runmat-runtime 0.5.0

//! MATLAB-compatible `string.empty` builtin for RunMat.

use runmat_builtins::{
    BuiltinCompletionPolicy, BuiltinDescriptor, BuiltinErrorDescriptor, BuiltinOutputMode,
    BuiltinParamArity, BuiltinParamDescriptor, BuiltinParamType, BuiltinSignatureDescriptor,
    StringArray, Value,
};
use runmat_macros::runtime_builtin;

use crate::builtins::common::map_control_flow_with_builtin;
use crate::builtins::common::random_args::{extract_dims, keyword_of};
use crate::builtins::common::spec::{
    BroadcastSemantics, BuiltinFusionSpec, BuiltinGpuSpec, ConstantStrategy, GpuOpKind,
    ReductionNaN, ResidencyPolicy, ShapeRequirements,
};
use crate::builtins::strings::type_resolvers::string_array_type;
use crate::{build_runtime_error, gather_if_needed_async, BuiltinResult, RuntimeError};

const LABEL: &str = "string.empty";

const STRING_EMPTY_OUTPUT: [BuiltinParamDescriptor; 1] = [BuiltinParamDescriptor {
    name: "S",
    ty: BuiltinParamType::Any,
    arity: BuiltinParamArity::Required,
    default: None,
    description: "Empty string array with at least one zero dimension.",
}];

const STRING_EMPTY_INPUT_SZ: [BuiltinParamDescriptor; 1] = [BuiltinParamDescriptor {
    name: "sz",
    ty: BuiltinParamType::SizeArg,
    arity: BuiltinParamArity::Required,
    default: None,
    description: "Size vector or scalar.",
}];

const STRING_EMPTY_INPUT_DIMS: [BuiltinParamDescriptor; 2] = [
    BuiltinParamDescriptor {
        name: "m",
        ty: BuiltinParamType::SizeArg,
        arity: BuiltinParamArity::Required,
        default: None,
        description: "First dimension.",
    },
    BuiltinParamDescriptor {
        name: "n...",
        ty: BuiltinParamType::SizeArg,
        arity: BuiltinParamArity::Variadic,
        default: None,
        description: "Additional dimensions.",
    },
];

const STRING_EMPTY_INPUT_LIKE: [BuiltinParamDescriptor; 3] = [
    BuiltinParamDescriptor {
        name: "dims...",
        ty: BuiltinParamType::SizeArg,
        arity: BuiltinParamArity::Variadic,
        default: None,
        description: "Optional explicit dimensions.",
    },
    BuiltinParamDescriptor {
        name: "like",
        ty: BuiltinParamType::StringScalar,
        arity: BuiltinParamArity::Required,
        default: Some("\"like\""),
        description: "Literal option keyword \"like\".",
    },
    BuiltinParamDescriptor {
        name: "p",
        ty: BuiltinParamType::LikePrototype,
        arity: BuiltinParamArity::Required,
        default: None,
        description: "Prototype supplying trailing dimensions.",
    },
];

const STRING_EMPTY_SIGNATURES: [BuiltinSignatureDescriptor; 4] = [
    BuiltinSignatureDescriptor {
        label: "S = string.empty()",
        inputs: &[],
        outputs: &STRING_EMPTY_OUTPUT,
    },
    BuiltinSignatureDescriptor {
        label: "S = string.empty(sz)",
        inputs: &STRING_EMPTY_INPUT_SZ,
        outputs: &STRING_EMPTY_OUTPUT,
    },
    BuiltinSignatureDescriptor {
        label: "S = string.empty(m, n...)",
        inputs: &STRING_EMPTY_INPUT_DIMS,
        outputs: &STRING_EMPTY_OUTPUT,
    },
    BuiltinSignatureDescriptor {
        label: "S = string.empty(___, \"like\", p)",
        inputs: &STRING_EMPTY_INPUT_LIKE,
        outputs: &STRING_EMPTY_OUTPUT,
    },
];

const STRING_EMPTY_ERROR_INVALID_SIZE: BuiltinErrorDescriptor = BuiltinErrorDescriptor {
    code: "RM.STRING_EMPTY.INVALID_SIZE",
    identifier: Some("RunMat:string.empty:InvalidSize"),
    when: "Size inputs are not valid numeric dimensions or vectors.",
    message: "string.empty: size inputs must be numeric scalars or size vectors",
};

const STRING_EMPTY_ERROR_LIKE_MISSING: BuiltinErrorDescriptor = BuiltinErrorDescriptor {
    code: "RM.STRING_EMPTY.LIKE_MISSING_PROTOTYPE",
    identifier: Some("RunMat:string.empty:LikeMissingPrototype"),
    when: "\"like\" keyword is present without a prototype.",
    message: "string.empty: expected prototype after 'like'",
};

const STRING_EMPTY_ERROR_LIKE_DUPLICATE: BuiltinErrorDescriptor = BuiltinErrorDescriptor {
    code: "RM.STRING_EMPTY.LIKE_DUPLICATE",
    identifier: Some("RunMat:string.empty:LikeDuplicate"),
    when: "Multiple \"like\" specifications are supplied.",
    message: "string.empty: multiple 'like' prototypes are not supported",
};

const STRING_EMPTY_ERROR_NOT_EMPTY_SHAPE: BuiltinErrorDescriptor = BuiltinErrorDescriptor {
    code: "RM.STRING_EMPTY.NONEMPTY_SHAPE",
    identifier: Some("RunMat:string.empty:NonEmptyShape"),
    when: "Parsed dimensions do not produce an empty array shape.",
    message: "string.empty: at least one dimension must be zero",
};

const STRING_EMPTY_ERROR_INTERNAL: BuiltinErrorDescriptor = BuiltinErrorDescriptor {
    code: "RM.STRING_EMPTY.INTERNAL",
    identifier: Some("RunMat:string.empty:InternalError"),
    when: "Internal empty string-array construction failed.",
    message: "string.empty: internal error",
};

const STRING_EMPTY_ERRORS: [BuiltinErrorDescriptor; 5] = [
    STRING_EMPTY_ERROR_INVALID_SIZE,
    STRING_EMPTY_ERROR_LIKE_MISSING,
    STRING_EMPTY_ERROR_LIKE_DUPLICATE,
    STRING_EMPTY_ERROR_NOT_EMPTY_SHAPE,
    STRING_EMPTY_ERROR_INTERNAL,
];

pub const STRING_EMPTY_DESCRIPTOR: BuiltinDescriptor = BuiltinDescriptor {
    signatures: &STRING_EMPTY_SIGNATURES,
    output_mode: BuiltinOutputMode::Fixed,
    completion_policy: BuiltinCompletionPolicy::MethodOnly,
    errors: &STRING_EMPTY_ERRORS,
};

fn string_empty_error(error: &'static BuiltinErrorDescriptor) -> RuntimeError {
    string_empty_error_with_message(error.message, error)
}

fn string_empty_error_with_message(
    message: impl Into<String>,
    error: &'static BuiltinErrorDescriptor,
) -> RuntimeError {
    let mut builder = build_runtime_error(message).with_builtin(LABEL);
    if let Some(identifier) = error.identifier {
        builder = builder.with_identifier(identifier);
    }
    builder.build()
}

fn remap_string_empty_flow(err: RuntimeError) -> RuntimeError {
    map_control_flow_with_builtin(err, LABEL)
}

#[runmat_macros::register_gpu_spec(builtin_path = "crate::builtins::strings::core::string_empty")]
pub const GPU_SPEC: BuiltinGpuSpec = BuiltinGpuSpec {
    name: "string.empty",
    op_kind: GpuOpKind::Custom("constructor"),
    supported_precisions: &[],
    broadcast: BroadcastSemantics::None,
    provider_hooks: &[],
    constant_strategy: ConstantStrategy::InlineLiteral,
    residency: ResidencyPolicy::NewHandle,
    nan_mode: ReductionNaN::Include,
    two_pass_threshold: None,
    workgroup_size: None,
    accepts_nan_mode: false,
    notes: "Host-only constructor that returns a new empty string array without contacting GPU providers.",
};

#[runmat_macros::register_fusion_spec(
    builtin_path = "crate::builtins::strings::core::string_empty"
)]
pub const FUSION_SPEC: BuiltinFusionSpec = BuiltinFusionSpec {
    name: "string.empty",
    shape: ShapeRequirements::Any,
    constant_strategy: ConstantStrategy::InlineLiteral,
    elementwise: None,
    reduction: None,
    emits_nan: false,
    notes: "Pure constructor; fusion planner treats calls as non-fusable sinks.",
};

#[runtime_builtin(
    name = "string.empty",
    category = "strings/core",
    summary = "Construct empty string arrays with MATLAB-compatible dimension semantics.",
    keywords = "string.empty,empty,string array,preallocate",
    accel = "none",
    type_resolver(string_array_type),
    descriptor(crate::builtins::strings::core::string_empty::STRING_EMPTY_DESCRIPTOR),
    builtin_path = "crate::builtins::strings::core::string_empty"
)]
async fn string_empty_builtin(rest: Vec<Value>) -> crate::BuiltinResult<Value> {
    let shape = parse_shape(&rest).await?;
    let total: usize = shape.iter().product();
    debug_assert_eq!(total, 0, "string.empty must produce an empty array");
    let data = Vec::<String>::new();
    let array = StringArray::new(data, shape)
        .map_err(|_| string_empty_error(&STRING_EMPTY_ERROR_INTERNAL))?;
    Ok(Value::StringArray(array))
}

async fn parse_shape(args: &[Value]) -> BuiltinResult<Vec<usize>> {
    if args.is_empty() {
        return Ok(vec![0, 0]);
    }

    let mut explicit_dims: Vec<usize> = Vec::new();
    let mut like_shape: Option<Vec<usize>> = None;
    let mut idx = 0;

    while idx < args.len() {
        let arg_host = gather_if_needed_async(&args[idx])
            .await
            .map_err(remap_string_empty_flow)?;

        if let Some(keyword) = keyword_of(&arg_host) {
            if keyword.as_str() == "like" {
                if like_shape.is_some() {
                    return Err(string_empty_error(&STRING_EMPTY_ERROR_LIKE_DUPLICATE));
                }
                let Some(proto_raw) = args.get(idx + 1) else {
                    return Err(string_empty_error(&STRING_EMPTY_ERROR_LIKE_MISSING));
                };
                let proto = gather_if_needed_async(proto_raw)
                    .await
                    .map_err(remap_string_empty_flow)?;
                like_shape = Some(prototype_dims(&proto));
                idx += 2;
                continue;
            }
            // Unrecognized keywords are treated as non-keyword inputs and will
            // be validated under numeric size parsing below.
        }

        if let Some(parsed) = extract_dims(&arg_host, LABEL).await.map_err(|message| {
            string_empty_error_with_message(message, &STRING_EMPTY_ERROR_INVALID_SIZE)
        })? {
            if explicit_dims.is_empty() {
                explicit_dims = parsed;
            } else {
                explicit_dims.extend(parsed);
            }
            idx += 1;
            continue;
        }

        return Err(string_empty_error(&STRING_EMPTY_ERROR_INVALID_SIZE));
    }

    let shape = if !explicit_dims.is_empty() {
        shape_from_explicit_dims(&explicit_dims)
    } else if let Some(proto_shape) = like_shape {
        shape_from_like(&proto_shape)
    } else {
        vec![0, 0]
    };
    ensure_empty_shape(&shape)?;
    Ok(shape)
}

fn shape_from_explicit_dims(dims: &[usize]) -> Vec<usize> {
    match dims.len() {
        0 => vec![0, 0],
        1 => vec![0, dims[0]],
        _ => {
            let mut shape = Vec::with_capacity(dims.len());
            shape.push(0);
            shape.extend_from_slice(&dims[1..]);
            shape
        }
    }
}

fn shape_from_like(proto: &[usize]) -> Vec<usize> {
    if proto.is_empty() {
        return vec![0, 0];
    }
    if proto.len() == 1 {
        return vec![0, proto[0]];
    }
    let mut shape = Vec::with_capacity(proto.len());
    shape.push(0);
    shape.extend_from_slice(&proto[1..]);
    shape
}

fn ensure_empty_shape(shape: &[usize]) -> BuiltinResult<()> {
    if shape.iter().product::<usize>() != 0 {
        return Err(string_empty_error(&STRING_EMPTY_ERROR_NOT_EMPTY_SHAPE));
    }
    Ok(())
}

fn prototype_dims(proto: &Value) -> Vec<usize> {
    match proto {
        Value::StringArray(sa) => sa.shape.clone(),
        Value::CharArray(ca) => vec![ca.rows, ca.cols],
        Value::Tensor(t) => t.shape.clone(),
        Value::ComplexTensor(t) => t.shape.clone(),
        Value::LogicalArray(l) => l.shape.clone(),
        Value::Cell(cell) => cell.shape.clone(),
        Value::GpuTensor(handle) => handle.shape.clone(),
        Value::Num(_) | Value::Int(_) | Value::Bool(_) | Value::Complex(_, _) => vec![1, 1],
        Value::String(_) => vec![1, 1],
        _ => vec![1, 1],
    }
}

#[cfg(test)]
pub(crate) mod tests {
    use super::*;
    use crate::builtins::common::test_support;
    use runmat_accelerate_api::HostTensorView;
    use runmat_builtins::{ResolveContext, StringArray, Tensor, Type, Value};

    fn string_empty_builtin(rest: Vec<Value>) -> BuiltinResult<Value> {
        futures::executor::block_on(super::string_empty_builtin(rest))
    }

    fn error_message(err: crate::RuntimeError) -> String {
        err.message().to_string()
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn default_is_zero_by_zero() {
        let result = string_empty_builtin(Vec::new()).expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 0]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn single_dimension_creates_zero_by_n() {
        let result = string_empty_builtin(vec![Value::from(5)]).expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 5]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn multiple_dimensions_respect_trailing_sizes() {
        let args = vec![Value::from(3), Value::from(4), Value::from(2)];
        let result = string_empty_builtin(args).expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 4, 2]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn size_vector_argument_supported() {
        let tensor = Tensor::new(vec![0.0, 5.0, 3.0], vec![1, 3]).unwrap();
        let result = string_empty_builtin(vec![Value::Tensor(tensor)]).expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 5, 3]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn size_vector_from_nonempty_array_drops_leading_extent() {
        let tensor = Tensor::new(vec![3.0, 2.0], vec![1, 2]).unwrap();
        let result = string_empty_builtin(vec![Value::Tensor(tensor)]).expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 2]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn accepts_zero_in_any_position() {
        let args = vec![Value::from(3), Value::from(4), Value::from(0)];
        let result = string_empty_builtin(args).expect("string.empty");
        match result {
            Value::StringArray(sa) => assert_eq!(sa.shape, vec![0, 4, 0]),
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[test]
    fn string_empty_type_is_string_array() {
        assert_eq!(
            string_array_type(&[], &ResolveContext::new(Vec::new())),
            Type::cell_of(Type::String)
        );
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn like_prototype_without_explicit_dims() {
        let proto = StringArray::new(vec!["alpha".to_string(); 6], vec![2, 3]).unwrap();
        let result = string_empty_builtin(vec![Value::from("like"), Value::StringArray(proto)])
            .expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 3]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn like_prototype_with_scalar_shape() {
        let proto = StringArray::new(vec!["foo".to_string()], vec![1, 1]).unwrap();
        let result = string_empty_builtin(vec![Value::from("like"), Value::StringArray(proto)])
            .expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 1]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn like_with_numeric_prototype() {
        let tensor = Tensor::new(vec![1.0, 2.0], vec![2, 1]).unwrap();
        let result = string_empty_builtin(vec![Value::from("like"), Value::Tensor(tensor)])
            .expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 1]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn like_with_explicit_dims_prefers_dimensions() {
        let proto = StringArray::new(Vec::new(), vec![0, 2]).unwrap();
        let args = vec![
            Value::from(0),
            Value::from(7),
            Value::from("like"),
            Value::StringArray(proto),
        ];
        let result = string_empty_builtin(args).expect("string.empty");
        match result {
            Value::StringArray(sa) => {
                assert_eq!(sa.shape, vec![0, 7]);
                assert_eq!(sa.data.len(), 0);
            }
            other => panic!("expected string array, got {other:?}"),
        }
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn missing_like_prototype_errors() {
        let err = error_message(
            string_empty_builtin(vec![Value::from("like")]).expect_err("expected error"),
        );
        assert!(
            err.contains("expected prototype"),
            "unexpected error: {err}"
        );
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn duplicate_like_errors() {
        let proto = StringArray::new(Vec::new(), vec![0, 2]).unwrap();
        let err = error_message(
            string_empty_builtin(vec![
                Value::from("like"),
                Value::StringArray(proto.clone()),
                Value::from("like"),
                Value::StringArray(proto),
            ])
            .expect_err("expected error"),
        );
        assert!(err.contains("multiple 'like'"), "unexpected error: {err}");
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn rejects_non_dimension_inputs() {
        let err = error_message(
            string_empty_builtin(vec![Value::String("oops".into())]).expect_err("expected error"),
        );
        assert!(
            err.contains("size inputs must be numeric"),
            "unexpected error: {err}"
        );
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn like_gathers_gpu_prototype() {
        test_support::with_test_provider(|provider| {
            let tensor =
                Tensor::new((1..=6).map(|v| v as f64).collect::<Vec<_>>(), vec![2, 3]).unwrap();
            let view = HostTensorView {
                data: &tensor.data,
                shape: &tensor.shape,
            };
            let handle = provider.upload(&view).expect("upload");
            let result =
                string_empty_builtin(vec![Value::from("like"), Value::GpuTensor(handle.clone())])
                    .expect("string.empty");
            match result {
                Value::StringArray(sa) => {
                    assert_eq!(sa.shape, vec![0, 3]);
                    assert_eq!(sa.data.len(), 0);
                }
                other => panic!("expected string array, got {other:?}"),
            }
            let _ = provider.free(&handle);
        });
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn gpu_dimension_arguments_are_gathered() {
        test_support::with_test_provider(|provider| {
            let dims = Tensor::new(vec![0.0, 5.0, 3.0], vec![1, 3]).unwrap();
            let view = HostTensorView {
                data: &dims.data,
                shape: &dims.shape,
            };
            let handle = provider.upload(&view).expect("upload");
            let result =
                string_empty_builtin(vec![Value::GpuTensor(handle.clone())]).expect("string.empty");
            match result {
                Value::StringArray(sa) => {
                    assert_eq!(sa.shape, vec![0, 5, 3]);
                    assert_eq!(sa.data.len(), 0);
                }
                other => panic!("expected string array, got {other:?}"),
            }
            let _ = provider.free(&handle);
        });
    }

    #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
    #[test]
    fn rejects_negative_dimension() {
        let err = error_message(
            string_empty_builtin(vec![Value::from(-1.0)]).expect_err("expected error"),
        );
        assert!(
            err.contains("matrix dimensions must be non-negative"),
            "unexpected error: {err}"
        );
    }
}