feat: support nullable scalar Function declarations (#5030)

Function declarations currently reject nullable scalar returns. Accept
`Optional[T]`, `T | None`, and explicit nullable Arrow fields,
preserving `nullable: true` in registration signatures, JSON round
trips, and remote `add_columns` requests.

The Rust planner accepts nullable scalar outputs, and binding readback
recognizes the server-owned assignment marker for a single nullable
scalar result. Clients do not allocate that marker. Non-nullable whole
results, whole structs, and Blob v2 retain their existing assignment
restrictions.

Validated with 127 Python tests, 30 documentation tests (1 skipped), and
seven focused Rust tests covering planning, binding readback, rejection
boundaries, and the remote request body. The final condition-collapse
change addresses strict Clippy; its delta check and CI are pending.
Python tests use a source overlay on a retained native extension, and
Rust remote coverage uses HTTP mocks.

Server support and end-to-end validation remain separate. This PR does
not claim that server publication distinguishes computed NULL values
from pending rows.
This commit is contained in:
Xuanwo
2026-11-14 20:59:20 +08:00
committed by GitHub
parent 005b33fa8b
commit e5a5b67212
8 changed files with 517 additions and 50 deletions
+21 -7
View File
@@ -154,13 +154,27 @@ def test_function_output_nullability():
("note", True),
]
# A scalar output, and a struct declared as one nullable field, may not be.
for nullable in (
pa.field("result", pa.float64(), nullable=True),
pa.field("result", pa.struct(fields), nullable=True),
):
with pytest.raises(ValueError, match="non-nullable"):
function(input_schema=inputs, output_schema=nullable)(describe)
# A scalar output may be nullable.
@function(input_schema=inputs)
def maybe_double(value: float) -> float | None:
return None if value == 0 else value * 2
output = maybe_double.registration_request.signature.output
assert (output.kind, output.arrow_type, output.nullable) == (
"scalar",
"float64",
True,
)
assert json.loads(maybe_double.registration_request.signature.to_canonical_json())[
"output"
] == {"kind": "scalar", "arrow_type": "float64", "nullable": True}
# A struct declared as one nullable field may not be.
with pytest.raises(ValueError, match="non-nullable"):
function(
input_schema=inputs,
output_schema=pa.field("result", pa.struct(fields), nullable=True),
)(describe)
def test_function_registration_is_remote(tmp_path):
+6 -5
View File
@@ -295,11 +295,12 @@ extract_frame_version = db.create_function_async(extract_frame).wait(
```
</CodeGroup>
An empty byte string is a filled result. A scalar output cannot be nullable,
because a physical NULL marks an unfilled computed row. To return a nullable
blob as a value, use a named-struct output schema with a nullable blob field;
the struct itself records whether the result has been filled. Read a nested
blob with `table.fetch_blobs("payload.image", hits)`.
An empty byte string is a filled result. A top-level Blob v2 output cannot be
nullable; to return a nullable blob as a value, use a named-struct output
schema with a nullable blob field. The struct itself records whether the
result has been filled. A nullable non-Blob scalar output is declared with
`Optional[T]` or `T | None`; see [Functions](/management/function#define).
Read a nested blob with `table.fetch_blobs("payload.image", hits)`.
A list or large-list output may also contain Blob v2 elements; inputs and
fixed-size lists do not support this layout. List-nested values are queryable
+37 -5
View File
@@ -66,11 +66,43 @@ your process, without the database, so unit tests stay ordinary tests:
Where annotations cannot express the Arrow type precisely, pass `input_schema`
and `output_schema` together: one without the other is rejected. A scalar
output must be non-nullable, because a physical NULL is what marks a
computed-column row as unfilled. A named-struct output may declare nullable
fields: whether a row has a result at all is the struct's own validity, which
Enterprise keeps in a separate internal column when it expands the result into
sibling columns, so a null field is a value rather than a missing row.
output is non-nullable by default. Annotate the return as `Optional[T]` or
`T | None`, or pass a nullable output field, to declare a nullable scalar:
```python
import pyarrow as pa
from lancedb import function
@function(
input_schema=pa.schema([pa.field("value", pa.int64(), nullable=False)]),
output_schema=pa.field("result", pa.int64(), nullable=True),
)
def maybe_double(value: int) -> int | None:
return None if value == 0 else value * 2
```
The signature sent in the registration request then carries the nullable
scalar:
```json
{
"inputs": [{"name": "value", "arrow_type": "int64", "nullable": false}],
"output": {"kind": "scalar", "arrow_type": "int64", "nullable": true}
}
```
<Note>
Nullable scalar outputs require an Enterprise server that supports them;
older servers reject the registration. On a supporting server, a computed
`NULL` counts as a filled result, independently of its stored value.
</Note>
A named-struct output may declare nullable fields: whether a row has a result
at all is the struct's own validity, which Enterprise keeps in a separate
internal column when it expands the result into sibling columns, so a null
field is a value rather than a missing row.
## Connect and create a table {#table}
+22 -13
View File
@@ -312,7 +312,11 @@ class FunctionResultField(_OpenRemoteValue):
class FunctionOutput(_OpenRemoteValue):
"""Scalar or ordered named-struct output; unknown kinds remain decodable."""
"""Scalar or ordered named-struct output; unknown kinds remain decodable.
A scalar output may be nullable. A named-struct output has no top-level
nullability; its fields carry their own flags.
"""
kind: str
arrow_type: Optional[str] = None
@@ -1399,12 +1403,12 @@ def _function_output(output: pa.DataType | pa.Field | pa.Schema) -> FunctionOutp
"output_schema must be a PyArrow DataType, Field, or Schema"
)
_validate_exact_arrow_field(field)
if field.nullable:
raise ValueError("Function output must be non-nullable")
if field.nullable and _is_blob_v2_field(field):
raise ValueError("Blob v2 Function output must be non-nullable")
return FunctionOutput(
kind="scalar",
arrow_type=_canonical_arrow_field(field, collection_blobs=True),
nullable=False,
nullable=field.nullable,
)
if not fields:
@@ -1504,9 +1508,9 @@ def _infer_signature(
output = _function_output(output_schema)
else:
output_type, output_nullable = _annotation_type(annotations["return"])
if output_nullable:
raise ValueError("Function output must be non-nullable")
output = _function_output(pa.field("result", output_type, nullable=False))
output = _function_output(
pa.field("result", output_type, nullable=output_nullable)
)
if inputs is not None:
return FunctionSignature(inputs=inputs, output=output)
@@ -2650,9 +2654,13 @@ def function(
``None`` for a null value. PyArrow fields are nullable unless declared
otherwise, so most columns need ``Optional``. For Arrow types annotations
cannot express precisely, pass ``input_schema``, ``output_schema``, or
both; the one omitted is inferred. Scalar outputs must be non-nullable.
Every named-struct field may be nullable; Enterprise preserves the
struct's validity when the result is expanded into sibling columns.
both; the one omitted is inferred. A scalar output is non-nullable by
default; annotate the return as ``Optional[T]`` or ``T | None``, or pass a
nullable ``output_schema`` field, to declare a nullable scalar. The
deployment must accept nullable scalar outputs; an older server rejects
the registration. Every named-struct field may be nullable; Enterprise
preserves the struct's validity when the result is expanded into sibling
columns.
**Large values.** A ``bytes`` parameter receives the whole value, which
is held in memory with its row and counts against the server's per-row
@@ -2757,9 +2765,10 @@ def function(
Required for batch annotations. When omitted, inputs are inferred from
the parameter annotations.
output_schema : pyarrow.DataType, pyarrow.Field, or pyarrow.Schema, optional
Explicit scalar or named-struct output. Scalar outputs must be
non-nullable. Required for batch annotations. When omitted, the output
is inferred from the return annotation.
Explicit scalar or named-struct output. A scalar field's ``nullable``
flag is preserved; a top-level named-struct field and a Blob v2 output
must be non-nullable. Required for batch annotations. When omitted, the
output is inferred from the return annotation.
pip : sequence of str, optional
Pip requirements for the remote environment. Syntax is validated at
decoration time, including extras and environment markers. Named direct
@@ -28,6 +28,7 @@ import lancedb
from lancedb import BlobReader
from lancedb.db import AsyncConnection
from lancedb.functions import (
FunctionSignature,
FunctionVersion,
HuggingFaceModel,
PythonRuntimeSpec,
@@ -1214,6 +1215,42 @@ def test_explicit_arrow_schema_is_deterministic():
assert signature.output.nullable is False
def test_nullable_scalar_output_annotation_and_explicit_schema():
input_schema = pa.schema([pa.field("value", pa.int64(), nullable=False)])
@function(input_schema=input_schema)
def optional_annotation(value: int) -> Optional[int]:
return None if value == 0 else value * 2
@function(input_schema=input_schema)
def optional_union(value: int) -> int | None:
return None if value == 0 else value * 2
@function(
input_schema=input_schema,
output_schema=pa.field("result", pa.int64(), nullable=True),
)
def explicit_schema(value: int):
return None if value == 0 else value * 2
for definition in (optional_annotation, optional_union, explicit_schema):
signature = definition.registration_request.signature
assert (signature.output.kind, signature.output.arrow_type) == (
"scalar",
"int64",
)
assert signature.output.nullable is True
assert json.loads(signature.to_canonical_json())["output"] == {
"kind": "scalar",
"arrow_type": "int64",
"nullable": True,
}
assert (
FunctionSignature.from_json(signature.to_canonical_json()).output.nullable
is True
)
def test_blob_fields_use_the_scalar_function_semantic_type():
@function(
input_schema=pa.schema([lancedb.blob("image", nullable=False)]),
@@ -1848,12 +1885,6 @@ def test_annotation_and_explicit_schema_validation_fail_closed():
def unsupported(value: set[str]) -> str:
return ""
with pytest.raises(ValueError, match="output must be non-nullable"):
@function
def nullable_output(value: int) -> Optional[int]:
return value
with pytest.raises(TypeError, match=r"missing Function annotations: \['return'\]"):
@function(input_schema=pa.schema([pa.field("value", pa.int64())]))
@@ -1873,10 +1904,25 @@ def test_annotation_and_explicit_schema_validation_fail_closed():
@function(
input_schema=pa.schema([pa.field("value", pa.int64())]),
output_schema=pa.field("result", pa.int64(), nullable=True),
output_schema=pa.field(
"result",
pa.struct([pa.field("value", pa.int64(), nullable=False)]),
nullable=True,
),
)
def nullable_explicit(value):
return value
def nullable_struct(value):
return {"value": value}
with pytest.raises(
ValueError, match="Blob v2 Function output must be non-nullable"
):
@function(
input_schema=pa.schema([lancedb.blob("image", nullable=False)]),
output_schema=lancedb.blob("result", nullable=True),
)
def nullable_blob(image):
return image
for invalid_field in [
pa.field("", pa.int32(), nullable=False),
+2 -2
View File
@@ -1032,8 +1032,8 @@ pub struct OutputMapping {
pub nullable: bool,
}
/// Internal physical column preserving the parent validity of a flattened
/// named-struct result.
/// Internal physical column preserving assignment of a nullable scalar result
/// or the parent validity of a flattened named-struct result.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct AssignmentMapping {
pub output_name: String,
+77
View File
@@ -10802,6 +10802,83 @@ mod tests {
assert_eq!(result.version, 8);
}
#[tokio::test]
async fn test_add_nullable_scalar_function_column_sends_atomic_null_declaration() {
let table = Table::new_with_handler("my_table", |request| {
match request.url().path() {
"/v1/table/my_table/describe/" => http::Response::builder()
.status(200)
.body(
r#"{"version":1,"schema":{"fields":[{"name":"description","nullable":true,"type":{"type":"string"}}]}}"#,
)
.unwrap(),
"/v1/table/my_table/add_columns/" => {
let actual: serde_json::Value = serde_json::from_slice(
request.body().unwrap().as_bytes().unwrap(),
)
.unwrap();
assert_eq!(
actual["new_columns"],
serde_json::json!([
{"name": "maybe", "all_null": true}
])
);
assert_eq!(
actual["function"]["application"]["output"],
serde_json::json!({
"kind": "scalar",
"arrow_type": "int64",
"nullable": true
})
);
assert_eq!(
actual["function"]["output_schema"],
serde_json::json!({
"fields": [{
"name": "maybe",
"nullable": true,
"type": {"type": "int64"}
}]
})
);
assert_eq!(
actual["function"]["outputs"],
serde_json::json!([
{
"result_field": "$value",
"output_name": "maybe",
"output_ordinal": 0
}
])
);
assert!(actual["function"].get("assignment").is_none());
assert_eq!(actual["auto_refresh_enabled"], serde_json::json!(true));
http::Response::builder()
.status(200)
.body(r#"{"version":8}"#)
.unwrap()
}
path => panic!("Unexpected path: {path}"),
}
});
let application = crate::function::FunctionApplication::from_json(
r#"{
"function":{"name":"nullable_scalar","version":"1","object_id":"fixture","location":"memory:///fixture","manifest_digest":"sha256:7e22f815b6648e14f093a3979a8e5a2082fa773ebe1ec84b135cae7e84d6f8e6"},
"inputs":[{"parameter":"text","kind":"column","value":{"path":"description"}}],
"output":{"kind":"scalar","arrow_type":"int64","nullable":true}
}"#,
)
.unwrap();
let result = table
.add_columns()
.function_as("maybe", application)
.execute()
.await
.unwrap();
assert_eq!(result.version, 8);
}
/// The fixture binding's table: `title` and `body` bound as inputs, its
/// two outputs declared, plus an unbound `spare`.
fn fixture_bound_schema() -> Schema {
+297 -9
View File
@@ -40,7 +40,7 @@ use lance_namespace::models::{JsonArrowDataType, JsonArrowField, JsonArrowSchema
use serde::{Deserialize, Serialize};
use serde_json::Value;
use crate::function::{FUNCTION_BLOB_V2_TYPE, FunctionApplication, FunctionBinding};
use crate::function::{FUNCTION_BLOB_V2_TYPE, FunctionApplication, FunctionBinding, OutputMapping};
use crate::materialized_view::without_declarations;
use crate::table::FunctionWorkerResourceBudget;
use crate::utils::resolve_arrow_field_path;
@@ -1289,6 +1289,21 @@ fn function_output_type_matches(expected: &DataType, actual: &DataType) -> bool
}
}
/// Whether a whole-result binding may carry the server-owned assignment marker.
///
/// Only a nullable scalar result needs a marker to distinguish a computed NULL
/// from an unfilled row. A non-nullable result, a whole named-struct column,
/// and a Blob v2 value keep their existing rejection.
fn nullable_scalar_allows_assignment(output: &OutputMapping) -> Result<bool> {
if !output.nullable || output.arrow_type == FUNCTION_BLOB_V2_TYPE {
return Ok(false);
}
let json = parse_output_arrow_type(&output.arrow_type)?;
let data_type = lance_namespace::schema::convert_json_arrow_type(&json)
.map_err(|e| invalid_function(format!("invalid Function output type: {e}")))?;
Ok(!matches!(data_type, DataType::Struct(_)))
}
fn ensure_binding_matches_schema(schema: &ArrowSchema, binding: &FunctionBinding) -> Result<()> {
let mut input_fields = Vec::with_capacity(binding.inputs().len());
for input in binding.inputs() {
@@ -1419,13 +1434,14 @@ fn ensure_binding_matches_schema(schema: &ArrowSchema, binding: &FunctionBinding
}
let mut declared_output_schema = binding.output_schema().cloned();
if let Some(assignment) = binding.assignment() {
if binding
if let Some(whole_result) = binding
.outputs()
.iter()
.any(|output| output.result_field == WHOLE_RESULT_FIELD)
.find(|output| output.result_field == WHOLE_RESULT_FIELD)
&& (binding.outputs().len() != 1 || !nullable_scalar_allows_assignment(whole_result)?)
{
return Err(invalid_function(format!(
"Function binding '{}' cannot attach an assignment column to a whole result",
"Function binding '{}' cannot attach an assignment column to a whole result; only a single nullable scalar output permits one",
binding.binding_id()
)));
}
@@ -1552,14 +1568,17 @@ pub(crate) fn plan_function_application(
"a scalar Function application must be mapped to one output column",
)
})?;
if output.nullable != Some(false) {
return Err(invalid_function(
"Function logical outputs must be non-nullable during NULL assignment",
));
}
let nullable = output.nullable.ok_or_else(|| {
invalid_function("scalar Function output must declare nullability")
})?;
let arrow_type = output.arrow_type.as_deref().ok_or_else(|| {
invalid_function("scalar Function output is missing its Arrow type")
})?;
if nullable && arrow_type == FUNCTION_BLOB_V2_TYPE {
return Err(invalid_function(
"Blob v2 Function outputs cannot be nullable",
));
}
outputs.push(FunctionOutputTarget {
result_field: WHOLE_RESULT_FIELD.to_string(),
output_name: name.to_string(),
@@ -4054,6 +4073,22 @@ mod tests {
.unwrap()
}
fn nullable_scalar_application() -> FunctionApplication {
FunctionApplication::from_json(
&serde_json::json!({
"function": {"name": "nullable_scalar", "version": "1", "object_id": "fixture", "location": "memory:///fixture", "manifest_digest": "sha256:7e22f815b6648e14f093a3979a8e5a2082fa773ebe1ec84b135cae7e84d6f8e6"},
"inputs": [{
"parameter": "value",
"kind": "column",
"value": {"path": "value"}
}],
"output": {"kind": "scalar", "arrow_type": "int64", "nullable": true}
})
.to_string(),
)
.unwrap()
}
fn binding_from_plan(plan: &FunctionDeclarationPlan) -> FunctionBinding {
let inputs = plan
.input_bindings
@@ -4363,6 +4398,259 @@ mod tests {
));
}
#[test]
fn test_nullable_scalar_application_plans_one_nullable_output() {
let input = ArrowField::new("value", DataType::Int64, false);
let plan = plan_function_application(
&ArrowSchema::new(vec![input]),
&nullable_scalar_application(),
Some("maybe"),
)
.unwrap();
assert_eq!(plan.application.output().nullable, Some(true));
assert_eq!(plan.outputs.len(), 1);
assert_eq!(plan.outputs[0].result_field, WHOLE_RESULT_FIELD);
assert_eq!(plan.outputs[0].output_name, "maybe");
assert_eq!(plan.outputs[0].output_ordinal, 0);
assert_eq!(plan.output_schema.fields.len(), 1);
assert!(plan.output_schema.fields[0].nullable);
assert_eq!(plan.output_schema.fields[0].r#type.r#type, "int64");
}
#[test]
fn test_invalid_nullable_scalar_applications_fail_closed() {
let input = ArrowField::new("value", DataType::Int64, false);
let schema = ArrowSchema::new(vec![input]);
let mut missing_nullability = serde_json::to_value(nullable_scalar_application()).unwrap();
missing_nullability["output"]
.as_object_mut()
.unwrap()
.remove("nullable");
let application = FunctionApplication::from_json(&missing_nullability.to_string()).unwrap();
let err = plan_function_application(&schema, &application, Some("maybe")).unwrap_err();
assert!(
matches!(&err, Error::InvalidInput { message } if message.contains("declare nullability")),
"{err:?}"
);
let nullable_blob =
blob_application(r#"{"kind":"scalar","arrow_type":"blob_v2","nullable":true}"#);
let err = plan_function_application(
&ArrowSchema::new(vec![crate::blob("image", false)]),
&nullable_blob,
Some("copy"),
)
.unwrap_err();
assert!(
matches!(&err, Error::InvalidInput { message } if message.contains("Blob v2") && message.contains("nullable")),
"{err:?}"
);
}
#[test]
fn test_nullable_scalar_binding_accepts_server_owned_assignment() {
let input = ArrowField::new("value", DataType::Int64, false);
let plan = plan_function_application(
&ArrowSchema::new(vec![input.clone()]),
&nullable_scalar_application(),
Some("maybe"),
)
.unwrap();
let assignment_name = "__function_assignment_nullable_scalar";
let mut raw = serde_json::to_value(binding_from_plan(&plan)).unwrap();
raw["outputs"][0]["nullable"] = Value::Bool(true);
raw["assignment"] = serde_json::json!({
"output_name": assignment_name,
"output_field_id": -1,
});
raw["output_schema"]["fields"]
.as_array_mut()
.unwrap()
.push(serde_json::json!({
"name": assignment_name,
"nullable": true,
"type": {"type": "bool"},
}));
let binding: FunctionBinding = serde_json::from_value(raw).unwrap();
let output = ArrowField::new("maybe", DataType::Int64, true).with_metadata(
function_computed_column_metadata(binding.binding_id(), 0, &["value".into()]),
);
let assignment = ArrowField::new(assignment_name, DataType::Boolean, true).with_metadata(
function_computed_column_metadata(
binding.binding_id(),
FUNCTION_ASSIGNMENT_OUTPUT_ORDINAL,
&["value".into()],
),
);
let schema = ArrowSchema::new(vec![input.clone(), output.clone(), assignment]);
ensure_binding_matches_schema(&schema, &binding).unwrap();
let hidden = ArrowSchema::new(vec![input, output]);
ensure_binding_matches_schema(&hidden, &binding).unwrap();
let schema = ArrowSchema::new_with_metadata(
schema.fields().to_vec(),
HashMap::from([(
FUNCTION_BINDINGS_META_KEY.to_string(),
function_bindings_metadata(std::slice::from_ref(&binding)).unwrap(),
)]),
);
ensure_supported_function_metadata(&schema).unwrap();
}
#[test]
fn test_non_nullable_scalar_binding_rejects_server_owned_assignment() {
let input = ArrowField::new("value", DataType::Int64, false);
let plan = plan_function_application(
&ArrowSchema::new(vec![input.clone()]),
&single_input_application("value"),
Some("maybe"),
)
.unwrap();
let assignment_name = "__function_assignment_non_nullable_scalar";
let mut raw = serde_json::to_value(binding_from_plan(&plan)).unwrap();
raw["assignment"] = serde_json::json!({
"output_name": assignment_name,
"output_field_id": -1,
});
raw["output_schema"]["fields"]
.as_array_mut()
.unwrap()
.push(serde_json::json!({
"name": assignment_name,
"nullable": true,
"type": {"type": "bool"},
}));
let binding: FunctionBinding = serde_json::from_value(raw).unwrap();
let output = ArrowField::new("maybe", DataType::Int64, true).with_metadata(
function_computed_column_metadata(binding.binding_id(), 0, &["value".into()]),
);
let assignment = ArrowField::new(assignment_name, DataType::Boolean, true).with_metadata(
function_computed_column_metadata(
binding.binding_id(),
FUNCTION_ASSIGNMENT_OUTPUT_ORDINAL,
&["value".into()],
),
);
let err = ensure_binding_matches_schema(
&ArrowSchema::new(vec![input, output, assignment]),
&binding,
)
.unwrap_err();
assert!(
matches!(&err, Error::InvalidInput { message } if message.contains("nullable scalar")),
"{err:?}"
);
}
#[test]
fn test_named_struct_whole_binding_rejects_server_owned_assignment() {
let plan = plan_function_application(
&function_input_schema(),
&named_struct_application("{}"),
Some("features"),
)
.unwrap();
let assignment_name = "__function_assignment_whole_struct";
let mut raw = serde_json::to_value(binding_from_plan(&plan)).unwrap();
raw["outputs"][0]["nullable"] = Value::Bool(true);
raw["assignment"] = serde_json::json!({
"output_name": assignment_name,
"output_field_id": -1,
});
raw["output_schema"]["fields"]
.as_array_mut()
.unwrap()
.push(serde_json::json!({
"name": assignment_name,
"nullable": true,
"type": {"type": "bool"},
}));
let binding: FunctionBinding = serde_json::from_value(raw).unwrap();
let output =
lance_namespace::schema::convert_json_arrow_field(&plan.output_schema.fields[0])
.unwrap()
.with_metadata(function_computed_column_metadata(
binding.binding_id(),
0,
&["title".into(), "body".into()],
));
let assignment = ArrowField::new(assignment_name, DataType::Boolean, true).with_metadata(
function_computed_column_metadata(
binding.binding_id(),
FUNCTION_ASSIGNMENT_OUTPUT_ORDINAL,
&["title".into(), "body".into()],
),
);
let mut fields = function_input_schema()
.fields()
.iter()
.map(|field| field.as_ref().clone())
.collect::<Vec<_>>();
fields.push(output);
fields.push(assignment);
let err = ensure_binding_matches_schema(&ArrowSchema::new(fields), &binding).unwrap_err();
assert!(
matches!(&err, Error::InvalidInput { message } if message.contains("nullable scalar")),
"{err:?}"
);
}
#[test]
fn test_blob_whole_binding_rejects_server_owned_assignment() {
let input = crate::blob("image", false);
let plan = plan_function_application(
&ArrowSchema::new(vec![input.clone()]),
&blob_application(r#"{"kind":"scalar","arrow_type":"blob_v2","nullable":false}"#),
Some("thumbnail"),
)
.unwrap();
let assignment_name = "__function_assignment_whole_blob";
let mut raw = serde_json::to_value(binding_from_plan(&plan)).unwrap();
raw["outputs"][0]["nullable"] = Value::Bool(true);
raw["assignment"] = serde_json::json!({
"output_name": assignment_name,
"output_field_id": -1,
});
raw["output_schema"]["fields"]
.as_array_mut()
.unwrap()
.push(serde_json::json!({
"name": assignment_name,
"nullable": true,
"type": {"type": "bool"},
}));
let binding: FunctionBinding = serde_json::from_value(raw).unwrap();
let mut metadata = full_blob_field("thumbnail", true).metadata().clone();
metadata.extend(function_computed_column_metadata(
binding.binding_id(),
0,
&["image".into()],
));
let output = full_blob_field("thumbnail", true).with_metadata(metadata);
let assignment = ArrowField::new(assignment_name, DataType::Boolean, true).with_metadata(
function_computed_column_metadata(
binding.binding_id(),
FUNCTION_ASSIGNMENT_OUTPUT_ORDINAL,
&["image".into()],
),
);
let err = ensure_binding_matches_schema(
&ArrowSchema::new(vec![input, output, assignment]),
&binding,
)
.unwrap_err();
assert!(
matches!(&err, Error::InvalidInput { message } if message.contains("nullable scalar")),
"{err:?}"
);
}
#[test]
fn test_nullable_function_input_cannot_bind_to_non_nullable_parameter() {
let mut raw_binding: Value = serde_json::from_str(include_str!(