X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/10620c71bb7999ed89ca45e545d9ce611a5e1476..1298d144e015aae98766ea708df031cefc282a8a:/website/search.php diff --git a/website/search.php b/website/search.php index cd567385..5e2ff371 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; @@ -20,6 +21,8 @@ $iOffset = isset($_GET['offset'])?(int)$_GET['offset']:0; $iMaxRank = 20; if ($iLimit > 100) $iLimit = 100; + $iMinAddressRank = 0; + $iMaxAddressRank = 30; // Format for output if (isset($_GET['format']) && ($_GET['format'] == 'html' || $_GET['format'] == 'xml' || $_GET['format'] == 'json' || $_GET['format'] == 'jsonv2')) @@ -33,9 +36,12 @@ // Show address breakdown $bShowAddressDetails = isset($_GET['addressdetails']) && $_GET['addressdetails']; - // Prefered language - $aLangPrefOrder = getPrefferedLangauges(); -// if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true; + // 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; + $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$aLangPrefOrder))."]"; if (isset($_GET['exclude_place_ids']) && $_GET['exclude_place_ids']) @@ -46,25 +52,67 @@ if ($iExcludedPlaceID) $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID; } } + + // Only certain ranks of feature + if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType']; + + if (isset($_GET['featuretype'])) + { + switch($_GET['featuretype']) + { + case 'country': + $iMinAddressRank = $iMaxAddressRank = 4; + break; + case 'state': + $iMinAddressRank = $iMaxAddressRank = 8; + break; + case 'city': + $iMinAddressRank = 14; + $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($_GET['q']):''); - if (!$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); } - 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, new york', $sQuery); + $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'])) + { + $sQuery = preg_replace('/,\s*il\s*(,|$)/',', 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 // Small is the actual view box, Large is double (on each axis) that @@ -114,8 +162,7 @@ $sViewboxSmallSQL = $oDB->getOne($sSQL); if (PEAR::isError($sViewboxSmallSQL)) { - var_dump($sViewboxSmallSQL); - exit; + failInternalError("Could not get small viewbox.", $sSQL, $sViewboxSmallSQL); } $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry"; @@ -123,8 +170,7 @@ $sViewboxLargeSQL = $oDB->getOne($sSQL); if (PEAR::isError($sViewboxLargeSQL)) { - var_dump($sViewboxLargeSQL); - exit; + failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL); } $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry"; } @@ -132,26 +178,40 @@ // 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) { - // Start with a blank search $aSearches = array( array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), @@ -184,6 +244,7 @@ $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) @@ -223,7 +284,8 @@ $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($aPhrase['string'])) @@ -237,7 +299,7 @@ { unset($aPhrases[$iPhrase]); } - } + } // reindex phrases - we make assumptions later on $aPhrases = array_values($aPhrases); @@ -260,9 +322,9 @@ $aDatabaseWords = array(); if (PEAR::IsError($aDatabaseWords)) { - var_dump($sSQL, $aDatabaseWords); - exit; + failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords); } + $aPossibleMainWordIDs = array(); foreach($aDatabaseWords as $aToken) { if (isset($aValidTokens[$aToken['word_token']])) @@ -273,6 +335,7 @@ { $aValidTokens[$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, $aValidTokens); @@ -304,11 +367,12 @@ } } if ($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) != ' ') { @@ -322,7 +386,16 @@ } } } -*/ + + 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(array('class'=>'place','type'=>'house')); + } + } + // Any words that have failed completely? // TODO: suggestions @@ -341,7 +414,6 @@ Score how good the search is so they can be ordered */ - foreach($aPhrases as $iPhrase => $sPhrase) { $aNewPhraseSearches = array(); @@ -374,12 +446,12 @@ if ($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) + // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation) if ($iWordset+1 != sizeof($aPhrases[$iPhrase]['wordsets']) || $iPhrase+1 != sizeof($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'] === '') { @@ -428,11 +500,11 @@ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } - else + elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { if (sizeof($aSearch['aName'])) { - if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4) + if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false) { $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; } @@ -450,14 +522,16 @@ } } } - if (isset($aValidTokens[$sToken]) && strlen($sToken) >= 4) + if (isset($aValidTokens[$sToken])) { // 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($aCurrentSearch['aName'])) + if (sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4) { $aSearch = $aCurrentSearch; $aSearch['iSearchRank'] += 1; @@ -468,11 +542,13 @@ if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) { $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 4; + $aSearch['iSearchRank'] += 2; + if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2; $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; $aSearch['iNamePhrase'] = $iPhrase; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } + } } } else @@ -491,6 +567,21 @@ $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches); usort($aNewPhraseSearches, 'bySearchRank'); + + $aSearchHash = array(); + foreach($aNewPhraseSearches as $iSearch => $aSearch) + { + $sHash = serialize($aSearch); + if (isset($aSearchHash[$sHash])) + { + unset($aNewPhraseSearches[$iSearch]); + } + else + { + $aSearchHash[$sHash] = 1; + } + } + $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50); } @@ -536,18 +627,24 @@ 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][] = $aReverseSearch; } } } @@ -586,16 +683,30 @@ // Must have a location term if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon']) { - if (!$bBoundingBoxSearch && !$aSearch['fLon']) continue; - if (!$aSearch['sClass']) continue; - if (CONST_Debug) var_dump('
',$aSearch); - if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens); - - $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'"; - if ($oDB->getOne($sSQL)) + if ($aSearch['sCountryCode'] && !$aSearch['sClass']) + { + if (4 >= $iMinAddressRank && 4 <= $iMaxAddressRank) + { + $sSQL = "select place_id from placex where country_code='".$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); + } + } + else { + if (!$bBoundingBoxSearch && !$aSearch['fLon']) continue; + if (!$aSearch['sClass']) continue; + if (CONST_Debug) var_dump('
',$aSearch); + if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens); + + $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'"; + if ($oDB->getOne($sSQL)) + { $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']; + if ($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); @@ -604,7 +715,9 @@ 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); @@ -615,11 +728,13 @@ { $sSQL = "select place_id from placex where class='".$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); $aPlaceIDs = $oDB->getCol($sSQL); } + } } else { @@ -630,6 +745,9 @@ // 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'])."'"; @@ -643,23 +761,27 @@ { $aTerms[] = "place_id not in (".join(',',$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"; @@ -670,22 +792,23 @@ $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? $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)) { @@ -693,7 +816,7 @@ $sPlaceIDs = join(',',$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)) { @@ -729,7 +852,7 @@ } // Fallback to the road - if (!sizeof($aPlaceIDs)) + if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) { $aPlaceIDs = $aRoadPlaceIDs; } @@ -744,6 +867,7 @@ { // 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); @@ -751,30 +875,70 @@ 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 || $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"; - $sSQL .= ",placex as f where "; - $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, st_centroid(f.geometry), $fRange) "; + $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) $sSQL .= " join placex as lp using (place_id)"; + if ($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 ($sNearPointSQL) $sSQL .= " order by ST_Distance($sNearPointSQL, l.centroid) ASC"; - else $sSQL .= " order by ST_Distance(l.centroid, f.geometry) asc"; + if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)"; + 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); @@ -782,19 +946,26 @@ else { 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) "; + + $sOrderBySQL = ''; + if ($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 ($sNearPointSQL) $sSQL .= " order by ST_Distance($sNearPointSQL, l.geometry) ASC"; - else $sSQL .= " order by ST_Distance(l.geometry, f.geometry) asc, l.rank_search ASC"; + if ($sCountryCodesSQL) $sSQL .= " and l.country_code in ($sCountryCodesSQL)"; + 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); } + } } } @@ -802,8 +973,7 @@ 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); @@ -832,48 +1002,123 @@ $sOrderSQL .= 'when min(place_id) = '.$iPlaceID.' then '.$iOrder.' '; } $sOrderSQL .= ' ELSE 10000000 END'; - $sSQL = "select osm_type,osm_id,class,type,rank_search,rank_address,min(place_id) as place_id,country_code,"; + $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 .= $sOrderSQL." as porder, "; + $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance "; $sSQL .= "from placex where place_id in ($sPlaceIDs) "; - $sSQL .= "group by osm_type,osm_id,class,type,rank_search,rank_address,country_code"; + $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) "; $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,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,"; + $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 .= $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 'T' as osm_type,place_id as osm_id,'place' as class,'house' as type,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,"; + $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 .= $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 rank_search,rank_address,porder asc"; + $sSQL .= "order by importance desc"; +// $sSQL .= "order by rank_search,rank_address,porder asc"; if (CONST_Debug) var_dump('
',$sSQL); $aSearchResults = $oDB->getAll($sSQL); //var_dump($sSQL,$aSearchResults);exit; if (PEAR::IsError($aSearchResults)) { - var_dump($sSQL, $aSearchResults); - exit; + failInternalError("Could not get details for place.", $sSQL, $aSearchResults); + } + } + } // end if ($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"; +// $sSQL .= "order by rank_search,rank_address,porder asc"; + if (CONST_Debug) var_dump('
',$sSQL); + $aSearchResults = $oDB->getAll($sSQL); +//var_dump($sSQL,$aSearchResults);exit; + + if (PEAR::IsError($aSearchResults)) + { + failInternalError("Could not get details for place (near).", $sSQL, $aSearchResults); } } } @@ -884,31 +1129,40 @@ { $sSearchResult = 'No Results Found'; } - +//var_Dump($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'],$aMatch)) { @@ -931,7 +1185,7 @@ } // Output data suitable for display (points and a bounding box) - if ($bShowPolygons) + if ($bShowPolygons && isset($aPolyPoints)) { $aResult['aPolyPoints'] = array(); foreach($aPolyPoints as $aPoint) @@ -991,6 +1245,12 @@ $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png'; } + if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label']) + && $aClassType[$aResult['class'].':'.$aResult['type']]['label']) + { + $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label']; + } + if ($bShowAddressDetails) { $aResult['address'] = getAddressDetails($oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code']); @@ -998,7 +1258,24 @@ //exit; } - if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance']) + // 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']) + && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance']) + { + $aResult['importance'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance']; + } + elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance']) && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']) { $aResult['importance'] = $aClassType[$aResult['class'].':'.$aResult['type']]['importance']; @@ -1007,16 +1284,15 @@ { $aResult['importance'] = 1000000000000000; } - +*/ $aResult['name'] = $aResult['langaddress']; $aResult['foundorder'] = $iResNum; $aSearchResults[$iResNum] = $aResult; } + uasort($aSearchResults, 'byImportance'); //var_dump($aSearchResults);exit; - uasort($aSearchResults, 'byImportance'); - $aOSMIDDone = array(); $aClassTypeNameDone = array(); $aToFilter = $aSearchResults; @@ -1035,12 +1311,15 @@ $bFirst = false; } if (!$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[] = $aResult; } + + // Absolute limit on number of results + if (sizeof($aSearchResults) >= $iLimit) break; } $sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '1 day'::interval,'YYYY/MM/DD') from import_status limit 1");