Author: momento-github-actions-machine-user

src/momento/internal/aio/_vector_index_data_client.py

@@ -98,19 +98,22 @@ async def upsert_item_batch(
     async def delete_item_batch(
         self,
         index_name: str,
-        filter: list[str],
+        filter: FilterExpression | list[str],
     ) -> DeleteItemBatchResponse:
         try:
             self._log_issuing_request("DeleteItemBatch", {"index_name": index_name})
             _validate_index_name(index_name)
 
-            if len(filter) == 0:
-                return DeleteItemBatch.Success()
+            filter_expression: vectorindex_pb._FilterExpression
 
-            request = vectorindex_pb._DeleteItemBatchRequest(
-                index_name=index_name,
-                filter=F.IdInSet(filter).to_filter_expression_proto(),
-            )
+            if isinstance(filter, FilterExpression):
+                filter_expression = filter.to_filter_expression_proto()
+            else:
+                if len(filter) == 0:
+                    return DeleteItemBatch.Success()
+                filter_expression = F.IdInSet(filter).to_filter_expression_proto()
+
+            request = vectorindex_pb._DeleteItemBatchRequest(index_name=index_name, filter=filter_expression)
 
             await self._build_stub().DeleteItemBatch(request, timeout=self._default_deadline_seconds)
 

src/momento/internal/synchronous/_vector_index_data_client.py

@@ -98,19 +98,22 @@ def upsert_item_batch(
     def delete_item_batch(
         self,
         index_name: str,
-        filter: list[str],
+        filter: FilterExpression | list[str],
     ) -> DeleteItemBatchResponse:
         try:
             self._log_issuing_request("DeleteItemBatch", {"index_name": index_name})
             _validate_index_name(index_name)
 
-            if len(filter) == 0:
-                return DeleteItemBatch.Success()
+            filter_expression: vectorindex_pb._FilterExpression
 
-            request = vectorindex_pb._DeleteItemBatchRequest(
-                index_name=index_name,
-                filter=F.IdInSet(filter).to_filter_expression_proto(),
-            )
+            if isinstance(filter, FilterExpression):
+                filter_expression = filter.to_filter_expression_proto()
+            else:
+                if len(filter) == 0:
+                    return DeleteItemBatch.Success()
+                filter_expression = F.IdInSet(filter).to_filter_expression_proto()
+
+            request = vectorindex_pb._DeleteItemBatchRequest(index_name=index_name, filter=filter_expression)
 
             self._build_stub().DeleteItemBatch(request, timeout=self._default_deadline_seconds)
 

src/momento/vector_index_client.py

@@ -194,14 +194,16 @@ def upsert_item_batch(self, index_name: str, items: list[Item]) -> UpsertItemBat
         """
         return self._data_client.upsert_item_batch(index_name, items)
 
-    def delete_item_batch(self, index_name: str, filter: list[str]) -> DeleteItemBatchResponse:
+    def delete_item_batch(self, index_name: str, filter: FilterExpression | list[str]) -> DeleteItemBatchResponse:
         """Deletes a batch of items from a vector index.
 
         Deletes any and all items with the given IDs from the index.
 
         Args:
             index_name (str): Name of the index to delete the items from.
-            filter (list[str]): The IDs of the items to be deleted from the index.
+            filter (FilterExpression | list[str]): A filter expression to filter
+                items by ID. If a list of strings is provided, it is treated as a
+                "IdInSet" filter expression.
 
         Returns:
             DeleteItemBatchResponse: The result of a delete item batch operation.

src/momento/vector_index_client_async.py

@@ -194,14 +194,16 @@ async def upsert_item_batch(self, index_name: str, items: list[Item]) -> UpsertI
         """
         return await self._data_client.upsert_item_batch(index_name, items)
 
-    async def delete_item_batch(self, index_name: str, filter: list[str]) -> DeleteItemBatchResponse:
+    async def delete_item_batch(self, index_name: str, filter: FilterExpression | list[str]) -> DeleteItemBatchResponse:
         """Deletes a batch of items from a vector index.
 
         Deletes any and all items with the given IDs from the index.
 
         Args:
             index_name (str): Name of the index to delete the items from.
-            filter (list[str]): The IDs of the items to be deleted from the index.
+            filter (FilterExpression | list[str]): A filter expression to filter
+                items by ID. If a list of strings is provided, it is treated as a
+                "IdInSet" filter expression.
 
         Returns:
             DeleteItemBatchResponse: The result of a delete item batch operation.

tests/momento/vector_index_client/test_data.py

@@ -688,7 +688,7 @@ def test_delete_validates_index_name(vector_index_client: PreviewVectorIndexClie
     assert response.error_code == MomentoErrorCode.INVALID_ARGUMENT_ERROR
 
 
-def test_delete_deletes_ids(
+def test_delete_items_by_id(
     vector_index_client: PreviewVectorIndexClient,
     unique_vector_index_name: TUniqueVectorIndexName,
 ) -> None:
@@ -701,10 +701,10 @@ def test_delete_deletes_ids(
     upsert_response = vector_index_client.upsert_item_batch(
         index_name,
         items=[
-            Item(id="test_item_1", vector=[1.0, 2.0]),
-            Item(id="test_item_2", vector=[3.0, 4.0]),
-            Item(id="test_item_3", vector=[5.0, 6.0]),
-            Item(id="test_item_3", vector=[7.0, 8.0]),
+            Item(id="test_item_1", vector=[1.0, 2.0], metadata={"key": "value1"}),
+            Item(id="test_item_2", vector=[3.0, 4.0], metadata={"key": "value2"}),
+            Item(id="test_item_3", vector=[5.0, 6.0], metadata={"key": "value3"}),
+            Item(id="test_item_4", vector=[7.0, 8.0], metadata={"key": "value4"}),
         ],
     )
     assert isinstance(upsert_response, UpsertItemBatch.Success)
@@ -713,10 +713,11 @@ def test_delete_deletes_ids(
 
     search_response = vector_index_client.search(index_name, query_vector=[1.0, 2.0], top_k=10)
     assert isinstance(search_response, Search.Success)
-    assert len(search_response.hits) == 3
+    assert len(search_response.hits) == 4
 
     assert search_response.hits == [
-        SearchHit(id="test_item_3", score=23.0),
+        SearchHit(id="test_item_4", score=23.0),
+        SearchHit(id="test_item_3", score=17.0),
         SearchHit(id="test_item_2", score=11.0),
         SearchHit(id="test_item_1", score=5.0),
     ]
@@ -728,13 +729,50 @@ def test_delete_deletes_ids(
 
     search_response = vector_index_client.search(index_name, query_vector=[1.0, 2.0], top_k=10)
     assert isinstance(search_response, Search.Success)
-    assert len(search_response.hits) == 1
+    assert len(search_response.hits) == 2
 
     assert search_response.hits == [
+        SearchHit(id="test_item_4", score=23.0),
         SearchHit(id="test_item_2", score=11.0),
     ]
 
 
+def test_delete_items_by_filter(
+    vector_index_client: PreviewVectorIndexClient,
+    unique_vector_index_name: TUniqueVectorIndexName,
+) -> None:
+    index_name = unique_vector_index_name(vector_index_client)
+    create_response = vector_index_client.create_index(index_name, num_dimensions=2)
+    assert isinstance(create_response, CreateIndex.Success)
+
+    items = [
+        Item(id="test_item_1", vector=[1.0, 1.0], metadata={"key": "value1"}),
+        Item(id="test_item_2", vector=[3.0, 4.0], metadata={"key": "value2"}),
+        Item(id="test_item_3", vector=[5.0, 6.0], metadata={"key": "value1"}),
+        Item(id="test_item_4", vector=[7.0, 8.0], metadata={"key": "value2"}),
+    ]
+    upsert_response = vector_index_client.upsert_item_batch(
+        index_name,
+        items=items,
+    )
+    assert isinstance(upsert_response, UpsertItemBatch.Success)
+
+    sleep(2)
+
+    search_response = vector_index_client.search(index_name, query_vector=[1.0, 1.0], top_k=10)
+    assert isinstance(search_response, Search.Success)
+    assert len(search_response.hits) == 4
+
+    delete_response = vector_index_client.delete_item_batch(index_name, filter=filters.Equals("key", "value1"))
+    assert isinstance(delete_response, DeleteItemBatch.Success)
+
+    sleep(2)
+
+    search_response = vector_index_client.search(index_name, query_vector=[1.0, 1.0], top_k=10)
+    assert isinstance(search_response, Search.Success)
+    assert len(search_response.hits) == 2
+
+
 @pytest.mark.parametrize(
     [
         "get_item_method_name",

tests/momento/vector_index_client/test_data_async.py

@@ -692,7 +692,7 @@ async def test_delete_validates_index_name(vector_index_client_async: PreviewVec
     assert response.error_code == MomentoErrorCode.INVALID_ARGUMENT_ERROR
 
 
-async def test_delete_deletes_ids(
+async def test_delete_items_by_id(
     vector_index_client_async: PreviewVectorIndexClientAsync,
     unique_vector_index_name_async: TUniqueVectorIndexNameAsync,
 ) -> None:
@@ -705,10 +705,10 @@ async def test_delete_deletes_ids(
     upsert_response = await vector_index_client_async.upsert_item_batch(
         index_name,
         items=[
-            Item(id="test_item_1", vector=[1.0, 2.0]),
-            Item(id="test_item_2", vector=[3.0, 4.0]),
-            Item(id="test_item_3", vector=[5.0, 6.0]),
-            Item(id="test_item_3", vector=[7.0, 8.0]),
+            Item(id="test_item_1", vector=[1.0, 2.0], metadata={"key": "value1"}),
+            Item(id="test_item_2", vector=[3.0, 4.0], metadata={"key": "value2"}),
+            Item(id="test_item_3", vector=[5.0, 6.0], metadata={"key": "value3"}),
+            Item(id="test_item_4", vector=[7.0, 8.0], metadata={"key": "value4"}),
         ],
     )
     assert isinstance(upsert_response, UpsertItemBatch.Success)
@@ -717,10 +717,11 @@ async def test_delete_deletes_ids(
 
     search_response = await vector_index_client_async.search(index_name, query_vector=[1.0, 2.0], top_k=10)
     assert isinstance(search_response, Search.Success)
-    assert len(search_response.hits) == 3
+    assert len(search_response.hits) == 4
 
     assert search_response.hits == [
-        SearchHit(id="test_item_3", score=23.0),
+        SearchHit(id="test_item_4", score=23.0),
+        SearchHit(id="test_item_3", score=17.0),
         SearchHit(id="test_item_2", score=11.0),
         SearchHit(id="test_item_1", score=5.0),
     ]
@@ -734,13 +735,52 @@ async def test_delete_deletes_ids(
 
     search_response = await vector_index_client_async.search(index_name, query_vector=[1.0, 2.0], top_k=10)
     assert isinstance(search_response, Search.Success)
-    assert len(search_response.hits) == 1
+    assert len(search_response.hits) == 2
 
     assert search_response.hits == [
+        SearchHit(id="test_item_4", score=23.0),
         SearchHit(id="test_item_2", score=11.0),
     ]
 
 
+async def test_delete_items_by_filter(
+    vector_index_client_async: PreviewVectorIndexClientAsync,
+    unique_vector_index_name_async: TUniqueVectorIndexNameAsync,
+) -> None:
+    index_name = unique_vector_index_name_async(vector_index_client_async)
+    create_response = await vector_index_client_async.create_index(index_name, num_dimensions=2)
+    assert isinstance(create_response, CreateIndex.Success)
+
+    items = [
+        Item(id="test_item_1", vector=[1.0, 1.0], metadata={"key": "value1"}),
+        Item(id="test_item_2", vector=[3.0, 4.0], metadata={"key": "value2"}),
+        Item(id="test_item_3", vector=[5.0, 6.0], metadata={"key": "value1"}),
+        Item(id="test_item_4", vector=[7.0, 8.0], metadata={"key": "value2"}),
+    ]
+    upsert_response = await vector_index_client_async.upsert_item_batch(
+        index_name,
+        items=items,
+    )
+    assert isinstance(upsert_response, UpsertItemBatch.Success)
+
+    await sleep_async(2)
+
+    search_response = await vector_index_client_async.search(index_name, query_vector=[1.0, 1.0], top_k=10)
+    assert isinstance(search_response, Search.Success)
+    assert len(search_response.hits) == 4
+
+    delete_response = await vector_index_client_async.delete_item_batch(
+        index_name, filter=filters.Equals("key", "value1")
+    )
+    assert isinstance(delete_response, DeleteItemBatch.Success)
+
+    await sleep_async(2)
+
+    search_response = await vector_index_client_async.search(index_name, query_vector=[1.0, 1.0], top_k=10)
+    assert isinstance(search_response, Search.Success)
+    assert len(search_response.hits) == 2
+
+
 @pytest.mark.parametrize(
     [
         "get_item_method_name",