type_helper.h

Provides the TensorType enumeration, the ToTypeString converter, and the SeqTypeOf/OptTypeOf/OptSeqTypeOf constexpr helpers used across the onnx-light library stack.

namespace onnx_light

Alias that makes onnx-light headers compatible with code that references ONNX_LIGHT_NAMESPACE (the macro used in the standard onnx package).

Set to ONNX_LIGHT_NAMESPACE so both names resolve to the same namespace.

Symbol-visibility attribute for the public onnx-light C++ API.

Maps the upstream compatibility macro to onnx-light’s explicit proto ABI annotation. This keeps declarations from vendored ONNX headers visible when lib_onnx_proto uses hidden visibility by default.

Namespace alias so that ONNX C++ code (and consumers such as onnxruntime) that refers to the literal onnx namespace — rather than the ONNX_NAMESPACE macro — resolves to the onnx-light namespace. The standard onnx package lives in namespace onnx; onnx-light uses onnx_light (via ONNX_LIGHT_NAMESPACE), so this alias keeps onnx-light a true drop-in. It is only introduced when the onnx-light namespace differs from onnx.

namespace onnx_proto

Enums

enum class TensorType : uint8_t

Identifies an element or sequence tensor type supported by onnx-light.

Each enumerator corresponds to a concrete ONNX element type or to a sequence-of-tensor or optional-tensor type used in type-constraint definitions. The mapping from an enumerator to its canonical ONNX type string is implemented exhaustively by ToTypeString(); the naming convention is:

• kXxx → "tensor(xxx)", e.g. kFloat → "tensor(float)", kInt64 → "tensor(int64)", kBfloat16 → "tensor(bfloat16)".

• kSeqXxx → "seq(tensor(xxx))", e.g. kSeqFloat → "seq(tensor(float))". The two map-valued sequences are kSeqMapStringFloat → "seq(map(string, float))" and kSeqMapInt64Float → "seq(map(int64, float))".

• kMapXxxYyy → "map(xxx, yyy)", e.g. kMapStringInt64 → "map(string, int64)". Used for operators like DictVectorizer whose inputs are dictionaries.

• kOptXxx → "optional(tensor(xxx))" and kOptSeqXxx → "optional(seq(tensor(xxx)))".

• kUndefined → "tensor(undefined)".

Values:

enumerator kBool

enumerator kString

enumerator kUint8

enumerator kUint16

enumerator kUint32

enumerator kUint64

enumerator kInt8

enumerator kInt16

enumerator kInt32

enumerator kInt64

enumerator kFloat16

enumerator kFloat

enumerator kDouble

enumerator kBfloat16

enumerator kFloat8e4m3fn

enumerator kFloat8e4m3fnuz

enumerator kFloat8e5m2

enumerator kFloat8e5m2fnuz

enumerator kFloat8e8m0

enumerator kFloat4e2m1

enumerator kUint4

enumerator kInt4

enumerator kUint2

enumerator kInt2

enumerator kComplex64

enumerator kComplex128

enumerator kSeqBool

enumerator kSeqString

enumerator kSeqUint8

enumerator kSeqUint16

enumerator kSeqUint32

enumerator kSeqUint64

enumerator kSeqInt8

enumerator kSeqInt16

enumerator kSeqInt32

enumerator kSeqInt64

enumerator kSeqFloat16

enumerator kSeqFloat

enumerator kSeqDouble

enumerator kSeqComplex64

enumerator kSeqComplex128

enumerator kSeqMapStringFloat

enumerator kSeqMapInt64Float

enumerator kMapStringInt64

enumerator kMapInt64String

enumerator kMapInt64Float

enumerator kMapInt64Double

enumerator kMapStringFloat

enumerator kMapStringDouble

enumerator kOptSeqBool

enumerator kOptSeqString

enumerator kOptSeqUint8

enumerator kOptSeqUint16

enumerator kOptSeqUint32

enumerator kOptSeqUint64

enumerator kOptSeqInt8

enumerator kOptSeqInt16

enumerator kOptSeqInt32

enumerator kOptSeqInt64

enumerator kOptSeqFloat16

enumerator kOptSeqFloat

enumerator kOptSeqDouble

enumerator kOptSeqComplex64

enumerator kOptSeqComplex128

enumerator kOptBool

enumerator kOptString

enumerator kOptUint8

enumerator kOptUint16

enumerator kOptUint32

enumerator kOptUint64

enumerator kOptInt8

enumerator kOptInt16

enumerator kOptInt32

enumerator kOptInt64

enumerator kOptFloat16

enumerator kOptFloat

enumerator kOptDouble

enumerator kOptComplex64

enumerator kOptComplex128

enumerator kUndefined

Functions

inline constexpr const char *ToTypeString(TensorType type)

Returns the ONNX type-string representation of a TensorType value.

Parameters:

type – Tensor type enumerator to convert.

Returns:

Null-terminated string such as "tensor(float)" or "seq(tensor(int64))".

inline constexpr TensorType SeqTypeOf(TensorType elem)

Maps a scalar tensor TensorType to the corresponding sequence TensorType (e.g. kFloat → kSeqFloat). Returns kUndefined when no sequence variant exists for the given element type (low-precision / extended float types that ONNX does not define sequence forms for).

inline constexpr TensorType OptTypeOf(TensorType elem)

Maps a scalar tensor TensorType to the corresponding optional TensorType (e.g. kFloat → kOptFloat). Returns kUndefined when no optional variant exists for the given element type.

inline constexpr TensorType OptSeqTypeOf(TensorType elem)

Maps a scalar tensor TensorType to the corresponding optional-sequence TensorType (e.g. kFloat → kOptSeqFloat). Returns kUndefined when no optional-sequence variant exists for the given element type.