X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/4ae5ba7fc4a3451c806d89a583158412ea26bd01..452021ef0c9ac746949a2ef7fd12db4c2d5fee35:/lib-php/tokenizer/legacy_tokenizer.php

diff --git a/lib-php/tokenizer/legacy_tokenizer.php b/lib-php/tokenizer/legacy_tokenizer.php
index e5ffbe02..d5686f64 100644
--- a/lib-php/tokenizer/legacy_tokenizer.php
+++ b/lib-php/tokenizer/legacy_tokenizer.php
@@ -48,6 +48,14 @@ class Tokenizer
     }
 
 
+    public function mostFrequentWords($iNum)
+    {
+        $sSQL = 'SELECT word FROM word WHERE word is not null ';
+        $sSQL .= 'ORDER BY search_name_count DESC LIMIT '.$iNum;
+        return $this->oDB->getCol($sSQL);
+    }
+
+
     public function tokensForSpecialTerm($sTerm)
     {
         $aResults = array();
@@ -87,6 +95,23 @@ class Tokenizer
             $sNormQuery .= ','.$this->normalizeString($oPhrase->getPhrase());
             $sSQL .= 'make_standard_name(:' .$iPhrase.') as p'.$iPhrase.',';
             $aParams[':'.$iPhrase] = $oPhrase->getPhrase();
+
+            // Conflicts between US state abbreviations and various words
+            // for 'the' in different languages
+            switch (strtolower($oPhrase->getPhrase())) {
+                case 'il':
+                    $aParams[':'.$iPhrase] = 'illinois';
+                    break;
+                case 'al':
+                    $aParams[':'.$iPhrase] = 'alabama';
+                    break;
+                case 'la':
+                    $aParams[':'.$iPhrase] = 'louisiana';
+                    break;
+                default:
+                    $aParams[':'.$iPhrase] = $oPhrase->getPhrase();
+                    break;
+            }
         }
         $sSQL = substr($sSQL, 0, -1);