Add support for multi target distances

.github/workflows/main.yaml

@@ -18,14 +18,14 @@ concurrency:
   cancel-in-progress: true
 
 env:
-  WEAVIATE_125: 1.25.34
   WEAVIATE_126: 1.26.17
   WEAVIATE_127: 1.27.27
   WEAVIATE_128: 1.28.16
   WEAVIATE_129: 1.29.9
   WEAVIATE_130: 1.30.12
   WEAVIATE_131: 1.31.5
   WEAVIATE_132: 1.32.0-rc.1-098c594
+  WEAVIATE_133: 1.33.0-dev-472b1e1.amd64
 
 jobs:
   lint-and-format:
@@ -278,7 +278,7 @@ jobs:
       fail-fast: false
       matrix:
         version: [
-          { extra: "1.62.0", server: $WEAVIATE_125},
+          { extra: "1.62.0", server: $WEAVIATE_133},
           { extra: "1.66.2", server: $WEAVIATE_126},
           { extra: "1.70.0", server: $WEAVIATE_127},
           { extra: "1.59.5", server: $WEAVIATE_128},

integration/test_vectors.py

@@ -684,6 +684,44 @@ def test_same_target_vector_multiple_input_combinations(
     assert sorted([obj.uuid for obj in objs]) == sorted([uuid2, uuid1])
 
 
+@pytest.mark.parametrize(
+    "near_vector,target_vector",
+    [
+        ({"first": [0, 0], "second": [[0, 1, 0], [0, 0, 1]]}, ["first", "second"]),
+    ],
+)
+def test_multi_target_vector_return(
+    collection_factory: CollectionFactory,
+    near_vector: Dict[str, Union[Sequence[float], Sequence[Sequence[float]], _ListOfVectorsQuery]],
+    target_vector: List[str],
+) -> None:
+    collection = collection_factory(
+        properties=[],
+        vector_config=[
+            wvc.config.Configure.Vectors.self_provided(name="first"),
+            wvc.config.Configure.Vectors.self_provided(name="second"),
+        ],
+    )
+
+    if collection._connection._weaviate_version.is_lower_than(1, 33, 0):
+        pytest.skip("Multi vector per target is not supported in versions lower than 1.33.0")
+
+    uuid1 = collection.data.insert({}, vector={"first": [1, 0], "second": [0, 1, 0]})
+    uuid2 = collection.data.insert({}, vector={"first": [0, 1], "second": [1, 0, 0]})
+    objs = collection.query.near_vector(
+        near_vector,
+        target_vector=wvc.query.TargetVectors.sum(target_vector),
+        return_metadata=wvc.query.MetadataQuery.full(),
+    ).objects
+
+    assert objs[0].uuid == uuid1
+    assert objs[1].uuid == uuid2
+    assert objs[0].metadata.distance == 2.0
+    assert objs[0].metadata.multi_target_distances == {"first": [1.0], "second": [0.0, 1.0]}
+    assert objs[1].metadata.distance == 3.0
+    assert objs[1].metadata.multi_target_distances == {"first": [1.0], "second": [1.0, 1.0]}
+
+
 def test_deprecated_syntax(collection_factory: CollectionFactory):
     dummy = collection_factory("dummy")
     if dummy._connection._weaviate_version.is_at_least(

weaviate/collections/classes/internal.py

@@ -69,6 +69,7 @@ class MetadataReturn:
     creation_time: Optional[datetime.datetime] = None
     last_update_time: Optional[datetime.datetime] = None
     distance: Optional[float] = None
+    multi_target_distances: Optional[Dict[str, List[float]]] = None
     certainty: Optional[float] = None
     score: Optional[float] = None
     explain_score: Optional[str] = None
@@ -86,6 +87,7 @@ def _is_empty(self) -> bool:
                 self.explain_score is None,
                 self.is_consistent is None,
                 self.rerank_score is None,
+                self.multi_target_distances is None,
             ]
         )
 

weaviate/collections/queries/base_executor.py

@@ -112,6 +112,7 @@ def __extract_metadata_for_object(
         meta = MetadataReturn(
             distance=add_props.distance if add_props.distance_present else None,
             certainty=add_props.certainty if add_props.certainty_present else None,
+            multi_target_distances=self.__extract_multi_target_distances(add_props),
             creation_time=(
                 self.__retrieve_timestamp(add_props.creation_time_unix)
                 if add_props.creation_time_unix_present
@@ -129,6 +130,18 @@ def __extract_metadata_for_object(
         )
         return meta
 
+    def __extract_multi_target_distances(
+        self,
+        add_props: "search_get_pb2.MetadataResult",
+    ) -> Optional[Dict[str, List[float]]]:
+        if add_props.multi_target_distances is None or len(add_props.multi_target_distances) == 0:
+            return None
+
+        distances: Dict[str, List[float]] = {}
+        for key, value in add_props.multi_target_distances.items():
+            distances[key] = _ByteOps.decode_float32s(value.multi_target_distances)
+        return distances
+
     def __extract_metadata_for_group_by_object(
         self,
         add_props: "search_get_pb2.MetadataResult",

weaviate/collections/queries/near_vector/query/executor.py

@@ -359,7 +359,7 @@ def resp(
         ) -> QuerySearchReturnType[Properties, References, TProperties, TReferences]:
             return cast(
                 Any,
-                self._result_to_generative_return(
+                self._result_to_query_return(
                     res,
                     _QueryOptions.from_input(
                         return_metadata,

weaviate/proto/v1/v1613/v1/file_replication_pb2.pyi

@@ -0,0 +1,116 @@
+from google.protobuf.internal import containers as _containers
+from google.protobuf.internal import enum_type_wrapper as _enum_type_wrapper
+from google.protobuf import descriptor as _descriptor
+from google.protobuf import message as _message
+from typing import ClassVar as _ClassVar, Iterable as _Iterable, Optional as _Optional, Union as _Union
+
+DESCRIPTOR: _descriptor.FileDescriptor
+
+class CompressionType(int, metaclass=_enum_type_wrapper.EnumTypeWrapper):
+    __slots__ = ()
+    COMPRESSION_TYPE_UNSPECIFIED: _ClassVar[CompressionType]
+    COMPRESSION_TYPE_GZIP: _ClassVar[CompressionType]
+    COMPRESSION_TYPE_ZLIB: _ClassVar[CompressionType]
+    COMPRESSION_TYPE_DEFLATE: _ClassVar[CompressionType]
+COMPRESSION_TYPE_UNSPECIFIED: CompressionType
+COMPRESSION_TYPE_GZIP: CompressionType
+COMPRESSION_TYPE_ZLIB: CompressionType
+COMPRESSION_TYPE_DEFLATE: CompressionType
+
+class PauseFileActivityRequest(_message.Message):
+    __slots__ = ("index_name", "shard_name", "schema_version")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    SCHEMA_VERSION_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    schema_version: int
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ..., schema_version: _Optional[int] = ...) -> None: ...
+
+class PauseFileActivityResponse(_message.Message):
+    __slots__ = ("index_name", "shard_name")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ...) -> None: ...
+
+class ResumeFileActivityRequest(_message.Message):
+    __slots__ = ("index_name", "shard_name")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ...) -> None: ...
+
+class ResumeFileActivityResponse(_message.Message):
+    __slots__ = ("index_name", "shard_name")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ...) -> None: ...
+
+class ListFilesRequest(_message.Message):
+    __slots__ = ("index_name", "shard_name")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ...) -> None: ...
+
+class ListFilesResponse(_message.Message):
+    __slots__ = ("index_name", "shard_name", "file_names")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    FILE_NAMES_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    file_names: _containers.RepeatedScalarFieldContainer[str]
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ..., file_names: _Optional[_Iterable[str]] = ...) -> None: ...
+
+class GetFileMetadataRequest(_message.Message):
+    __slots__ = ("index_name", "shard_name", "file_name")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    FILE_NAME_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    file_name: str
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ..., file_name: _Optional[str] = ...) -> None: ...
+
+class FileMetadata(_message.Message):
+    __slots__ = ("index_name", "shard_name", "file_name", "size", "crc32")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    FILE_NAME_FIELD_NUMBER: _ClassVar[int]
+    SIZE_FIELD_NUMBER: _ClassVar[int]
+    CRC32_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    file_name: str
+    size: int
+    crc32: int
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ..., file_name: _Optional[str] = ..., size: _Optional[int] = ..., crc32: _Optional[int] = ...) -> None: ...
+
+class GetFileRequest(_message.Message):
+    __slots__ = ("index_name", "shard_name", "file_name", "compression")
+    INDEX_NAME_FIELD_NUMBER: _ClassVar[int]
+    SHARD_NAME_FIELD_NUMBER: _ClassVar[int]
+    FILE_NAME_FIELD_NUMBER: _ClassVar[int]
+    COMPRESSION_FIELD_NUMBER: _ClassVar[int]
+    index_name: str
+    shard_name: str
+    file_name: str
+    compression: CompressionType
+    def __init__(self, index_name: _Optional[str] = ..., shard_name: _Optional[str] = ..., file_name: _Optional[str] = ..., compression: _Optional[_Union[CompressionType, str]] = ...) -> None: ...
+
+class FileChunk(_message.Message):
+    __slots__ = ("offset", "data", "eof")
+    OFFSET_FIELD_NUMBER: _ClassVar[int]
+    DATA_FIELD_NUMBER: _ClassVar[int]
+    EOF_FIELD_NUMBER: _ClassVar[int]
+    offset: int
+    data: bytes
+    eof: bool
+    def __init__(self, offset: _Optional[int] = ..., data: _Optional[bytes] = ..., eof: bool = ...) -> None: ...