X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/1787892d32351988231a60b71900c6a623609c54..64ace51e02790bf88e162a15236c5ac2af865c5e:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index 9daaff84..0c93e0a9 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -7,6 +7,7 @@ require_once(CONST_BasePath.'/lib/Phrase.php'); require_once(CONST_BasePath.'/lib/ReverseGeocode.php'); require_once(CONST_BasePath.'/lib/SearchDescription.php'); require_once(CONST_BasePath.'/lib/SearchContext.php'); +require_once(CONST_BasePath.'/lib/TokenList.php'); class Geocode { @@ -16,8 +17,6 @@ class Geocode protected $aLangPrefOrder = array(); - protected $bIncludeAddressDetails = false; - protected $aExcludePlaceIDs = array(); protected $bReverseInPlan = false; @@ -86,7 +85,6 @@ class Geocode $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs); } - if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1'; if ($this->bBoundedSearch) $aParams['bounded'] = '1'; if ($this->aCountryCodes) { @@ -151,6 +149,10 @@ class Geocode private function viewboxImportanceFactor($fX, $fY) { + if (!$this->aViewBox) { + return 1; + } + $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2; $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2; @@ -182,9 +184,6 @@ class Geocode public function loadParamArray($oParams, $sForceGeometryType = null) { - $this->bIncludeAddressDetails - = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails); - $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch); $this->setLimit($oParams->getInt('limit', $this->iFinalLimit)); @@ -246,8 +245,7 @@ class Geocode } $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType); - $this->oPlaceLookup->setIncludeAddressDetails(false); - $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon')); + $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false)); } public function setQueryFromParams($oParams) @@ -332,10 +330,10 @@ class Geocode return false; } - public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured) + public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured) { /* - Calculate all searches using aValidTokens i.e. + Calculate all searches using oValidTokens i.e. 'Wodsworth Road, Sheffield' => Phrase Wordset @@ -349,10 +347,7 @@ class Geocode $aNewPhraseSearches = array(); $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : ''; - foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) { - // Too many permutations - too expensive - if ($iWordSet > 120) break; - + foreach ($oPhrase->getWordSets() as $aWordset) { $aWordsetSearches = $aSearches; // Add all words from this wordset @@ -365,38 +360,37 @@ class Geocode //var_dump($oCurrentSearch); //echo ""; - // If the token is valid - if (isset($aValidTokens[' '.$sToken])) { - foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) { - $aNewSearches = $oCurrentSearch->extendWithFullTerm( - $aSearchTerm, - isset($aValidTokens[$sToken]) - && strpos($sToken, ' ') === false, - $sPhraseType, - $iToken == 0 && $iPhrase == 0, - $iPhrase == 0, - $iToken + 1 == count($aWordset) - && $iPhrase + 1 == count($aPhrases) - ); - - foreach ($aNewSearches as $oSearch) { - if ($oSearch->getRank() < $this->iMaxRank) { - $aNewWordsetSearches[] = $oSearch; - } + // Tokens with full name matches. + foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) { + $aNewSearches = $oCurrentSearch->extendWithFullTerm( + $oSearchTerm, + $oValidTokens->contains($sToken) + && strpos($sToken, ' ') === false, + $sPhraseType, + $iToken == 0 && $iPhrase == 0, + $iPhrase == 0, + $iToken + 1 == count($aWordset) + && $iPhrase + 1 == count($aPhrases) + ); + + foreach ($aNewSearches as $oSearch) { + if ($oSearch->getRank() < $this->iMaxRank) { + $aNewWordsetSearches[] = $oSearch; } } } // Look for partial matches. // Note that there is no point in adding country terms here // because country is omitted in the address. - if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') { + if ($sPhraseType != 'country') { // Allow searching for a word - but at extra cost - foreach ($aValidTokens[$sToken] as $aSearchTerm) { + foreach ($oValidTokens->get($sToken) as $oSearchTerm) { $aNewSearches = $oCurrentSearch->extendWithPartialTerm( - $aSearchTerm, + $sToken, + $oSearchTerm, $bIsStructured, $iPhrase, - isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array() + $oValidTokens->get(' '.$sToken) ); foreach ($aNewSearches as $oSearch) { @@ -446,8 +440,6 @@ class Geocode $aSearches = array_merge($aSearches, $aNewSearches); if ($iSearchCount > 50) break; } - - //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); } // Revisit searches, drop bad searches and give penalty to unlikely combinations. @@ -502,8 +494,11 @@ class Geocode public function lookup() { + Debug::newFunction('Geocode::lookup'); if (!$this->sQuery && !$this->aStructuredQuery) return array(); + Debug::printDebugArray('Geocode', $this); + $oCtx = new SearchContext(); if ($this->aRoutePoints) { @@ -523,9 +518,13 @@ class Geocode $oCtx->setCountryList($this->aCountryCodes); } + Debug::newSection('Query Preprocessing'); + $sNormQuery = $this->normTerm($this->sQuery); - $sLanguagePrefArraySQL = getArraySQL( - array_map('getDBQuoted', $this->aLangPrefOrder) + Debug::printVar('Normalized query', $sNormQuery); + + $sLanguagePrefArraySQL = $this->oDB->getArraySQL( + $this->oDB->getDBQuotedList($this->aLangPrefOrder) ); $sQuery = $this->sQuery; @@ -543,7 +542,6 @@ class Geocode // Do we have anything that looks like a lat/lon pair? $sQuery = $oCtx->setNearPointFromQuery($sQuery); - $aResults = array(); if ($sQuery || $this->aStructuredQuery) { // Start with a single blank search $aSearches = array(new SearchDescription($oCtx)); @@ -560,6 +558,10 @@ class Geocode $aSpecialTermsRaw, PREG_SET_ORDER ); + if (!empty($aSpecialTermsRaw)) { + Debug::printVar('Special terms', $aSpecialTermsRaw); + } + foreach ($aSpecialTermsRaw as $aSpecialTerm) { $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery); if (!$sSpecialTerm) { @@ -575,15 +577,17 @@ class Geocode if ($sSpecialTerm && !$aSearches[0]->hasOperator()) { $sSpecialTerm = pg_escape_string($sSpecialTerm); - $sToken = chksql( - $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"), + $sToken = $this->oDB->getOne( + 'SELECT make_standard_name(:term)', + array(':term' => $sSpecialTerm), 'Cannot decode query. Wrong encoding?' ); $sSQL = 'SELECT class, type FROM word '; $sSQL .= ' WHERE word_token in (\' '.$sToken.'\')'; $sSQL .= ' AND class is not null AND class not in (\'place\')'; - if (CONST_Debug) var_Dump($sSQL); - $aSearchWords = chksql($this->oDB->getAll($sSQL)); + + Debug::printSQL($sSQL); + $aSearchWords = $this->oDB->getAll($sSQL); $aNewSearches = array(); foreach ($aSearches as $oSearch) { foreach ($aSearchWords as $aSearchTerm) { @@ -609,15 +613,21 @@ class Geocode $bStructuredPhrases = false; } + Debug::printDebugArray('Search context', $oCtx); + Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]); + Debug::printVar('Final query phrases', $aInPhrases); + // Convert each phrase to standard form // Create a list of standard words // Get all 'sets' of words // Generate a complete list of all + Debug::newSection('Tokenization'); $aTokens = array(); $aPhrases = array(); foreach ($aInPhrases as $iPhrase => $sPhrase) { - $sPhrase = chksql( - $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'), + $sPhrase = $this->oDB->getOne( + 'SELECT make_standard_name(:phrase)', + array(':phrase' => $sPhrase), 'Cannot normalize query string (is it a UTF-8 string?)' ); if (trim($sPhrase)) { @@ -627,74 +637,58 @@ class Geocode } } + Debug::printVar('Tokens', $aTokens); + + $oValidTokens = new TokenList(); + if (!empty($aTokens)) { - // 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 '; - $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')'; + $sSQL .= ' WHERE word_token in ('.join(',', $this->oDB->getDBQuotedList($aTokens)).')'; - if (CONST_Debug) var_Dump($sSQL); + Debug::printSQL($sSQL); - $aValidTokens = array(); - $aDatabaseWords = chksql( - $this->oDB->getAll($sSQL), - 'Could not get word tokens.' + $oValidTokens->addTokensFromDB( + $this->oDB, + $aTokens, + $this->aCountryCodes, + $sNormQuery, + $this->oNormalizer ); - $aWordFrequencyScores = array(); - foreach ($aDatabaseWords as $aToken) { - // Filter country tokens that do not match restricted countries. - if ($this->aCountryCodes - && $aToken['country_code'] - && !in_array($aToken['country_code'], $this->aCountryCodes) - ) { - continue; - } - - // Special terms need to appear in their normalized form. - if ($aToken['word'] && $aToken['class']) { - $sNormWord = $this->normTerm($aToken['word']); - if (strpos($sNormQuery, $sNormWord) === false) { - continue; - } - } - - if (isset($aValidTokens[$aToken['word_token']])) { - $aValidTokens[$aToken['word_token']][] = $aToken; - } else { - $aValidTokens[$aToken['word_token']] = array($aToken); - } - $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1; - } - if (CONST_Debug) var_Dump($aPhrases, $aValidTokens); - // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code + // Try more interpretations for Tokens that could not be matched. foreach ($aTokens as $sToken) { - if (!isset($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']) { - if (isset($aValidTokens[$sToken])) { - $aValidTokens[$sToken][] = $aToken; - } else { - $aValidTokens[$sToken] = array($aToken); - } - } - } + if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) { + if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) { + // US ZIP+4 codes - merge in the 5-digit ZIP code + $oValidTokens->addToken( + $sToken, + new Token\Postcode(null, $aData[1], 'us') + ); + } elseif (preg_match('/^ [0-9]+$/', $sToken)) { + // Unknown single word token with a number. + // Assume it is a house number. + $oValidTokens->addToken( + $sToken, + new Token\HouseNumber(null, trim($sToken)) + ); } } } - foreach ($aTokens as $sToken) { - // Unknown single word token with a number - assume it is a house number - if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) { - $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken)); - } - } - // Any words that have failed completely? // TODO: suggestions - $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases); + Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo()); + + foreach ($aPhrases as $oPhrase) { + $oPhrase->computeWordSets($oValidTokens); + } + Debug::printDebugTable('Phrases', $aPhrases); + + Debug::newSection('Search candidates'); + + $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bStructuredPhrases); if ($this->bReverseInPlan) { // Reverse phrase array and also reverse the order of the wordsets in @@ -705,7 +699,7 @@ class Geocode if (count($aPhrases) > 1) { $aPhrases[count($aPhrases)-1]->invertWordSets(); } - $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false); + $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false); foreach ($aGroupedSearches as $aSearches) { foreach ($aSearches as $aSearch) { @@ -745,32 +739,63 @@ class Geocode } } - if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); + Debug::printGroupedSearch( + $aGroupedSearches, + $oValidTokens->debugTokenByWordIdList() + ); // Start the search process $iGroupLoop = 0; $iQueryLoop = 0; + $aNextResults = array(); foreach ($aGroupedSearches as $iGroupedRank => $aSearches) { $iGroupLoop++; + $aResults = $aNextResults; foreach ($aSearches as $oSearch) { $iQueryLoop++; - if (CONST_Debug) { - echo "