gnu: onnxruntime: Fix integration into projects.

* gnu/packages/patches/onnxruntime-1.22.0-splittosequence-bool.patch: New file.
* gnu/local.mk (dist_patch_DATA): Register it.
* gnu/packages/machine-learning.scm (onnxruntime)[source]: Use patch.
[arguments]<#:configure-flags>: Enable onnxruntime_USE_FULL_PROTOBUF,
Protobuf_USE_STATIC_LIBS=ON.
[inputs]: Replace onnx by onnx-for-onnxruntime.
Replace protobuf by protobuf-static-for-onnxruntime.

Change-Id: I20d8172b9c4e81ce351cf40102be7db388387a0d

2 files changed, 81 insertions, 4 deletions

diff --git a/gnu/packages/machine-learning.scm b/gnu/packages/machine-learning.scm
index 005fca18ab8..74084019b64 100644
--- a/gnu/packages/machine-learning.scm
+++ b/gnu/packages/machine-learning.scm
@@ -2203,7 +2203,14 @@ with a single function call.")
               (commit (string-append "v" version))))
        (file-name (git-file-name name version))
        (sha256
-        (base32 "0z2s79l4wdilssw9lmj319ypyyqi2y0dx0fpwr2yhq8bax3ci50n"))))
+        (base32 "0z2s79l4wdilssw9lmj319ypyyqi2y0dx0fpwr2yhq8bax3ci50n"))
+       (patches
+        (search-patches
+         ;; SplitToSequence operator was missing bool type support, despite
+         ;; the ONNX spec allowing it in models.  Fixed upstream in 1.23+,
+         ;; but we cannot upgrade because ROCm support was removed in 1.23.
+         ;; <https://github.com/microsoft/onnxruntime/pull/24929>
+         "onnxruntime-1.22.0-splittosequence-bool.patch"))))
     (build-system cmake-build-system)
     (arguments
      (list
@@ -2217,7 +2224,12 @@ with a single function call.")
               "-Donnxruntime_BUILD_SHARED_LIB=ON"
               "-Donnxruntime_ENABLE_LTO=ON"
               "-Donnxruntime_ENABLE_PYTHON=ON"
-              "-Donnxruntime_USE_FULL_PROTOBUF=OFF"
+              "-Donnxruntime_USE_FULL_PROTOBUF=ON"
+              ;; Link static protobuf, matching upstream's vcpkg approach.
+              ;; This keeps the protobuf descriptor pool inside
+              ;; onnxruntime_pybind11_state.so separate from the shared
+              ;; libprotobuf.so used by the regular onnx Python package.
+              "-DProtobuf_USE_STATIC_LIBS=ON"
               ;; XXX: Fixes build with gcc@14.
               "-DCMAKE_CXX_FLAGS=-Wl,-z,noexecstack")
       #:phases
@@ -2296,8 +2308,8 @@ with a single function call.")
            flatbuffers-23.5
            googletest
            nlohmann-json
-           onnx
-           protobuf
+           onnx-for-onnxruntime
+           protobuf-static-for-onnxruntime
            pybind11-2
            re2-next
            safeint
diff --git a/gnu/packages/patches/onnxruntime-1.22.0-splittosequence-bool.patch b/gnu/packages/patches/onnxruntime-1.22.0-splittosequence-bool.patch
new file mode 100644
index 00000000000..373471f4844
--- /dev/null
+++ b/gnu/packages/patches/onnxruntime-1.22.0-splittosequence-bool.patch
@@ -0,0 +1,65 @@
+From 3b18e8ac863f817cddf0a2fc1ff89aa05fe60557 Mon Sep 17 00:00:00 2001
+From: Mauricio Cortazar <mcortazar@truora.com>
+Date: Mon, 2 Jun 2025 12:51:41 -0500
+Subject: [PATCH] Add support for bool type in SplitToSequence.
+
+Add support for `bool` type to address the below issue.
+This PR fixes <https://github.com/microsoft/onnxruntime/issues/12286>
+---
+ docs/OperatorKernels.md                          |  2 +-
+ .../core/providers/cpu/sequence/sequence_ops.cc  |  2 +-
+ .../providers/cpu/sequence/sequence_ops_test.cc  | 16 +++++++++++++++-
+ 3 files changed, 17 insertions(+), 3 deletions(-)
+
+diff --git a/docs/OperatorKernels.md b/docs/OperatorKernels.md
+index 8c1ab002bce67..fc7b1ed21575c 100644
+--- a/docs/OperatorKernels.md
++++ b/docs/OperatorKernels.md
+@@ -437,7 +437,7 @@ Do not modify directly.*
+ |||[13, 17]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+ |||[11, 12]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+ |||[2, 10]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+-|SplitToSequence|*in* input:**T**<br> *in* split:**I**<br> *out* output_sequence:**S**|11+|**I** = tensor(int32), tensor(int64)<br/> **S** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8))<br/> **T** = tensor(double), tensor(float), tensor(float16), tensor(int32), tensor(int64), tensor(string)|
++|SplitToSequence|*in* input:**T**<br> *in* split:**I**<br> *out* output_sequence:**S**|11+|**I** = tensor(int32), tensor(int64)<br/> **S** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8))<br/> **T** = tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int32), tensor(int64), tensor(string)|
+ |Sqrt|*in* X:**T**<br> *out* Y:**T**|13+|**T** = tensor(double), tensor(float)|
+ |||[6, 12]|**T** = tensor(double), tensor(float)|
+ |Squeeze|*in* data:**T**<br> *in* axes:**tensor(int64)**<br> *out* squeezed:**T**<br><br>or<br><br>*in* data:**T**<br> *out* squeezed:**T**|23+|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+diff --git a/onnxruntime/core/providers/cpu/sequence/sequence_ops.cc b/onnxruntime/core/providers/cpu/sequence/sequence_ops.cc
+index 7a27b04ece7cf..d2541cf3d35ce 100644
+--- a/onnxruntime/core/providers/cpu/sequence/sequence_ops.cc
++++ b/onnxruntime/core/providers/cpu/sequence/sequence_ops.cc
+@@ -339,7 +339,7 @@ ONNX_CPU_OPERATOR_KERNEL(
+     11,
+     KernelDefBuilder()
+         .TypeConstraint("T",
+-                        BuildKernelDefConstraints<float, MLFloat16, double, int32_t, int64_t, std::string>())
++                        BuildKernelDefConstraints<float, MLFloat16, double, int32_t, int64_t, bool, std::string>())
+         .TypeConstraint("S", DataTypeImpl::AllSequenceTensorTypes())
+         .TypeConstraint("I", BuildKernelDefConstraints<int32_t, int64_t>()),
+     SplitToSequence);
+diff --git a/onnxruntime/test/providers/cpu/sequence/sequence_ops_test.cc b/onnxruntime/test/providers/cpu/sequence/sequence_ops_test.cc
+index c2d64b8e5ee4a..d819b0973adc2 100644
+--- a/onnxruntime/test/providers/cpu/sequence/sequence_ops_test.cc
++++ b/onnxruntime/test/providers/cpu/sequence/sequence_ops_test.cc
+@@ -536,5 +536,19 @@ TEST(SequenceOpsTest, SplitToSequence_PositiveAxisDontKeepDims) {
+   test.AddSeqOutput("S2", output);
+   test.Run();
+ }
++
++TEST(SequenceOpsTest, SplitToSequence_BoolSplit) {
++  OpTester test("SplitToSequence", 11);
++  test.AddInput<bool>("input", {4, 2}, std::initializer_list<bool>({1, 1, 1, 1, 0, 0, 0, 0}));
++  int64_t axis = 0;
++  test.AddAttribute("axis", axis);
++  SeqTensors<bool> output;
++  output.AddTensor({1, 2}, {1, 1});
++  output.AddTensor({1, 2}, {1, 1});
++  output.AddTensor({1, 2}, {0, 0});
++  output.AddTensor({1, 2}, {0, 0});
++  test.AddSeqOutput("S2", output);
++  test.Run();
++}
+ }  // namespace test
+-}  // namespace onnxruntime
+\ No newline at end of file
++}  // namespace onnxruntime