php: make word list a first-class object

[nominatim.git] / 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 9bd9828cb0703ce388108713856c6f176c15feb4..ca224a224861f5bd2e0f126cf10be5ea0edf5488 100644 (file)
--- a/lib-php/tokenizer/legacy_icu_tokenizer.php
+++ b/lib-php/tokenizer/legacy_icu_tokenizer.php
@@ -2,13 +2,14 @@
 
 namespace Nominatim;
 
+require_once(CONST_LibDir.'/SimpleWordList.php');
+
 class Tokenizer
 {
     private $oDB;
 
     private $oNormalizer;
     private $oTransliterator;
-    private $aCountryRestriction;
 
     public function __construct(&$oDB)
     {
@@ -19,23 +20,17 @@ class Tokenizer
 
     public function checkStatus()
     {
-        $sSQL = "SELECT word_id FROM word limit 1";
+        $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) {
@@ -88,13 +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);
@@ -103,7 +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 $oValidTokens;
@@ -145,8 +137,7 @@ class Tokenizer
     private function addTokensFromDB(&$oValidTokens, $aTokens, $sNormQuery)
     {
         // Check which tokens we have, get the ID numbers
-        $sSQL = 'SELECT word_id, word_token, type,';
-        $sSQL .= "      info->>'cc' as country, info->>'postcode' as postcode,";
+        $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";
@@ -163,11 +154,11 @@ class Tokenizer
 
             switch ($aWord['type']) {
                 case 'C':  // country name tokens
-                    if ($aWord['country'] !== null
-                        && (!$this->aCountryRestriction
-                            || in_array($aWord['country'], $this->aCountryRestriction))
-                    ) {
-                        $oValidTokens->addToken($sTok, new Token\Country($iId, $aWord['country']));
+                    if ($aWord['word'] !== null) {
+                        $oValidTokens->addToken(
+                            $sTok,
+                            new Token\Country($iId, $aWord['word'])
+                        );
                     }
                     break;
                 case 'H':  // house number tokens
@@ -177,12 +168,15 @@ class Tokenizer
                     // Postcodes are not normalized, so they may have content
                     // that makes SQL injection possible. Reject postcodes
                     // that would need special escaping.
-                    if ($aWord['postcode'] !== null
-                        && pg_escape_string($aWord['postcode']) == $aWord['postcode']
+                    if ($aWord['word'] !== null
+                        && pg_escape_string($aWord['word']) == $aWord['word']
                     ) {
-                        $sNormPostcode = $this->normalizeString($aWord['postcode']);
+                        $sNormPostcode = $this->normalizeString($aWord['word']);
                         if (strpos($sNormQuery, $sNormPostcode) !== false) {
-                            $oValidTokens->addToken($sTok, new Token\Postcode($iId, $aWord['postcode'], null));
+                            $oValidTokens->addToken(
+                                $sTok,
+                                new Token\Postcode($iId, $aWord['word'], null)
+                            );
                         }
                     }
                     break;
@@ -192,7 +186,7 @@ class Tokenizer
                             $iId,
                             $aWord['class'],
                             $aWord['ctype'],
-                            (isset($aWord['op'])) ? Operator::NEAR : Operator::NONE
+                            (isset($aWord['operator'])) ? Operator::NEAR : Operator::NONE
                         ));
                     }
                     break;
@@ -215,27 +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;
-            }
-        }
-    }
 }