Struct ShapeData 

pub struct ShapeData {
    pub dtype: DataType,
    pub dims: Vec<usize>,
    pub elems: Vec<DimExpr>,
    pub float_elems: Option<Vec<f64>>,
}

The known element values of a rank-0 or rank-1 integer tensor flowing through a shape-computation subgraph.

Integer tensors carry their values in elems as DimExprs. Floating-point scalar constants (needed by Range, whose float form the CPU kernel supports) are carried additively in float_elems; for those, elems is empty. Eq is not derived because f64 is not Eq — ShapeData is only ever used as a map value, so structural equality is not required.

Fields

dtype: DataType

The element type (Int64/Int32, or Bool for masks; Float32/ Float64 for the float-scalar side-channel).

dims: Vec<usize>

Static dimensions. Empty == rank-0 scalar; [n] == rank-1 vector.

elems: Vec<DimExpr>

Row-major integer elements. Length is 1 for a scalar, dims[0] for a vector. Empty for floating-point shape-data (see float_elems).

float_elems: Option<Vec<f64>>

Row-major floating-point elements for rank-0/rank-1 floating-point constants used by Range and Resize; None otherwise.

Implementations

impl ShapeData

pub fn scalar(dtype: DataType, value: DimExpr) -> Self

A rank-0 scalar holding value.

pub fn vector(dtype: DataType, elems: Vec<DimExpr>) -> Self

A rank-1 vector holding elems.

pub fn float_scalar(dtype: DataType, value: f64) -> Self

A rank-0 floating-point scalar holding value.

pub fn float_vector(dtype: DataType, values: Vec<f64>) -> Self

A rank-1 floating-point vector holding values.

pub fn as_float_scalar(&self) -> Option<f64>

This value interpreted as a floating-point scalar, if it is one.

pub fn as_float_vector(&self) -> Option<&[f64]>

This value interpreted as a floating-point vector, if it is one.

pub fn is_scalar(&self) -> bool

Whether this is a rank-0 scalar.

pub fn within_bounds(&self) -> bool

Whether the element count is within MAX_SHAPE_DATA_ELEMS.

pub fn as_shape(&self) -> Vec<DimExpr>

This value’s contents interpreted as a shape (each element is a dim).

Used by Reshape/Expand/ConstantOfShape to turn a resolved shape vector into the output shape.

pub fn from_tensor(dtype: DataType, dims: &[usize], data: &[u8]) -> Option<Self>

Extract shape-data from a concrete constant tensor, if it is a small (rank ≤ 1, within MAX_SHAPE_DATA_ELEMS) integer or boolean tensor.

Non-integer tensors, higher-rank tensors, and over-large tensors are not tracked (returns None) — only the shape-computation operands matter.

Trait Implementations

impl Clone for ShapeData

fn clone(&self) -> ShapeData

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ShapeData

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for ShapeData

fn eq(&self, other: &ShapeData) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for ShapeData