X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/6a7e0d652b1d40a397e1c1386d500101796676c4..1e40d65aa93e011815d07a406bcc46448512d7e3:/lib-php/Phrase.php diff --git a/lib-php/Phrase.php b/lib-php/Phrase.php index e2643e87..d14c842d 100644 --- a/lib-php/Phrase.php +++ b/lib-php/Phrase.php @@ -16,8 +16,6 @@ class Phrase private $sPhrase; // Element type for structured searches. private $sPhraseType; - // Space-separated words of the phrase. - private $aWords; // Possible segmentations of the phrase. private $aWordSets; @@ -38,7 +36,14 @@ class Phrase { $this->sPhrase = trim($sPhrase); $this->sPhraseType = $sPhraseType; - $this->aWords = explode(' ', $this->sPhrase); + } + + /** + * Get the orginal phrase of the string. + */ + public function getPhrase() + { + return $this->sPhrase; } /** @@ -63,30 +68,6 @@ class Phrase return $this->aWordSets; } - /** - * Add the tokens from this phrase to the given list of tokens. - * - * @param string[] $aTokens List of tokens to append. - * - * @return void - */ - public function addTokens(&$aTokens) - { - $iNumWords = count($this->aWords); - - for ($i = 0; $i < $iNumWords; $i++) { - $sPhrase = $this->aWords[$i]; - $aTokens[' '.$sPhrase] = ' '.$sPhrase; - $aTokens[$sPhrase] = $sPhrase; - - for ($j = $i + 1; $j < $iNumWords; $j++) { - $sPhrase .= ' '.$this->aWords[$j]; - $aTokens[' '.$sPhrase] = ' '.$sPhrase; - $aTokens[$sPhrase] = $sPhrase; - } - } - } - /** * Invert the set of possible segmentations. * @@ -99,21 +80,27 @@ class Phrase } } - public function computeWordSets($oTokens) + public function computeWordSets($aWords, $oTokens) { - $iNumWords = count($this->aWords); + $iNumWords = count($aWords); + + if ($iNumWords == 0) { + $this->aWordSets = null; + return; + } + // Caches the word set for the partial phrase up to word i. $aSetCache = array_fill(0, $iNumWords, array()); // Initialise first element of cache. There can only be the word. - if ($oTokens->containsAny($this->aWords[0])) { - $aSetCache[0][] = array($this->aWords[0]); + if ($oTokens->containsAny($aWords[0])) { + $aSetCache[0][] = array($aWords[0]); } // Now do the next elements using what we already have. for ($i = 1; $i < $iNumWords; $i++) { for ($j = $i; $j > 0; $j--) { - $sPartial = $j == $i ? $this->aWords[$j] : $this->aWords[$j].' '.$sPartial; + $sPartial = $j == $i ? $aWords[$j] : $aWords[$j].' '.$sPartial; if (!empty($aSetCache[$j - 1]) && $oTokens->containsAny($sPartial)) { $aPartial = array($sPartial); foreach ($aSetCache[$j - 1] as $aSet) { @@ -136,7 +123,7 @@ class Phrase } // finally the current full phrase - $sPartial = $this->aWords[0].' '.$sPartial; + $sPartial = $aWords[0].' '.$sPartial; if ($oTokens->containsAny($sPartial)) { $aSetCache[$i][] = array($sPartial); } @@ -153,7 +140,6 @@ class Phrase return array( 'Type' => $this->sPhraseType, 'Phrase' => $this->sPhrase, - 'Words' => $this->aWords, 'WordSets' => $this->aWordSets ); }