X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/95d50f35d046e2ae540eadf7ab9af40ac119e06f..c8756ce92160289fe45bac8095d4fb3466b1ea7e:/website/search.php diff --git a/website/search.php b/website/search.php index d6007ef5..6b492414 100755 --- a/website/search.php +++ b/website/search.php @@ -13,6 +13,7 @@ $sOutputFormat = 'html'; $aSearchResults = array(); $aExcludePlaceIDs = array(); + $sCountryCodesSQL = false; $sSuggestion = $sSuggestionURL = false; $bDeDupe = isset($_GET['dedupe'])?(bool)$_GET['dedupe']:true; $bReverseInPlan = false; @@ -37,7 +38,10 @@ // Prefered language $aLangPrefOrder = getPrefferedLangauges(); -// if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true; + if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true; + if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true; + if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true; + $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$aLangPrefOrder))."]"; if (isset($_GET['exclude_place_ids']) && $_GET['exclude_place_ids']) @@ -50,6 +54,8 @@ } // Only certain ranks of feature + if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType']; + if (isset($_GET['featuretype'])) { switch($_GET['featuretype']) @@ -62,10 +68,27 @@ break; case 'city': $iMinAddressRank = 14; - $iMaxAddressRank = 18; + $iMaxAddressRank = 16; + break; + case 'settlement': + $iMinAddressRank = 8; + $iMaxAddressRank = 20; break; } } + + if (isset($_GET['countrycodes'])) + { + $aCountryCodes = array(); + foreach(explode(',',$_GET['countrycodes']) as $sCountryCode) + { + if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) + { + $aCountryCodes[] = "'".strtolower($sCountryCode)."'"; + } + } + $sCountryCodesSQL = join(',', $aCountryCodes); + } // Search query $sQuery = (isset($_GET['q'])?trim($_GET['q']):''); @@ -85,6 +108,12 @@ // 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); + } // If we have a view box create the SQL // Small is the actual view box, Large is double (on each axis) that @@ -162,7 +191,7 @@ $_GET['nearlon'] = ($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'] = $aData[3]; @@ -171,7 +200,6 @@ if ($sQuery) { - // Start with a blank search $aSearches = array( array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), @@ -257,7 +285,7 @@ { unset($aPhrases[$iPhrase]); } - } + } // reindex phrases - we make assumptions later on $aPhrases = array_values($aPhrases); @@ -283,6 +311,7 @@ var_dump($sSQL, $aDatabaseWords); exit; } + $aPossibleMainWordIDs = array(); foreach($aDatabaseWords as $aToken) { if (isset($aValidTokens[$aToken['word_token']])) @@ -293,6 +322,7 @@ { $aValidTokens[$aToken['word_token']] = array($aToken); } + if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1; } if (CONST_Debug) var_Dump($aPhrases, $aValidTokens); @@ -343,6 +373,16 @@ } } */ + + foreach($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'); + } + } + // Any words that have failed completely? // TODO: suggestions @@ -572,18 +612,24 @@ if (CONST_Debug) var_Dump($aGroupedSearches); - if ($bReverseInPlan && false) + if ($bReverseInPlan) { - foreach($aGroupedSearches as $iGroup => $aSearches) + $aCopyGroupedSearches = $aGroupedSearches; + foreach($aCopyGroupedSearches as $iGroup => $aSearches) { foreach($aSearches as $iSearch => $aSearch) { if (sizeof($aSearch['aAddress'])) { - $aReverseSearch = $aSearch; $iReverseItem = array_pop($aSearch['aAddress']); - $aReverseSearch['aName'][$iReverseItem] = $iReverseItem; - $aGroupedSearches[$iGroup][] = $aReverseSearch; + if (isset($aPossibleMainWordIDs[$iReverseItem])) + { + $aSearch['aAddress'] = array_merge($aSearch['aAddress'], $aSearch['aName']); + $aSearch['aName'] = array($iReverseItem); + $aGroupedSearches[$iGroup][] = $aSearch; + } +// $aReverseSearch['aName'][$iReverseItem] = $iReverseItem; + // $aGroupedSearches[$iGroup][] = $aReverseSearch; } } } @@ -626,7 +672,9 @@ { 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); } } @@ -641,7 +689,9 @@ if ($oDB->getOne($sSQL)) { $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']; + if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)"; $sSQL .= " where st_contains($sViewboxSmallSQL, centroid)"; + if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); @@ -650,7 +700,9 @@ if (!sizeof($aPlaceIDs)) { $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']; + if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)"; $sSQL .= " where st_contains($sViewboxLargeSQL, centroid)"; + if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); @@ -661,6 +713,7 @@ { $sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'"; $sSQL .= " and st_contains($sViewboxSmallSQL, centroid)"; + if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); @@ -690,23 +743,27 @@ { $aTerms[] = "place_id not in (".join(',',$aExcludePlaceIDs).")"; } + if ($sCountryCodesSQL) + { + $aTerms[] = "country_code in ($sCountryCodesSQL)"; + } + if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL"; if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc"; - if ($sViewboxSmallSQL) $aOrder[] = "ST_Contains($sViewboxSmallSQL, centroid) desc"; - if ($sViewboxLargeSQL) $aOrder[] = "ST_Contains($sViewboxLargeSQL, centroid) desc"; - $aOrder[] = "search_rank ASC"; + + $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.92-(search_rank::float/33) else importance end'; + + if ($sViewboxSmallSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END"; + if ($sViewboxLargeSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END"; + $aOrder[] = "$sImportanceSQL DESC"; if (sizeof($aTerms)) { $sSQL = "select place_id"; - if ($sViewboxSmallSQL) $sSQL .= ",ST_Contains($sViewboxSmallSQL, centroid) as in_small"; - else $sSQL .= ",false as in_small"; - if ($sViewboxLargeSQL) $sSQL .= ",ST_Contains($sViewboxLargeSQL, centroid) as in_large"; - else $sSQL .= ",false as in_large"; $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']) $sSQL .= " limit 1"; @@ -720,19 +777,21 @@ var_dump($sSQL, $aViewBoxPlaceIDs); exit; } - +//var_dump($aViewBoxPlaceIDs); // Did we have an viewbox matches? $aPlaceIDs = array(); $bViewBoxMatch = false; foreach($aViewBoxPlaceIDs as $aViewBoxRow) { - if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break; - if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break; - if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1; - else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2; +// if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break; +// if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break; +// if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1; +// else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2; $aPlaceIDs[] = $aViewBoxRow['place_id']; } } +//var_Dump($aPlaceIDs); +//exit; if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs)) { @@ -776,7 +835,7 @@ } // Fallback to the road - if (!sizeof($aPlaceIDs)) + if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) { $aPlaceIDs = $aRoadPlaceIDs; } @@ -791,6 +850,7 @@ { // If they were searching for a named class (i.e. 'Kings Head pub') then we might have an extra match $sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'"; + if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; $sSQL .= " order by rank_search asc limit $iLimit"; if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $oDB->getCol($sSQL); @@ -798,30 +858,65 @@ 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 || $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; + $fRange = 0.05; + $sSQL = "select l.place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l"; - $sSQL .= ",placex as f where "; - $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, st_centroid(f.geometry), $fRange) "; + if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)"; + if ($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 $sSQL .= " order by ST_Distance(l.centroid, f.geometry) asc"; + else if ($sPlaceIDs) $sSQL .= " order by ST_Distance(l.centroid, f.geometry) asc"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $oDB->getCol($sSQL); @@ -830,18 +925,20 @@ { if (isset($aSearch['fRadius']) && $aSearch['fRadius']) $fRange = $aSearch['fRadius']; $sSQL = "select l.place_id 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 .= "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"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $oDB->getCol($sSQL); } + } } } @@ -884,10 +981,11 @@ $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 .= $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"; + $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance"; if (!$bDeDupe) $sSQL .= ",place_id"; $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) "; $sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) "; @@ -898,24 +996,96 @@ $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 .= $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('