X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/500c61685b16fd1d51e7b5f4738c788a9d00eab4..1e9f37ab82db1758235bedf83c659693f4ca6c3e:/lib-php/tokenizer/legacy_tokenizer.php

diff --git a/lib-php/tokenizer/legacy_tokenizer.php b/lib-php/tokenizer/legacy_tokenizer.php
index 50207c31..570b8828 100644
--- a/lib-php/tokenizer/legacy_tokenizer.php
+++ b/lib-php/tokenizer/legacy_tokenizer.php
@@ -19,20 +19,20 @@ 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);
         }
     }
 
@@ -137,14 +137,14 @@ 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(
@@ -212,17 +212,29 @@ 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'], ' ')
                 );
+            // 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'], $oToken);
+                // remove any leading spaces
+                if ($aWord['word_token'][0] == ' ') {
+                    $oValidTokens->addToken(substr($aWord['word_token'], 1), $oToken);
+                } else {
+                    $oValidTokens->addToken($aWord['word_token'], $oToken);
+                }
             }
         }
     }