X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/38e40cd01440bc6d98fd16bcf8e0598695957830..b4c87a09a194a7d966724ce48ba2d8ef0d044ca0:/website/search.php diff --git a/website/search.php b/website/search.php index 5e2ff371..387a79de 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')) @@ -31,12 +36,44 @@ } // 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']; - // Preferred language + // Preferred language $aLangPrefOrder = getPreferredLanguages(); if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true; if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true; @@ -89,7 +126,7 @@ } $sCountryCodesSQL = join(',', $aCountryCodes); } - + // Search query $sQuery = (isset($_GET['q'])?trim($_GET['q']):''); if (!$sQuery && isset($_SERVER['PATH_INFO']) && $_SERVER['PATH_INFO'][0] == '/') @@ -97,17 +134,57 @@ $sQuery = substr($_SERVER['PATH_INFO'], 1); // reverse order of '/' separated string - $aPhrases = explode('/', $sQuery); - $aPhrases = array_reverse($aPhrases); + $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', 5, 11), + ); + $aStructuredQuery = array(); + $sAllowedTypesSQLList = ''; + foreach($aStructuredOptions as $aStructuredOption) + { + loadStructuredAddressElement($aStructuredQuery, $iMinAddressRank, $iMaxAddressRank, $_GET, $aStructuredOption[0], $aStructuredOption[1], $aStructuredOption[2]); + } + if (sizeof($aStructuredQuery) > 0) + { + $sQuery = join(', ', $aStructuredQuery); + if ($iMaxAddressRank < 30) + { + $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; + } + } + if ($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 = 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); @@ -115,7 +192,7 @@ } // If we have a view box create the SQL - // Small is the actual view box, Large is double (on each axis) that + // Small is the actual view box, Large is double (on each axis) that $sViewboxCentreSQL = $sViewboxSmallSQL = $sViewboxLargeSQL = false; if (isset($_GET['viewboxlbrt']) && $_GET['viewboxlbrt']) { @@ -134,12 +211,16 @@ $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("; @@ -173,6 +254,7 @@ failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL); } $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry"; + $bBoundingBoxSearch = true; } // Do we have anything that looks like a lat/lon pair? @@ -210,12 +292,13 @@ } } - if ($sQuery) + if ($sQuery || $aStructuredQuery) { // Start with a blank search $aSearches = array( - array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), - 'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'') + array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), 'aFullNameAddress'=>array(), + 'aNameNonSearch'=>array(), 'aAddressNonSearch'=>array(), + 'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'') ); $sNearPointSQL = false; @@ -236,14 +319,19 @@ $aSpecialTerms[strtolower($aSpecialTerm[1])] = $aSpecialTerm[2]; } - preg_match_all('/\\[([a-zA-Z]*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER); + preg_match_all('/\\[([\\w ]*)\\]/u', $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 = 'select * from (select word_id,word_token, word, class, type, country_code, operator'; + $sSQL .= ' from word where word_token in (\' '.$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); $aNewSearches = array(); @@ -251,7 +339,7 @@ { foreach($aSearchWords as $aSearchTerm) { - $aNewSearch = $aSearch; + $aNewSearch = $aSearch; if ($aSearchTerm['country_code']) { $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']); @@ -272,19 +360,29 @@ // 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 // Get all 'sets' of words - // Generate a complete list of all + // Generate a complete list of all $aTokens = array(); foreach($aPhrases as $iPhrase => $sPhrase) { $aPhrase = $oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string"); if (PEAR::isError($aPhrase)) { - echo "Illegal query string (not an UTF-8 string): ".$sPhrase; + userError("Illegal query string (not an UTF-8 string): ".$sPhrase); if (CONST_Debug) var_dump($aPhrase); exit; } @@ -292,7 +390,7 @@ { $aPhrases[$iPhrase] = $aPhrase; $aPhrases[$iPhrase]['words'] = explode(' ',$aPhrases[$iPhrase]['string']); - $aPhrases[$iPhrase]['wordsets'] = getWordSets($aPhrases[$iPhrase]['words']); + $aPhrases[$iPhrase]['wordsets'] = getWordSets($aPhrases[$iPhrase]['words'], 0); $aTokens = array_merge($aTokens, getTokensFromSets($aPhrases[$iPhrase]['wordsets'])); } else @@ -302,137 +400,147 @@ } // reindex phrases - we make assumptions later on + $aPhraseTypes = array_keys($aPhrases); $aPhrases = array_values($aPhrases); if (sizeof($aTokens)) { - // 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 .= ' group by word_token, word, class, type, location, country_code'; + // Check which tokens we have, get the ID numbers + $sSQL = 'select word_id,word_token, word, class, type, country_code, operator, search_name_count'; + $sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$aTokens)).')'; + //$sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency; + //$sSQL .= ' group by word_token, word, class, type, country_code'; - if (CONST_Debug) var_Dump($sSQL); + if (CONST_Debug) var_Dump($sSQL); - $aValidTokens = array(); - if (sizeof($aTokens)) - $aDatabaseWords = $oDB->getAll($sSQL); - else - $aDatabaseWords = array(); - if (PEAR::IsError($aDatabaseWords)) - { - failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords); - } - $aPossibleMainWordIDs = array(); - foreach($aDatabaseWords as $aToken) - { - if (isset($aValidTokens[$aToken['word_token']])) + $aValidTokens = array(); + if (sizeof($aTokens)) $aDatabaseWords = $oDB->getAll($sSQL); + else $aDatabaseWords = array(); + if (PEAR::IsError($aDatabaseWords)) { - $aValidTokens[$aToken['word_token']][] = $aToken; + failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords); } - else + $aPossibleMainWordIDs = array(); + $aWordFrequencyScores = array(); + foreach($aDatabaseWords as $aToken) { - $aValidTokens[$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, $aValidTokens); + // Very special case - require 2 letter country param to match the country code found + if ($bStructuredPhrases && $aToken['country_code'] && !empty($aStructuredQuery['country']) + && strlen($aStructuredQuery['country']) == 2 && strtolower($aStructuredQuery['country']) != $aToken['country_code']) + { + continue; + } - $aSuggestion = array(); - $bSuggestion = false; - if (CONST_Suggestions_Enabled) - { - foreach($aPhrases as $iPhrase => $aPhrase) - { - if (!isset($aValidTokens[' '.$aPhrase['wordsets'][0][0]])) - { - $sQuotedPhrase = getDBQuoted(' '.$aPhrase['wordsets'][0][0]); - $aSuggestionWords = getWordSuggestions($oDB, $aPhrase['wordsets'][0][0]); - $aRow = $aSuggestionWords[0]; - if ($aRow && $aRow['word']) - { - $aSuggestion[] = $aRow['word']; - $bSuggestion = true; - } - else - { - $aSuggestion[] = $aPhrase['string']; - } - } - else - { - $aSuggestion[] = $aPhrase['string']; - } + if (isset($aValidTokens[$aToken['word_token']])) + { + $aValidTokens[$aToken['word_token']][] = $aToken; + } + else + { + $aValidTokens[$aToken['word_token']] = array($aToken); + } + if (!$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1; + $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1; } - } - if ($bSuggestion) $sSuggestion = join(', ',$aSuggestion); + if (CONST_Debug) var_Dump($aPhrases, $aValidTokens); - // Try and calculate GB postcodes we might be missing - foreach($aTokens as $sToken) - { - // Source of gb postcodes is now definitive - always use - if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) + $aSuggestion = array(); + $bSuggestion = false; + if (CONST_Suggestions_Enabled) { - if (substr($aData[1],-2,1) != ' ') + foreach($aPhrases as $iPhrase => $aPhrase) { - $aData[0] = substr($aData[0],0,strlen($aData[1]-1)).' '.substr($aData[0],strlen($aData[1]-1)); - $aData[1] = substr($aData[1],0,-1).' '.substr($aData[1],-1,1); + if (!isset($aValidTokens[' '.$aPhrase['wordsets'][0][0]])) + { + $sQuotedPhrase = getDBQuoted(' '.$aPhrase['wordsets'][0][0]); + $aSuggestionWords = getWordSuggestions($oDB, $aPhrase['wordsets'][0][0]); + $aRow = $aSuggestionWords[0]; + if ($aRow && $aRow['word']) + { + $aSuggestion[] = $aRow['word']; + $bSuggestion = true; + } + else + { + $aSuggestion[] = $aPhrase['string']; + } + } + else + { + $aSuggestion[] = $aPhrase['string']; + } } - $aGBPostcodeLocation = gbPostcodeCalculate($aData[0], $aData[1], $aData[2], $oDB); - if ($aGBPostcodeLocation) + } + if ($bSuggestion) $sSuggestion = join(', ',$aSuggestion); + + // Try and calculate GB postcodes we might be missing + foreach($aTokens as $sToken) + { + // Source of gb postcodes is now definitive - always use + if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) { - $aValidTokens[$sToken] = $aGBPostcodeLocation; + if (substr($aData[1],-2,1) != ' ') + { + $aData[0] = substr($aData[0],0,strlen($aData[1]-1)).' '.substr($aData[0],strlen($aData[1]-1)); + $aData[1] = substr($aData[1],0,-1).' '.substr($aData[1],-1,1); + } + $aGBPostcodeLocation = gbPostcodeCalculate($aData[0], $aData[1], $aData[2], $oDB); + if ($aGBPostcodeLocation) + { + $aValidTokens[$sToken] = $aGBPostcodeLocation; + } } } - } - foreach($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]/', $sToken)) + foreach($aTokens as $sToken) { - $aValidTokens[' '.$sToken] = array(array('class'=>'place','type'=>'house')); + // 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 + // Any words that have failed completely? + // TODO: suggestions + + // Start the search process + $aResultPlaceIDs = array(); + + /* + Calculate all searches using aValidTokens i.e. - // Start the search process - $aResultPlaceIDs = array(); + 'Wodsworth Road, Sheffield' => - /* - Calculate all searches using aValidTokens i.e. + Phrase Wordset + 0 0 (wodsworth road) + 0 1 (wodsworth)(road) + 1 0 (sheffield) - 'Wodsworth Road, Sheffield' => - - Phrase Wordset - 0 0 (wodsworth road) - 0 1 (wodsworth)(road) - 1 0 (sheffield) - - Score how good the search is so they can be ordered - */ + Score how good the search is so they can be ordered + */ foreach($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"; + //echo "
$sToken"; $aNewWordsetSearches = array(); - + foreach($aWordsetSearches as $aCurrentSearch) { -//echo ""; -//var_dump($aCurrentSearch); -//echo ""; + //echo ""; + //var_dump($aCurrentSearch); + //echo ""; // If the token is valid if (isset($aValidTokens[' '.$sToken])) @@ -441,13 +549,18 @@ { $aSearch = $aCurrentSearch; $aSearch['iSearchRank']++; - if ($aSearchTerm['country_code'] !== null && $aSearchTerm['country_code'] != '0') + if (($sPhraseType == '' || $sPhraseType == 'country') && !empty($aSearchTerm['country_code']) && $aSearchTerm['country_code'] != '0') { if ($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 (optimisation) - if ($iWordset+1 != sizeof($aPhrases[$iPhrase]['wordsets']) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5; + // If reverse order is enabled, it may appear at the beginning as well. + if (($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) && + (!$bReverseInPlan || $iToken > 0 || $iPhrase > 0)) + { + $aSearch['iSearchRank'] += 5; + } if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } @@ -461,21 +574,21 @@ if ($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'] === '') { $aSearch['sHouseNumber'] = $sToken; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; -/* + /* // Fall back to not searching for this item (better than nothing) $aSearch = $aCurrentSearch; $aSearch['iSearchRank'] += 1; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; -*/ + */ } } - elseif ($aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) + elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) { if ($aSearch['sClass'] === '') { @@ -504,19 +617,20 @@ { if (sizeof($aSearch['aName'])) { - if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false) + if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false)) { $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; } else { + $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; $aSearch['iSearchRank'] += 1000; // skip; } } else { $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; -// $aSearch['iNamePhrase'] = $iPhrase; + //$aSearch['iNamePhrase'] = $iPhrase; } if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } @@ -529,25 +643,49 @@ { if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { -//var_Dump('
',$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[] = $aSearch; - } + if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4) + { + $aSearch = $aCurrentSearch; + $aSearch['iSearchRank'] += 1; + if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) + { + $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; + } + elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version? + { + foreach($aValidTokens[' '.$sToken] as $aSearchTermToken) + { + if (empty($aSearchTermToken['country_code']) + && empty($aSearchTermToken['lat']) + && empty($aSearchTermToken['class'])) + { + $aSearch = $aCurrentSearch; + $aSearch['iSearchRank'] += 1; + $aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id']; + if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; + } + } + } + else + { + $aSearch['aAddressNonSearch'][$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)) $aSearch['iSearchRank'] += 2; + if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) + $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + else + $aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + $aSearch['iNamePhrase'] = $iPhrase; + if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; + } } } } @@ -562,25 +700,19 @@ // Sort and cut usort($aNewWordsetSearches, 'bySearchRank'); $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50); - } -// var_Dump('
',sizeof($aWordsetSearches)); exit; + } + //var_Dump('
',sizeof($aWordsetSearches)); exit; $aNewPhraseSearches = array_merge($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; - } - } + $aSearchHash = array(); + foreach($aNewPhraseSearches as $iSearch => $aSearch) + { + $sHash = serialize($aSearch); + if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]); + else $aSearchHash[$sHash] = 1; + } $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50); } @@ -606,119 +738,164 @@ if ($iSearchCount > 50) break; } -// if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); + //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); } } else { - // Re-group the searches by their score, junk anything over 20 as just not worth trying - $aGroupedSearches = array(); - foreach($aSearches as $aSearch) + // Re-group the searches by their score, junk anything over 20 as just not worth trying + $aGroupedSearches = array(); + foreach($aSearches as $aSearch) + { + if ($aSearch['iSearchRank'] < $iMaxRank) { - if ($aSearch['iSearchRank'] < $iMaxRank) + if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array(); + $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch; + } + } + ksort($aGroupedSearches); + } + + if (CONST_Debug) var_Dump($aGroupedSearches); + + if ($bReverseInPlan) + { + $aCopyGroupedSearches = $aGroupedSearches; + foreach($aCopyGroupedSearches as $iGroup => $aSearches) + { + foreach($aSearches as $iSearch => $aSearch) + { + if (sizeof($aSearch['aAddress'])) { - if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array(); - $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch; + $iReverseItem = array_pop($aSearch['aAddress']); + 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][] = $aReverseSearch; } } - ksort($aGroupedSearches); + } } - - if (CONST_Debug) var_Dump($aGroupedSearches); - if ($bReverseInPlan) + if (CONST_Search_TryDroppedAddressTerms && sizeof($aStructuredQuery) > 0) + { + $aCopyGroupedSearches = $aGroupedSearches; + foreach($aCopyGroupedSearches as $iGroup => $aSearches) { - $aCopyGroupedSearches = $aGroupedSearches; - foreach($aCopyGroupedSearches as $iGroup => $aSearches) + foreach($aSearches as $iSearch => $aSearch) { - foreach($aSearches as $iSearch => $aSearch) + $aReductionsList = array($aSearch['aAddress']); + $iSearchRank = $aSearch['iSearchRank']; + while(sizeof($aReductionsList) > 0) { - if (sizeof($aSearch['aAddress'])) + $iSearchRank += 5; + if ($iSearchRank > iMaxRank) break 3; + $aNewReductionsList = array(); + foreach($aReductionsList as $aReductionsWordList) { - $iReverseItem = array_pop($aSearch['aAddress']); - if (isset($aPossibleMainWordIDs[$iReverseItem])) + for ($iReductionWord = 0; $iReductionWord < sizeof($aReductionsWordList); $iReductionWord++) { - $aSearch['aAddress'] = array_merge($aSearch['aAddress'], $aSearch['aName']); - $aSearch['aName'] = array($iReverseItem); - $aGroupedSearches[$iGroup][] = $aSearch; + $aReductionsWordListResult = array_merge(array_slice($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; + } } -// $aReverseSearch['aName'][$iReverseItem] = $iReverseItem; - // $aGroupedSearches[$iGroup][] = $aReverseSearch; } + $aReductionsList = $aNewReductionsList; } } } + ksort($aGroupedSearches); + } - // Filter out duplicate searches - $aSearchHash = array(); - foreach($aGroupedSearches as $iGroup => $aSearches) + // Filter out duplicate searches + $aSearchHash = array(); + foreach($aGroupedSearches as $iGroup => $aSearches) + { + foreach($aSearches as $iSearch => $aSearch) { - foreach($aSearches as $iSearch => $aSearch) + $sHash = serialize($aSearch); + if (isset($aSearchHash[$sHash])) { - $sHash = serialize($aSearch); - if (isset($aSearchHash[$sHash])) - { - unset($aGroupedSearches[$iGroup][$iSearch]); - if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]); - } - else - { - $aSearchHash[$sHash] = 1; - } + unset($aGroupedSearches[$iGroup][$iSearch]); + if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]); + } + else + { + $aSearchHash[$sHash] = 1; } } + } - if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); + if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); - $iGroupLoop = 0; - $iQueryLoop = 0; - foreach($aGroupedSearches as $iGroupedRank => $aSearches) + $iGroupLoop = 0; + $iQueryLoop = 0; + foreach($aGroupedSearches as $iGroupedRank => $aSearches) + { + $iGroupLoop++; + foreach($aSearches as $aSearch) { - $iGroupLoop++; - foreach($aSearches as $aSearch) - { - $iQueryLoop++; + $iQueryLoop++; + + if (CONST_Debug) { echo "
Search Loop, group $iGroupLoop, loop $iQueryLoop"; } + if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens); - // Must have a location term - if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon']) + + // Must have a location term + if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon']) + { + if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) { - if ($aSearch['sCountryCode'] && !$aSearch['sClass']) + if (4 >= $iMinAddressRank && 4 <= $iMaxAddressRank) { - 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); - } + $sSQL = "select place_id from placex where calculated_country_code='".$aSearch['sCountryCode']."' and rank_search = 4"; + if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; + $sSQL .= " order by st_area(geometry) desc limit 1"; + if (CONST_Debug) var_dump($sSQL); + $aPlaceIDs = $oDB->getCol($sSQL); } - else + } + else + { + if (!$bBoundingBoxSearch && !$aSearch['fLon']) continue; + if (!$aSearch['sClass']) continue; + $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'"; + if ($oDB->getOne($sSQL)) { - if (!$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']; + $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct"; if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)"; - $sSQL .= " where st_contains($sViewboxSmallSQL, centroid)"; - if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; - if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; + $sSQL .= " where st_contains($sViewboxSmallSQL, ct.centroid)"; + if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; + if (sizeof($aExcludePlaceIDs)) + { + $sSQL .= " and place_id not in (".join(',',$aExcludePlaceIDs).")"; + } + 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($aPlaceIDs)) + // If excluded place IDs are given, it is fair to assume that + // there have been results in the small box, so no further + // expansion in that case. + if (!sizeof($aPlaceIDs) && !sizeof($aExcludePlaceIDs)) { - $sSQL = "select place_id from place_classtype_".$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)"; - if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; - if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc"; + $sSQL .= " where st_contains($sViewboxLargeSQL, ct.centroid)"; + if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; + if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc"; $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $oDB->getCol($sSQL); @@ -727,205 +904,221 @@ else { $sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'"; - $sSQL .= " and st_contains($sViewboxSmallSQL, centroid)"; - if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)"; + $sSQL .= " and st_contains($sViewboxSmallSQL, geometry) and linked_place_id is null"; + if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($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 + } + else + { + $aPlaceIDs = array(); + + // 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['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]"; + if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) { - if (CONST_Debug) var_dump('
',$aSearch); - if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens); - $aPlaceIDs = array(); - - // 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'])."'"; - if ($aSearch['sHouseNumber']) $aTerms[] = "address_rank in (26,27)"; - if ($aSearch['fLon'] && $aSearch['fLat']) + // For infrequent name terms disable index usage for address + if (CONST_Search_NameOnlySearchFrequencyThreshold && + sizeof($aSearch['aName']) == 1 && + $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold) { - $aTerms[] = "ST_DWithin(centroid, ST_SetSRID(ST_Point(".$aSearch['fLon'].",".$aSearch['fLat']."),4326), ".$aSearch['fRadius'].")"; - $aOrder[] = "ST_Distance(centroid, ST_SetSRID(ST_Point(".$aSearch['fLon'].",".$aSearch['fLat']."),4326)) ASC"; + $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]"; } - if (sizeof($aExcludePlaceIDs)) + else { - $aTerms[] = "place_id not in (".join(',',$aExcludePlaceIDs).")"; + $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]"; + if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]"; } - if ($sCountryCodesSQL) + } + if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'"; + if ($aSearch['sHouseNumber']) $aTerms[] = "address_rank in (26,27)"; + if ($aSearch['fLon'] && $aSearch['fLat']) + { + $aTerms[] = "ST_DWithin(centroid, ST_SetSRID(ST_Point(".$aSearch['fLon'].",".$aSearch['fLat']."),4326), ".$aSearch['fRadius'].")"; + $aOrder[] = "ST_Distance(centroid, ST_SetSRID(ST_Point(".$aSearch['fLon'].",".$aSearch['fLat']."),4326)) ASC"; + } + if (sizeof($aExcludePlaceIDs)) + { + $aTerms[] = "place_id not in (".join(',',$aExcludePlaceIDs).")"; + } + if ($sCountryCodesSQL) + { + $aTerms[] = "country_code in ($sCountryCodesSQL)"; + } + + if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL"; + if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc"; + + $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)'; + if ($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"; + $aOrder[] = "$sImportanceSQL DESC"; + if (sizeof($aSearch['aFullNameAddress'])) + $aOrder[] = '(select count(*) from (select unnest(ARRAY['.join($aSearch['aFullNameAddress'],",").']) INTERSECT select unnest(nameaddress_vector))s) DESC'; + + if (sizeof($aTerms)) + { + $sSQL = "select place_id"; + $sSQL .= " from search_name"; + $sSQL .= " where ".join(' and ',$aTerms); + $sSQL .= " order by ".join(', ',$aOrder); + if ($aSearch['sHouseNumber'] || $aSearch['sClass']) + $sSQL .= " limit 50"; + elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass']) + $sSQL .= " limit 1"; + else + $sSQL .= " limit ".$iLimit; + + if (CONST_Debug) { var_dump($sSQL); } + $aViewBoxPlaceIDs = $oDB->getAll($sSQL); + if (PEAR::IsError($aViewBoxPlaceIDs)) { - $aTerms[] = "country_code in ($sCountryCodesSQL)"; + 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; + $aPlaceIDs[] = $aViewBoxRow['place_id']; + } + } + //var_Dump($aPlaceIDs); + //exit; - if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL"; - if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc"; - - $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.92-(search_rank::float/33) else importance end'; + if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs)) + { + $aRoadPlaceIDs = $aPlaceIDs; + $sPlaceIDs = join(',',$aPlaceIDs); - 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)) + // 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)) { - $sSQL = "select place_id"; - $sSQL .= " from search_name"; - $sSQL .= " where ".join(' and ',$aTerms); - $sSQL .= " order by ".join(', ',$aOrder); - if ($aSearch['sHouseNumber'] || $aSearch['sClass']) - $sSQL .= " limit 50"; - elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass']) - $sSQL .= " limit 1"; - else - $sSQL .= " limit ".$iLimit; + $sSQL .= " and place_id not in (".join(',',$aExcludePlaceIDs).")"; + } + $sSQL .= " limit $iLimit"; + if (CONST_Debug) var_dump($sSQL); + $aPlaceIDs = $oDB->getCol($sSQL); - if (CONST_Debug) var_dump($sSQL); - $aViewBoxPlaceIDs = $oDB->getAll($sSQL); - if (PEAR::IsError($aViewBoxPlaceIDs)) - { - 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 $aViewBoxRow) + // If not try the aux fallback table + if (!sizeof($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)) { -// 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']; + $sSQL .= " and place_id not in (".join(',',$aExcludePlaceIDs).")"; } + //$sSQL .= " limit $iLimit"; + if (CONST_Debug) var_dump($sSQL); + $aPlaceIDs = $oDB->getCol($sSQL); } -//var_Dump($aPlaceIDs); -//exit; - if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs)) + if (!sizeof($aPlaceIDs)) { - $aRoadPlaceIDs = $aPlaceIDs; - $sPlaceIDs = join(',',$aPlaceIDs); - - // 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."'"; + $sSQL = "select place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'"; if (sizeof($aExcludePlaceIDs)) { $sSQL .= " and place_id not in (".join(',',$aExcludePlaceIDs).")"; } - $sSQL .= " limit $iLimit"; + //$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); - } + // Fallback to the road + if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) + { + $aPlaceIDs = $aRoadPlaceIDs; + } - if (!sizeof($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)) + { + $sPlaceIDs = join(',',$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) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; + $sSQL .= " order by rank_search asc limit $iLimit"; + if (CONST_Debug) var_dump($sSQL); + $aClassPlaceIDs = $oDB->getCol($sSQL); } - - if ($aSearch['sClass'] && sizeof($aPlaceIDs)) + + if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') // & in { - $sPlaceIDs = join(',',$aPlaceIDs); + $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'"; + $bCacheTable = $oDB->getOne($sSQL); - 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']."'"; - 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); - } - - if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') // & in - { - $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'"; - $bCacheTable = $oDB->getOne($sSQL); + $sSQL = "select min(rank_search) from placex where place_id in ($sPlaceIDs)"; - $sSQL = "select min(rank_search) from placex where place_id in ($sPlaceIDs)"; + if (CONST_Debug) var_dump($sSQL); + $iMaxRank = ((int)$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); - $iMaxRank = ((int)$oDB->getOne($sSQL)); + $sPlaceGeom = $oDB->getOne($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 - { - $iMaxRank += 5; + if ($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 || $sPlaceGeom) - { + if ($sPlaceIDs || $sPlaceGeom) + { $fRange = 0.01; if ($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)"; - - $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from place_classtype_".$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) { $sSQL .= ",placex as f where "; - $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, st_centroid(f.geometry), $fRange) "; + $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, f.centroid, $fRange) "; } if ($sPlaceGeom) { @@ -936,12 +1129,12 @@ { $sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")"; } - if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)"; - if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc"; + if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)"; + 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 { @@ -952,187 +1145,207 @@ 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 .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, f.centroid, $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 ($sCountryCodesSQL) $sSQL .= " and l.country_code in ($sCountryCodesSQL)"; + if ($sCountryCodesSQL) $sSQL .= " and l.calculated_country_code in ($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); - } + $aClassPlaceIDs = array_merge($aClassPlaceIDs, $oDB->getCol($sSQL)); } } } - - } - if (PEAR::IsError($aPlaceIDs)) - { - failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs); - } - - if (CONST_Debug) var_Dump($aPlaceIDs); + $aPlaceIDs = $aClassPlaceIDs; - foreach($aPlaceIDs as $iPlaceID) - { - $aResultPlaceIDs[$iPlaceID] = $iPlaceID; } - if ($iQueryLoop > 20) break; + } - //exit; - if (sizeof($aResultPlaceIDs)) break; - if ($iGroupLoop > 4) break; - if ($iQueryLoop > 30) break; - } -//exit; - // Did we find anything? - if (sizeof($aResultPlaceIDs)) - { -//var_Dump($aResultPlaceIDs);exit; - // Get the details for display (is this a redundant extra step?) - $sPlaceIDs = join(',',$aResultPlaceIDs); - $sOrderSQL = 'CASE '; - foreach(array_keys($aResultPlaceIDs) as $iOrder => $iPlaceID) + + if (PEAR::IsError($aPlaceIDs)) { - $sOrderSQL .= 'when min(place_id) = '.$iPlaceID.' then '.$iOrder.' '; + failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs); } - $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 .= "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,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($aSearchResults)) + + if (CONST_Debug) { echo "
Place IDs: "; var_Dump($aPlaceIDs); } + + foreach($aPlaceIDs as $iPlaceID) { - failInternalError("Could not get details for place.", $sSQL, $aSearchResults); + $aResultPlaceIDs[$iPlaceID] = $iPlaceID; } + if ($iQueryLoop > 20) break; } - } // end if ($sQuery) - else + + //exit; + if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) break; + if ($iGroupLoop > 4) break; + if ($iQueryLoop > 30) break; + } + + // Did we find anything? + if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) { - if (isset($_GET['nearlat']) && trim($_GET['nearlat'])!=='' && isset($_GET['nearlon']) && trim($_GET['nearlon']) !== '') + //var_Dump($aResultPlaceIDs);exit; + // Get the details for display (is this a redundant extra step?) + $sPlaceIDs = join(',',$aResultPlaceIDs); + $sOrderSQL = 'CASE '; + foreach(array_keys($aResultPlaceIDs) as $iOrder => $iPlaceID) { - $iPlaceID = geocodeReverse($_GET['nearlat'], $_GET['nearlon']); - $aResultPlaceIDs = array($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,calculated_country_code as 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,"; + $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; + //$sSQL .= $sOrderSQL." as porder, "; + $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance, "; + $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(placex.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $sSQL .= "(extratags->'place') as extra_place "; + $sSQL .= "from placex where place_id in ($sPlaceIDs) "; + $sSQL .= "and (placex.rank_address between $iMinAddressRank and $iMaxAddressRank "; + if (14 >= $iMinAddressRank && 14 <= $iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'"; + $sSQL .= ") "; + if ($sAllowedTypesSQLList) $sSQL .= "and placex.class in $sAllowedTypesSQLList "; + $sSQL .= "and linked_place_id is null "; + $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance"; + if (!$bDeDupe) $sSQL .= ",place_id"; + $sSQL .= ",langaddress "; + $sSQL .= ",placename "; + $sSQL .= ",ref "; + $sSQL .= ",extratags->'place' "; + $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 .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_tiger.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $sSQL .= "null as extra_place "; + $sSQL .= "from location_property_tiger where place_id in ($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 .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $sSQL .= "null as extra_place "; + $sSQL .= "from location_property_aux where place_id in ($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) { echo "
"; var_dump($sSQL); } + $aSearchResults = $oDB->getAll($sSQL); + //var_dump($sSQL,$aSearchResults);exit; + + if (PEAR::IsError($aSearchResults)) + { + 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! + // 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); - } + // 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,calculated_country_code as 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,"; + $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; + //$sSQL .= $sOrderSQL." as porder, "; + $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance, "; + $sSQL .= "(extratags->'place') as extra_place "; + $sSQL .= "from placex where place_id in ($sPlaceIDs) "; + $sSQL .= "and (placex.rank_address between $iMinAddressRank and $iMaxAddressRank "; + if (14 >= $iMinAddressRank && 14 <= $iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'"; + $sSQL .= ") "; + $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance"; + if (!$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 .= ",extratags->'place' "; + $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 .= "null as extra_place "; + $sSQL .= "from location_property_tiger where place_id in ($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 .= "null as extra_place "; + $sSQL .= "from location_property_aux where place_id in ($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) { echo "
", 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); } } } - + } + $sSearchResult = ''; if (!sizeof($aSearchResults) && isset($_GET['q']) && $_GET['q']) { $sSearchResult = 'No Results Found'; } -//var_Dump($aSearchResults); -//exit; + //var_Dump($aSearchResults); + //exit; $aClassType = getClassTypesWithImportance(); - $aRecheckWords = preg_split('/\b/',$sQuery); + $aRecheckWords = preg_split('/\b/u',$sQuery); foreach($aRecheckWords as $i => $sWord) { if (!$sWord) unset($aRecheckWords[$i]); @@ -1150,8 +1363,12 @@ $sSQL = "select place_id,0 as numfeatures,st_area(geometry) as area,"; $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\''; + $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon"; + if ($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($aPointPolygon)) { @@ -1159,29 +1376,42 @@ } if ($aPointPolygon['place_id']) { - if ($aPointPolygon['centrelon'] !== null && $aPointPolygon['centrelat'] !== null ) { + if ($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'] = $aPointPolygon['centrelon']; } - // Translate geometary string to point array - if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['outlinestring'],$aMatch)) - { - preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER); - } - elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$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)) { - $aPolyPoints[] = array('',$aMatch[1]+($fRadius*sin($f)),$aMatch[2]+($fRadius*cos($f))); + preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER); + } + elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch)) + { + preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$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'] + $fRadius; } // Output data suitable for display (points and a bounding box) @@ -1197,17 +1427,24 @@ } } + if ($aResult['extra_place'] == 'city') + { + $aResult['class'] = 'place'; + $aResult['type'] = 'city'; + $aResult['rank_search'] = 16; + } + if (!isset($aResult['aBoundingBox'])) { // Default $fDiameter = 0.0001; - if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['defdiameter']) + if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['defdiameter']) && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['defdiameter']) { $fDiameter = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['defzoom']; } - elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['defdiameter']) + elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['defdiameter']) && $aClassType[$aResult['class'].':'.$aResult['type']]['defdiameter']) { $fDiameter = $aClassType[$aResult['class'].':'.$aResult['type']]['defdiameter']; @@ -1239,14 +1476,14 @@ } // Is there an icon set for this type of result? - if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon']) - && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']) + if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon']) + && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']) { $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png'; } - if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label']) - && $aClassType[$aResult['class'].':'.$aResult['type']]['label']) + if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label']) + && $aClassType[$aResult['class'].':'.$aResult['type']]['label']) { $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label']; } @@ -1254,8 +1491,13 @@ if ($bShowAddressDetails) { $aResult['address'] = getAddressDetails($oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code']); -//var_dump($aResult['address']); -//exit; + if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) + { + $aResult['address'] = array_merge(array('city' => array_shift(array_values($aResult['address']))), $aResult['address']); + } + + //var_dump($aResult['address']); + //exit; } // Adjust importance for the number of exact string matches in the result @@ -1266,33 +1508,32 @@ { 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']; - } - else - { - $aResult['importance'] = 1000000000000000; - } -*/ + $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']) + && $aClassType[$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']; + } + else + { + $aResult['importance'] = 1000000000000000; + } + */ $aResult['name'] = $aResult['langaddress']; - $aResult['foundorder'] = $iResNum; + $aResult['foundorder'] = -$aResult['addressimportance']; $aSearchResults[$iResNum] = $aResult; } uasort($aSearchResults, 'byImportance'); -//var_dump($aSearchResults);exit; - $aOSMIDDone = array(); $aClassTypeNameDone = array(); $aToFilter = $aSearchResults; @@ -1311,18 +1552,18 @@ $bFirst = false; } if (!$bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']]) - && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']]))) + && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))) { $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true; - $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']] = true; + $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = 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"); + $sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '2 minutes'::interval,'YYYY/MM/DD HH24:MI')||' GMT' from import_status limit 1"); if (isset($_GET['nearlat']) && isset($_GET['nearlon'])) { @@ -1333,8 +1574,8 @@ { logEnd($oDB, $hLog, sizeof($aToFilter)); } - $sMoreURL = CONST_Website_BaseURL.'search?format='.urlencode($sOutputFormat).'&exclude_place_ids='.join(',',$aExcludePlaceIDs); - $sMoreURL .= '&accept-language='.$_SERVER["HTTP_ACCEPT_LANGUAGE"]; + $sMoreURL = CONST_Website_BaseURL.'search?format='.urlencode($sOutputFormat).'&exclude_place_ids='.join(',',$aExcludePlaceIDs); + if (isset($_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']);