X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/cb40aa41d5b0efefdb2b9263e87ed016cce96277..14e6dab005c49a77ac49b6503f7807e620ea0ed3:/website/search.php diff --git a/website/search.php b/website/search.php index a1052394..d9adf6b8 100755 --- a/website/search.php +++ b/website/search.php @@ -13,12 +13,16 @@ $sOutputFormat = 'html'; $aSearchResults = array(); $aExcludePlaceIDs = array(); + $sCountryCodesSQL = false; $sSuggestion = $sSuggestionURL = false; $bDeDupe = isset($_GET['dedupe'])?(bool)$_GET['dedupe']:true; $bReverseInPlan = false; $iLimit = isset($_GET['limit'])?(int)$_GET['limit']:10; + $iOffset = isset($_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')) @@ -35,6 +39,9 @@ // Prefered language $aLangPrefOrder = getPrefferedLangauges(); if (isset($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']) && $_GET['exclude_place_ids']) @@ -45,6 +52,43 @@ if ($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']):''); @@ -57,11 +101,20 @@ $aPhrases = array_reverse($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 $sViewboxCentreSQL = $sViewboxSmallSQL = $sViewboxLargeSQL = false; @@ -128,29 +181,43 @@ // 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' => 0, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), + array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), 'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'') ); @@ -179,7 +246,7 @@ $sQuery = str_replace($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 != \'place\') 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\',\'highway\')) or country_code is not null'; $aSearchWords = $oDB->getAll($sSQL); $aNewSearches = array(); foreach($aSearches as $aSearch) @@ -219,7 +286,8 @@ $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($aPhrase['string'])) @@ -233,7 +301,7 @@ { unset($aPhrases[$iPhrase]); } - } + } // reindex phrases - we make assumptions later on $aPhrases = array_values($aPhrases); @@ -244,7 +312,8 @@ // 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 .= ' group by word_token, word, class, type, location,country_code'; + $sSQL .= ' and (class is null or class not in (\'highway\'))'; +// $sSQL .= ' group by word_token, word, class, type, location, country_code'; if (CONST_Debug) var_Dump($sSQL); @@ -258,6 +327,7 @@ var_dump($sSQL, $aDatabaseWords); exit; } + $aPossibleMainWordIDs = array(); foreach($aDatabaseWords as $aToken) { if (isset($aValidTokens[$aToken['word_token']])) @@ -268,6 +338,7 @@ { $aValidTokens[$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, $aValidTokens); @@ -318,6 +389,15 @@ } } + foreach($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 @@ -348,10 +428,15 @@ // Add all words from this wordset foreach($aWordset as $sToken) { +//echo "
$sToken"; $aNewWordsetSearches = array(); foreach($aWordsetSearches as $aCurrentSearch) { +//echo ""; +//var_dump($aCurrentSearch); +//echo ""; + // If the token is valid if (isset($aValidTokens[' '.$sToken])) { @@ -385,11 +470,12 @@ { $aSearch['sHouseNumber'] = $sToken; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; - +/* // Fall back to not searching for this item (better than nothing) $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 2; + $aSearch['iSearchRank'] += 1; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; +*/ } } elseif ($aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) @@ -421,12 +507,19 @@ { if (sizeof($aSearch['aName'])) { - $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false) + { + $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + } + else + { + $aSearch['iSearchRank'] += 1000; // skip; + } } else { $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - $aSearch['iNamePhrase'] = $iPhrase; +// $aSearch['iNamePhrase'] = $iPhrase; } if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } @@ -437,15 +530,25 @@ // Allow searching for a word - but at extra cost foreach($aValidTokens[$sToken] as $aSearchTerm) { - $aSearch = $aCurrentSearch; - $aSearch['iSearchRank']+=5; - $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - if (!sizeof($aSearch['aName']) || $aSearch['iNamePhrase'] == $iPhrase) +//var_Dump('
',$aSearch['aName']); + + if (sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4) + { + $aSearch = $aCurrentSearch; + $aSearch['iSearchRank'] += 1; + $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; + } + + if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) { + $aSearch = $aCurrentSearch; + $aSearch['iSearchRank'] += 2; + if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2; $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; $aSearch['iNamePhrase'] = $iPhrase; + if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } - if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } else @@ -464,6 +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); } @@ -487,6 +605,9 @@ $aSearches = array_merge($aSearches, $aNewSearches); if ($iSearchCount > 50) break; } + +// if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); + } } else @@ -508,56 +629,22 @@ if ($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][] = $aReverseSearch; - } - } - } - } - -//var_Dump($aGroupedSearches); exit; - - // Filter out duplicate searches - $aSearchHash = array(); - foreach($aGroupedSearches as $iGroup => $aSearches) - { - foreach($aSearches as $iSearch => $aSearch) - { - $sHash = serialize($aSearch); - if (isset($aSearchHash[$sHash])) - { - unset($aGroupedSearches[$iGroup][$iSearch]); - if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]); - } - else - { - $aSearchHash[$sHash] = 1; - } - } - } - - if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); - - if ($bReverseInPlan) - { - foreach($aGroupedSearches 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][] = $aReverseSearch; } } } @@ -592,19 +679,34 @@ foreach($aSearches as $aSearch) { $iQueryLoop++; + // Must have a location term - if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress'] && !$aSearch['fLon'])) + 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); @@ -613,7 +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); @@ -624,16 +728,16 @@ { $sSQL = "select place_id from placex where class='".$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 { - if ($aSearch['aName'] == array(282=>'282')) continue; - if (CONST_Debug) var_dump('
',$aSearch); if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens); $aPlaceIDs = array(); @@ -654,23 +758,27 @@ { $aTerms[] = "place_id not in (".join(',',$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']) $sSQL .= " limit 1"; @@ -684,35 +792,29 @@ var_dump($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)) { + $aRoadPlaceIDs = $aPlaceIDs; $sPlaceIDs = join(',',$aPlaceIDs); - $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-, ]',$aSearch['sHouseNumber']).'\\\\M'; - - // Make sure everything nearby is indexed (if we pre-indexed houses this wouldn't be needed!) - $sSQL = "update placex set indexed = true from placex as f where placex.indexed = false"; - $sSQL .= " and f.place_id in (".$sPlaceIDs.") and ST_DWithin(placex.geometry, f.geometry, 0.004)"; - $sSQL .= " and placex.housenumber ~* E'".$sHouseNumberRegex."'"; - $sSQL .= " and placex.class='place' and placex.type='house'"; - if (CONST_Debug) var_dump($sSQL); - $oDB->query($sSQL); - // Now they are indexed look for a house attached to a street we found + $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-, ]',$aSearch['sHouseNumber']).'\\\\M'; $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'"; if (sizeof($aExcludePlaceIDs)) { @@ -721,6 +823,38 @@ $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $oDB->getCol($sSQL); + + // If not try the aux fallback table + if (!sizeof($aPlaceIDs)) + { + $sSQL = "select place_id from location_property_aux where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'"; + if (sizeof($aExcludePlaceIDs)) + { + $sSQL .= " and place_id not in (".join(',',$aExcludePlaceIDs).")"; + } +// $sSQL .= " limit $iLimit"; + if (CONST_Debug) var_dump($sSQL); + $aPlaceIDs = $oDB->getCol($sSQL); + } + + if (!sizeof($aPlaceIDs)) + { + $sSQL = "select place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'"; + if (sizeof($aExcludePlaceIDs)) + { + $sSQL .= " and place_id not in (".join(',',$aExcludePlaceIDs).")"; + } +// $sSQL .= " limit $iLimit"; + if (CONST_Debug) var_dump($sSQL); + $aPlaceIDs = $oDB->getCol($sSQL); + } + + // Fallback to the road + if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) + { + $aPlaceIDs = $aRoadPlaceIDs; + } + } if ($aSearch['sClass'] && sizeof($aPlaceIDs)) @@ -731,6 +865,7 @@ { // 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); @@ -738,30 +873,70 @@ 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.geometry) 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); @@ -769,19 +944,26 @@ 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); } + } } } @@ -818,19 +1000,117 @@ { $sOrderSQL .= 'when min(place_id) = '.$iPlaceID.' then '.$iOrder.' '; } - $sOrderSQL .= ' ELSE 10000000 END ASC'; - $sSQL = "select osm_type,osm_id,class,type,rank_search,rank_address,min(place_id) as place_id,country_code,"; + $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($aSearchResults)) + { + var_dump($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 .= "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 .= "group by osm_type,osm_id,class,type,rank_search,rank_address,country_code"; + $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']) "; - $sSQL .= "order by rank_search,rank_address,".$sOrderSQL; + $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; @@ -849,9 +1129,9 @@ { $sSearchResult = 'No Results Found'; } - +//var_Dump($aSearchResults); +//exit; $aClassType = getClassTypesWithImportance(); - foreach($aSearchResults as $iResNum => $aResult) { if (CONST_Search_AreaPolygons || true) @@ -861,6 +1141,11 @@ $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,"; $sSQL .= "ST_AsText(outline) as outlinestring from get_place_boundingbox_quick(".$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\''; $aPointPolygon = $oDB->getRow($sSQL); if (PEAR::IsError($aPointPolygon)) { @@ -891,7 +1176,7 @@ } // Output data suitable for display (points and a bounding box) - if ($bShowPolygons) + if ($bShowPolygons && isset($aPolyPoints)) { $aResult['aPolyPoints'] = array(); foreach($aPolyPoints as $aPoint) @@ -951,12 +1236,27 @@ $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$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'], $aResult['country_code']); +//var_dump($aResult['address']); +//exit; } - if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance']) +//if (CONST_Debug) var_dump($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']; @@ -965,13 +1265,15 @@ { $aResult['importance'] = 1000000000000000; } - +*/ $aResult['name'] = $aResult['langaddress']; $aResult['foundorder'] = $iResNum; $aSearchResults[$iResNum] = $aResult; } uasort($aSearchResults, 'byImportance'); + +//var_dump($aSearchResults);exit; $aOSMIDDone = array(); $aClassTypeNameDone = array(); @@ -997,6 +1299,9 @@ $aClassTypeNameDone[$aResult['osm_type'].$aResult['osm_class'].$aResult['name']] = true; $aSearchResults[] = $aResult; } + + // Absolute limit on number of results + if (sizeof($aSearchResults) >= $iLimit) break; } $sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '1 day'::interval,'YYYY/MM/DD') from import_status limit 1"); @@ -1024,4 +1329,4 @@ if (CONST_Debug) exit; - include('.htlib/output/search-'.$sOutputFormat.'.php'); + include(CONST_BasePath.'/lib/template/search-'.$sOutputFormat.'.php');