rust/lancedb
1
0
Fork 0
mirror of https://github.com/lancedb/lancedb.git synced 2026-09-03 20:18:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: support nested blob function signatures

Browse Source
This commit is contained in:
Xuanwo
2026-09-01 16:54:15 +08:00
parent 5cbd979455
commit d383c37967
3 changed files with 446 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/python/lancedb/functions.py
+97 -10
View File
@@ -521,6 +521,12 @@ class RefreshColumnResult(_RemoteValue):
_FUNCTION_NAME = re.compile(r"^[A-Za-z_][A-Za-z0-9_.-]*$")
_FUNCTION_BLOB_V2_TYPE = "blob_v2"
_ARROW_EXTENSION_NAME_KEY = "ARROW:extension:name"
_BLOB_V2_EXTENSION_NAME = "lance.blob.v2"
_NESTED_BLOB_COLLECTION_ERROR = (
"unsupported Arrow type for Function signature: Blob v2 fields nested under "
"collection types are not supported"
)
_GRAMMAR_PRIMITIVES = (
@@ -591,6 +597,19 @@ def _validate_exact_arrow_field(field: pa.Field) -> None:
"unsupported Arrow type for Function signature: lance.blob.v2 "
f"requires a supported Blob storage layout, got {field}"
)
metadata = {
(key.decode() if isinstance(key, bytes) else key): (
value.decode() if isinstance(value, bytes) else value
)
for key, value in (field.metadata or {}).items()
}
if metadata and metadata != {
_ARROW_EXTENSION_NAME_KEY: _BLOB_V2_EXTENSION_NAME
}:
raise TypeError(
"unsupported Arrow type for Function signature: lance.blob.v2 "
"field metadata must contain only its canonical extension marker"
)
elif field.metadata:
raise TypeError(
"unsupported Arrow type for Function signature: field metadata "
@@ -655,23 +674,84 @@ def _canonical_arrow_field(field: pa.Field) -> str:
return _canonical_arrow_type(field.type)
def _exact_arrow_field(field: pa.Field) -> dict[str, Any]:
def _blob_storage_type(field: pa.Field) -> pa.DataType:
data_type = field.type
if isinstance(data_type, pa.ExtensionType):
return data_type.storage_type
return data_type
def _exact_blob_storage_type(field: pa.Field) -> dict[str, Any]:
storage = _blob_storage_type(field)
if not pa.types.is_struct(storage):
raise TypeError(
"unsupported Arrow type for Function signature: lance.blob.v2 "
"requires struct storage"
)
return {
"type": "struct",
"fields": [
{
"name": child.name,
"nullable": child.nullable,
"type": (
{"type": "large_binary"}
if pa.types.is_large_binary(child.type)
else _exact_arrow_type(child.type)
),
}
for child in storage
],
}
def _data_type_has_blob_v2(data_type: pa.DataType) -> bool:
if pa.types.is_struct(data_type):
return any(
_is_blob_v2_field(field) or _data_type_has_blob_v2(field.type)
for field in data_type
)
if (
pa.types.is_list(data_type)
or pa.types.is_large_list(data_type)
or pa.types.is_fixed_size_list(data_type)
):
field = data_type.value_field
return _is_blob_v2_field(field) or _data_type_has_blob_v2(field.type)
if pa.types.is_map(data_type):
return any(
_is_blob_v2_field(field) or _data_type_has_blob_v2(field.type)
for field in (data_type.key_field, data_type.item_field)
)
return False
def _exact_arrow_field(
field: pa.Field, *, inside_collection: bool = False
) -> dict[str, Any]:
_validate_exact_arrow_field(field)
if _is_blob_v2_field(field):
raise TypeError(
"unsupported Arrow type for Function signature: nested Blob v2 "
"fields are not supported; declare Blob parameters or named result "
"fields directly"
)
if inside_collection:
raise TypeError(_NESTED_BLOB_COLLECTION_ERROR)
return {
"name": field.name,
"nullable": field.nullable,
"type": _exact_blob_storage_type(field),
"metadata": {
_ARROW_EXTENSION_NAME_KEY: _BLOB_V2_EXTENSION_NAME,
},
}
value = {
"name": field.name,
"nullable": field.nullable,
"type": _exact_arrow_type(field.type),
"type": _exact_arrow_type(field.type, inside_collection=inside_collection),
}
return value
def _exact_arrow_type(data_type: pa.DataType) -> dict[str, Any]:
def _exact_arrow_type(
data_type: pa.DataType, *, inside_collection: bool = False
) -> dict[str, Any]:
for candidate, name in _GRAMMAR_PRIMITIVES:
if data_type == candidate:
return {"type": name}
@@ -685,7 +765,10 @@ def _exact_arrow_type(data_type: pa.DataType) -> dict[str, Any]:
)
return {
"type": "struct",
"fields": [_exact_arrow_field(field) for field in fields],
"fields": [
_exact_arrow_field(field, inside_collection=inside_collection)
for field in fields
],
}
if (
pa.types.is_list(data_type)
@@ -710,11 +793,15 @@ def _exact_arrow_type(data_type: pa.DataType) -> dict[str, Any]:
if pa.types.is_large_list(data_type)
else "fixed_size_list"
),
"fields": [_exact_arrow_field(data_type.value_field)],
"fields": [
_exact_arrow_field(data_type.value_field, inside_collection=True)
],
}
if pa.types.is_fixed_size_list(data_type):
value["length"] = data_type.list_size
return value
if pa.types.is_map(data_type) and _data_type_has_blob_v2(data_type):
raise TypeError(_NESTED_BLOB_COLLECTION_ERROR)
raise TypeError(f"unsupported Arrow type for Function signature: {data_type}")
python/python/tests/test_first_class_function_slice2.py
+161 -16
View File
@@ -668,6 +668,167 @@ def test_blob_fields_use_the_scalar_function_semantic_type():
assert signature.output.arrow_type == "blob_v2"
def test_whole_named_struct_function_can_include_a_blob_result_field():
@udf(
input_schema=pa.schema([lancedb.blob("image", nullable=False)]),
output_schema=pa.field(
"payload",
pa.struct(
[
pa.field("mime_type", pa.string(), nullable=False),
lancedb.blob("image", nullable=False),
]
),
nullable=False,
),
)
def inspect_blob(image):
return {"mime_type": "image/png", "image": image}
output = inspect_blob.registration_request.signature.output
assert output.kind == "named_struct"
assert [(field.name, field.arrow_type) for field in output.fields] == [
("mime_type", "utf8"),
("image", "blob_v2"),
]
def test_struct_blob_signature_fields_preserve_exact_metadata_and_nullability():
nested_input = pa.field(
"payload",
pa.struct(
[
pa.field("mime_type", pa.string(), nullable=False),
pa.field(
"nested",
pa.struct([lancedb.blob("image", nullable=True)]),
nullable=True,
),
]
),
nullable=True,
)
nested_output = pa.field(
"result",
pa.struct(
[
pa.field("mime_type", pa.string(), nullable=False),
pa.field(
"nested",
pa.struct([lancedb.blob("image", nullable=True)]),
nullable=False,
),
]
),
nullable=False,
)
@udf(input_schema=pa.schema([nested_input]), output_schema=nested_output)
def copy_payload(payload):
return payload
signature = copy_payload.registration_request.signature
input_type = json.loads(signature.inputs[0].arrow_type)
assert input_type["fields"][1]["nullable"] is True
input_blob = input_type["fields"][1]["type"]["fields"][0]
assert input_blob["nullable"] is True
assert input_blob["metadata"] == {"ARROW:extension:name": "lance.blob.v2"}
assert signature.output.kind == "named_struct"
nested_result = next(
field for field in signature.output.fields if field.name == "nested"
)
output_type = json.loads(nested_result.arrow_type)
output_blob = output_type["fields"][0]
assert output_blob["nullable"] is True
assert output_blob["metadata"] == {"ARROW:extension:name": "lance.blob.v2"}
def test_struct_blob_signature_supports_multiple_struct_levels():
recursive = pa.field(
"value",
pa.struct(
[
pa.field(
"level_1",
pa.struct(
[
pa.field(
"level_2",
pa.struct([lancedb.blob("image", nullable=False)]),
nullable=False,
)
]
),
nullable=False,
)
]
),
nullable=False,
)
@udf(
input_schema=pa.schema([recursive]),
output_schema=pa.field("size", pa.int64(), nullable=False),
)
def blob_size(value):
return len(value["level_1"]["level_2"]["image"])
encoded = json.loads(blob_size.registration_request.signature.inputs[0].arrow_type)
blob = encoded["fields"][0]["type"]["fields"][0]["type"]["fields"][0]
assert blob["metadata"]["ARROW:extension:name"] == "lance.blob.v2"
@pytest.mark.parametrize(
"data_type",
[
pa.list_(lancedb.blob("item", nullable=False)),
pa.large_list(lancedb.blob("item", nullable=False)),
pa.list_(lancedb.blob("item", nullable=False), 2),
pa.map_(pa.string(), lancedb.blob("value", nullable=False).type),
],
)
def test_blob_signature_rejects_collection_ancestors(data_type):
with pytest.raises(
TypeError,
match="Blob v2 fields nested under collection types are not supported",
):
@udf(
input_schema=pa.schema([pa.field("value", data_type, nullable=False)]),
output_schema=pa.field("size", pa.int64(), nullable=False),
)
def blob_size(value):
return len(value)
def test_blob_signature_rejects_collection_below_a_struct():
nested = pa.field(
"value",
pa.struct(
[
pa.field(
"images",
pa.list_(lancedb.blob("item", nullable=False)),
nullable=False,
)
]
),
nullable=False,
)
with pytest.raises(
TypeError,
match="Blob v2 fields nested under collection types are not supported",
):
@udf(
input_schema=pa.schema([nested]),
output_schema=pa.field("size", pa.int64(), nullable=False),
)
def blob_size(value):
return len(value["images"])
def test_named_struct_function_can_include_a_blob_result_field():
@udf(
input_schema=pa.schema([lancedb.blob("image", nullable=False)]),
@@ -729,22 +890,6 @@ def test_blob_marker_rejects_invalid_storage_layout():
return len(image)
def test_nested_blob_signature_field_has_a_clear_error():
nested = pa.field(
"value",
pa.struct([lancedb.blob("image", nullable=False)]),
nullable=False,
)
with pytest.raises(TypeError, match="nested Blob v2 fields are not supported"):
@udf(
input_schema=pa.schema([nested]),
output_schema=pa.field("size", pa.int64(), nullable=False),
)
def blob_size(value):
return len(value["image"])
def test_nested_non_blob_extension_is_not_silently_unwrapped():
class TestExtension(pa.ExtensionType):
def __init__(self):