Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib-php / tokenizer / icu_tokenizer.php

diff --git a/lib-php/tokenizer/icu_tokenizer.php b/lib-php/tokenizer/icu_tokenizer.php
index ca224a224861f5bd2e0f126cf10be5ea0edf5488..ccce99ca1330d7a42a6976d7fb7c9eaf3d8a84d7 100644 (file)
--- a/lib-php/tokenizer/icu_tokenizer.php
+++ b/lib-php/tokenizer/icu_tokenizer.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;
 
@@ -40,6 +48,15 @@ class Tokenizer
         return $this->oNormalizer->transliterate($sTerm);
     }
 
+
+    public function mostFrequentWords($iNum)
+    {
+        $sSQL = "SELECT word FROM word WHERE type = 'W'";
+        $sSQL .= "ORDER BY info->'count' DESC LIMIT ".$iNum;
+        return $this->oDB->getCol($sSQL);
+    }
+
+
     private function makeStandardWord($sTerm)
     {
         return trim($this->oTransliterator->transliterate(' '.$sTerm.' '));
@@ -140,7 +157,8 @@ class Tokenizer
         $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";
+        $sSQL .= "      info->>'count' as count,";
+        $sSQL .= "      info->>'lookup' as lookup";
         $sSQL .= ' FROM word WHERE word_token in (';
         $sSQL .= join(',', $this->oDB->getDBQuotedList($aTokens)).')';
 
@@ -162,7 +180,8 @@ class Tokenizer
                     }
                     break;
                 case 'H':  // house number tokens
-                    $oValidTokens->addToken($sTok, new Token\HouseNumber($iId, $aWord['word_token']));
+                    $sLookup = $aWord['lookup'] ?? $aWord['word_token'];
+                    $oValidTokens->addToken($sTok, new Token\HouseNumber($iId, $sLookup));
                     break;
                 case 'P':  // postcode tokens
                     // Postcodes are not normalized, so they may have content