keep inherited address parts after indexing

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

diff --git a/lib-php/TokenHousenumber.php b/lib-php/TokenHousenumber.php
index 5c7c6e9b633a4af458acbc637249ffad2453f7ac..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;
 
@@ -8,9 +16,9 @@ namespace Nominatim\Token;
 class HouseNumber
 {
     /// Database word id, if available.
-    public $iId;
+    private $iId;
     /// Normalized house number.
-    public $sToken;
+    private $sToken;
 
     public function __construct($iId, $sToken)
     {
@@ -18,6 +26,80 @@ class HouseNumber
         $this->sToken = $sToken;
     }
 
+    public function getId()
+    {
+        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.
+     *
+     * @param object  $oSearch      Partial search description derived so far.
+     * @param object  $oPosition    Description of the token position within
+                                    the query.
+     *
+     * @return SearchDescription[] List of derived search descriptions.
+     */
+    public function extendSearch($oSearch, $oPosition)
+    {
+        $aNewSearches = array();
+
+        // 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) > 3) {
+            $iSearchCost += strlen($this->sToken) - 1;
+        }
+        if (!$oSearch->hasOperator(\Nominatim\Operator::NONE)) {
+            $iSearchCost++;
+        }
+        if (empty($this->iId)) {
+            $iSearchCost++;
+        }
+        // also must not appear in the middle of the address
+        if ($oSearch->hasAddress() || $oSearch->hasPostcode()) {
+            $iSearchCost++;
+        }
+
+        $oNewSearch = $oSearch->clone($iSearchCost);
+        $oNewSearch->setHousenumber($this->sToken);
+        $aNewSearches[] = $oNewSearch;
+
+        // Housenumbers may appear in the name when the place has its own
+        // address terms.
+        if ($this->iId !== null
+            && ($oSearch->getNamePhrase() >= 0 || !$oSearch->hasName())
+            && !$oSearch->hasAddress()
+        ) {
+            $oNewSearch = $oSearch->clone($iSearchCost);
+            $oNewSearch->setHousenumberAsName($this->iId);
+
+            $aNewSearches[] = $oNewSearch;
+        }
+
+        return $aNewSearches;
+    }
+
+
     public function debugInfo()
     {
         return array(
@@ -26,4 +108,9 @@ class HouseNumber
                 'Info' => array('nr' => $this->sToken)
                );
     }
+
+    public function debugCode()
+    {
+        return 'H';
+    }
 }