X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/c988be1ddf10c7838c4e057adb351b1c076c6cca..c36e3d9c7ea9f0fe61377c0769e14d1c8e90f45a:/website/search.php diff --git a/website/search.php b/website/search.php index ba7df0f0..a1d0db72 100755 --- a/website/search.php +++ b/website/search.php @@ -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']) @@ -99,13 +101,19 @@ $aPhrases = array_reverse($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 @@ -155,8 +163,7 @@ $sViewboxSmallSQL = $oDB->getOne($sSQL); if (PEAR::isError($sViewboxSmallSQL)) { - var_dump($sViewboxSmallSQL); - exit; + failInternalError("Could not get small viewbox.", $sSQL, $sViewboxSmallSQL); } $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry"; @@ -164,8 +171,7 @@ $sViewboxLargeSQL = $oDB->getOne($sSQL); if (PEAR::isError($sViewboxLargeSQL)) { - var_dump($sViewboxLargeSQL); - exit; + failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL); } $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry"; } @@ -173,21 +179,36 @@ // Do we have anything that looks like a lat/lon pair? if (preg_match('/\\b([NS])[ ]+([0-9]+[0-9.]*)[ ]+([0-9.]+)?[, ]+([EW])[ ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?\\b/', $sQuery, $aData)) { - $_GET['nearlat'] = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60); - $_GET['nearlon'] = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60); - $sQuery = trim(str_replace($aData[0], ' ', $sQuery)); + $fQueryLat = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60); + $fQueryLon = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60); + if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1) + { + $_GET['nearlat'] = $fQueryLat; + $_GET['nearlon'] = $fQueryLon; + $sQuery = trim(str_replace($aData[0], ' ', $sQuery)); + } } elseif (preg_match('/\\b([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([NS])[, ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([EW])\\b/', $sQuery, $aData)) { - $_GET['nearlat'] = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60); - $_GET['nearlon'] = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60); - $sQuery = trim(str_replace($aData[0], ' ', $sQuery)); + $fQueryLat = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60); + $fQueryLon = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60); + if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1) + { + $_GET['nearlat'] = $fQueryLat; + $_GET['nearlon'] = $fQueryLon; + $sQuery = trim(str_replace($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]; - $sQuery = trim(str_replace($aData[0], ' ', $sQuery)); + $fQueryLat = $aData[2]; + $fQueryLon = $aData[3]; + if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1) + { + $_GET['nearlat'] = $fQueryLat; + $_GET['nearlon'] = $fQueryLon; + $sQuery = trim(str_replace($aData[0], ' ', $sQuery)); + } } if ($sQuery) @@ -224,6 +245,7 @@ $sToken = $oDB->getOne("select make_standard_name('".$aSpecialTerm[1]."') as string"); $sSQL = 'select * from (select word_id,word_token, word, class, type, location, country_code, operator'; $sSQL .= ' from word where word_token in (\' '.$sToken.'\')) as x where (class is not null and class not in (\'place\',\'highway\')) or country_code is not null'; + if (CONST_Debug) var_Dump($sSQL); $aSearchWords = $oDB->getAll($sSQL); $aNewSearches = array(); foreach($aSearches as $aSearch) @@ -263,7 +285,8 @@ $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'])) @@ -300,8 +323,7 @@ $aDatabaseWords = array(); if (PEAR::IsError($aDatabaseWords)) { - var_dump($sSQL, $aDatabaseWords); - exit; + failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords); } $aPossibleMainWordIDs = array(); foreach($aDatabaseWords as $aToken) @@ -346,11 +368,12 @@ } } if ($bSuggestion) $sSuggestion = join(', ',$aSuggestion); -/* + // Try and calculate GB postcodes we might be missing foreach($aTokens as $sToken) { - if (!isset($aValidTokens[$sToken]) && !isset($aValidTokens[' '.$sToken]) && preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) + // Source of gb postcodes is now definitive - always use + if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) { if (substr($aData[1],-2,1) != ' ') { @@ -364,14 +387,13 @@ } } } -*/ 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')); } } @@ -393,7 +415,6 @@ Score how good the search is so they can be ordered */ - foreach($aPhrases as $iPhrase => $sPhrase) { $aNewPhraseSearches = array(); @@ -480,7 +501,7 @@ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } - else + elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { if (sizeof($aSearch['aName'])) { @@ -507,6 +528,8 @@ // Allow searching for a word - but at extra cost foreach($aValidTokens[$sToken] as $aSearchTerm) { + if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) + { //var_Dump('
',$aSearch['aName']); if (sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4) @@ -526,6 +549,7 @@ $aSearch['iNamePhrase'] = $iPhrase; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } + } } } else @@ -664,7 +688,9 @@ { 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); } } @@ -720,6 +746,9 @@ // First we need a position, either aName or fLat or both $aTerms = array(); $aOrder = array(); + + // TODO: filter out the pointless search terms (2 letter name tokens and less) + // they might be right - but they are just too darned expensive to run if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]"; if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]"; if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'"; @@ -764,8 +793,7 @@ $aViewBoxPlaceIDs = $oDB->getAll($sSQL); if (PEAR::IsError($aViewBoxPlaceIDs)) { - var_dump($sSQL, $aViewBoxPlaceIDs); - exit; + failInternalError("Could not get places for search terms.", $sSQL, $aViewBoxPlaceIDs); } //var_dump($aViewBoxPlaceIDs); // Did we have an viewbox matches? @@ -789,7 +817,7 @@ $sPlaceIDs = join(',',$aPlaceIDs); // Now they are indexed look for a house attached to a street we found - $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-, ]',$aSearch['sHouseNumber']).'\\\\M'; + $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-,/ ]',$aSearch['sHouseNumber']).'\\\\M'; $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'"; if (sizeof($aExcludePlaceIDs)) { @@ -848,36 +876,70 @@ 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); @@ -885,7 +947,12 @@ 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)) @@ -893,8 +960,8 @@ $sSQL .= " and l.place_id not in (".join(',',$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); @@ -907,8 +974,7 @@ if (PEAR::IsError($aPlaceIDs)) { - var_dump($sSQL, $aPlaceIDs); - exit; + failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs); } if (CONST_Debug) var_Dump($aPlaceIDs); @@ -946,6 +1012,7 @@ $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 .= "and linked_place_id is null "; $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance"; if (!$bDeDupe) $sSQL .= ",place_id"; $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) "; @@ -984,8 +1051,7 @@ if (PEAR::IsError($aSearchResults)) { - var_dump($sSQL, $aSearchResults); - exit; + failInternalError("Could not get details for place.", $sSQL, $aSearchResults); } } } // end if ($sQuery) @@ -1053,8 +1119,7 @@ if (PEAR::IsError($aSearchResults)) { - var_dump($sSQL, $aSearchResults); - exit; + failInternalError("Could not get details for place (near).", $sSQL, $aSearchResults); } } } @@ -1063,35 +1128,42 @@ $sSearchResult = ''; if (!sizeof($aSearchResults) && isset($_GET['q']) && $_GET['q']) { -echo "here"; -exit; $sSearchResult = 'No Results Found'; } //var_Dump($aSearchResults); //exit; $aClassType = getClassTypesWithImportance(); + $aRecheckWords = preg_split('/\b/',$sQuery); + foreach($aRecheckWords as $i => $sWord) + { + if (!$sWord) unset($aRecheckWords[$i]); + } foreach($aSearchResults as $iResNum => $aResult) { - if (CONST_Search_AreaPolygons || true) + if (CONST_Search_AreaPolygons) { // Get the bounding box and outline polygon $sSQL = "select place_id,numfeatures,area,outline,"; - $sSQL .= "ST_Y(ST_PointN(ExteriorRing(ST_Box2D(outline)),4)) as minlat,ST_Y(ST_PointN(ExteriorRing(ST_Box2D(outline)),2)) as maxlat,"; - $sSQL .= "ST_X(ST_PointN(ExteriorRing(ST_Box2D(outline)),1)) as minlon,ST_X(ST_PointN(ExteriorRing(ST_Box2D(outline)),3)) as maxlon,"; + $sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(outline)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(outline)),2)) as maxlat,"; + $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(outline)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(outline)),3)) as maxlon,"; $sSQL .= "ST_AsText(outline) as outlinestring from get_place_boundingbox_quick(".$aResult['place_id'].")"; $sSQL = "select place_id,0 as numfeatures,st_area(geometry) as area,"; - $sSQL .= "ST_Y(ST_PointN(ExteriorRing(ST_Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ExteriorRing(ST_Box2D(geometry)),2)) as maxlat,"; - $sSQL .= "ST_X(ST_PointN(ExteriorRing(ST_Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ExteriorRing(ST_Box2D(geometry)),3)) as maxlon,"; - $sSQL .= "ST_AsText(geometry) as outlinestring from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(ST_Box2D(geometry)) = \'ST_Polygon\''; + $sSQL .= "ST_Y(centroid) as centrelat,ST_X(centroid) as centrelon,"; + $sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),2)) as maxlat,"; + $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon,"; + $sSQL .= "ST_AsText(geometry) as outlinestring from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(Box2D(geometry)) = \'ST_Polygon\''; $aPointPolygon = $oDB->getRow($sSQL); if (PEAR::IsError($aPointPolygon)) { - var_dump($sSQL, $aPointPolygon); - exit; + failInternalError("Could not get outline.", $sSQL, $aPointPolygon); } if ($aPointPolygon['place_id']) { + if ($aPointPolygon['centrelon'] !== null && $aPointPolygon['centrelat'] !== null ) { + $aResult['lat'] = $aPointPolygon['centrelat']; + $aResult['lon'] = $aPointPolygon['centrelon']; + } // Translate geometary string to point array if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['outlinestring'],$aMatch)) { @@ -1114,7 +1186,7 @@ exit; } // Output data suitable for display (points and a bounding box) - if ($bShowPolygons) + if ($bShowPolygons && isset($aPolyPoints)) { $aResult['aPolyPoints'] = array(); foreach($aPolyPoints as $aPoint) @@ -1187,6 +1259,16 @@ exit; //exit; } + // Adjust importance for the number of exact string matches in the result + $aResult['importance'] = max(0.001,$aResult['importance']); + $iCountWords = 0; + $sAddress = $aResult['langaddress']; + foreach($aRecheckWords as $i => $sWord) + { + if (stripos($sAddress, $sWord)!==false) $iCountWords++; + } + $aResult['importance'] = $aResult['importance'] + $iCountWords; + //if (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']); /* if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance']) @@ -1208,7 +1290,6 @@ exit; $aResult['foundorder'] = $iResNum; $aSearchResults[$iResNum] = $aResult; } - uasort($aSearchResults, 'byImportance'); //var_dump($aSearchResults);exit;