introduce a separate token type for partials

[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 804f69e20da034134e130b63d18d75d023231758..ec2d7e68cbeb5ed4baa011db99ec13295f9cc1aa 100644 (file)
--- a/lib-php/tokenizer/legacy_tokenizer.php
+++ b/lib-php/tokenizer/legacy_tokenizer.php
@@ -15,6 +15,27 @@ class Tokenizer
         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
     }
 
+    public function checkStatus()
+    {
+        $sStandardWord = $this->oDB->getOne("SELECT make_standard_name('a')");
+        if ($sStandardWord === false) {
+            throw new Exception('Module failed', 701);
+        }
+
+        if ($sStandardWord != 'a') {
+            throw new Exception('Module call failed', 702);
+        }
+
+        $sSQL = "SELECT word_id FROM word WHERE word_token IN (' a')";
+        $iWordID = $this->oDB->getOne($sSQL);
+        if ($iWordID === false) {
+            throw new Exception('Query failed', 703);
+        }
+        if (!$iWordID) {
+            throw new Exception('No value', 704);
+        }
+    }
+
 
     public function setCountryRestriction($aCountries)
     {
@@ -84,7 +105,7 @@ class Tokenizer
         // now compute all possible tokens
         $aWordLists = array();
         $aTokens = array();
-        foreach ($aNormPhrases as $sTitle => $sPhrase) {
+        foreach ($aNormPhrases as $sPhrase) {
             if (strlen($sPhrase) > 0) {
                 $aWords = explode(' ', $sPhrase);
                 Tokenizer::addTokens($aTokens, $aWords);
@@ -191,17 +212,26 @@ class Tokenizer
                 ) {
                     $oToken = new Token\Country($iId, $aWord['country_code']);
                 }
-            } else {
+            } elseif ($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) {
-                $oValidTokens->addToken($aWord['word_token'], $oToken);
+                // remove any leading spaces
+                if ($aWord['word_token'][0] == ' ') {
+                    $oValidTokens->addToken(substr($aWord['word_token'], 1), $oToken);
+                } else {
+                    $oValidTokens->addToken($aWord['word_token'], $oToken);
+                }
             }
         }
     }