postcodes: strip leading country codes

[nominatim.git] / lib-php / TokenHousenumber.php

diff --git a/lib-php/TokenHousenumber.php b/lib-php/TokenHousenumber.php
index 0cc67a1212d17fc8bbd5a6251f5b86827fc05efe..62c2a624a6e4d7bed6fe55dae4afe8a88c5ba076 100644 (file)
--- a/lib-php/TokenHousenumber.php
+++ b/lib-php/TokenHousenumber.php
@@ -1,4 +1,12 @@
 <?php
+/**
+ * SPDX-License-Identifier: GPL-2.0-only
+ *
+ * This file is part of Nominatim. (https://nominatim.org)
+ *
+ * Copyright (C) 2022 by the Nominatim developer community.
+ * For a full list of authors see the git log.
+ */
 
 namespace Nominatim\Token;
 
@@ -23,6 +31,24 @@ class HouseNumber
         return $this->iId;
     }
 
+    /**
+     * Check if the token can be added to the given search.
+     * Derive new searches by adding this token to an existing search.
+     *
+     * @param object  $oSearch      Partial search description derived so far.
+     * @param object  $oPosition    Description of the token position within
+                                    the query.
+     *
+     * @return True if the token is compatible with the search configuration
+     *         given the position.
+     */
+    public function isExtendable($oSearch, $oPosition)
+    {
+        return !$oSearch->hasHousenumber()
+               && !$oSearch->hasOperator(\Nominatim\Operator::POSTCODE)
+               && $oPosition->maybePhrase('street');
+    }
+
     /**
      * Derive new searches by adding this token to an existing search.
      *
@@ -36,19 +62,12 @@ class HouseNumber
     {
         $aNewSearches = array();
 
-        if ($oSearch->hasHousenumber()
-            || $oSearch->hasOperator(\Nominatim\Operator::POSTCODE)
-            || !$oPosition->maybePhrase('street')
-        ) {
-            return $aNewSearches;
-        }
-
         // sanity check: if the housenumber is not mainly made
         // up of numbers, add a penalty
         $iSearchCost = 1;
         if (preg_match('/\\d/', $this->sToken) === 0
-            || preg_match_all('/[^0-9]/', $this->sToken, $aMatches) > 2) {
-            $iSearchCost++;
+            || preg_match_all('/[^0-9 ]/', $this->sToken, $aMatches) > 3) {
+            $iSearchCost += strlen($this->sToken) - 1;
         }
         if (!$oSearch->hasOperator(\Nominatim\Operator::NONE)) {
             $iSearchCost++;