[Node.js] Fix float16 tensor input support

Accept both Uint16Array and Float16Array for float16 tensors in the
Node.js binding. Float16Array is a newer JavaScript type (ES2024) that
N-API supports as napi_float16_array (type 11) in Node.js 23+.

For older Node.js versions, define napi_float16_array to ensure
compatibility when users pass Float16Array data.

Fixes #26791

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Author: ingyukoh

js/node/src/tensor_helper.cc

@@ -11,6 +11,12 @@
 #include "tensor_helper.h"
 #include "inference_session_wrap.h"
 
+// napi_float16_array was added in Node.js 23 (N-API version 10).
+// Define it for older Node.js versions to support Float16Array input tensors.
+#ifndef napi_float16_array
+#define napi_float16_array static_cast<napi_typedarray_type>(11)
+#endif
+
 // make sure consistent with origin definition
 static_assert(ONNX_TENSOR_ELEMENT_DATA_TYPE_UNDEFINED == 0, "definition not consistent with OnnxRuntime");
 static_assert(ONNX_TENSOR_ELEMENT_DATA_TYPE_FLOAT == 1, "definition not consistent with OnnxRuntime");
@@ -196,9 +202,19 @@ Ort::Value NapiValueToOrtValue(Napi::Env env, Napi::Value value, OrtMemoryInfo*
 
       auto tensorDataTypedArray = tensorDataValue.As<Napi::TypedArray>();
       std::underlying_type_t<napi_typedarray_type> typedArrayType = tensorDataValue.As<Napi::TypedArray>().TypedArrayType();
-      ORT_NAPI_THROW_TYPEERROR_IF(DATA_TYPE_TYPEDARRAY_MAP[elemType] != typedArrayType, env,
-                                  "Tensor.data must be a typed array (", DATA_TYPE_TYPEDARRAY_MAP[elemType], ") for ",
-                                  tensorTypeString, " tensors, but got typed array (", typedArrayType, ").");
+
+      // For float16 tensors, accept both Uint16Array and Float16Array.
+      // Float16Array is a newer JavaScript type (ES2024) that may be passed by users.
+      // Both use 16-bit storage, so they are compatible at the binary level.
+      bool isValidTypedArray = (DATA_TYPE_TYPEDARRAY_MAP[elemType] == typedArrayType);
+      if (!isValidTypedArray && elemType == ONNX_TENSOR_ELEMENT_DATA_TYPE_FLOAT16) {
+        // Accept Float16Array (napi_float16_array = 11) for float16 tensors
+        isValidTypedArray = (typedArrayType == napi_float16_array);
+      }
+
+      ORT_NAPI_THROW_TYPEERROR_IF(!isValidTypedArray, env,
+                                  "Tensor.data must be a typed array (", DATA_TYPE_TYPEDARRAY_MAP[elemType],
+                                  " or Float16Array) for ", tensorTypeString, " tensors, but got typed array (", typedArrayType, ").");
 
       char* buffer = reinterpret_cast<char*>(tensorDataTypedArray.ArrayBuffer().Data());
       size_t bufferByteOffset = tensorDataTypedArray.ByteOffset();