X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/496a0f1384594aad954f7abcf124e8da86c4b3a1..b610118a88489c11249e9c0d30b252d3d98bd287:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index ecd5be1c..3c537777 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -51,10 +51,22 @@ class Geocode protected $sQuery = false; protected $aStructuredQuery = false; + protected $oNormalizer = null; + public function __construct(&$oDB) { $this->oDB =& $oDB; + $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules); + } + + private function normTerm($sTerm) + { + if ($this->oNormalizer === null) { + return $sTerm; + } + + return $this->oNormalizer->transliterate($sTerm); } public function setReverseInPlan($bReverse) @@ -292,7 +304,7 @@ class Geocode $aViewbox = $oParams->getStringList('viewboxlbrt'); if ($aViewbox) { if (count($aViewbox) != 4) { - userError("Bad parmater 'viewbox'. Expected 4 coordinates."); + userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates."); } $this->setViewbox($aViewbox); } else { @@ -301,12 +313,7 @@ class Geocode if (count($aViewbox) != 4) { userError("Bad parmater 'viewbox'. Expected 4 coordinates."); } - $this->setViewBox(array( - $aViewbox[0], - $aViewbox[3], - $aViewbox[2], - $aViewbox[1] - )); + $this->setViewBox($aViewbox); } else { $aRoute = $oParams->getStringList('route'); $fRouteWidth = $oParams->getFloat('routewidth'); @@ -360,7 +367,7 @@ class Geocode $this->aAddressRankList = array(); $this->aStructuredQuery = array(); - $this->sAllowedTypesSQLList = ''; + $this->sAllowedTypesSQLList = False; $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false); $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false); @@ -373,7 +380,7 @@ class Geocode if (sizeof($this->aStructuredQuery) > 0) { $this->sQuery = join(', ', $this->aStructuredQuery); if ($this->iMaxAddressRank < 30) { - $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; + $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')'; } } } @@ -410,8 +417,15 @@ class Geocode $sPlaceIDs = join(',', array_keys($aPlaceIDs)); $sImportanceSQL = ''; - if ($this->sViewboxSmallSQL) $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * "; - if ($this->sViewboxLargeSQL) $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * "; + $sImportanceSQLGeom = ''; + if ($this->sViewboxSmallSQL) { + $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * "; + $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, geometry) THEN 1 ELSE 0.75 END * "; + } + if ($this->sViewboxLargeSQL) { + $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * "; + $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, geometry) THEN 1 ELSE 0.75 END * "; + } $sSQL = "SELECT "; $sSQL .= " osm_type,"; @@ -476,6 +490,35 @@ class Geocode if ($this->bIncludeNameDetails) $sSQL .= "name, "; $sSQL .= " extratags->'place' "; + // postcode table + $sSQL .= "UNION "; + $sSQL .= "SELECT"; + $sSQL .= " 'P' as osm_type,"; + $sSQL .= " (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,"; + $sSQL .= " 'place' as class, 'postcode' as type,"; + $sSQL .= " null as admin_level, rank_search, rank_address,"; + $sSQL .= " place_id, parent_place_id, country_code,"; + $sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,"; + $sSQL .= " postcode as placename,"; + $sSQL .= " postcode as ref,"; + if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,"; + if ($this->bIncludeNameDetails) $sSQL .= "null AS names,"; + $sSQL .= " ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,"; + $sSQL .= $sImportanceSQLGeom."(0.75-(rank_search::float/40)) AS importance, "; + $sSQL .= " ("; + $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; + $sSQL .= " FROM "; + $sSQL .= " place_addressline s, "; + $sSQL .= " placex p"; + $sSQL .= " WHERE s.place_id = lp.parent_place_id"; + $sSQL .= " AND p.place_id = s.address_place_id "; + $sSQL .= " AND s.isaddress"; + $sSQL .= " AND p.importance is not null"; + $sSQL .= " ) AS addressimportance, "; + $sSQL .= " null AS extra_place "; + $sSQL .= "FROM location_postcode lp"; + $sSQL .= " WHERE place_id in ($sPlaceIDs) "; + if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) { // only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines // with start- and endnumber, the common osm housenumbers are usually saved as points @@ -653,7 +696,7 @@ class Geocode return $aSearchResults; } - public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases) + public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery) { /* Calculate all searches using aValidTokens i.e. @@ -666,12 +709,12 @@ class Geocode Score how good the search is so they can be ordered */ - foreach ($aPhrases as $iPhrase => $sPhrase) { + foreach ($aPhrases as $iPhrase => $aPhrase) { $aNewPhraseSearches = array(); if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase]; else $sPhraseType = ''; - foreach ($aPhrases[$iPhrase]['wordsets'] as $iWordSet => $aWordset) { + foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) { // Too many permutations - too expensive if ($iWordSet > 120) break; @@ -701,42 +744,29 @@ class Geocode } if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } - } elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) { - if ($aSearch['oNear'] === false) { - $aSearch['oNear'] = new NearPoint( - $aSearchTerm['lat'], - $aSearchTerm['lon'], - $aSearchTerm['radius'] - ); - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } elseif ($sPhraseType == 'postalcode') { + } elseif ($sPhraseType == 'postalcode' || ($aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode')) { // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both - if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { - // If we already have a name try putting the postcode first - if (sizeof($aSearch['aName'])) { + if ($aSearch['sPostcode'] === '' && $aSearch['sHouseNumber'] === '' && + isset($aSearchTerm['word']) && $aSearchTerm['word'] && strpos($sNormQuery, $this->normTerm($aSearchTerm['word'])) !== false) { + // If we have structured search or this is the first term, + // make the postcode the primary search element. + if ($aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) { $aNewSearch = $aSearch; + $aNewSearch['sOperator'] = 'postcode'; $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']); - $aNewSearch['aName'] = array(); - $aNewSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + $aNewSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word']; if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch; } - if (sizeof($aSearch['aName'])) { - if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($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'] = $iPhrase; + // If we have a structured search or this is not the first term, + // add the postcode as an addendum. + if ($aSearch['sOperator'] !== 'postcode' && ($sPhraseType == 'postalcode' || sizeof($aSearch['aName']))) { + $aSearch['sPostcode'] = $aSearchTerm['word']; + if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } } elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') { - if ($aSearch['sHouseNumber'] === '') { + if ($aSearch['sHouseNumber'] === '' && $aSearch['sOperator'] !== 'postcode') { $aSearch['sHouseNumber'] = $sToken; // sanity check: if the housenumber is not mainly made // up of numbers, add a penalty @@ -752,13 +782,19 @@ class Geocode */ } } elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) { - if ($aSearch['sClass'] === '') { - $aSearch['sOperator'] = $aSearchTerm['operator']; + // require a normalized exact match of the term + // if we have the normalizer version of the query + // available + if ($aSearch['sClass'] === '' + && ($sNormQuery === null || !($aSearchTerm['word'] && strpos($sNormQuery, $aSearchTerm['word']) === false))) { $aSearch['sClass'] = $aSearchTerm['class']; $aSearch['sType'] = $aSearchTerm['type']; - if (sizeof($aSearch['aName'])) $aSearch['sOperator'] = 'name'; - else $aSearch['sOperator'] = 'near'; // near = in for the moment - if (strlen($aSearchTerm['operator']) == 0) $aSearch['iSearchRank'] += 1; + if ($aSearchTerm['operator'] == '') { + $aSearch['sOperator'] = sizeof($aSearch['aName']) ? 'name' : 'near'; + $aSearch['iSearchRank'] += 2; + } else { + $aSearch['sOperator'] = 'near'; // near = in for the moment + } if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } @@ -874,6 +910,19 @@ class Geocode //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); } + + // Revisit searches, giving penalty to unlikely combinations + $aGroupedSearches = array(); + foreach ($aSearches as $aSearch) { + if (!$aSearch['aName']) { + if ($aSearch['sHouseNumber']) { + continue; + } + } + $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch; + } + ksort($aGroupedSearches); + return $aGroupedSearches; } @@ -913,6 +962,7 @@ class Geocode { if (!$this->sQuery && !$this->aStructuredQuery) return array(); + $sNormQuery = $this->normTerm($this->sQuery); $sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]"; $sCountryCodesSQL = false; if ($this->aCountryCodes) { @@ -972,21 +1022,31 @@ class Geocode 'sClass' => '', 'sType' => '', 'sHouseNumber' => '', + 'sPostcode' => '', 'oNear' => $oNearPoint ) ); // Any 'special' terms in the search? $bSpecialTerms = false; - preg_match_all('/\\[(.*)=(.*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER); - $aSpecialTerms = array(); + preg_match_all('/\\[([\\w_]*)=([\\w_]*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER); foreach ($aSpecialTermsRaw as $aSpecialTerm) { $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery); - $aSpecialTerms[strtolower($aSpecialTerm[1])] = $aSpecialTerm[2]; + if (!$bSpecialTerms) { + $aNewSearches = array(); + foreach ($aSearches as $aSearch) { + $aNewSearch = $aSearch; + $aNewSearch['sClass'] = $aSpecialTerm[1]; + $aNewSearch['sType'] = $aSpecialTerm[2]; + $aNewSearches[] = $aNewSearch; + } + + $aSearches = $aNewSearches; + $bSpecialTerms = true; + } } preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER); - $aSpecialTerms = array(); if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity']) { $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']); unset($this->aStructuredQuery['amenity']); @@ -994,6 +1054,10 @@ class Geocode foreach ($aSpecialTermsRaw as $aSpecialTerm) { $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery); + if ($bSpecialTerms) { + continue; + } + $sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".$aSpecialTerm[1]."') AS string")); $sSQL = 'SELECT * '; $sSQL .= 'FROM ( '; @@ -1001,25 +1065,17 @@ class Geocode $sSQL .= ' FROM word '; $sSQL .= ' WHERE word_token in (\' '.$sToken.'\')'; $sSQL .= ') AS x '; - $sSQL .= ' WHERE (class is not null AND class not in (\'place\')) '; - $sSQL .= ' OR country_code is not null'; + $sSQL .= ' WHERE (class is not null AND class not in (\'place\'))'; if (CONST_Debug) var_Dump($sSQL); $aSearchWords = chksql($this->oDB->getAll($sSQL)); $aNewSearches = array(); foreach ($aSearches as $aSearch) { foreach ($aSearchWords as $aSearchTerm) { $aNewSearch = $aSearch; - if ($aSearchTerm['country_code']) { - $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']); - $aNewSearches[] = $aNewSearch; - $bSpecialTerms = true; - } - if ($aSearchTerm['class']) { - $aNewSearch['sClass'] = $aSearchTerm['class']; - $aNewSearch['sType'] = $aSearchTerm['type']; - $aNewSearches[] = $aNewSearch; - $bSpecialTerms = true; - } + $aNewSearch['sClass'] = $aSearchTerm['class']; + $aNewSearch['sType'] = $aSearchTerm['type']; + $aNewSearches[] = $aNewSearch; + $bSpecialTerms = true; } } $aSearches = $aNewSearches; @@ -1068,14 +1124,10 @@ class Geocode if (CONST_Debug) var_Dump($sSQL); $aValidTokens = array(); - if (sizeof($aTokens)) { - $aDatabaseWords = chksql( - $this->oDB->getAll($sSQL), - "Could not get word tokens." - ); - } else { - $aDatabaseWords = array(); - } + $aDatabaseWords = chksql( + $this->oDB->getAll($sSQL), + "Could not get word tokens." + ); $aPossibleMainWordIDs = array(); $aWordFrequencyScores = array(); foreach ($aDatabaseWords as $aToken) { @@ -1096,21 +1148,9 @@ class Geocode } if (CONST_Debug) var_Dump($aPhrases, $aValidTokens); - // Try and calculate GB postcodes we might be missing + // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code foreach ($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)), $aData)) { - if (substr($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], $this->oDB); - if ($aGBPostcodeLocation) { - $aValidTokens[$sToken] = $aGBPostcodeLocation; - } - } elseif (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) { - // US ZIP+4 codes - if there is no token, - // merge in the 5-digit ZIP code + if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) { if (isset($aValidTokens[$aData[1]])) { foreach ($aValidTokens[$aData[1]] as $aToken) { if (!$aToken['class']) { @@ -1139,7 +1179,7 @@ class Geocode // array with: placeid => -1 | tiger-housenumber $aResultPlaceIDs = array(); - $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases); + $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery); if ($this->bReverseInPlan) { // Reverse phrase array and also reverse the order of the wordsets in @@ -1151,7 +1191,7 @@ class Geocode $aFinalPhrase = end($aPhrases); $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0); } - $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false); + $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery); foreach ($aGroupedSearches as $aSearches) { foreach ($aSearches as $aSearch) { @@ -1177,7 +1217,6 @@ class Geocode ksort($aGroupedSearches); } - if (CONST_Debug) var_Dump($aGroupedSearches); if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0) { $aCopyGroupedSearches = $aGroupedSearches; foreach ($aCopyGroupedSearches as $iGroup => $aSearches) { @@ -1234,13 +1273,16 @@ class Geocode if (CONST_Debug) echo "