ' AND indexed_status = 0 and linked_place_id is null';

-        $sSQL .= ' ORDER BY rank_address DESC LIMIT 50 ) as a';
-        $sSQL .= ' WHERE ST_CONTAINS(geometry, '.$sPointSQL.' )';
-        $sSQL .= ' ORDER BY rank_address DESC LIMIT 1';
+        $sSQL .= ' ORDER BY rank_search DESC LIMIT 50 ) as a';
+        $sSQL .= ' WHERE ST_Contains(geometry, '.$sPointSQL.' )';
+        $sSQL .= ' ORDER BY rank_search DESC LIMIT 1';

         Debug::printSQL($sSQL);
 
         $aPoly = $this->oDB->getRow($sSQL, null, 'Could not determine polygon containing the point.');
         Debug::printSQL($sSQL);
 
         $aPoly = $this->oDB->getRow($sSQL, null, 'Could not determine polygon containing the point.');


@@ -190,33 +216,28 @@ class ReverseGeocode
 
         if ($aPoly) {
         // if a polygon is found, search for placenodes begins ...
 
         if ($aPoly) {
         // if a polygon is found, search for placenodes begins ...

-            $iParentPlaceID = $aPoly['parent_place_id'];

             $iRankAddress = $aPoly['rank_address'];
             $iRankSearch = $aPoly['rank_search'];
             $iPlaceID = $aPoly['place_id'];
 
             $iRankAddress = $aPoly['rank_address'];
             $iRankSearch = $aPoly['rank_search'];
             $iPlaceID = $aPoly['place_id'];
 

-            if ($iRankAddress != $iMaxRank) {
+            if ($iRankSearch != $iMaxRank) {

                 $sSQL = 'SELECT place_id FROM ';
                 $sSQL .= '(SELECT place_id, rank_search, country_code, geometry,';
                 $sSQL .= ' ST_distance('.$sPointSQL.', geometry) as distance';
                 $sSQL .= ' FROM placex';
                 $sSQL .= ' WHERE osm_type = \'N\'';
                 $sSQL = 'SELECT place_id FROM ';
                 $sSQL .= '(SELECT place_id, rank_search, country_code, geometry,';
                 $sSQL .= ' ST_distance('.$sPointSQL.', geometry) as distance';
                 $sSQL .= ' FROM placex';
                 $sSQL .= ' WHERE osm_type = \'N\'';

-                // using rank_search because of a better differentiation
-                // for place nodes at rank_address 16

                 $sSQL .= ' AND rank_search > '.$iRankSearch;
                 $sSQL .= ' AND rank_search <= '.$iMaxRank;
                 $sSQL .= ' AND rank_search > '.$iRankSearch;
                 $sSQL .= ' AND rank_search <= '.$iMaxRank;

-                $sSQL .= ' AND rank_address > 0';
-                $sSQL .= ' AND class = \'place\'';
+                $sSQL .= ' AND rank_address between 4 and 25';  // needed to select right index

                 $sSQL .= ' AND type != \'postcode\'';
                 $sSQL .= ' AND name IS NOT NULL ';
                 $sSQL .= ' AND indexed_status = 0 AND linked_place_id is null';
                 $sSQL .= ' AND type != \'postcode\'';
                 $sSQL .= ' AND name IS NOT NULL ';
                 $sSQL .= ' AND indexed_status = 0 AND linked_place_id is null';

-                $sSQL .= ' AND ST_DWithin('.$sPointSQL.', geometry, reverse_place_diameter('.$iRankSearch.'::smallint))';
-                $sSQL .= ' ORDER BY distance ASC,';
-                $sSQL .= ' rank_address DESC';
-                $sSQL .= ' limit 500) as a';
-                $sSQL .= ' WHERE ST_CONTAINS((SELECT geometry FROM placex WHERE place_id = '.$iPlaceID.'), geometry )';
+                $sSQL .= ' AND ST_Buffer(geometry, reverse_place_diameter(rank_search)) && '.$sPointSQL;
+                $sSQL .= ' ORDER BY rank_search DESC, distance ASC';
+                $sSQL .= ' limit 100) as a';
+                $sSQL .= ' WHERE ST_Contains((SELECT geometry FROM placex WHERE place_id = '.$iPlaceID.'), geometry )';

                 $sSQL .= ' AND distance <= reverse_place_diameter(rank_search)';
                 $sSQL .= ' AND distance <= reverse_place_diameter(rank_search)';

-                $sSQL .= ' ORDER BY distance ASC, rank_search DESC';
+                $sSQL .= ' ORDER BY rank_search DESC, distance ASC';

                 $sSQL .= ' LIMIT 1';
                 Debug::printSQL($sSQL);
 
                 $sSQL .= ' LIMIT 1';
                 Debug::printSQL($sSQL);
 


@@ -242,26 +263,24 @@ class ReverseGeocode
     public function lookupPoint($sPointSQL, $bDoInterpolation = true)
     {
         Debug::newFunction('lookupPoint');
     public function lookupPoint($sPointSQL, $bDoInterpolation = true)
     {
         Debug::newFunction('lookupPoint');

-        // starts if the search is on POI or street level,
-        // searches for the nearest POI or street,
-        // if a street is found and a POI is searched for,
-        // the nearest POI which the found street is a parent of is choosen.
-        $iMaxRank = $this->iMaxRank;
-

         // Find the nearest point
         $fSearchDiam = 0.006;
         $oResult = null;
         $aPlace = null;
 
         // for POI or street level
         // Find the nearest point
         $fSearchDiam = 0.006;
         $oResult = null;
         $aPlace = null;
 
         // for POI or street level

-        if ($iMaxRank >= 26) {
+        if ($this->iMaxRank >= 26) {
+            // starts if the search is on POI or street level,
+            // searches for the nearest POI or street,
+            // if a street is found and a POI is searched for,
+            // the nearest POI which the found street is a parent of is chosen.

             $sSQL = 'select place_id,parent_place_id,rank_address,country_code,';
             $sSQL .= ' ST_distance('.$sPointSQL.', geometry) as distance';
             $sSQL .= ' FROM ';
             $sSQL .= ' placex';
             $sSQL .= '   WHERE ST_DWithin('.$sPointSQL.', geometry, '.$fSearchDiam.')';
             $sSQL .= '   AND';
             $sSQL = 'select place_id,parent_place_id,rank_address,country_code,';
             $sSQL .= ' ST_distance('.$sPointSQL.', geometry) as distance';
             $sSQL .= ' FROM ';
             $sSQL .= ' placex';
             $sSQL .= '   WHERE ST_DWithin('.$sPointSQL.', geometry, '.$fSearchDiam.')';
             $sSQL .= '   AND';

-            $sSQL .= ' rank_address between 26 and '.$iMaxRank;
+            $sSQL .= ' rank_address between 26 and '.$this->iMaxRank;

             $sSQL .= ' and (name is not null or housenumber is not null';
             $sSQL .= ' or rank_address between 26 and 27)';
             $sSQL .= ' and (rank_address between 26 and 27';
             $sSQL .= ' and (name is not null or housenumber is not null';
             $sSQL .= ' or rank_address between 26 and 27)';
             $sSQL .= ' and (rank_address between 26 and 27';


@@ -280,34 +299,11 @@ class ReverseGeocode
                 $iPlaceID = $aPlace['place_id'];
                 $oResult = new Result($iPlaceID);
                 $iRankAddress = $aPlace['rank_address'];
                 $iPlaceID = $aPlace['place_id'];
                 $oResult = new Result($iPlaceID);
                 $iRankAddress = $aPlace['rank_address'];

-                $iParentPlaceID = $aPlace['parent_place_id'];
-            }
-
-            if ($bDoInterpolation && $iMaxRank >= 30) {
-                $fDistance = $fSearchDiam;
-                if ($aPlace) {
-                    // We can't reliably go from the closest street to an
-                    // interpolation line because the closest interpolation
-                    // may have a different street segments as a parent.
-                    // Therefore allow an interpolation line to take precendence
-                    // even when the street is closer.
-                    $fDistance = $iRankAddress < 28 ? 0.001 : $aPlace['distance'];
-                }
-
-                $aHouse = $this->lookupInterpolation($sPointSQL, $fDistance);
-                Debug::printVar('Interpolation result', $aPlace);
-
-                if ($aHouse) {
-                    $oResult = new Result($aHouse['place_id'], Result::TABLE_OSMLINE);
-                    $oResult->iHouseNumber = closestHouseNumber($aHouse);
-                    $aPlace = $aHouse;
-                    $iRankAddress = 30;
-                }

             }
 
             if ($aPlace) {
                 // if street and maxrank > streetlevel
             }
 
             if ($aPlace) {
                 // if street and maxrank > streetlevel

-                if ($iRankAddress <= 27 && $iMaxRank > 27) {
+                if ($iRankAddress <= 27 && $this->iMaxRank > 27) {

                     // find the closest object (up to a certain radius) of which the street is a parent of
                     $sSQL = ' select place_id,';
                     $sSQL .= ' ST_distance('.$sPointSQL.', geometry) as distance';
                     // find the closest object (up to a certain radius) of which the street is a parent of
                     $sSQL = ' select place_id,';
                     $sSQL .= ' ST_distance('.$sPointSQL.', geometry) as distance';


@@ -328,7 +324,9 @@ class ReverseGeocode
                     Debug::printVar('Closest POI result', $aStreet);
 
                     if ($aStreet) {
                     Debug::printVar('Closest POI result', $aStreet);
 
                     if ($aStreet) {

+                        $aPlace = $aStreet;

                         $oResult = new Result($aStreet['place_id']);
                         $oResult = new Result($aStreet['place_id']);

+                        $iRankAddress = 30;

                     }
                 }
 
                     }
                 }
 


@@ -339,9 +337,9 @@ class ReverseGeocode
                     && $this->iMaxRank >= 28
                 ) {
                     $sSQL = 'SELECT place_id,parent_place_id,30 as rank_search,';
                     && $this->iMaxRank >= 28
                 ) {
                     $sSQL = 'SELECT place_id,parent_place_id,30 as rank_search,';

-                    $sSQL .= 'ST_LineLocatePoint(linegeo,'.$sPointSQL.') as fraction,';
-                    $sSQL .= 'ST_distance('.$sPointSQL.', linegeo) as distance,';
-                    $sSQL .= 'startnumber,endnumber,interpolationtype';
+                    $sSQL .= '      (endnumber - startnumber) * ST_LineLocatePoint(linegeo,'.$sPointSQL.') as fhnr,';
+                    $sSQL .= '      startnumber, endnumber, step,';
+                    $sSQL .= '      ST_Distance('.$sPointSQL.', linegeo) as distance';

                     $sSQL .= ' FROM location_property_tiger WHERE parent_place_id = '.$oResult->iId;
                     $sSQL .= ' AND ST_DWithin('.$sPointSQL.', linegeo, 0.001)';
                     $sSQL .= ' ORDER BY distance ASC limit 1';
                     $sSQL .= ' FROM location_property_tiger WHERE parent_place_id = '.$oResult->iId;
                     $sSQL .= ' AND ST_DWithin('.$sPointSQL.', linegeo, 0.001)';
                     $sSQL .= ' ORDER BY distance ASC limit 1';


@@ -351,17 +349,50 @@ class ReverseGeocode
                     Debug::printVar('Tiger house number result', $aPlaceTiger);
 
                     if ($aPlaceTiger) {
                     Debug::printVar('Tiger house number result', $aPlaceTiger);
 
                     if ($aPlaceTiger) {

+                        $aPlace = $aPlaceTiger;

                         $oResult = new Result($aPlaceTiger['place_id'], Result::TABLE_TIGER);
                         $oResult = new Result($aPlaceTiger['place_id'], Result::TABLE_TIGER);

-                        $oResult->iHouseNumber = closestHouseNumber($aPlaceTiger);
+                        $iRndNum = max(0, round($aPlaceTiger['fhnr'] / $aPlaceTiger['step']) * $aPlaceTiger['step']);
+                        $oResult->iHouseNumber = $aPlaceTiger['startnumber'] + $iRndNum;
+                        if ($oResult->iHouseNumber > $aPlaceTiger['endnumber']) {
+                            $oResult->iHouseNumber = $aPlaceTiger['endnumber'];
+                        }
+                        $iRankAddress = 30;

                     }
                 }
                     }
                 }

-            } else {
+            }
+
+            if ($bDoInterpolation && $this->iMaxRank >= 30) {
+                $fDistance = $fSearchDiam;
+                if ($aPlace) {
+                    // We can't reliably go from the closest street to an
+                    // interpolation line because the closest interpolation
+                    // may have a different street segments as a parent.
+                    // Therefore allow an interpolation line to take precedence
+                    // even when the street is closer.
+                    $fDistance = $iRankAddress < 28 ? 0.001 : $aPlace['distance'];
+                }
+
+                $aHouse = $this->lookupInterpolation($sPointSQL, $fDistance);
+                Debug::printVar('Interpolation result', $aPlace);
+
+                if ($aHouse) {
+                    $oResult = new Result($aHouse['place_id'], Result::TABLE_OSMLINE);
+                    $iRndNum = max(0, round($aHouse['fhnr'] / $aHouse['step']) * $aHouse['step']);
+                    $oResult->iHouseNumber = $aHouse['startnumber'] + $iRndNum;
+                    if ($oResult->iHouseNumber > $aHouse['endnumber']) {
+                        $oResult->iHouseNumber = $aHouse['endnumber'];
+                    }
+                    $aPlace = $aHouse;
+                }
+            }
+
+            if (!$aPlace) {

                 // if no POI or street is found ...
                 $oResult = $this->lookupLargeArea($sPointSQL, 25);
             }
         } else {
             // lower than street level ($iMaxRank < 26 )
                 // if no POI or street is found ...
                 $oResult = $this->lookupLargeArea($sPointSQL, 25);
             }
         } else {
             // lower than street level ($iMaxRank < 26 )

-            $oResult = $this->lookupLargeArea($sPointSQL, $iMaxRank);
+            $oResult = $this->lookupLargeArea($sPointSQL, $this->iMaxRank);

         }
 
         Debug::printVar('Final result', $oResult);
         }
 
         Debug::printVar('Final result', $oResult);