namespace Nominatim;
+require_once(CONST_LibDir.'/SimpleWordList.php');
+
class Tokenizer
{
private $oDB;
private $oNormalizer;
private $oTransliterator;
- private $aCountryRestriction;
public function __construct(&$oDB)
{
public function checkStatus()
{
- $sSQL = "SELECT word_id FROM word WHERE word_token IN (' a')";
+ $sSQL = 'SELECT word_id FROM word WHERE word_id is not null limit 1';
$iWordID = $this->oDB->getOne($sSQL);
if ($iWordID === false) {
- throw new Exception('Query failed', 703);
+ throw new \Exception('Query failed', 703);
}
if (!$iWordID) {
- throw new Exception('No value', 704);
+ throw new \Exception('No value', 704);
}
}
- public function setCountryRestriction($aCountries)
- {
- $this->aCountryRestriction = $aCountries;
- }
-
-
public function normalizeString($sTerm)
{
if ($this->oNormalizer === null) {
{
$aResults = array();
- $sSQL = 'SELECT word_id, class, type FROM word ';
- $sSQL .= ' WHERE word_token = \' \' || :term';
- $sSQL .= ' AND class is not null AND class not in (\'place\')';
+ $sSQL = "SELECT word_id, info->>'class' as class, info->>'type' as type ";
+ $sSQL .= ' FROM word WHERE word_token = :term and type = \'S\'';
Debug::printVar('Term', $sTerm);
Debug::printSQL($sSQL);
$sNormQuery .= ','.$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);
$oValidTokens = $this->computeValidTokens($aTokens, $sNormQuery);
foreach ($aPhrases as $iPhrase => $oPhrase) {
- $oPhrase->computeWordSets($aWordLists[$iPhrase], $oValidTokens);
+ $oPhrase->setWordSets($aWordLists[$iPhrase]->getWordSets($oValidTokens));
}
return $oValidTokens;
private function addTokensFromDB(&$oValidTokens, $aTokens, $sNormQuery)
{
// Check which tokens we have, get the ID numbers
- $sSQL = 'SELECT word_id, word_token, word, class, type, country_code,';
- $sSQL .= ' operator, coalesce(search_name_count, 0) as count';
+ $sSQL = 'SELECT word_id, word_token, type, word,';
+ $sSQL .= " info->>'op' as operator,";
+ $sSQL .= " info->>'class' as class, info->>'type' as ctype,";
+ $sSQL .= " info->>'count' as count";
$sSQL .= ' FROM word WHERE word_token in (';
$sSQL .= join(',', $this->oDB->getDBQuotedList($aTokens)).')';
$aDBWords = $this->oDB->getAll($sSQL, null, 'Could not get word tokens.');
foreach ($aDBWords as $aWord) {
- $oToken = null;
$iId = (int) $aWord['word_id'];
+ $sTok = $aWord['word_token'];
- if ($aWord['class']) {
- // Special terms need to appear in their normalized form.
- // (postcodes are not normalized in the word table)
- $sNormWord = $this->normalizeString($aWord['word']);
- if ($aWord['word'] && strpos($sNormQuery, $sNormWord) === false) {
- continue;
- }
-
- if ($aWord['class'] == 'place' && $aWord['type'] == 'house') {
- $oToken = new Token\HouseNumber($iId, trim($aWord['word_token']));
- } elseif ($aWord['class'] == 'place' && $aWord['type'] == 'postcode') {
- if ($aWord['word']
+ switch ($aWord['type']) {
+ case 'C': // country name tokens
+ if ($aWord['word'] !== null) {
+ $oValidTokens->addToken(
+ $sTok,
+ new Token\Country($iId, $aWord['word'])
+ );
+ }
+ break;
+ case 'H': // house number tokens
+ $oValidTokens->addToken($sTok, new Token\HouseNumber($iId, $aWord['word_token']));
+ break;
+ case 'P': // postcode tokens
+ // Postcodes are not normalized, so they may have content
+ // that makes SQL injection possible. Reject postcodes
+ // that would need special escaping.
+ if ($aWord['word'] !== null
&& pg_escape_string($aWord['word']) == $aWord['word']
) {
- $oToken = new Token\Postcode(
+ $sNormPostcode = $this->normalizeString($aWord['word']);
+ if (strpos($sNormQuery, $sNormPostcode) !== false) {
+ $oValidTokens->addToken(
+ $sTok,
+ new Token\Postcode($iId, $aWord['word'], null)
+ );
+ }
+ }
+ break;
+ case 'S': // tokens for classification terms (special phrases)
+ if ($aWord['class'] !== null && $aWord['ctype'] !== null) {
+ $oValidTokens->addToken($sTok, new Token\SpecialTerm(
$iId,
- $aWord['word'],
- $aWord['country_code']
- );
+ $aWord['class'],
+ $aWord['ctype'],
+ (isset($aWord['operator'])) ? Operator::NEAR : Operator::NONE
+ ));
}
- } else {
- // near and in operator the same at the moment
- $oToken = new Token\SpecialTerm(
+ break;
+ case 'W': // full-word tokens
+ $oValidTokens->addToken($sTok, new Token\Word(
$iId,
- $aWord['class'],
- $aWord['type'],
- $aWord['operator'] ? Operator::NEAR : Operator::NONE
- );
- }
- } elseif ($aWord['country_code']) {
- // Filter country tokens that do not match restricted countries.
- if (!$this->aCountryRestriction
- || in_array($aWord['country_code'], $this->aCountryRestriction)
- ) {
- $oToken = new Token\Country($iId, $aWord['country_code']);
- }
- } elseif ($aWord['word_token'][0] == ' ') {
- $oToken = new Token\Word(
- $iId,
- $aWord['word_token'][0] != ' ',
- (int) $aWord['count'],
- substr_count($aWord['word_token'], ' ')
- );
- } else {
- $oToken = new Token\Partial(
- $iId,
- (int) $aWord['count']
- );
- }
-
- if ($oToken) {
- // remove any leading spaces
- if ($aWord['word_token'][0] == ' ') {
- $oValidTokens->addToken(substr($aWord['word_token'], 1), $oToken);
- } else {
- $oValidTokens->addToken($aWord['word_token'], $oToken);
- }
- }
- }
- }
-
-
- /**
- * 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;
- $aTokens[$sPhrase] = $sPhrase;
-
- for ($j = $i + 1; $j < $iNumWords; $j++) {
- $sPhrase .= ' '.$aWords[$j];
- $aTokens[' '.$sPhrase] = ' '.$sPhrase;
- $aTokens[$sPhrase] = $sPhrase;
+ (int) $aWord['count'],
+ substr_count($aWord['word_token'], ' ')
+ ));
+ break;
+ case 'w': // partial word terms
+ $oValidTokens->addToken($sTok, new Token\Partial(
+ $iId,
+ $aWord['word_token'],
+ (int) $aWord['count']
+ ));
+ break;
+ default:
+ break;
}
}
}
projects / nominatim.git / blobdiff