X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/0fb8eade136ea03e7853aca0795ca69833c33661..1147b83b22a073db2f81ea177f9caa12e727e249:/lib-php/tokenizer/legacy_icu_tokenizer.php diff --git a/lib-php/tokenizer/legacy_icu_tokenizer.php b/lib-php/tokenizer/legacy_icu_tokenizer.php index 690ef136..ca224a22 100644 --- a/lib-php/tokenizer/legacy_icu_tokenizer.php +++ b/lib-php/tokenizer/legacy_icu_tokenizer.php @@ -2,6 +2,8 @@ namespace Nominatim; +require_once(CONST_LibDir.'/SimpleWordList.php'); + class Tokenizer { private $oDB; @@ -81,13 +83,10 @@ class Tokenizer $sNormQuery .= ','.$this->normalizeString($oPhrase->getPhrase()); $sPhrase = $this->makeStandardWord($oPhrase->getPhrase()); Debug::printVar('Phrase', $sPhrase); - if (strlen($sPhrase) > 0) { - $aWords = explode(' ', $sPhrase); - Tokenizer::addTokens($aTokens, $aWords); - $aWordLists[] = $aWords; - } else { - $aWordLists[] = array(); - } + + $oWordList = new SimpleWordList($sPhrase); + $aTokens = array_merge($aTokens, $oWordList->getTokens()); + $aWordLists[] = $oWordList; } Debug::printVar('Tokens', $aTokens); @@ -96,7 +95,7 @@ class Tokenizer $oValidTokens = $this->computeValidTokens($aTokens, $sNormQuery); foreach ($aPhrases as $iPhrase => $oPhrase) { - $oPhrase->computeWordSets($aWordLists[$iPhrase], $oValidTokens); + $oPhrase->setWordSets($aWordLists[$iPhrase]->getWordSets($oValidTokens)); } return $oValidTokens; @@ -210,27 +209,4 @@ class Tokenizer } } } - - - /** - * Add the tokens from this phrase to the given list of tokens. - * - * @param string[] $aTokens List of tokens to append. - * - * @return void - */ - private static function addTokens(&$aTokens, $aWords) - { - $iNumWords = count($aWords); - - for ($i = 0; $i < $iNumWords; $i++) { - $sPhrase = $aWords[$i]; - $aTokens[$sPhrase] = $sPhrase; - - for ($j = $i + 1; $j < $iNumWords; $j++) { - $sPhrase .= ' '.$aWords[$j]; - $aTokens[$sPhrase] = $sPhrase; - } - } - } }