Merge pull request #3122 from miku0/sanitizer-final

[nominatim.git] / nominatim / tokenizer / base.py

diff --git a/nominatim/tokenizer/base.py b/nominatim/tokenizer/base.py
index afbd1914b35d84219812afdd64f3061d306944f5..f0fd9dd0e178c5a1319ce6e79a560cb438567f12 100644 (file)
--- a/nominatim/tokenizer/base.py
+++ b/nominatim/tokenizer/base.py
@@ -13,6 +13,7 @@ from typing import List, Tuple, Dict, Any, Optional, Iterable
 from pathlib import Path
 
 from nominatim.config import Configuration
+from nominatim.db.connection import Connection
 from nominatim.data.place_info import PlaceInfo
 from nominatim.typing import Protocol
 
@@ -233,6 +234,13 @@ class AbstractTokenizer(ABC):
         """
 
 
+    @abstractmethod
+    def most_frequent_words(self, conn: Connection, num: int) -> List[str]:
+        """ Return a list of the `num` most frequent full words
+            in the database.
+        """
+
+
 class TokenizerModule(Protocol):
     """ Interface that must be exported by modules that implement their
         own tokenizer.