]> git.openstreetmap.org Git - nominatim.git/blobdiff - website/search.php
create proper token array for unknown housenumbers
[nominatim.git] / website / search.php
index 070ca5245a05ec50da87bc4984020d7d5f9575b9..c6aa05d17039a8d64806d0b668b58d7fa9df16cd 100755 (executable)
@@ -54,6 +54,8 @@
        }
 
   // Only certain ranks of feature
+       if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType'];
+
        if (isset($_GET['featuretype']))
        {
                switch($_GET['featuretype'])
                $aPhrases = array_reverse($aPhrases); 
                $sQuery = join(', ',$aPhrases);
        }
-
        if ($sQuery)
        {
                $hLog = logStart($oDB, 'search', $sQuery, $aLangPrefOrder);
 
                // Hack to make it handle "new york, ny" (and variants) correctly
                 $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $sQuery);
+               if (isset($aLangPrefOrder['name:en']))          
+               {
+                       $sQuery = preg_replace('/\\bil\\b/','illinois', $sQuery);
+                       $sQuery = preg_replace('/\\bal\\b/','alabama', $sQuery);
+                       $sQuery = preg_replace('/\\bla\\b/','louisiana', $sQuery);
+                       $sQuery = preg_replace('/\\bde louisiana\\b/','de la', $sQuery);
+               }
 
                // If we have a view box create the SQL
                // Small is the actual view box, Large is double (on each axis) that 
                        $_GET['nearlon'] = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60);
                        $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
                }
-               elseif (preg_match('/(\\[|\\b)(-?[0-9]+[0-9.]*)[, ]+(-?[0-9]+[0-9.]*)(\\]|\\b])/', $sQuery, $aData))
+               elseif (preg_match('/(\\[|^|\\b)(-?[0-9]+[0-9.]*)[, ]+(-?[0-9]+[0-9.]*)(\\]|$|\\b)/', $sQuery, $aData))
                {
                        $_GET['nearlat'] = $aData[2];
                        $_GET['nearlon'] = $aData[3];
 
                if ($sQuery)
                {
-
                        // Start with a blank search
                        $aSearches = array(
                                array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), 
                                $aPhrase = $oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string");
                                if (PEAR::isError($aPhrase))
                                {
-                                       var_dump($aPhrase);
+                                       echo "Illegal query string (not an UTF-8 string): ".$sPhrase;
+                                       if (CONST_Debug) var_dump($aPhrase);
                                        exit;
                                }
                                if (trim($aPhrase['string']))
                                }
                        }
                        if ($bSuggestion) $sSuggestion = join(', ',$aSuggestion);
-/*
+
                        // Try and calculate GB postcodes we might be missing
                        foreach($aTokens as $sToken)
                        {
                                        }
                                }
                        }
-*/
 
                        foreach($aTokens as $sToken)
                        {
                                // Unknown single word token with a number - assume it is a house number
                                if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken,' ') === false && preg_match('/[0-9]/', $sToken))
                                {
-                                       $aValidTokens[' '.$sToken] = array('class'=>'place','type'=>'house');
+                                       $aValidTokens[' '.$sToken] = array(array('class'=>'place','type'=>'house'));
                                }
                        }
 
                                                        {
                                                                if (4 >= $iMinAddressRank && 4 <= $iMaxAddressRank)
                                                                {
-                                                                       $sSQL = "select place_id from placex where country_code='".$aSearch['sCountryCode']."' and rank_search = 4 order by st_area(geometry) desc limit 1";
+                                                                       $sSQL = "select place_id from placex where country_code='".$aSearch['sCountryCode']."' and rank_search = 4";
+                                                                       if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";                                                             
+                                                                       $sSQL .= " order by st_area(geometry) desc limit 1";
                                                                        $aPlaceIDs = $oDB->getCol($sSQL);
                                                                }
                                                        }
                                                                $sSQL .= " from search_name";
                                                                $sSQL .= " where ".join(' and ',$aTerms);
                                                                $sSQL .= " order by ".join(', ',$aOrder);
-                                                               if ($aSearch['sHouseNumber'])
+                                                               if ($aSearch['sHouseNumber'] || $aSearch['sClass'])
                                                                        $sSQL .= " limit 50";
                                                                elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass'])
                                                                        $sSQL .= " limit 1";
                                                                
                                                                if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') // & in
                                                                {
-                                                                       $sSQL = "select rank_search from placex where place_id in ($sPlaceIDs) order by rank_search asc limit 1";
+                                                                       $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
+                                                                       $bCacheTable = $oDB->getOne($sSQL);
+
+                                                                       $sSQL = "select min(rank_search) from placex where place_id in ($sPlaceIDs)";
 
                                                                        if (CONST_Debug) var_dump($sSQL);
-                                                                       $iMaxRank = ((int)$oDB->getOne($sSQL)) + 5;
+                                                                       $iMaxRank = ((int)$oDB->getOne($sSQL));
 
+                                                                       // For state / country level searches the normal radius search doesn't work very well
+                                                                       $sPlaceGeom = false;
+                                                                       if ($iMaxRank < 9 && $bCacheTable)
+                                                                       {
+                                                                               // Try and get a polygon to search in instead
+       $sSQL = "select geometry from placex where place_id in ($sPlaceIDs) and rank_search < $iMaxRank + 5 and st_geometrytype(geometry) in ('ST_Polygon','ST_MultiPolygon') order by rank_search asc limit 1";
+       if (CONST_Debug) var_dump($sSQL);
+       $sPlaceGeom = $oDB->getOne($sSQL);
+                                                                       }
+                                                                       
+                                                                       if ($sPlaceGeom)
+                                                                       {
+                                                                               $sPlaceIDs = false;
+                                                                       }
+                                                                       else
+                                                                       {
+                                                                               $iMaxRank += 5;
                                                                        $sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and rank_search < $iMaxRank";
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $oDB->getCol($sSQL);
                                                                        $sPlaceIDs = join(',',$aPlaceIDs);
+                                                                       }
 
-                                                                       if ($sPlaceIDs)
+                                                                       if ($sPlaceIDs || $sPlaceGeom)
                                                                        {
 
                                                                        $fRange = 0.01;
-                                                                       $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
-                                                                       if ($oDB->getOne($sSQL))
+                                                                       if ($bCacheTable)
                                                                        {
                                                                                // More efficient - can make the range bigger
-                                                                       $fRange = 0.05;
-                                                                               $sSQL = "select l.place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
+                                                                               $fRange = 0.05;
+
+                                                                               $sOrderBySQL = '';
+                                                                               if ($sNearPointSQL) $sOrderBySQL = "ST_Distance($sNearPointSQL, l.centroid)";
+                                                                               else if ($sPlaceIDs) $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
+                                                                               else if ($sPlaceGeom) $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
+                                                                               
+                                                                               $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
                                                                                if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)";
-                                                                               $sSQL .= ",placex as f where ";
-                                                                               $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, st_centroid(f.geometry), $fRange) ";
+                                                                               if ($sPlaceIDs)
+                                                                               {
+                                                                                       $sSQL .= ",placex as f where ";
+                                                                                       $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, st_centroid(f.geometry), $fRange) ";
+                                                                               }
+                                                                               if ($sPlaceGeom)
+                                                                               {
+                                                                                       $sSQL .= " where ";
+                                                                                       $sSQL .= "ST_Contains('".$sPlaceGeom."', l.centroid) ";
+                                                                               }
                                                                                if (sizeof($aExcludePlaceIDs))
                                                                                {
                                                                                        $sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
                                                                                }
                                                                                if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)";
-                                                                               if ($sNearPointSQL) $sSQL .= " order by ST_Distance($sNearPointSQL, l.centroid) ASC";
-                                                                               else $sSQL .= " order by ST_Distance(l.centroid, f.geometry) asc";
+                                                                               if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc";
+                                                                               if ($iOffset) $sSQL .= " offset $iOffset";
                                                                                $sSQL .= " limit $iLimit";
                                                                                if (CONST_Debug) var_dump($sSQL);
                                                                                $aPlaceIDs = $oDB->getCol($sSQL);
                                                                        else
                                                                        {
                                                                                if (isset($aSearch['fRadius']) && $aSearch['fRadius']) $fRange = $aSearch['fRadius'];
-                                                                               $sSQL = "select l.place_id from placex as l,placex as f where ";
+
+                                                                               $sOrderBySQL = '';
+                                                                               if ($sNearPointSQL) $sOrderBySQL = "ST_Distance($sNearPointSQL, l.geometry)";
+                                                                               else $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
+
+                                                                               $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from placex as l,placex as f where ";
                                                                                $sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, st_centroid(f.geometry), $fRange) ";
                                                                                $sSQL .= "and l.class='".$aSearch['sClass']."' and l.type='".$aSearch['sType']."' ";
                                                                                if (sizeof($aExcludePlaceIDs))
                                                                                        $sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
                                                                                }
                                                                                if ($sCountryCodesSQL) $sSQL .= " and l.country_code in ($sCountryCodesSQL)";                                                           
-                                                                               if ($sNearPointSQL) $sSQL .= " order by ST_Distance($sNearPointSQL, l.geometry) ASC";
-                                                                               else $sSQL .= " order by ST_Distance(l.geometry, f.geometry) asc, l.rank_search ASC";
+                                                                               if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc";
+                                                                               if ($iOffset) $sSQL .= " offset $iOffset";
                                                                                $sSQL .= " limit $iLimit";
                                                                                if (CONST_Debug) var_dump($sSQL);
                                                                                $aPlaceIDs = $oDB->getCol($sSQL);
                                        $aSearchResults = $oDB->getAll($sSQL);
 //var_dump($sSQL,$aSearchResults);exit;
 
+                                       if (PEAR::IsError($aSearchResults))
+                                       {
+                                               var_dump($sSQL, $aSearchResults);                                       
+                                               exit;
+                                       }
+                               }
+                       } // end if ($sQuery)
+                       else
+                       {
+                               if (isset($_GET['nearlat']) && trim($_GET['nearlat'])!=='' && isset($_GET['nearlon']) && trim($_GET['nearlon']) !== '')
+                               {
+                                       $iPlaceID = geocodeReverse($_GET['nearlat'], $_GET['nearlon']);
+                                       $aResultPlaceIDs = array($iPlaceID);
+
+                                       // TODO: this needs refactoring!
+
+                                       // Get the details for display (is this a redundant extra step?)
+                                       $sPlaceIDs = join(',',$aResultPlaceIDs);
+                                       $sOrderSQL = 'CASE ';
+                                       foreach(array_keys($aResultPlaceIDs) as $iOrder => $iPlaceID)
+                                       {
+                                               $sOrderSQL .= 'when min(place_id) = '.$iPlaceID.' then '.$iOrder.' ';
+                                       }
+                                       $sOrderSQL .= ' ELSE 10000000 END';
+                                       $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,country_code,";
+                                       $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
+                                       $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,";
+                                       $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
+                                       $sSQL .= "avg(ST_X(ST_Centroid(geometry))) as lon,avg(ST_Y(ST_Centroid(geometry))) as lat, ";
+//                                     $sSQL .= $sOrderSQL." as porder, ";
+                                       $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance ";
+                                       $sSQL .= "from placex where place_id in ($sPlaceIDs) ";
+                                       $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank ";
+                                       $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance";
+                                       if (!$bDeDupe) $sSQL .= ",place_id";
+                                       $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
+                                       $sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
+                                       $sSQL .= ",get_name_by_language(name, ARRAY['ref']) ";
+                                       $sSQL .= " union ";
+                                       $sSQL .= "select 'T' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,";
+                                       $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
+                                       $sSQL .= "null as placename,";
+                                       $sSQL .= "null as ref,";
+                                       $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
+//                                     $sSQL .= $sOrderSQL." as porder, ";
+                                       $sSQL .= "-0.15 as importance ";
+                                       $sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
+                                       $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
+                                       $sSQL .= "group by place_id";
+                                       if (!$bDeDupe) $sSQL .= ",place_id";
+                                       $sSQL .= " union ";
+                                       $sSQL .= "select 'L' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,";
+                                       $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
+                                       $sSQL .= "null as placename,";
+                                       $sSQL .= "null as ref,";
+                                       $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
+//                                     $sSQL .= $sOrderSQL." as porder, ";
+                                       $sSQL .= "-0.10 as importance ";
+                                       $sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
+                                       $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
+                                       $sSQL .= "group by place_id";
+                                       if (!$bDeDupe) $sSQL .= ",place_id";
+                                       $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
+                                       $sSQL .= "order by importance desc";
+//                                     $sSQL .= "order by rank_search,rank_address,porder asc";
+                                       if (CONST_Debug) var_dump('<hr>',$sSQL);
+                                       $aSearchResults = $oDB->getAll($sSQL);
+//var_dump($sSQL,$aSearchResults);exit;
+
                                        if (PEAR::IsError($aSearchResults))
                                        {
                                                var_dump($sSQL, $aSearchResults);                                       
                                }
 
                                // Output data suitable for display (points and a bounding box)
-                               if ($bShowPolygons)
+                               if ($bShowPolygons && isset($aPolyPoints))
                                {
                                        $aResult['aPolyPoints'] = array();
                                        foreach($aPolyPoints as $aPoint)
                        $aClassTypeNameDone[$aResult['osm_type'].$aResult['osm_class'].$aResult['name']] = true;
                        $aSearchResults[] = $aResult;
                }
+
+               // Absolute limit on number of results
+               if (sizeof($aSearchResults) >= $iLimit) break;
        }
 
        $sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '1 day'::interval,'YYYY/MM/DD') from import_status limit 1");