feat: suppot search with highlighter in pymilvus

pymilvus/__init__.py

@@ -45,7 +45,14 @@
 from .orm.index import Index
 from .orm.partition import Partition
 from .orm.role import Role
-from .orm.schema import CollectionSchema, FieldSchema, Function, FunctionScore, StructFieldSchema
+from .orm.schema import (
+    CollectionSchema,
+    FieldSchema,
+    Function,
+    FunctionScore,
+    LexicalHighlighter,
+    StructFieldSchema,
+)
 from .orm.utility import (
     create_resource_group,
     create_user,
@@ -96,6 +103,7 @@
     "Hits",
     "Index",
     "IndexType",
+    "LexicalHighlighter",
     "MilvusClient",
     "MilvusException",
     "MilvusUnavailableException",

pymilvus/client/async_grpc_handler.py

@@ -23,7 +23,7 @@
 )
 from pymilvus.grpc_gen import common_pb2, milvus_pb2_grpc
 from pymilvus.grpc_gen import milvus_pb2 as milvus_types
-from pymilvus.orm.schema import Function
+from pymilvus.orm.schema import Function, Highlighter
 from pymilvus.settings import Config
 
 from . import entity_helper, ts_utils, utils
@@ -830,6 +830,7 @@ async def search(
         round_decimal: int = -1,
         timeout: Optional[float] = None,
         ranker: Optional[Function] = None,
+        highlighter: Optional[Highlighter] = None,
         **kwargs,
     ):
         await self.ensure_channel_ready()
@@ -860,6 +861,7 @@ async def search(
             output_fields,
             round_decimal,
             ranker=ranker,
+            highlighter=highlighter,
             **kwargs,
         )
         return await self._execute_search(request, timeout, round_decimal=round_decimal, **kwargs)

pymilvus/client/grpc_handler.py

@@ -23,7 +23,7 @@
 )
 from pymilvus.grpc_gen import common_pb2, milvus_pb2_grpc
 from pymilvus.grpc_gen import milvus_pb2 as milvus_types
-from pymilvus.orm.schema import Function, FunctionScore
+from pymilvus.orm.schema import Function, FunctionScore, Highlighter
 from pymilvus.settings import Config
 
 from . import entity_helper, interceptor, ts_utils, utils
@@ -1010,6 +1010,7 @@ def search(
         round_decimal: int = -1,
         timeout: Optional[float] = None,
         ranker: Union[Function, FunctionScore] = None,
+        highlighter: Optional[Highlighter] = None,
         **kwargs,
     ):
         check_pass_param(
@@ -1034,6 +1035,7 @@ def search(
             output_fields,
             round_decimal,
             ranker=ranker,
+            highlighter=highlighter,
             **kwargs,
         )
         return self._execute_search(request, timeout, round_decimal=round_decimal, **kwargs)

pymilvus/client/prepare.py

@@ -16,6 +16,7 @@
     FieldSchema,
     Function,
     FunctionScore,
+    Highlighter,
     isVectorDataType,
 )
 from pymilvus.orm.types import infer_dtype_by_scalar_data
@@ -1370,6 +1371,7 @@ def search_requests_with_expr(
         output_fields: Optional[List[str]] = None,
         round_decimal: int = -1,
         ranker: Optional[Union[Function, FunctionScore]] = None,
+        highlighter: Optional[Highlighter] = None,
         **kwargs,
     ) -> milvus_types.SearchRequest:
         use_default_consistency = ts_utils.construct_guarantee_ts(collection_name, kwargs)
@@ -1524,6 +1526,9 @@ def search_requests_with_expr(
         elif ranker is not None:
             raise ParamError(message="The search ranker must be a Function or FunctionScore.")
 
+        if highlighter is not None:
+            request.highlighter.CopyFrom(Prepare.highlighter_schema(highlighter))
+
         return request
 
     @classmethod
@@ -1611,6 +1616,16 @@ def common_kv_value(v: Any) -> str:
             return json.dumps(v)
         return str(v)
 
+    @staticmethod
+    def highlighter_schema(highlighter: Highlighter) -> common_types.Highlighter:
+        return common_types.Highlighter(
+            type=highlighter.type,
+            params=[
+                common_types.KeyValuePair(key=str(k), value=Prepare.common_kv_value(v))
+                for k, v in highlighter.params.items()
+            ],
+        )
+
     @staticmethod
     def function_score_schema(function_score: FunctionScore) -> schema_types.FunctionScore:
         functions = [

pymilvus/client/search_result.py

@@ -30,6 +30,7 @@ def __init__(
         all_scores: List[float],
         fields_data: List[schema_pb2.FieldData],
         output_fields: List[str],
+        highlight_results: List[common_pb2.HighlightResult],
         pk_name: str,
     ):
         self.ids = all_pks[start:end]
@@ -93,6 +94,14 @@ def __init__(
             else:
                 msg = f"Unsupported field type: {field_data.type}"
                 raise MilvusException(msg)
+
+        if len(highlight_results) > 0:
+            for i, hit in enumerate(top_k_res):
+                hit["highlight"] = {
+                    result.field_name: list(result.datas[i + start].fragments)
+                    for result in highlight_results
+                }
+
         super().__init__(top_k_res)
 
     def __str__(self) -> str:
@@ -331,9 +340,11 @@ def _parse_search_result_data(
                     all_scores,
                     res.fields_data,
                     res.output_fields,
+                    res.highlight_results,
                     _pk_name,
                 )
             )
+
             nq_thres += topk
         return data
 
@@ -723,6 +734,10 @@ def score(self) -> float:
         """Alias of distance, will be deprecated soon"""
         return self.distance
 
+    @property
+    def highlight(self) -> Dict[str, Any]:
+        return self.data.get("highlight")
+
     @property
     def fields(self) -> Dict[str, Any]:
         """Patch for orm, will be deprecated soon"""

pymilvus/client/types.py

@@ -144,6 +144,11 @@ class FunctionType(IntEnum):
     RERANK = 3
 
 
+class HighlightType(IntEnum):
+    LEXICAL = 0
+    SEMANTIC = 1
+
+
 class RangeType(IntEnum):
     LT = 0  # less than
     LTE = 1  # less than or equal

pymilvus/grpc_gen/common_pb2.pyi

@@ -355,9 +355,6 @@ class ObjectPrivilege(int, metaclass=_enum_type_wrapper.EnumTypeWrapper):
     PrivilegeAddFileResource: _ClassVar[ObjectPrivilege]
     PrivilegeRemoveFileResource: _ClassVar[ObjectPrivilege]
     PrivilegeListFileResources: _ClassVar[ObjectPrivilege]
-    PrivilegeAddCollectionFunction: _ClassVar[ObjectPrivilege]
-    PrivilegeAlterCollectionFunction: _ClassVar[ObjectPrivilege]
-    PrivilegeDropCollectionFunction: _ClassVar[ObjectPrivilege]
     PrivilegeUpdateReplicateConfiguration: _ClassVar[ObjectPrivilege]
 
 class StateCode(int, metaclass=_enum_type_wrapper.EnumTypeWrapper):
@@ -706,9 +703,6 @@ PrivilegeAddCollectionField: ObjectPrivilege
 PrivilegeAddFileResource: ObjectPrivilege
 PrivilegeRemoveFileResource: ObjectPrivilege
 PrivilegeListFileResources: ObjectPrivilege
-PrivilegeAddCollectionFunction: ObjectPrivilege
-PrivilegeAlterCollectionFunction: ObjectPrivilege
-PrivilegeDropCollectionFunction: ObjectPrivilege
 PrivilegeUpdateReplicateConfiguration: ObjectPrivilege
 Initializing: StateCode
 Healthy: StateCode

pymilvus/grpc_gen/milvus_pb2.pyi

@@ -158,7 +158,7 @@ class ListAliasesResponse(_message.Message):
     def __init__(self, status: _Optional[_Union[_common_pb2.Status, _Mapping]] = ..., db_name: _Optional[str] = ..., collection_name: _Optional[str] = ..., aliases: _Optional[_Iterable[str]] = ...) -> None: ...
 
 class CreateCollectionRequest(_message.Message):
-    __slots__ = ("base", "db_name", "collection_name", "schema", "shards_num", "consistency_level", "properties", "num_partitions")
+    __slots__ = ("base", "db_name", "collection_name", "schema", "shards_num", "consistency_level", "properties", "num_partitions", "external_source", "external_spec")
     BASE_FIELD_NUMBER: _ClassVar[int]
     DB_NAME_FIELD_NUMBER: _ClassVar[int]
     COLLECTION_NAME_FIELD_NUMBER: _ClassVar[int]
@@ -167,6 +167,8 @@ class CreateCollectionRequest(_message.Message):
     CONSISTENCY_LEVEL_FIELD_NUMBER: _ClassVar[int]
     PROPERTIES_FIELD_NUMBER: _ClassVar[int]
     NUM_PARTITIONS_FIELD_NUMBER: _ClassVar[int]
+    EXTERNAL_SOURCE_FIELD_NUMBER: _ClassVar[int]
+    EXTERNAL_SPEC_FIELD_NUMBER: _ClassVar[int]
     base: _common_pb2.MsgBase
     db_name: str
     collection_name: str
@@ -175,7 +177,9 @@ class CreateCollectionRequest(_message.Message):
     consistency_level: _common_pb2.ConsistencyLevel
     properties: _containers.RepeatedCompositeFieldContainer[_common_pb2.KeyValuePair]
     num_partitions: int
-    def __init__(self, base: _Optional[_Union[_common_pb2.MsgBase, _Mapping]] = ..., db_name: _Optional[str] = ..., collection_name: _Optional[str] = ..., schema: _Optional[bytes] = ..., shards_num: _Optional[int] = ..., consistency_level: _Optional[_Union[_common_pb2.ConsistencyLevel, str]] = ..., properties: _Optional[_Iterable[_Union[_common_pb2.KeyValuePair, _Mapping]]] = ..., num_partitions: _Optional[int] = ...) -> None: ...
+    external_source: str
+    external_spec: str
+    def __init__(self, base: _Optional[_Union[_common_pb2.MsgBase, _Mapping]] = ..., db_name: _Optional[str] = ..., collection_name: _Optional[str] = ..., schema: _Optional[bytes] = ..., shards_num: _Optional[int] = ..., consistency_level: _Optional[_Union[_common_pb2.ConsistencyLevel, str]] = ..., properties: _Optional[_Iterable[_Union[_common_pb2.KeyValuePair, _Mapping]]] = ..., num_partitions: _Optional[int] = ..., external_source: _Optional[str] = ..., external_spec: _Optional[str] = ...) -> None: ...
 
 class DropCollectionRequest(_message.Message):
     __slots__ = ("base", "db_name", "collection_name")