X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/535c3d60d1cd45a03358d54b91af39c23b90bb39..b86d5c6a42624708af4e57121caf5cc02241544f:/website/search.php diff --git a/website/search.php b/website/search.php index fc03a58d..5ba2571b 100755 --- a/website/search.php +++ b/website/search.php @@ -1,8 +1,11 @@ 100) $iLimit = 100; + if ($iFinalLimit > 50) $iFinalLimit = 50; + $iLimit = $iFinalLimit + min($iFinalLimit, 10); $iMinAddressRank = 0; $iMaxAddressRank = 30; + $sAllowedTypesSQLList = false; // Format for output if (isset($_GET['format']) && ($_GET['format'] == 'html' || $_GET['format'] == 'xml' || $_GET['format'] == 'json' || $_GET['format'] == 'jsonv2')) @@ -31,13 +36,38 @@ } // Show / use polygons - $bShowPolygons = isset($_GET['polygon']) && $_GET['polygon']; + $bShowPolygons = (boolean)isset($_GET['polygon']) && $_GET['polygon']; + if ($sOutputFormat == 'html') { + $bAsText = $bShowPolygons; + $bShowPolygons = false; + $bAsGeoJSON = false; + $bAsKML = false; + $bAsSVG = false; + } else { + $bAsGeoJSON = (boolean)isset($_GET['polygon_geojson']) && $_GET['polygon_geojson']; + $bAsKML = (boolean)isset($_GET['polygon_kml']) && $_GET['polygon_kml']; + $bAsSVG = (boolean)isset($_GET['polygon_svg']) && $_GET['polygon_svg']; + $bAsText = (boolean)isset($_GET['polygon_text']) && $_GET['polygon_text']; + if ((($bShowPolygons?1:0) + + ($bAsGeoJSON?1:0) + + ($bAsKML?1:0) + + ($bAsSVG?1:0) + + ($bAsText?1:0) + ) > CONST_PolygonOutput_MaximumTypes) { + if (CONST_PolygonOutput_MaximumTypes) { + userError("Select only ".CONST_PolygonOutput_MaximumTypes." polgyon output option"); + } else { + userError("Polygon output is disabled"); + } + exit; + } + } // 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'])) $bReverseInPlan = true; @@ -53,7 +83,7 @@ } } - // Only certain ranks of feature + // Only certain ranks of feature if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType']; if (isset($_GET['featuretype'])) @@ -92,16 +122,50 @@ // 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); } + function structuredAddressElement(&$aStructuredQuery, &$iMinAddressRank, &$iMaxAddressRank, $aParams, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank) + { + if (!isset($_GET[$sKey])) return false; + $sValue = trim($_GET[$sKey]); + if (!$sValue) return false; + $aStructuredQuery[$sKey] = $sValue; + if ($iMinAddressRank == 0 && $iMaxAddressRank == 30) { + $iMinAddressRank = $iNewMinAddressRank; + $iMaxAddressRank = $iNewMaxAddressRank; + } + return true; + } + + // Structured query? + $aStructuredOptions = array( + array('amenity', 26, 30), + array('street', 26, 30), + array('city', 14, 24), + array('county', 9, 13), + array('state', 8, 8), + array('country', 4, 4), + array('postalcode', 16, 25), + ); + $aStructuredQuery = array(); + $sAllowedTypesSQLList = ''; + foreach($aStructuredOptions as $aStructuredOption) + { + loadStructuredAddressElement($aStructuredQuery, $iMinAddressRank, $iMaxAddressRank, $_GET, $aStructuredOption[0], $aStructuredOption[1], $aStructuredOption[2]); + } + if (sizeof($aStructuredQuery) > 0) { + $sQuery = join(', ', $aStructuredQuery); + $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; + } + if ($sQuery) { $hLog = logStart($oDB, 'search', $sQuery, $aLangPrefOrder); @@ -110,9 +174,9 @@ $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); + $sQuery = preg_replace('/,\s*il\s*(,|$)/',', 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 @@ -134,13 +198,15 @@ $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']) { $aPoints = explode(',',$_GET['route']); if (sizeof($aPoints) % 2 != 0) { - echo "Uneven number of points"; + userError("Uneven number of points"); exit; } $sViewboxCentreSQL = "ST_SetSRID('LINESTRING("; @@ -163,8 +229,7 @@ $sViewboxSmallSQL = $oDB->getOne($sSQL); if (PEAR::isError($sViewboxSmallSQL)) { - var_dump($sViewboxSmallSQL); - exit; + failInternalError("Could not get small viewbox.", $sSQL, $sViewboxSmallSQL); } $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry"; @@ -172,33 +237,48 @@ $sViewboxLargeSQL = $oDB->getOne($sSQL); if (PEAR::isError($sViewboxLargeSQL)) { - var_dump($sViewboxLargeSQL); - exit; + failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL); } $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry"; + $bBoundingBoxSearch = true; } // 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) + if ($sQuery || $aStructuredQuery) { // Start with a blank search $aSearches = array( @@ -226,12 +306,18 @@ preg_match_all('/\\[([a-zA-Z]*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER); $aSpecialTerms = array(); + if (isset($aStructuredQuery['amenity']) && $aStructuredQuery['amenity']) + { + $aSpecialTermsRaw[] = array('['.$aStructuredQuery['amenity'].']', $aStructuredQuery['amenity']); + unset($aStructuredQuery['amenity']); + } foreach($aSpecialTermsRaw as $aSpecialTerm) { $sQuery = str_replace($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) @@ -259,7 +345,17 @@ // Split query into phrases // Commas are used to reduce the search space by indicating where phrases split - $aPhrases = explode(',',$sQuery); + if (sizeof($aStructuredQuery) > 0) + { + $aPhrases = $aStructuredQuery; + $bStructuredPhrases = true; + } + else + { + $aPhrases = explode(',',$sQuery); + $bStructuredPhrases = false; + } + // Convert each phrase to standard form // Create a list of standard words @@ -271,7 +367,8 @@ $aPhrase = $oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string"); if (PEAR::isError($aPhrase)) { - var_dump($aPhrase); + userError("Illegal query string (not an UTF-8 string): ".$sPhrase); + if (CONST_Debug) var_dump($aPhrase); exit; } if (trim($aPhrase['string'])) @@ -288,6 +385,7 @@ } // reindex phrases - we make assumptions later on + $aPhraseTypes = array_keys($aPhrases); $aPhrases = array_values($aPhrases); if (sizeof($aTokens)) @@ -296,7 +394,7 @@ // 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); @@ -308,8 +406,7 @@ $aDatabaseWords = array(); if (PEAR::IsError($aDatabaseWords)) { - var_dump($sSQL, $aDatabaseWords); - exit; + failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords); } $aPossibleMainWordIDs = array(); foreach($aDatabaseWords as $aToken) @@ -354,11 +451,12 @@ } } if ($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) != ' ') { @@ -372,14 +470,13 @@ } } } -*/ foreach($aTokens as $sToken) { // Unknown single word token with a number - assume it is a house number if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken,' ') === false && preg_match('/[0-9]/', $sToken)) { - $aValidTokens[' '.$sToken] = array('class'=>'place','type'=>'house'); + $aValidTokens[' '.$sToken] = array(array('class'=>'place','type'=>'house')); } } @@ -393,29 +490,30 @@ Calculate all searches using aValidTokens i.e. 'Wodsworth Road, Sheffield' => - + Phrase Wordset 0 0 (wodsworth road) 0 1 (wodsworth)(road) 1 0 (sheffield) - + Score how good the search is so they can be ordered */ - foreach($aPhrases as $iPhrase => $sPhrase) { $aNewPhraseSearches = array(); + if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase]; + else $sPhraseType = ''; - foreach($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"; $aNewWordsetSearches = array(); - + foreach($aWordsetSearches as $aCurrentSearch) { //echo ""; @@ -429,17 +527,17 @@ { $aSearch = $aCurrentSearch; $aSearch['iSearchRank']++; - if ($aSearchTerm['country_code'] !== null && $aSearchTerm['country_code'] != '0') + if (($sPhraseType == '' || $sPhraseType == 'country') && $aSearchTerm['country_code'] !== null && $aSearchTerm['country_code'] != '0') { if ($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'] === '') { @@ -449,7 +547,7 @@ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } - elseif ($aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') + elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') { if ($aSearch['sHouseNumber'] === '') { @@ -463,7 +561,7 @@ */ } } - elseif ($aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) + elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) { if ($aSearch['sClass'] === '') { @@ -488,11 +586,11 @@ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } - else + elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { if (sizeof($aSearch['aName'])) { - if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false) + if (($sPhraseType != 'street' && $sPhraseType != 'country') && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false)) { $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; } @@ -515,24 +613,25 @@ // Allow searching for a word - but at extra cost foreach($aValidTokens[$sToken] as $aSearchTerm) { -//var_Dump('
',$aSearch['aName']); - - if (sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4) + if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { - $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 1; - $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; - } + if (($sPhraseType != 'street') && 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[] = $aSearch; + if (!sizeof($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[] = $aSearch; + } } } } @@ -635,6 +734,42 @@ } } + if (CONST_Search_TryDroppedAddressTerms && sizeof($aStructuredQuery) > 0) + { + $aCopyGroupedSearches = $aGroupedSearches; + foreach($aCopyGroupedSearches as $iGroup => $aSearches) + { + foreach($aSearches as $iSearch => $aSearch) + { + $aReductionsList = array($aSearch['aAddress']); + $iSearchRank = $aSearch['iSearchRank']; + while(sizeof($aReductionsList) > 0) + { + $iSearchRank += 5; + if ($iSearchRank > iMaxRank) break 3; + $aNewReductionsList = array(); + foreach($aReductionsList as $aReductionsWordList) + { + for ($iReductionWord = 0; $iReductionWord < sizeof($aReductionsWordList); $iReductionWord++) + { + $aReductionsWordListResult = array_merge(array_slice($aReductionsWordList, 0, $iReductionWord), array_slice($aReductionsWordList, $iReductionWord+1)); + $aReverseSearch = $aSearch; + $aSearch['aAddress'] = $aReductionsWordListResult; + $aSearch['iSearchRank'] = $iSearchRank; + $aGroupedSearches[$iSearchRank][] = $aReverseSearch; + if (sizeof($aReductionsWordListResult) > 0) + { + $aNewReductionsList[] = $aReductionsWordListResult; + } + } + } + $aReductionsList = $aNewReductionsList; + } + } + } + ksort($aGroupedSearches); + } + // Filter out duplicate searches $aSearchHash = array(); foreach($aGroupedSearches as $iGroup => $aSearches) @@ -668,7 +803,7 @@ // Must have a location term if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon']) { - if ($aSearch['sCountryCode'] && !$aSearch['sClass']) + if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) { if (4 >= $iMinAddressRank && 4 <= $iMaxAddressRank) { @@ -684,26 +819,25 @@ if (!$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']; + $sSQL = "select place_id from place_classtype_".$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 +846,7 @@ 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) and linked_place_id is null"; if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; $sSQL .= " limit $iLimit"; @@ -730,6 +864,9 @@ // 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'])."'"; @@ -751,7 +888,7 @@ if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL"; if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc"; - $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.92-(search_rank::float/33) else importance end'; + $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end'; if ($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"; @@ -770,12 +907,11 @@ else $sSQL .= " limit ".$iLimit; - if (CONST_Debug) var_dump($sSQL); + if (CONST_Debug) { var_dump($sSQL); } $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 +935,7 @@ $sPlaceIDs = join(',',$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)) { @@ -846,14 +982,17 @@ { $sPlaceIDs = join(',',$aPlaceIDs); + $aClassPlaceIDs = array(); + if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'name') { // 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']."'"; + $sSQL .= " and linked_place_id is null"; if ($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); + $aClassPlaceIDs = $oDB->getCol($sSQL); } if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') // & in @@ -871,9 +1010,9 @@ if ($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); + $sSQL = "select geometry from placex where place_id in ($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) @@ -883,10 +1022,10 @@ 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); + $sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and rank_search < $iMaxRank"; + if (CONST_Debug) var_dump($sSQL); + $aPlaceIDs = $oDB->getCol($sSQL); + $sPlaceIDs = join(',',$aPlaceIDs); } if ($sPlaceIDs || $sPlaceGeom) @@ -896,9 +1035,14 @@ if ($bCacheTable) { // More efficient - can make the range bigger - $fRange = 0.05; + $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)"; - $sSQL = "select l.place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l"; + $sSQL = "select distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l"; if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)"; if ($sPlaceIDs) { @@ -915,18 +1059,21 @@ $sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")"; } if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)"; - if ($sNearPointSQL) $sSQL .= " order by ST_Distance($sNearPointSQL, l.centroid) ASC"; - else if ($sPlaceIDs) $sSQL .= " order by ST_Distance(l.centroid, f.geometry) asc"; - else if ($sPlaceGeom) $sSQL .= " order by ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid) asc"; - + if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc"; + if ($iOffset) $sSQL .= " offset $iOffset"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); - $aPlaceIDs = $oDB->getCol($sSQL); + $aClassPlaceIDs = array_merge($aClassPlaceIDs, $oDB->getCol($sSQL)); } else { if (isset($aSearch['fRadius']) && $aSearch['fRadius']) $fRange = $aSearch['fRadius']; - $sSQL = "select l.place_id from placex as l,placex as f where "; + + $sOrderBySQL = ''; + if ($sNearPointSQL) $sOrderBySQL = "ST_Distance($sNearPointSQL, l.geometry)"; + else $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)"; + + $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from placex as l,placex as f where "; $sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, st_centroid(f.geometry), $fRange) "; $sSQL .= "and l.class='".$aSearch['sClass']."' and l.type='".$aSearch['sType']."' "; if (sizeof($aExcludePlaceIDs)) @@ -934,22 +1081,24 @@ $sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")"; } if ($sCountryCodesSQL) $sSQL .= " and l.country_code in ($sCountryCodesSQL)"; - if ($sNearPointSQL) $sSQL .= " order by ST_Distance($sNearPointSQL, l.geometry) ASC"; - else $sSQL .= " order by ST_Distance(l.geometry, f.geometry) asc, l.rank_search ASC"; + if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc"; + if ($iOffset) $sSQL .= " offset $iOffset"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); - $aPlaceIDs = $oDB->getCol($sSQL); + $aClassPlaceIDs = array_merge($aClassPlaceIDs, $oDB->getCol($sSQL)); } } } + + $aPlaceIDs = $aClassPlaceIDs; + } } if (PEAR::IsError($aPlaceIDs)) { - var_dump($sSQL, $aPlaceIDs); - exit; + failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs); } if (CONST_Debug) var_Dump($aPlaceIDs); @@ -960,14 +1109,15 @@ } if ($iQueryLoop > 20) break; } + //exit; - if (sizeof($aResultPlaceIDs)) break; + if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) break; if ($iGroupLoop > 4) break; if ($iQueryLoop > 30) break; } //exit; // Did we find anything? - if (sizeof($aResultPlaceIDs)) + if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) { //var_Dump($aResultPlaceIDs);exit; // Get the details for display (is this a redundant extra step?) @@ -984,9 +1134,11 @@ $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 .= "coalesce(importance,0.75-(rank_search::float/40)) as importance "; $sSQL .= "from placex where place_id in ($sPlaceIDs) "; $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank "; + if ($sAllowedTypesSQLList) $sSQL .= "and placex.class in $sAllowedTypesSQLList "; + $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, $sLanguagePrefArraySQL) "; @@ -1025,8 +1177,7 @@ if (PEAR::IsError($aSearchResults)) { - var_dump($sSQL, $aSearchResults); - exit; + failInternalError("Could not get details for place.", $sSQL, $aSearchResults); } } } // end if ($sQuery) @@ -1053,7 +1204,7 @@ $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 .= "coalesce(importance,0.75-(rank_search::float/40)) as importance "; $sSQL .= "from placex where place_id in ($sPlaceIDs) "; $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank "; $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance"; @@ -1094,8 +1245,7 @@ if (PEAR::IsError($aSearchResults)) { - var_dump($sSQL, $aSearchResults); - exit; + failInternalError("Could not get details for place (near).", $sSQL, $aSearchResults); } } } @@ -1109,51 +1259,76 @@ //var_Dump($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,"; + $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon"; + if ($bAsGeoJSON) $sSQL .= ",ST_AsGeoJSON(geometry) as asgeojson"; + if ($bAsKML) $sSQL .= ",ST_AsKML(geometry) as askml"; + if ($bAsSVG) $sSQL .= ",ST_AsSVG(geometry) as assvg"; + if ($bAsText || $bShowPolygons) $sSQL .= ",ST_AsText(geometry) as astext"; + $sSQL .= " from placex where place_id = ".$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']) { - // Translate geometary string to point array - if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['outlinestring'],$aMatch)) - { - preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER); + if ($bAsGeoJSON) $aResult['asgeojson'] = $aPointPolygon['asgeojson']; + if ($bAsKML) $aResult['askml'] = $aPointPolygon['askml']; + if ($bAsSVG) $aResult['assvg'] = $aPointPolygon['assvg']; + if ($bAsText) $aResult['astext'] = $aPointPolygon['astext']; + + if ($aPointPolygon['centrelon'] !== null && $aPointPolygon['centrelat'] !== null ) { + $aResult['lat'] = $aPointPolygon['centrelat']; + $aResult['lon'] = $aPointPolygon['centrelon']; } - elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['outlinestring'],$aMatch)) + if ($bShowPolygons) { - $fRadius = 0.01; - $iSteps = ($fRadius * 40000)^2; - $fStepSize = (2*pi())/$iSteps; - $aPolyPoints = array(); - for($f = 0; $f < 2*pi(); $f += $fStepSize) + // Translate geometary string to point array + if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch)) + { + preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER); + } + elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch)) { - $aPolyPoints[] = array('',$aMatch[1]+($fRadius*sin($f)),$aMatch[2]+($fRadius*cos($f))); + preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER); + } + elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['astext'],$aMatch)) + { + $fRadius = 0.01; + $iSteps = ($fRadius * 40000)^2; + $fStepSize = (2*pi())/$iSteps; + $aPolyPoints = array(); + for($f = 0; $f < 2*pi(); $f += $fStepSize) + { + $aPolyPoints[] = array('',$aMatch[1]+($fRadius*sin($f)),$aMatch[2]+($fRadius*cos($f))); + } + $aPointPolygon['minlat'] = $aPointPolygon['minlat'] - $fRadius; + $aPointPolygon['maxlat'] = $aPointPolygon['maxlat'] + $fRadius; + $aPointPolygon['minlon'] = $aPointPolygon['minlon'] - $fRadius; + $aPointPolygon['maxlon'] = $aPointPolygon['maxlon'] + $fRadius; } - $aPointPolygon['minlat'] = $aPointPolygon['minlat'] - $fRadius; - $aPointPolygon['maxlat'] = $aPointPolygon['maxlat'] + $fRadius; - $aPointPolygon['minlon'] = $aPointPolygon['minlon'] - $fRadius; - $aPointPolygon['maxlon'] = $aPointPolygon['maxlon'] + $fRadius; } // Output data suitable for display (points and a bounding box) - if ($bShowPolygons) + if ($bShowPolygons && isset($aPolyPoints)) { $aResult['aPolyPoints'] = array(); foreach($aPolyPoints as $aPoint) @@ -1226,6 +1401,16 @@ //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']) @@ -1247,11 +1432,8 @@ $aResult['foundorder'] = $iResNum; $aSearchResults[$iResNum] = $aResult; } - uasort($aSearchResults, 'byImportance'); -//var_dump($aSearchResults);exit; - $aOSMIDDone = array(); $aClassTypeNameDone = array(); $aToFilter = $aSearchResults; @@ -1270,15 +1452,15 @@ $bFirst = false; } if (!$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) >= $iLimit) break; + if (sizeof($aSearchResults) >= $iFinalLimit) break; } $sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '1 day'::interval,'YYYY/MM/DD') from import_status limit 1"); @@ -1293,7 +1475,7 @@ logEnd($oDB, $hLog, sizeof($aToFilter)); } $sMoreURL = CONST_Website_BaseURL.'search?format='.urlencode($sOutputFormat).'&exclude_place_ids='.join(',',$aExcludePlaceIDs); - $sMoreURL .= '&accept-language='.$_SERVER["HTTP_ACCEPT_LANGUAGE"]; + if (isset($_SERVER["HTTP_ACCEPT_LANGUAGE"])) $sMoreURL .= '&accept-language='.$_SERVER["HTTP_ACCEPT_LANGUAGE"]; if ($bShowPolygons) $sMoreURL .= '&polygon=1'; if ($bShowAddressDetails) $sMoreURL .= '&addressdetails=1'; if (isset($_GET['viewbox']) && $_GET['viewbox']) $sMoreURL .= '&viewbox='.urlencode($_GET['viewbox']);