corvic-engine 0.3.0rc78__cp38-abi3-win_amd64.whl → 0.3.0rc80__cp38-abi3-win_amd64.whl

corvic/orm/__init__.py

@@ -139,7 +139,7 @@ class Base(sa_orm.MappedAsDataclass, sa_orm.DeclarativeBase):
         return cls._updated_at
 
 
-class Org(Base, kw_only=True):
+class Org(SoftDeleteMixin, Base, kw_only=True):
     """An organization is a top-level grouping of resources."""
 
     __tablename__ = "org"

corvic/orm/_soft_delete.py

@@ -81,6 +81,11 @@ class SoftDeleteMixin(sa_orm.MappedAsDataclass):
             return None
         return self._deleted_at.replace(tzinfo=UTC)
 
+    @deleted_at.inplace.expression
+    @classmethod
+    def _deleted_at_at_expression(cls):
+        return cls._deleted_at
+
     def reset_delete(self):
         self._deleted_at = None
 

corvic/system/__init__.py

@@ -29,7 +29,6 @@ from corvic.system.client import Client
 from corvic.system.in_memory_executor import (
     InMemoryExecutionResult,
     InMemoryExecutor,
-    batch_to_proto_struct,  # pyright: ignore[reportDeprecated]
     get_polars_embedding,
     get_polars_embedding_length,
     make_dict_bytes_human_readable,
@@ -96,7 +95,6 @@ __all__ = [
     "UmapDimensionReducer",
     "ValidateFirstExecutor",
     "VectorSimilarityMetric",
-    "batch_to_proto_struct",
     "get_polars_embedding",
     "get_polars_embedding_length",
     "make_dict_bytes_human_readable",

corvic/system/_image_embedder.py

@@ -180,17 +180,53 @@ class Clip(ImageEmbedder):
         )
 
 
+class SigLIP2(Clip):
+    """SigLIP2 image embedder."""
+
+    def _load_models(self):
+        from transformers.models.auto.modeling_auto import AutoModel
+        from transformers.models.auto.processing_auto import AutoProcessor
+        from transformers.models.clip import (
+            CLIPModel,
+            CLIPProcessor,
+        )
+
+        model = cast(
+            CLIPModel,
+            AutoModel.from_pretrained(  # pyright: ignore[reportUnknownMemberType]
+                pretrained_model_name_or_path="google/siglip2-base-patch16-512",
+                revision="a89f5c5093f902bf39d3cd4d81d2c09867f0724b",
+                device_map="auto",
+            ),
+        )
+        processor = cast(
+            CLIPProcessor,
+            AutoProcessor.from_pretrained(  # pyright: ignore[reportUnknownMemberType]
+                pretrained_model_name_or_path="google/siglip2-base-patch16-512",
+                revision="a89f5c5093f902bf39d3cd4d81d2c09867f0724b",
+            ),
+        )
+        return ClipModels(model=model, processor=processor)
+
+
 class CombinedImageEmbedder(ImageEmbedder):
     def __init__(self):
         self._clip_embedder = Clip()
+        self._siglip2_embedder = SigLIP2()
         self._random_embedder = RandomImageEmbedder()
 
     def embed(
         self, context: EmbedImageContext
     ) -> Ok[EmbedImageResult] | InvalidArgumentError | InternalError:
-        if context.model_name == "random":
-            return self._random_embedder.embed(context)
-        return self._clip_embedder.embed(context)
+        match context.model_name:
+            case "random":
+                return self._random_embedder.embed(context)
+            case "clip":
+                return self._clip_embedder.embed(context)
+            case "siglip2":
+                return self._siglip2_embedder.embed(context)
+            case _:
+                return InvalidArgumentError(f"Unknown model name {context.model_name}")
 
     async def aembed(
         self,

corvic/system/in_memory_executor.py

@@ -18,9 +18,7 @@ import polars.selectors as cs
 import pyarrow as pa
 import pyarrow.parquet as pq
 import structlog
-from google.protobuf import json_format, struct_pb2
 from more_itertools import flatten
-from typing_extensions import deprecated
 
 import corvic.system._column_encoding as column_encoding
 from corvic import embed, embedding_metric, op_graph, sql
@@ -99,17 +97,6 @@ def get_polars_embedding(
     )
 
 
-@deprecated("use pa_scalar.batch_to_structs instead")
-def batch_to_proto_struct(batch: pa.RecordBatch) -> list[struct_pb2.Struct]:
-    """Converts a RecordBatch to protobuf Structs safely."""
-    data = batch.to_pylist()
-    structs = [struct_pb2.Struct() for _ in range(len(data))]
-    for idx, datum in enumerate(data):
-        make_dict_bytes_human_readable(datum)
-        json_format.ParseDict(datum, structs[idx])
-    return structs
-
-
 def make_list_bytes_human_readable(data: list[Any]) -> None:
     """Utility function to cleanup list data types.
 

{corvic_engine-0.3.0rc78.dist-info → corvic_engine-0.3.0rc80.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: corvic-engine
-Version: 0.3.0rc78
+Version: 0.3.0rc80
 Classifier: Environment :: Console
 Classifier: License :: Other/Proprietary License
 Classifier: Programming Language :: Python :: Implementation :: CPython

{corvic_engine-0.3.0rc78.dist-info → corvic_engine-0.3.0rc80.dist-info}/RECORD

@@ -11,7 +11,7 @@ corvic/embedding_metric/__init__.py,sha256=8a-QKSQNbiksueHk5LdkugjZr6wasP4ff8A-J
 corvic/embedding_metric/embeddings.py,sha256=XCiMzoGdRSmCOJnBDnxm3xlU0L_vrXwUxEjwdMv1FMI,14036
 corvic/embedding_metric/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 corvic/engine/__init__.py,sha256=XL4Vg7rNcBi29ccVelpeFizR9oJtGYXDn84W9zok9d4,975
-corvic/engine/_native.pyd,sha256=BnJ0aIVFyA4rV9MW1QJgQzI0sGDfe1Gbsl5lyG3C4N8,438272
+corvic/engine/_native.pyd,sha256=V8crGQ6LIdFBAZ_TZYIZbv_ULgLHuF8Kll7xPX7ELGE,438272
 corvic/engine/_native.pyi,sha256=KYMPtvXqHZ-jMgZohLf4se3rr-rBpCihmjANcr6s8ag,1390
 corvic/engine/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 corvic/eorm/__init__.py,sha256=b4dFnu4fW7wj3Y0SMNVXOp8KoKOp_HAL4GyDN-S8fOY,13704
@@ -42,9 +42,9 @@ corvic/op_graph/row_filters/__init__.py,sha256=1sibH_kLw7t_9bpRccnEGWqdCiN0VaUh9
 corvic/op_graph/row_filters/_jsonlogic.py,sha256=0UdwOZmIGp4yuExHM3qqAnJYmcGv7iuc3vLub3GD-9Y,7685
 corvic/op_graph/row_filters/_row_filters.py,sha256=p3O7tJbLsy65Vs7shAiDjpdM4RzYA4-fyzwskt15pPk,9469
 corvic/op_graph/sample_strategy.py,sha256=DrbtJ3ORkIRfyIE_FdlOh_UMnCW_K9jL1LeonVYb3bU,3007
-corvic/orm/__init__.py,sha256=YGBnwYtPGop9-MWHTMetILlwse9FStTDcme28VCrHHE,9006
+corvic/orm/__init__.py,sha256=FJIFVj8zOOTz0YR0Moq_lgvnb5yLpWnhuLYNTzZw42o,9023
 corvic/orm/_proto_columns.py,sha256=tcOu92UjFJFYZLasS6sWJQBDRK26yrnmpTii_LDY4iw,913
-corvic/orm/_soft_delete.py,sha256=TolZ0j-1yL8yQW39Obe0c0af0jyT3sluk1oInqzxmGU,7919
+corvic/orm/_soft_delete.py,sha256=PpNrWAQaGZhfLOYKgdMvUUcwHyqVtvyiWEg_pratk4o,8048
 corvic/orm/errors.py,sha256=uFhFXpVG6pby1lndJZHGHxv3Y0Fbt0RiaZ-CqDfuY1o,545
 corvic/orm/func/__init__.py,sha256=X47bbG7G-rDGmRkEGMq4Vn7mPKePdx724xQIwd_pUc0,471
 corvic/orm/func/utc_func.py,sha256=-FC6w9wBWXejMv1AICT2Gg7tdkSo7gqL2dFT-YKPGQ4,4518
@@ -68,15 +68,15 @@ corvic/result/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 corvic/sql/__init__.py,sha256=kZ1a39KVZ08P8Bg6XuXDLD_dTQX0k620u4nwxZF4SnY,303
 corvic/sql/parse_ops.py,sha256=5jm2CHycTqzdu9apXTgcvwwyBVpjf7n5waqKfIa84JA,29940
 corvic/sql/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-corvic/system/__init__.py,sha256=U28LyDwpirtG0WDXqE6dzkx8PipvY2rtZSex03C7xlY,2792
+corvic/system/__init__.py,sha256=Qddpk3SoHgoiBJyr9w2bPkp477NZv5MDR8701A17nCY,2697
 corvic/system/_column_encoding.py,sha256=feSWIv4vKstVq-aavWPk53YucUiq7rZvuyofqTicXBE,7574
 corvic/system/_dimension_reduction.py,sha256=2tg5SIHY4P480DJQj6PSjW1VgAJCAVJAH8D3BY-ZYXA,2964
 corvic/system/_embedder.py,sha256=Jr-f4rwNdFRZuMaW-prPCDtjNkkNZVn7zCGMNi_hEYw,5424
-corvic/system/_image_embedder.py,sha256=i595n22L8rmiZQ0sT3JTu_RBJrcU_5Vl95K2HPJtGGw,10447
+corvic/system/_image_embedder.py,sha256=jTuSjxwh8n1peSidKCQFV_LLJDH4XelX9NV4uvGBBtw,11829
 corvic/system/_planner.py,sha256=ecL-HW8PVz5eWJ1Ktf-RAD2IdZkHu3GuBtXdqElo4ts,8210
 corvic/system/_text_embedder.py,sha256=NDi--3_tzwIWVImjhFWmp8dHmydGGXNu6GYH8qODsIc,4000
 corvic/system/client.py,sha256=JcA-fPraqDkl9f8BiClS0qeGY6wzKcEDPymutWrJo54,812
-corvic/system/in_memory_executor.py,sha256=Ds2A_RLoQeGVMpvKDz4IXTq0r5SgPcpMKHYm8vW6R7k,66622
+corvic/system/in_memory_executor.py,sha256=KzX8baGFROpv-5hqVC1qOMf7AlPD22ffYtyBesV-W-M,66076
 corvic/system/op_graph_executor.py,sha256=tSKro-yb_y1_sgajZluM-6FCvDqO1oUPsiWw2DRxyMQ,3641
 corvic/system/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 corvic/system/staging.py,sha256=K5P5moiuAMfPx7lxK4mArxeURBwKoyB6x9HGu9JJ16E,1846
@@ -94,9 +94,9 @@ corvic/version/__init__.py,sha256=JlkRLvKXsu3zIxhdynO_0Ub5NfQOvGjfwCRkNnaOu9U,11
 corvic/version/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 corvic/well_known_types/__init__.py,sha256=Btbeqieik2AcmijeOXeqBptzueBpgNitvH9J5VNm12w,1289
 corvic/well_known_types/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-corvic_engine-0.3.0rc78.dist-info/METADATA,sha256=h3-wxx4uit-9adJbG11D8-YSOvmTM_4ZSRK6bFYjQpc,1814
-corvic_engine-0.3.0rc78.dist-info/WHEEL,sha256=qo08K5WTt1v9liGoFGXfI182ciKs5521XAErJtzFynQ,94
-corvic_engine-0.3.0rc78.dist-info/licenses/LICENSE,sha256=DSS1OD0oIgssKOmAzkMRBv5jvvVuZQbrIv8lpl9DXY8,1035
+corvic_engine-0.3.0rc80.dist-info/METADATA,sha256=h5FHy0y1OwCZu1dkPk1gY1Fb60l5z6wMe5DV9qR3aH8,1814
+corvic_engine-0.3.0rc80.dist-info/WHEEL,sha256=qo08K5WTt1v9liGoFGXfI182ciKs5521XAErJtzFynQ,94
+corvic_engine-0.3.0rc80.dist-info/licenses/LICENSE,sha256=DSS1OD0oIgssKOmAzkMRBv5jvvVuZQbrIv8lpl9DXY8,1035
 corvic_generated/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 corvic_generated/algorithm/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 corvic_generated/algorithm/graph/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -207,4 +207,4 @@ corvic_generated/status/v1/service_pb2.py,sha256=CKXPX2ahq8O4cFhPpt6wo6l--6VZcgj
 corvic_generated/status/v1/service_pb2.pyi,sha256=iXLR2FOKQJpBgvBzpD2kVwcYOCksP2aRwK4JYaI9CBw,558
 corvic_generated/status/v1/service_pb2_grpc.py,sha256=y-a5ldrphWlNJW-yKswyjNmXokK4-5bbEEfczjagJHo,2736
 corvic_generated/status/v1/service_pb2_grpc.pyi,sha256=OoAnaZ64FD0UTzPoRhYvQU8ecoilhHj3ySjSfHbVDaU,1501
-corvic_engine-0.3.0rc78.dist-info/RECORD,,
+corvic_engine-0.3.0rc80.dist-info/RECORD,,