add support for QuerySet.distinct()

Author: WaVEV

django_mongodb/compiler.py

@@ -245,7 +245,9 @@ def execute_sql(
         try:
             query = self.build_query(
                 # Avoid $project (columns=None) if unneeded.
-                columns if self.query.annotations or not self.query.default_cols else None
+                columns
+                if self.query.annotations or not self.query.default_cols or self.query.distinct
+                else None
             )
         except EmptyResultSet:
             return iter([]) if result_type == MULTI else None
@@ -331,13 +333,7 @@ def cursor_iter(self, cursor, chunk_size, columns):
 
     def check_query(self):
         """Check if the current query is supported by the database."""
-        if self.query.distinct or getattr(
-            # In the case of Query.distinct().count(), the distinct attribute
-            # will be set on the inner_query.
-            getattr(self.query, "inner_query", None),
-            "distinct",
-            None,
-        ):
+        if self.query.distinct:
             # This is a heuristic to detect QuerySet.datetimes() and dates().
             # "datetimefield" and "datefield" are the names of the annotations
             # the methods use. A user could annotate with the same names which
@@ -346,7 +342,6 @@ def check_query(self):
                 raise NotSupportedError("QuerySet.datetimes() is not supported on MongoDB.")
             if "datefield" in self.query.annotations:
                 raise NotSupportedError("QuerySet.dates() is not supported on MongoDB.")
-            raise NotSupportedError("QuerySet.distinct() is not supported on MongoDB.")
         if self.query.extra:
             if any(key.startswith("_prefetch_related_") for key in self.query.extra):
                 raise NotSupportedError("QuerySet.prefetch_related() is not supported on MongoDB.")
@@ -365,7 +360,19 @@ def build_query(self, columns=None):
                 )
             query.combinator_pipeline = self.get_combinator_queries()
         else:
-            query.project_fields = self.get_project_fields(columns, ordering_fields)
+            if self.query.distinct:
+                # If query is distinct, build a $group stage for distinct
+                # fields, then set project fields based on the grouped _id.
+                distinct_fields = self.get_project_fields(
+                    columns, ordering_fields, force_expression=True
+                )
+                if not query.aggregation_pipeline:
+                    query.aggregation_pipeline = []
+                query.aggregation_pipeline.append({"$group": {"_id": distinct_fields}})
+                query.project_fields = {key: f"$_id.{key}" for key in distinct_fields}
+            else:
+                # Otherwise, project fields without grouping.
+                query.project_fields = self.get_project_fields(columns, ordering_fields)
             # If columns is None, then get_project_fields() won't add
             # ordering_fields to $project. Use $addFields (extra_fields) instead.
             if columns is None:
@@ -750,7 +757,7 @@ def build_query(self, columns=None):
         # Avoid $project (columns=None) if unneeded.
         columns = (
             compiler.get_columns()
-            if compiler.query.annotations or not compiler.query.default_cols
+            if self.query.annotations or not self.query.default_cols or self.query.distinct
             else None
         )
         subquery = compiler.build_query(columns)

django_mongodb/features.py

@@ -231,39 +231,13 @@ def django_test_expected_failures(self):
             "datetimes.tests.DateTimesTests.test_related_model_traverse",
             "model_inheritance_regress.tests.ModelInheritanceTest.test_issue_7105",
             "queries.tests.Queries1Tests.test_ticket7155",
+            "queries.tests.Queries1Tests.test_ticket7791",
+            "queries.tests.Queries1Tests.test_tickets_6180_6203",
             "queries.tests.Queries1Tests.test_tickets_7087_12242",
             "timezones.tests.LegacyDatabaseTests.test_query_datetimes",
             "timezones.tests.NewDatabaseTests.test_query_datetimes",
             "timezones.tests.NewDatabaseTests.test_query_datetimes_in_other_timezone",
         },
-        "QuerySet.distinct() is not supported.": {
-            "aggregation.tests.AggregateTestCase.test_sum_distinct_aggregate",
-            "aggregation_regress.tests.AggregationTests.test_annotate_distinct_aggregate",
-            "aggregation_regress.tests.AggregationTests.test_conditional_aggregate_on_complex_condition",
-            "aggregation_regress.tests.AggregationTests.test_distinct_conditional_aggregate",
-            "lookup.tests.LookupTests.test_lookup_collision_distinct",
-            "many_to_many.tests.ManyToManyTests.test_reverse_selects",
-            "many_to_many.tests.ManyToManyTests.test_selects",
-            "many_to_one.tests.ManyToOneTests.test_reverse_selects",
-            "ordering.tests.OrderingTests.test_orders_nulls_first_on_filtered_subquery",
-            "queries.tests.ExcludeTest17600.test_exclude_plain_distinct",
-            "queries.tests.ExcludeTest17600.test_exclude_with_q_is_equal_to_plain_exclude",
-            "queries.tests.ExcludeTest17600.test_exclude_with_q_is_equal_to_plain_exclude_variation",
-            "queries.tests.ExcludeTest17600.test_exclude_with_q_object_distinct",
-            "queries.tests.ExcludeTests.test_exclude_m2m_through",
-            "queries.tests.ExistsSql.test_distinct_exists",
-            "queries.tests.ExistsSql.test_sliced_distinct_exists",
-            "queries.tests.ExistsSql.test_ticket_18414",
-            "queries.tests.Queries1Tests.test_ticket4464",
-            "queries.tests.Queries1Tests.test_ticket7096",
-            "queries.tests.Queries1Tests.test_ticket7791",
-            "queries.tests.Queries1Tests.test_tickets_1878_2939",
-            "queries.tests.Queries1Tests.test_tickets_5321_7070",
-            "queries.tests.Queries1Tests.test_tickets_5324_6704",
-            "queries.tests.Queries1Tests.test_tickets_6180_6203",
-            "queries.tests.Queries6Tests.test_distinct_ordered_sliced_subquery_aggregation",
-            "update.tests.AdvancedTests.test_update_all",
-        },
         "QuerySet.extra() is not supported.": {
             "aggregation.tests.AggregateTestCase.test_exists_extra_where_with_aggregate",
             "annotations.tests.NonAggregateAnnotationTestCase.test_column_field_ordering",
@@ -281,6 +255,7 @@ def django_test_expected_failures(self):
             "queries.test_qs_combinators.QuerySetSetOperationTests.test_union_multiple_models_with_values_list_and_order_by_extra_select",
             "queries.test_qs_combinators.QuerySetSetOperationTests.test_union_with_extra_and_values_list",
             "queries.tests.EscapingTests.test_ticket_7302",
+            "queries.tests.Queries1Tests.test_tickets_1878_2939",
             "queries.tests.Queries5Tests.test_extra_select_literal_percent_s",
             "queries.tests.Queries5Tests.test_ticket7256",
             "queries.tests.ValuesQuerysetTests.test_extra_multiple_select_params_values_order_by",