X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/6f9d3a9d8004e3e9d40d07e9fd67fd9eaf6a7d4f..9d7c51a81332c3d10b6ca41e2f17f4ca2920738e:/website/search.php diff --git a/website/search.php b/website/search.php index f4b9b924..91869c3e 100755 --- a/website/search.php +++ b/website/search.php @@ -36,8 +36,8 @@ // Show address breakdown $bShowAddressDetails = isset($_GET['addressdetails']) && $_GET['addressdetails']; - // Prefered language - $aLangPrefOrder = getPrefferedLangauges(); + // Preferred language + $aLangPrefOrder = getPreferredLanguages(); if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true; if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true; if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true; @@ -53,7 +53,7 @@ } } - // Only certain ranks of feature + // Only certain ranks of feature if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType']; if (isset($_GET['featuretype'])) @@ -92,11 +92,11 @@ // Search query $sQuery = (isset($_GET['q'])?trim($_GET['q']):''); - if (!$sQuery && $_SERVER['PATH_INFO'] && $_SERVER['PATH_INFO'][0] == '/') + if (!$sQuery && isset($_SERVER['PATH_INFO']) && $_SERVER['PATH_INFO'][0] == '/') { $sQuery = substr($_SERVER['PATH_INFO'], 1); - // reverse order of '/' seperated string + // reverse order of '/' separated string $aPhrases = explode('/', $sQuery); $aPhrases = array_reverse($aPhrases); $sQuery = join(', ',$aPhrases); @@ -109,10 +109,9 @@ $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); + $sQuery = preg_replace('/,\s*il\s*(,|$)/',', illinois\1', $sQuery); + $sQuery = preg_replace('/,\s*al\s*(,|$)/',', alabama\1', $sQuery); + $sQuery = preg_replace('/,\s*la\s*(,|$)/',', louisiana\1', $sQuery); } // If we have a view box create the SQL @@ -134,6 +133,8 @@ $aCoOrdinates[1] += $fWidth; $aCoOrdinates[3] -= $fWidth; $sViewboxLargeSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$aCoOrdinates[0].",".(float)$aCoOrdinates[1]."),ST_Point(".(float)$aCoOrdinates[2].",".(float)$aCoOrdinates[3].")),4326)"; + } else { + $bBoundingBoxSearch = false; } if (isset($_GET['route']) && $_GET['route'] && isset($_GET['routewidth']) && $_GET['routewidth']) { @@ -163,8 +164,7 @@ $sViewboxSmallSQL = $oDB->getOne($sSQL); if (PEAR::isError($sViewboxSmallSQL)) { - var_dump($sViewboxSmallSQL); - exit; + failInternalError("Could not get small viewbox.", $sSQL, $sViewboxSmallSQL); } $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry"; @@ -172,8 +172,7 @@ $sViewboxLargeSQL = $oDB->getOne($sSQL); if (PEAR::isError($sViewboxLargeSQL)) { - var_dump($sViewboxLargeSQL); - exit; + failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL); } $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry"; } @@ -181,21 +180,36 @@ // 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)) { - $_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) @@ -231,7 +245,8 @@ $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery); $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'; + $sSQL .= ' from word where word_token in (\' '.$sToken.'\')) as x where (class is not null and class not in (\'place\')) or country_code is not null'; + if (CONST_Debug) var_Dump($sSQL); $aSearchWords = $oDB->getAll($sSQL); $aNewSearches = array(); foreach($aSearches as $aSearch) @@ -297,7 +312,7 @@ // Check which tokens we have, get the ID numbers $sSQL = 'select word_id,word_token, word, class, type, location, country_code, operator'; $sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$aTokens)).')'; - $sSQL .= ' and (class is null or class not in (\'highway\'))'; + $sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency; // $sSQL .= ' group by word_token, word, class, type, location, country_code'; if (CONST_Debug) var_Dump($sSQL); @@ -309,8 +324,7 @@ $aDatabaseWords = array(); if (PEAR::IsError($aDatabaseWords)) { - var_dump($sSQL, $aDatabaseWords); - exit; + failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords); } $aPossibleMainWordIDs = array(); foreach($aDatabaseWords as $aToken) @@ -359,7 +373,8 @@ // Try and calculate GB postcodes we might be missing foreach($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) != ' ') { @@ -379,7 +394,7 @@ // 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')); } } @@ -401,17 +416,16 @@ Score how good the search is so they can be ordered */ - foreach($aPhrases as $iPhrase => $sPhrase) { $aNewPhraseSearches = array(); - foreach($aPhrases[$iPhrase]['wordsets'] as $iWordset => $aWordset) + foreach($aPhrases[$iPhrase]['wordsets'] as $aWordset) { $aWordsetSearches = $aSearches; // Add all words from this wordset - foreach($aWordset as $sToken) + foreach($aWordset as $iToken => $sToken) { //echo "
$sToken"; $aNewWordsetSearches = array(); @@ -434,12 +448,12 @@ if ($aSearch['sCountryCode'] === false) { $aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']); - // Country is almost always at the end of the string - increase score for finding it anywhere else (opimisation) - if ($iWordset+1 != sizeof($aPhrases[$iPhrase]['wordsets']) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5; + // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation) + if ($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } - elseif ($aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) + elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) { if ($aSearch['fLat'] === '') { @@ -488,7 +502,7 @@ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } - else + elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { if (sizeof($aSearch['aName'])) { @@ -515,6 +529,8 @@ // 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) @@ -534,6 +550,7 @@ $aSearch['iNamePhrase'] = $iPhrase; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } + } } } else @@ -688,22 +705,22 @@ $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'"; if ($oDB->getOne($sSQL)) { - $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']; + $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct"; if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)"; - $sSQL .= " where st_contains($sViewboxSmallSQL, centroid)"; + $sSQL .= " where st_contains($sViewboxSmallSQL, ct.centroid)"; if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; - if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; + if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $oDB->getCol($sSQL); if (!sizeof($aPlaceIDs)) { - $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']; + $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct"; if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)"; - $sSQL .= " where st_contains($sViewboxLargeSQL, centroid)"; + $sSQL .= " where st_contains($sViewboxLargeSQL, ct.centroid)"; if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; - if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; + if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $oDB->getCol($sSQL); @@ -712,7 +729,7 @@ else { $sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'"; - $sSQL .= " and st_contains($sViewboxSmallSQL, centroid)"; + $sSQL .= " and st_contains($sViewboxSmallSQL, geometry)"; if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; $sSQL .= " limit $iLimit"; @@ -730,6 +747,9 @@ // 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'])."'"; @@ -774,8 +794,7 @@ $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? @@ -799,7 +818,7 @@ $sPlaceIDs = join(',',$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)) { @@ -956,8 +975,7 @@ 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); @@ -995,6 +1013,7 @@ $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, $sLanguagePrefArraySQL) "; @@ -1033,8 +1052,7 @@ if (PEAR::IsError($aSearchResults)) { - var_dump($sSQL, $aSearchResults); - exit; + failInternalError("Could not get details for place.", $sSQL, $aSearchResults); } } } // end if ($sQuery) @@ -1102,8 +1120,7 @@ if (PEAR::IsError($aSearchResults)) { - var_dump($sSQL, $aSearchResults); - exit; + failInternalError("Could not get details for place (near).", $sSQL, $aSearchResults); } } } @@ -1117,28 +1134,37 @@ //var_Dump($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)) { @@ -1234,6 +1260,16 @@ //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*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right + //if (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']); /* if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance']) @@ -1255,7 +1291,6 @@ $aResult['foundorder'] = $iResNum; $aSearchResults[$iResNum] = $aResult; } - uasort($aSearchResults, 'byImportance'); //var_dump($aSearchResults);exit; @@ -1278,10 +1313,10 @@ $bFirst = false; } if (!$bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']]) - && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['osm_class'].$aResult['name']]))) + && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']]))) { $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true; - $aClassTypeNameDone[$aResult['osm_type'].$aResult['osm_class'].$aResult['name']] = true; + $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']] = true; $aSearchResults[] = $aResult; }