use runmat_builtins::{CharArray, ComplexTensor, LogicalArray, StringArray, Tensor, Value};
use runmat_macros::runtime_builtin;
use crate::builtins::common::spec::{
BroadcastSemantics, BuiltinFusionSpec, BuiltinGpuSpec, ConstantStrategy, FusionError,
FusionExprContext, FusionKernelTemplate, GpuOpKind, ProviderHook, ReductionNaN,
ResidencyPolicy, ScalarType, ShapeRequirements,
};
use crate::builtins::common::{gpu_helpers, tensor};
use crate::builtins::logical::type_resolvers::logical_unary_type;
use crate::{build_runtime_error, BuiltinResult, RuntimeError};
#[runmat_macros::register_gpu_spec(builtin_path = "crate::builtins::logical::tests::isinf")]
pub const GPU_SPEC: BuiltinGpuSpec = BuiltinGpuSpec {
name: "isinf",
op_kind: GpuOpKind::Elementwise,
supported_precisions: &[ScalarType::F32, ScalarType::F64],
broadcast: BroadcastSemantics::Matlab,
provider_hooks: &[ProviderHook::Unary {
name: "logical_isinf",
}],
constant_strategy: ConstantStrategy::InlineLiteral,
residency: ResidencyPolicy::NewHandle,
nan_mode: ReductionNaN::Include,
two_pass_threshold: None,
workgroup_size: None,
accepts_nan_mode: false,
notes:
"Dispatches to the provider `logical_isinf` hook when available; otherwise the runtime gathers to host and builds the logical mask on the CPU.",
};
#[runmat_macros::register_fusion_spec(builtin_path = "crate::builtins::logical::tests::isinf")]
pub const FUSION_SPEC: BuiltinFusionSpec = BuiltinFusionSpec {
name: "isinf",
shape: ShapeRequirements::BroadcastCompatible,
constant_strategy: ConstantStrategy::InlineLiteral,
elementwise: Some(FusionKernelTemplate {
scalar_precisions: &[ScalarType::F32, ScalarType::F64],
wgsl_body: |ctx: &FusionExprContext| {
let input = ctx.inputs.first().ok_or(FusionError::MissingInput(0))?;
let (zero, one) = match ctx.scalar_ty {
ScalarType::F32 => ("0.0", "1.0"),
ScalarType::F64 => ("f64(0.0)", "f64(1.0)"),
other => return Err(FusionError::UnsupportedPrecision(other)),
};
Ok(format!("select({zero}, {one}, isInf({input}))"))
},
}),
reduction: None,
emits_nan: false,
notes: "Fused kernels emit 0/1 masks; providers can override with native logical-isinf implementations.",
};
const BUILTIN_NAME: &str = "isinf";
const IDENTIFIER_INVALID_INPUT: &str = "RunMat:isinf:InvalidInput";
const IDENTIFIER_INTERNAL: &str = "RunMat:isinf:InternalError";
#[runtime_builtin(
name = "isinf",
category = "logical/tests",
summary = "Return a logical mask indicating which elements of the input are ±Inf.",
keywords = "isinf,infinity,logical,gpu",
accel = "elementwise",
type_resolver(logical_unary_type),
builtin_path = "crate::builtins::logical::tests::isinf"
)]
async fn isinf_builtin(value: Value) -> BuiltinResult<Value> {
match value {
Value::GpuTensor(handle) => {
if let Some(provider) = runmat_accelerate_api::provider() {
if let Ok(mask) = provider.logical_isinf(&handle) {
return Ok(gpu_helpers::logical_gpu_value(mask));
}
}
let tensor = gpu_helpers::gather_tensor_async(&handle)
.await
.map_err(|err| internal_error(BUILTIN_NAME, format!("{BUILTIN_NAME}: {err}")))?;
isinf_tensor(BUILTIN_NAME, tensor)
}
other => isinf_host(other),
}
}
fn isinf_host(value: Value) -> BuiltinResult<Value> {
match value {
Value::Num(x) => Ok(Value::Bool(x.is_infinite())),
Value::Int(_) | Value::Bool(_) => Ok(Value::Bool(false)),
Value::Complex(re, im) => Ok(Value::Bool(re.is_infinite() || im.is_infinite())),
Value::Tensor(tensor) => isinf_tensor(BUILTIN_NAME, tensor),
Value::ComplexTensor(tensor) => isinf_complex_tensor(BUILTIN_NAME, tensor),
Value::LogicalArray(array) => {
let LogicalArray { shape, .. } = array;
logical_zeros(BUILTIN_NAME, shape)
}
Value::CharArray(array) => {
let CharArray { rows, cols, .. } = array;
logical_zeros(BUILTIN_NAME, vec![rows, cols])
}
Value::String(_) => Ok(Value::Bool(false)),
Value::StringArray(array) => {
let StringArray { shape, .. } = array;
logical_zeros(BUILTIN_NAME, shape)
}
_ => Err(build_runtime_error(format!(
"{BUILTIN_NAME}: expected numeric, logical, char, or string input"
))
.with_identifier(IDENTIFIER_INVALID_INPUT)
.with_builtin(BUILTIN_NAME)
.build()),
}
}
fn isinf_tensor(name: &str, tensor: Tensor) -> BuiltinResult<Value> {
let data = tensor
.data
.iter()
.map(|&x| if x.is_infinite() { 1u8 } else { 0u8 })
.collect::<Vec<_>>();
logical_result(name, data, tensor.shape)
}
fn isinf_complex_tensor(name: &str, tensor: ComplexTensor) -> BuiltinResult<Value> {
let data = tensor
.data
.iter()
.map(|&(re, im)| {
if re.is_infinite() || im.is_infinite() {
1u8
} else {
0u8
}
})
.collect::<Vec<_>>();
logical_result(name, data, tensor.shape)
}
fn logical_zeros(name: &str, shape: Vec<usize>) -> BuiltinResult<Value> {
let total = tensor::element_count(&shape);
if total == 0 {
return LogicalArray::new(Vec::new(), shape)
.map(Value::LogicalArray)
.map_err(|e| logical_array_error(name, e));
}
let data = vec![0u8; total];
logical_result(name, data, shape)
}
fn logical_result(name: &str, bits: Vec<u8>, shape: Vec<usize>) -> BuiltinResult<Value> {
let total = tensor::element_count(&shape);
if total != bits.len() {
return Err(internal_error(
name,
format!(
"{name}: internal error, mask length {} does not match shape {:?}",
bits.len(),
shape
),
));
}
if total == 1 {
Ok(Value::Bool(bits[0] != 0))
} else {
LogicalArray::new(bits, shape)
.map(Value::LogicalArray)
.map_err(|e| logical_array_error(name, e))
}
}
fn logical_array_error(name: &str, err: impl std::fmt::Display) -> RuntimeError {
internal_error(name, format!("{name}: {err}"))
}
fn internal_error(name: &str, message: impl Into<String>) -> RuntimeError {
build_runtime_error(message)
.with_identifier(IDENTIFIER_INTERNAL)
.with_builtin(name)
.build()
}
#[cfg(test)]
pub(crate) mod tests {
use super::*;
use crate::builtins::common::test_support;
use futures::executor::block_on;
use runmat_builtins::{ResolveContext, Type};
#[test]
fn isinf_type_returns_logical() {
let out = logical_unary_type(
&[Type::Tensor { shape: None }],
&ResolveContext::new(Vec::new()),
);
assert_eq!(out, Type::logical());
}
use runmat_builtins::IntValue;
fn run_isinf(value: Value) -> BuiltinResult<Value> {
block_on(super::isinf_builtin(value))
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_scalar_positive() {
let result = run_isinf(Value::Num(f64::INFINITY)).expect("isinf");
assert_eq!(result, Value::Bool(true));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_scalar_negative() {
let result = run_isinf(Value::Num(f64::NEG_INFINITY)).expect("isinf");
assert_eq!(result, Value::Bool(true));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_scalar_finite() {
let result = run_isinf(Value::Num(42.0)).expect("isinf");
assert_eq!(result, Value::Bool(false));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_scalar_nan_false() {
let result = run_isinf(Value::Num(f64::NAN)).expect("isinf");
assert_eq!(result, Value::Bool(false));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_scalar_bool_false() {
let result = run_isinf(Value::Bool(true)).expect("isinf");
assert_eq!(result, Value::Bool(false));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_scalar_int_false() {
let result = run_isinf(Value::Int(IntValue::I32(7))).expect("isinf");
assert_eq!(result, Value::Bool(false));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_complex_scalar_detects_infinite_components() {
let finite = run_isinf(Value::Complex(1.0, 2.0)).expect("isinf");
assert_eq!(finite, Value::Bool(false));
let inf_real = run_isinf(Value::Complex(f64::INFINITY, 0.0)).expect("isinf");
assert_eq!(inf_real, Value::Bool(true));
let inf_imag = run_isinf(Value::Complex(0.0, f64::NEG_INFINITY)).expect("isinf");
assert_eq!(inf_imag, Value::Bool(true));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_tensor_mask() {
let tensor =
Tensor::new(vec![1.0, f64::INFINITY, -f64::INFINITY, 0.0], vec![2, 2]).unwrap();
let result = run_isinf(Value::Tensor(tensor)).expect("isinf");
match result {
Value::LogicalArray(mask) => {
assert_eq!(mask.shape, vec![2, 2]);
assert_eq!(mask.data, vec![0, 1, 1, 0]);
}
other => panic!("expected logical array, got {other:?}"),
}
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_logical_array_returns_zeros() {
let logical = LogicalArray::new(vec![1, 0, 1], vec![3, 1]).unwrap();
let result = run_isinf(Value::LogicalArray(logical)).expect("isinf");
match result {
Value::LogicalArray(mask) => {
assert_eq!(mask.shape, vec![3, 1]);
assert!(mask.data.iter().all(|&bit| bit == 0));
}
other => panic!("expected logical array, got {other:?}"),
}
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_complex_tensor_mask() {
let tensor = ComplexTensor::new(
vec![(0.0, 0.0), (f64::INFINITY, 1.0), (2.0, f64::NEG_INFINITY)],
vec![3, 1],
)
.unwrap();
let result = run_isinf(Value::ComplexTensor(tensor)).expect("isinf");
match result {
Value::LogicalArray(mask) => {
assert_eq!(mask.shape, vec![3, 1]);
assert_eq!(mask.data, vec![0, 1, 1]);
}
other => panic!("expected logical array, got {other:?}"),
}
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_string_scalar_false() {
let result = run_isinf(Value::String("Inf".to_string())).expect("isinf");
assert_eq!(result, Value::Bool(false));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_string_array_returns_all_false() {
let strings = StringArray::new(vec!["foo".into(), "bar".into()], vec![1, 2]).unwrap();
let result = run_isinf(Value::StringArray(strings)).expect("isinf");
match result {
Value::LogicalArray(mask) => {
assert_eq!(mask.shape, vec![1, 2]);
assert_eq!(mask.data, vec![0, 0]);
}
other => panic!("expected logical array, got {other:?}"),
}
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_empty_tensor_preserves_shape() {
let tensor = Tensor::new(Vec::new(), vec![0, 3]).unwrap();
let result = run_isinf(Value::Tensor(tensor)).expect("isinf");
match result {
Value::LogicalArray(mask) => {
assert_eq!(mask.shape, vec![0, 3]);
assert!(mask.data.is_empty());
}
other => panic!("expected logical array, got {other:?}"),
}
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_singleton_tensor_returns_scalar_bool() {
let tensor = Tensor::new(vec![f64::INFINITY], vec![1, 1]).unwrap();
let result = run_isinf(Value::Tensor(tensor)).expect("isinf");
assert_eq!(result, Value::Bool(true));
let finite = Tensor::new(vec![0.0], vec![1, 1]).unwrap();
let result = run_isinf(Value::Tensor(finite)).expect("isinf");
assert_eq!(result, Value::Bool(false));
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_rejects_unsupported_types() {
let err = run_isinf(Value::FunctionHandle("foo".to_string()))
.expect_err("isinf should reject function handles");
assert!(
err.message()
.contains("expected numeric, logical, char, or string input"),
"unexpected error message: {err:?}"
);
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_char_array_returns_zeros() {
let array = CharArray::new("Inf".chars().collect(), 1, 3).unwrap();
let result = run_isinf(Value::CharArray(array)).expect("isinf");
match result {
Value::LogicalArray(mask) => {
assert_eq!(mask.shape, vec![1, 3]);
assert_eq!(mask.data, vec![0, 0, 0]);
}
other => panic!("expected logical array, got {other:?}"),
}
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
fn isinf_gpu_roundtrip() {
test_support::with_test_provider(|provider| {
let tensor = Tensor::new(vec![1.0, f64::INFINITY, -f64::INFINITY], vec![3, 1]).unwrap();
let view = runmat_accelerate_api::HostTensorView {
data: &tensor.data,
shape: &tensor.shape,
};
let handle = provider.upload(&view).expect("upload");
let result = run_isinf(Value::GpuTensor(handle)).expect("isinf");
let gathered = test_support::gather(result).expect("gather");
assert_eq!(gathered.shape, vec![3, 1]);
assert_eq!(gathered.data, vec![0.0, 1.0, 1.0]);
});
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[test]
#[cfg(feature = "wgpu")]
fn isinf_wgpu_matches_host_path() {
let _ = runmat_accelerate::backend::wgpu::provider::register_wgpu_provider(
runmat_accelerate::backend::wgpu::provider::WgpuProviderOptions::default(),
);
let tensor =
Tensor::new(vec![1.0, f64::INFINITY, -f64::INFINITY, 0.0], vec![2, 2]).unwrap();
let cpu = isinf_tensor("isinf", tensor.clone()).expect("cpu path");
let view = runmat_accelerate_api::HostTensorView {
data: &tensor.data,
shape: &tensor.shape,
};
let handle = runmat_accelerate_api::provider()
.unwrap()
.upload(&view)
.expect("upload");
let gpu = run_isinf(Value::GpuTensor(handle)).expect("gpu path");
let gathered = test_support::gather(gpu).expect("gather");
match (cpu, gathered) {
(Value::LogicalArray(expected), Tensor { data, shape, .. }) => {
assert_eq!(shape, expected.shape);
let expected_f64: Vec<f64> = expected
.data
.iter()
.map(|&b| if b != 0 { 1.0 } else { 0.0 })
.collect();
assert_eq!(data, expected_f64);
}
(Value::Bool(flag), Tensor { data, .. }) => {
assert_eq!(data, vec![if flag { 1.0 } else { 0.0 }]);
}
other => panic!("unexpected results {other:?}"),
}
}
}