✨ Add BaseQueryPreprocessor for customizable search query preprocessing

Author: noeddl

froide/helper/search/__init__.py

@@ -55,6 +55,12 @@ def get_default_ngram_analyzer():
     )
 
 
+def get_default_query_preprocessor():
+    from froide.helper.search.filters import BaseQueryPreprocessor
+
+    return BaseQueryPreprocessor()
+
+
 def get_func(config_name, default_func):
     def get_it():
         from django.conf import settings
@@ -73,3 +79,4 @@ def get_it():
 get_search_analyzer = get_func("search_analyzer", get_default_text_analyzer)
 get_search_quote_analyzer = get_func("search_quote_analyzer", get_default_text_analyzer)
 get_ngram_analyzer = get_func("ngram_analyzer", get_default_ngram_analyzer)
+get_query_preprocessor = get_func("query_preprocessor", get_default_query_preprocessor)

froide/helper/search/filters.py

@@ -4,6 +4,8 @@
 import django_filters
 from elasticsearch_dsl.query import Q
 
+from froide.helper.search import get_query_preprocessor
+
 
 class BaseSearchFilterSet(django_filters.FilterSet):
     query_fields = ["content"]
@@ -19,6 +21,7 @@ class BaseSearchFilterSet(django_filters.FilterSet):
     def __init__(self, *args, **kwargs):
         self.facet_config = kwargs.pop("facet_config", {})
         self.view = kwargs.pop("view", None)
+        self.query_preprocessor = get_query_preprocessor()
         super().__init__(*args, **kwargs)
 
     def apply_filter(self, qs, name, *args, **kwargs):
@@ -42,13 +45,29 @@ def filter_queryset(self, queryset):
 
     def auto_query(self, qs, name, value):
         if value:
+            query = self.query_preprocessor.prepare_query(value)
             return qs.set_query(
                 Q(
                     "simple_query_string",
-                    query=value,
+                    query=query,
                     fields=self.query_fields,
                     default_operator="and",
                     lenient=True,
                 )
             )
         return qs
+
+
+class BaseQueryPreprocessor:
+    """
+    Base class that can be overridden for custom search query preprocessing.
+    """
+
+    def prepare_query(self, text: str):
+        """
+        Preprocess the given search query text and return the processed text.
+
+        This method can be overridden in subclasses to implement custom
+        preprocessing logic.
+        """
+        return text

froide/helper/tests/test_search.py

@@ -0,0 +1,55 @@
+from unittest.mock import MagicMock
+
+from froide.helper.search.filters import BaseSearchFilterSet
+
+
+class DummyModel:
+    pass
+
+
+class DummyQS:
+    def __init__(self):
+        self.model = DummyModel()
+        self.query = None
+
+    def set_query(self, q):
+        self.query = q
+        return self
+
+
+class TestBaseSearchFilterSetQueryPreprocessing:
+    def test_auto_query_without_query_value(self):
+        qs = DummyQS()
+        fs = BaseSearchFilterSet(queryset=qs)
+
+        result = fs.auto_query(qs, "q", "")
+
+        assert result is qs
+        assert result.query is None
+
+    def test_auto_query_with_default_query_preprocessor(self):
+        qs = DummyQS()
+        fs = BaseSearchFilterSet(queryset=qs)
+
+        result = fs.auto_query(qs, "q", "test query")
+
+        assert result is qs
+        assert result.query is not None
+        assert result.query.query == "test query"
+
+    def test_auto_query_with_custom_query_preprocessor(self, monkeypatch):
+        # Mock custom query preprocessor.
+        mock_preprocessor = MagicMock()
+        mock_preprocessor.prepare_query.return_value = "processed query"
+        monkeypatch.setattr(
+            "froide.helper.search.filters.get_query_preprocessor",
+            lambda: mock_preprocessor,
+        )
+
+        qs = DummyQS()
+        fs = BaseSearchFilterSet(queryset=qs)
+
+        result = fs.auto_query(qs, "q", "original query")
+
+        assert result.query.query == "processed query"
+        mock_preprocessor.prepare_query.assert_called_once_with("original query")