add tests for sanitizer tagging language

[nominatim.git] / lib-php / tokenizer / legacy_tokenizer.php

diff --git a/lib-php/tokenizer/legacy_tokenizer.php b/lib-php/tokenizer/legacy_tokenizer.php
index 50207c31785d6c3579418fa42931f8d6a39be81c..b508d220aca3f07e861fd0bbb6efc39a13cf5528 100644 (file)
--- a/lib-php/tokenizer/legacy_tokenizer.php
+++ b/lib-php/tokenizer/legacy_tokenizer.php
@@ -2,12 +2,13 @@
 
 namespace Nominatim;
 
+require_once(CONST_LibDir.'/SimpleWordList.php');
+
 class Tokenizer
 {
     private $oDB;
 
     private $oNormalizer = null;
-    private $aCountryRestriction = null;
 
     public function __construct(&$oDB)
     {
@@ -19,30 +20,24 @@ class Tokenizer
     {
         $sStandardWord = $this->oDB->getOne("SELECT make_standard_name('a')");
         if ($sStandardWord === false) {
-            throw new Exception('Module failed', 701);
+            throw new \Exception('Module failed', 701);
         }
 
         if ($sStandardWord != 'a') {
-            throw new Exception('Module call failed', 702);
+            throw new \Exception('Module call failed', 702);
         }
 
         $sSQL = "SELECT word_id FROM word WHERE word_token IN (' a')";
         $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) {
@@ -92,6 +87,23 @@ class Tokenizer
             $sNormQuery .= ','.$this->normalizeString($oPhrase->getPhrase());
             $sSQL .= 'make_standard_name(:' .$iPhrase.') as p'.$iPhrase.',';
             $aParams[':'.$iPhrase] = $oPhrase->getPhrase();
+
+            // Conflicts between US state abbreviations and various words
+            // for 'the' in different languages
+            switch (strtolower($oPhrase->getPhrase())) {
+                case 'il':
+                    $aParams[':'.$iPhrase] = 'illinois';
+                    break;
+                case 'al':
+                    $aParams[':'.$iPhrase] = 'alabama';
+                    break;
+                case 'la':
+                    $aParams[':'.$iPhrase] = 'louisiana';
+                    break;
+                default:
+                    $aParams[':'.$iPhrase] = $oPhrase->getPhrase();
+                    break;
+            }
         }
         $sSQL = substr($sSQL, 0, -1);
 
@@ -106,13 +118,14 @@ class Tokenizer
         $aWordLists = array();
         $aTokens = array();
         foreach ($aNormPhrases as $sPhrase) {
-            if (strlen($sPhrase) > 0) {
-                $aWords = explode(' ', $sPhrase);
-                Tokenizer::addTokens($aTokens, $aWords);
-                $aWordLists[] = $aWords;
-            } else {
-                $aWordLists[] = array();
+            $oWordList = new SimpleWordList($sPhrase);
+
+            foreach ($oWordList->getTokens() as $sToken) {
+                $aTokens[' '.$sToken] = ' '.$sToken;
+                $aTokens[$sToken] = $sToken;
             }
+
+            $aWordLists[] = $oWordList;
         }
 
         Debug::printVar('Tokens', $aTokens);
@@ -121,7 +134,7 @@ 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;
@@ -137,14 +150,14 @@ class Tokenizer
 
             // Try more interpretations for Tokens that could not be matched.
             foreach ($aTokens as $sToken) {
-                if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) {
-                    if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
+                if ($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, $aData[1], 'us')
                         );
-                    } elseif (preg_match('/^ [0-9]+$/', $sToken)) {
+                    } elseif (preg_match('/^[0-9]+$/', $sToken)) {
                         // Unknown single word token with a number.
                         // Assume it is a house number.
                         $oValidTokens->addToken(
@@ -206,48 +219,30 @@ class Tokenizer
                     );
                 }
             } elseif ($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']);
-                }
-            } else {
+                $oToken = new Token\Country($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'], ' ')
                 );
+            // For backward compatibility: ignore all partial tokens with more
+            // than one word.
+            } elseif (strpos($aWord['word_token'], ' ') === false) {
+                $oToken = new Token\Partial(
+                    $iId,
+                    $aWord['word_token'],
+                    (int) $aWord['count']
+                );
             }
 
             if ($oToken) {
-                $oValidTokens->addToken($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] = $sPhrase;
+                // remove any leading spaces
+                if ($aWord['word_token'][0] == ' ') {
+                    $oValidTokens->addToken(substr($aWord['word_token'], 1), $oToken);
+                } else {
+                    $oValidTokens->addToken($aWord['word_token'], $oToken);
+                }
             }
         }
     }