diff --git a/.gitignore b/.gitignore
index 36ef4dbb..130a530f 100644
--- a/.gitignore
+++ b/.gitignore
@@ -5,6 +5,7 @@ build_host_protoc
 build_android
 build_ios
 build_*
+_subbuild/
 .build_debug/*
 .build_release/*
 distribute/*
diff --git a/CMakeLists.txt b/CMakeLists.txt
index a783f43b..ee4be1f9 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -38,7 +38,9 @@ option(OCOS_ENABLE_SELECTED_OPLIST "Enable including the selected_ops tool file"
 
 
 function(disable_all_operators)
+  set(OCOS_ENABLE_RE2_REGEX OFF CACHE INTERNAL "")
   set(OCOS_ENABLE_TF_STRING OFF CACHE INTERNAL "")
+  set(OCOS_ENABLE_WORDPIECE_TOKENIZER OFF CACHE INTERNAL "")
   set(OCOS_ENABLE_GPT2_TOKENIZER OFF CACHE INTERNAL "")
   set(OCOS_ENABLE_SPM_TOKENIZER OFF CACHE INTERNAL "")
   set(OCOS_ENABLE_BERT_TOKENIZER OFF CACHE INTERNAL "")
@@ -122,6 +124,11 @@ if (OCOS_ENABLE_TF_STRING)
   list(APPEND TARGET_SRC ${TARGET_SRC_KERNELS} ${TARGET_SRC_HASH})
 endif()
 
+if (OCOS_ENABLE_RE2_REGEX)
+  file(GLOB TARGET_SRC_RE2_KERNELS "operators/text/re2_strings/*.cc" "operators/text/re2_strings/*.h*")
+  list(APPEND TARGET_SRC ${TARGET_SRC_RE2_KERNELS})
+endif()
+
 if (OCOS_ENABLE_MATH)
   set(DLIB_ISO_CPP_ONLY ON CACHE INTERNAL "")
   set(DLIB_NO_GUI_SUPPORT ON CACHE INTERNAL "")
diff --git a/includes/ocos.h b/includes/ocos.h
index 6f0f7c36..a396beaa 100644
--- a/includes/ocos.h
+++ b/includes/ocos.h
@@ -59,9 +59,6 @@ struct OrtTensorDimensions : std::vector<int64_t> {
 };
 
 
-struct CustomOpClassBegin{
-};
-
 template <typename... Args>
 class CuopContainer {
  public:
@@ -86,6 +83,9 @@ class CuopContainer {
   std::vector<OrtCustomOp*> ocos_list_;
 };
 
+struct CustomOpClassBegin{
+};
+
 typedef std::function<const OrtCustomOp**()> FxLoadCustomOpFactory;
 
 template <typename _Begin_place_holder, typename... Args>
@@ -99,3 +99,15 @@ const OrtCustomOp* FetchPyCustomOps(size_t& count);
 OrtStatusPtr RegisterPythonDomainAndOps(OrtSessionOptions*, const OrtApi*);
 bool EnablePyCustomOps(bool enable = true);
 #endif
+
+#ifdef ENABLE_MATH
+extern FxLoadCustomOpFactory LoadCustomOpClasses_Math;
+#endif  // ENABLE_MATH
+
+#ifdef ENABLE_TOKENIZER
+extern FxLoadCustomOpFactory LoadCustomOpClasses_Tokenizer;
+#endif // ENABLE_TOKENIZER
+
+#ifdef ENABLE_TF_STRING
+extern FxLoadCustomOpFactory LoadCustomOpClasses_Text;
+#endif  // ENABLE_TF_STRING
diff --git a/operators/math/math.cc b/operators/math/math.cc
index f9eeec9e..6ca2896e 100644
--- a/operators/math/math.cc
+++ b/operators/math/math.cc
@@ -1,6 +1,13 @@
 #include "ocos.h"
 #include "negpos.hpp"
 #include "inverse.hpp"
+#include "segment_extraction.hpp"
+#include "segment_sum.hpp"
 
 
-FxLoadCustomOpFactory LoadCustomOpClasses_Math = LoadCustomOpClasses<CustomOpClassBegin, CustomOpNegPos, CustomOpInverse>;
+FxLoadCustomOpFactory LoadCustomOpClasses_Math = 
+    LoadCustomOpClasses<CustomOpClassBegin, 
+                        CustomOpNegPos,
+                        CustomOpInverse,
+                        CustomOpSegmentExtraction,
+                        CustomOpSegmentSum>;
diff --git a/operators/text/op_segement_extraction.cc b/operators/math/segement_extraction.cc
similarity index 98%
rename from operators/text/op_segement_extraction.cc
rename to operators/math/segement_extraction.cc
index 3ae086da..461168bd 100644
--- a/operators/text/op_segement_extraction.cc
+++ b/operators/math/segement_extraction.cc
@@ -1,7 +1,7 @@
 // Copyright (c) Microsoft Corporation. All rights reserved.
 // Licensed under the MIT License.
 
-#include "op_segment_extraction.hpp"
+#include "segment_extraction.hpp"
 
 KernelSegmentExtraction::KernelSegmentExtraction(OrtApi api) : BaseKernel(api) {
 }
diff --git a/operators/text/op_segment_extraction.hpp b/operators/math/segment_extraction.hpp
similarity index 96%
rename from operators/text/op_segment_extraction.hpp
rename to operators/math/segment_extraction.hpp
index bcdb9c3e..9425b33d 100644
--- a/operators/text/op_segment_extraction.hpp
+++ b/operators/math/segment_extraction.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelSegmentExtraction : BaseKernel {
diff --git a/operators/text/op_segment_sum.cc b/operators/math/segment_sum.cc
similarity index 99%
rename from operators/text/op_segment_sum.cc
rename to operators/math/segment_sum.cc
index 79ffd72a..b81e5126 100644
--- a/operators/text/op_segment_sum.cc
+++ b/operators/math/segment_sum.cc
@@ -1,7 +1,7 @@
 // Copyright (c) Microsoft Corporation. All rights reserved.
 // Licensed under the MIT License.
 
-#include "op_segment_sum.hpp"
+#include "segment_sum.hpp"
 
 template <typename T>
 void KernelSegmentSum_Compute(Ort::CustomOpApi& ort_, OrtKernelContext* context) {
diff --git a/operators/text/op_segment_sum.hpp b/operators/math/segment_sum.hpp
similarity index 96%
rename from operators/text/op_segment_sum.hpp
rename to operators/math/segment_sum.hpp
index b985ffe1..c3f4505a 100644
--- a/operators/text/op_segment_sum.hpp
+++ b/operators/math/segment_sum.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelSegmentSum : BaseKernel {
diff --git a/operators/text/kernels.h b/operators/text/kernels.h
deleted file mode 100644
index 7168bac7..00000000
--- a/operators/text/kernels.h
+++ /dev/null
@@ -1,9 +0,0 @@
-// Copyright (c) Microsoft Corporation. All rights reserved.
-// Licensed under the MIT License.
-
-#pragma once
-
-#include "ocos.h"
-
-
-// TO BE DELETED.
diff --git a/operators/text/op_equal.hpp b/operators/text/op_equal.hpp
index 0bed3107..18a4b261 100644
--- a/operators/text/op_equal.hpp
+++ b/operators/text/op_equal.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringEqual : BaseKernel {
diff --git a/operators/text/op_equal_impl.hpp b/operators/text/op_equal_impl.hpp
index de97a8a3..9186cf46 100644
--- a/operators/text/op_equal_impl.hpp
+++ b/operators/text/op_equal_impl.hpp
@@ -13,7 +13,7 @@ class BroadcastIteratorRight {
                          const std::vector<int64_t>& shape2,
                          const T1* p1, const T2* p2, T3* p3) : p1_(p1), p2_(p2), p3_(p3), shape1_(shape1) {
     if (shape2.size() > shape1.size())
-      throw std::runtime_error("shape2 must have less dimensions than shape1");
+      ORT_CXX_API_THROW("shape2 must have less dimensions than shape1", ORT_INVALID_ARGUMENT);
     shape2_.resize(shape1_.size());
     cum_shape2_.resize(shape1_.size());
     total_ = 1;
@@ -26,8 +26,8 @@ class BroadcastIteratorRight {
         shape2_[i] = shape2[i];
       }
       if (shape2[i] != 1 && shape1[i] != shape2[i]) {
-        throw std::runtime_error(MakeString(
-            "Cannot broadcast dimension ", i, " left:", shape1[i], " right:", shape2[i]));
+        ORT_CXX_API_THROW(MakeString(
+            "Cannot broadcast dimension ", i, " left:", shape1[i], " right:", shape2[i]), ORT_INVALID_ARGUMENT);
       }
     }
     cum_shape2_[shape2_.size() - 1] = 1;
@@ -84,7 +84,7 @@ class BroadcastIteratorRight {
     template <typename TCMP>
     void loop(TCMP& cmp, BroadcastIteratorRightState& it, int64_t pos = 0) {
       if (pos != 0)
-        throw std::runtime_error("Not implemented yet.");
+        ORT_CXX_API_THROW("Not implemented yet.", ORT_NOT_IMPLEMENTED);
       while (!end()) {
         *p3 = cmp(*p1, *p2);
         next();
diff --git a/operators/text/op_ragged_tensor.cc b/operators/text/op_ragged_tensor.cc
index c8c7bded..7e312548 100644
--- a/operators/text/op_ragged_tensor.cc
+++ b/operators/text/op_ragged_tensor.cc
@@ -14,8 +14,8 @@ void KernelRaggedTensorToSparse::Compute(OrtKernelContext* context) {
   OrtTensorDimensions d_length(ort_, n_elements);
 
   if (d_length.size() != 1)
-    throw std::runtime_error(MakeString(
-        "First input must have one dimension not ", d_length.size(), "."));
+    ORT_CXX_API_THROW(MakeString(
+        "First input must have one dimension not ", d_length.size(), "."), ORT_INVALID_ARGUMENT);
   int64_t n_els = d_length[0] - 1;
   int64_t n_values = p_n_elements[n_els];
   std::vector<int64_t> shape{n_values, 2};
@@ -110,9 +110,9 @@ void KernelRaggedTensorToDense::Compute(OrtKernelContext* context) {
   for (int64_t i = 0; i < size - 1; ++i) {
     pos_end = pos + max_col;
     if (pos_end > shape_out_size)
-      throw std::runtime_error(MakeString(
+      ORT_CXX_API_THROW(MakeString(
           "Unexpected index ", pos_end, " greather than ", shape_out[0], "x", shape_out[1],
-          " - i=", i, " size=", size, "."));
+          " - i=", i, " size=", size, "."), ORT_INVALID_ARGUMENT);
     for (j = p_indices[i]; j < p_indices[i + 1]; ++j, ++pos) {
       dense[pos] = p_values[j];
     }
@@ -168,9 +168,9 @@ void KernelStringRaggedTensorToDense::Compute(OrtKernelContext* context) {
   for (int64_t i = 0; i < size - 1; ++i) {
     pos_end = pos + max_col;
     if (pos_end > shape_out_size)
-      throw std::runtime_error(MakeString(
+      ORT_CXX_API_THROW(MakeString(
           "Unexpected index ", pos_end, " greather than ", shape_out[0], "x", shape_out[1],
-          " - i=", i, " size=", size, "."));
+          " - i=", i, " size=", size, "."), ORT_INVALID_ARGUMENT);
     for (j = p_indices[i]; j < p_indices[i + 1]; ++j, ++pos) {
       dense[pos] = input[j];
     }
@@ -210,6 +210,6 @@ ONNXTensorElementDataType CustomOpStringRaggedTensorToDense::GetInputType(size_t
     case 3:
       return ONNX_TENSOR_ELEMENT_DATA_TYPE_INT64;
     default:
-      throw std::runtime_error(MakeString("[StringRaggedTensorToDense] Unexpected output index ", index, "."));
+      ORT_CXX_API_THROW(MakeString("[StringRaggedTensorToDense] Unexpected output index ", index, "."), ORT_INVALID_ARGUMENT);
   }
 };
diff --git a/operators/text/op_ragged_tensor.hpp b/operators/text/op_ragged_tensor.hpp
index 7fb3c8dd..eb0241b9 100644
--- a/operators/text/op_ragged_tensor.hpp
+++ b/operators/text/op_ragged_tensor.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 
 struct KernelRaggedTensorToSparse : BaseKernel {
   KernelRaggedTensorToSparse(OrtApi api);
diff --git a/operators/text/string_regex_replace.cc b/operators/text/re2_strings/string_regex_replace.cc
similarity index 100%
rename from operators/text/string_regex_replace.cc
rename to operators/text/re2_strings/string_regex_replace.cc
diff --git a/operators/text/string_regex_replace.hpp b/operators/text/re2_strings/string_regex_replace.hpp
similarity index 97%
rename from operators/text/string_regex_replace.hpp
rename to operators/text/re2_strings/string_regex_replace.hpp
index 2fb63fb6..6d3cb982 100644
--- a/operators/text/string_regex_replace.hpp
+++ b/operators/text/re2_strings/string_regex_replace.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringRegexReplace : BaseKernel {
diff --git a/operators/text/string_regex_split.cc b/operators/text/re2_strings/string_regex_split.cc
similarity index 100%
rename from operators/text/string_regex_split.cc
rename to operators/text/re2_strings/string_regex_split.cc
index 91650306..544a8cc8 100644
--- a/operators/text/string_regex_split.cc
+++ b/operators/text/re2_strings/string_regex_split.cc
@@ -3,9 +3,9 @@
 
 #include "string_regex_split.hpp"
 #include "string_regex_split_re.hpp"
+#include "string_tensor.h"
 #include <vector>
 #include <cmath>
-#include "string_tensor.h"
 
 KernelStringRegexSplitWithOffsets::KernelStringRegexSplitWithOffsets(OrtApi api, const OrtKernelInfo* info) : BaseKernel(api, info) {
 }
diff --git a/operators/text/string_regex_split.hpp b/operators/text/re2_strings/string_regex_split.hpp
similarity index 97%
rename from operators/text/string_regex_split.hpp
rename to operators/text/re2_strings/string_regex_split.hpp
index bc182729..9d54ada2 100644
--- a/operators/text/string_regex_split.hpp
+++ b/operators/text/re2_strings/string_regex_split.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 // See https://github.com/tensorflow/text/blob/master/docs/api_docs/python/text/regex_split_with_offsets.md.
diff --git a/operators/text/string_regex_split_re.hpp b/operators/text/re2_strings/string_regex_split_re.hpp
similarity index 100%
rename from operators/text/string_regex_split_re.hpp
rename to operators/text/re2_strings/string_regex_split_re.hpp
diff --git a/operators/text/string_concat.cc b/operators/text/string_concat.cc
index 15699b0c..39093714 100644
--- a/operators/text/string_concat.cc
+++ b/operators/text/string_concat.cc
@@ -20,7 +20,7 @@ void KernelStringConcat::Compute(OrtKernelContext* context) {
   OrtTensorDimensions right_dim(ort_, right);
 
   if (left_dim != right_dim) {
-    throw std::runtime_error("Two input tensor should have the same dimension.");
+    ORT_CXX_API_THROW("Two input tensor should have the same dimension.", ORT_INVALID_ARGUMENT);
   }
 
   std::vector<std::string> left_value;
diff --git a/operators/text/string_concat.hpp b/operators/text/string_concat.hpp
index 2be1dcf9..a96fa73e 100644
--- a/operators/text/string_concat.hpp
+++ b/operators/text/string_concat.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringConcat : BaseKernel {
diff --git a/operators/text/string_ecmaregex_replace.hpp b/operators/text/string_ecmaregex_replace.hpp
index 5e9b34e9..1098f1e3 100644
--- a/operators/text/string_ecmaregex_replace.hpp
+++ b/operators/text/string_ecmaregex_replace.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringECMARegexReplace : BaseKernel {
diff --git a/operators/text/string_ecmaregex_split.hpp b/operators/text/string_ecmaregex_split.hpp
index 01a6c8ab..16fe8934 100644
--- a/operators/text/string_ecmaregex_split.hpp
+++ b/operators/text/string_ecmaregex_split.hpp
@@ -4,7 +4,7 @@
 #pragma once
 
 #include <regex>
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 // See https://github.com/tensorflow/text/blob/master/docs/api_docs/python/text/regex_split_with_offsets.md.
diff --git a/operators/text/string_hash.cc b/operators/text/string_hash.cc
index 9a6f7482..74f3e9c8 100644
--- a/operators/text/string_hash.cc
+++ b/operators/text/string_hash.cc
@@ -23,9 +23,9 @@ void KernelStringHash::Compute(OrtKernelContext* context) {
   // Verifications
   OrtTensorDimensions num_buckets_dimensions(ort_, num_buckets);
   if (num_buckets_dimensions.size() != 1 || num_buckets_dimensions[0] != 1)
-    throw std::runtime_error(MakeString(
+    ORT_CXX_API_THROW(MakeString(
         "num_buckets must contain only one element. It has ",
-        num_buckets_dimensions.size(), " dimensions."));
+        num_buckets_dimensions.size(), " dimensions."), ORT_INVALID_ARGUMENT);
 
   // Setup output
   OrtTensorDimensions dimensions(ort_, input);
@@ -60,7 +60,7 @@ ONNXTensorElementDataType CustomOpStringHash::GetInputType(size_t index) const {
     case 1:
       return ONNX_TENSOR_ELEMENT_DATA_TYPE_INT64;
     default:
-      throw std::runtime_error(MakeString("Unexpected input index ", index));
+      ORT_CXX_API_THROW(MakeString("Unexpected input index ", index), ORT_INVALID_ARGUMENT);
   }
 };
 
@@ -86,9 +86,9 @@ void KernelStringHashFast::Compute(OrtKernelContext* context) {
   // Verifications
   OrtTensorDimensions num_buckets_dimensions(ort_, num_buckets);
   if (num_buckets_dimensions.size() != 1 || num_buckets_dimensions[0] != 1)
-    throw std::runtime_error(MakeString(
+    ORT_CXX_API_THROW(MakeString(
         "num_buckets must contain only one element. It has ",
-        num_buckets_dimensions.size(), " dimensions."));
+        num_buckets_dimensions.size(), " dimensions."), ORT_INVALID_ARGUMENT);
 
   // Setup output
   OrtTensorDimensions dimensions(ort_, input);
@@ -123,7 +123,7 @@ ONNXTensorElementDataType CustomOpStringHashFast::GetInputType(size_t index) con
     case 1:
       return ONNX_TENSOR_ELEMENT_DATA_TYPE_INT64;
     default:
-      throw std::runtime_error(MakeString("Unexpected input index ", index));
+      ORT_CXX_API_THROW(MakeString("Unexpected input index ", index), ORT_INVALID_ARGUMENT);
   }
 };
 
diff --git a/operators/text/string_hash.hpp b/operators/text/string_hash.hpp
index 82c73be3..6bf87d78 100644
--- a/operators/text/string_hash.hpp
+++ b/operators/text/string_hash.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringHash : BaseKernel {
diff --git a/operators/text/string_join.cc b/operators/text/string_join.cc
index 9b35e485..5f20a03c 100644
--- a/operators/text/string_join.cc
+++ b/operators/text/string_join.cc
@@ -20,13 +20,13 @@ void KernelStringJoin::Compute(OrtKernelContext* context) {
   // Setup output
   OrtTensorDimensions dimensions_sep(ort_, input_sep);
   if (dimensions_sep.size() != 1 || dimensions_sep[0] != 1)
-    throw std::runtime_error("Input 2 is the separator, it has 1 element.");
+    ORT_CXX_API_THROW("Input 2 is the separator, it has 1 element.", ORT_INVALID_ARGUMENT);
   OrtTensorDimensions dimensions_axis(ort_, input_axis);
   if (dimensions_axis.size() != 1 || dimensions_axis[0] != 1)
-    throw std::runtime_error("Input 3 is the axis, it has 1 element.");
+    ORT_CXX_API_THROW("Input 3 is the axis, it has 1 element.", ORT_INVALID_ARGUMENT);
   OrtTensorDimensions dimensions(ort_, input_X);
   if (*axis < 0 || *axis >= dimensions.size())
-    throw std::runtime_error(MakeString("axis must be positive and smaller than the number of dimension but it is ", *axis));
+    ORT_CXX_API_THROW(MakeString("axis must be positive and smaller than the number of dimension but it is ", *axis), ORT_INVALID_ARGUMENT);
 
   std::vector<int64_t> dimensions_out(dimensions.size() > 1 ? dimensions.size() - 1 : 1);
   if (dimensions.size() > 1) {
@@ -89,7 +89,7 @@ ONNXTensorElementDataType CustomOpStringJoin::GetInputType(size_t index) const {
     case 2:
       return ONNX_TENSOR_ELEMENT_DATA_TYPE_INT64;
     default:
-      throw std::runtime_error(MakeString("Unexpected input index ", index));
+      ORT_CXX_API_THROW(MakeString("Unexpected input index ", index), ORT_INVALID_ARGUMENT);
   }
 };
 
diff --git a/operators/text/string_join.hpp b/operators/text/string_join.hpp
index abcc5a50..76511c77 100644
--- a/operators/text/string_join.hpp
+++ b/operators/text/string_join.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringJoin : BaseKernel {
diff --git a/operators/text/string_length.hpp b/operators/text/string_length.hpp
index 0589abfc..f5b0822e 100644
--- a/operators/text/string_length.hpp
+++ b/operators/text/string_length.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringLength : BaseKernel {
diff --git a/operators/text/string_lower.hpp b/operators/text/string_lower.hpp
index 496161fb..31b79210 100644
--- a/operators/text/string_lower.hpp
+++ b/operators/text/string_lower.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringLower : BaseKernel {
diff --git a/operators/text/string_split.cc b/operators/text/string_split.cc
index fb8430c1..ce623a4a 100644
--- a/operators/text/string_split.cc
+++ b/operators/text/string_split.cc
@@ -20,13 +20,13 @@ void KernelStringSplit::Compute(OrtKernelContext* context) {
   // Setup output
   OrtTensorDimensions dimensions_sep(ort_, input_sep);
   if (dimensions_sep.size() != 1 || dimensions_sep[0] != 1)
-    throw std::runtime_error("Input 2 is the delimiter, it has 1 element.");
+    ORT_CXX_API_THROW("Input 2 is the delimiter, it has 1 element.", ORT_INVALID_ARGUMENT);
   OrtTensorDimensions dimensions_skip_empty(ort_, input_skip_empty);
   if (dimensions_skip_empty.size() != 1 || dimensions_skip_empty[0] != 1)
-    throw std::runtime_error("Input 3 is skip_empty, it has 1 element.");
+    ORT_CXX_API_THROW("Input 3 is skip_empty, it has 1 element.", ORT_INVALID_ARGUMENT);
   OrtTensorDimensions dimensions(ort_, input_X);
   if (dimensions.size() != 1)
-    throw std::runtime_error("Only 1D tensor are supported as input.");
+    ORT_CXX_API_THROW("Only 1D tensor are supported as input.", ORT_INVALID_ARGUMENT);
 
   std::vector<std::string> words;
   std::vector<int64_t> indices;
@@ -116,7 +116,7 @@ ONNXTensorElementDataType CustomOpStringSplit::GetInputType(size_t index) const
     case 2:
       return ONNX_TENSOR_ELEMENT_DATA_TYPE_BOOL;
     default:
-      throw std::runtime_error(MakeString("Unexpected input index ", index));
+      ORT_CXX_API_THROW(MakeString("Unexpected input index ", index), ORT_INVALID_ARGUMENT);
   }
 };
 
@@ -132,6 +132,6 @@ ONNXTensorElementDataType CustomOpStringSplit::GetOutputType(size_t index) const
     case 1:
       return ONNX_TENSOR_ELEMENT_DATA_TYPE_STRING;
     default:
-      throw std::runtime_error(MakeString("[StringSplit] Unexpected output index ", index));
+      ORT_CXX_API_THROW(MakeString("[StringSplit] Unexpected output index ", index), ORT_INVALID_ARGUMENT);
   }
 };
diff --git a/operators/text/string_split.hpp b/operators/text/string_split.hpp
index bc14a144..7edfc1ab 100644
--- a/operators/text/string_split.hpp
+++ b/operators/text/string_split.hpp
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "kernels.h"
+#include "ocos.h"
 #include "string_utils.h"
 
 struct KernelStringSplit : BaseKernel {
diff --git a/operators/text/string_to_vector.cc b/operators/text/string_to_vector.cc
index 4814e082..58e94586 100644
--- a/operators/text/string_to_vector.cc
+++ b/operators/text/string_to_vector.cc
@@ -41,7 +41,7 @@ void StringToVectorImpl::ParseMappingTable(std::string& map) {
 
   vector_len_ = ParseVectorLen(lines[0]);
   if (vector_len_ == 0) {
-    throw std::runtime_error(MakeString("The mapped value of string input cannot be empty: ", lines[0]));
+    ORT_CXX_API_THROW(MakeString("The mapped value of string input cannot be empty: ", lines[0]), ORT_INVALID_ARGUMENT);
   }
 
   std::vector<int64_t> values(vector_len_);
@@ -49,7 +49,7 @@ void StringToVectorImpl::ParseMappingTable(std::string& map) {
     auto kv = SplitString(line, "\t", true);
 
     if (kv.size() != 2) {
-      throw std::runtime_error(MakeString("Failed to parse mapping_table when processing the line: ", line));
+      ORT_CXX_API_THROW(MakeString("Failed to parse mapping_table when processing the line: ", line), ORT_INVALID_ARGUMENT);
     }
 
     ParseValues(kv[1], values);
@@ -62,14 +62,14 @@ void StringToVectorImpl::ParseMappingTable(std::string& map) {
 void StringToVectorImpl::ParseUnkownValue(std::string& unk) {
   auto unk_strs = SplitString(unk, " ", true);
   if (unk_strs.size() != vector_len_) {
-    throw std::runtime_error(MakeString("Incompatible dimension: required vector length of unknown_value should be: ", vector_len_));
+    ORT_CXX_API_THROW(MakeString("Incompatible dimension: required vector length of unknown_value should be: ", vector_len_), ORT_INVALID_ARGUMENT);
   }
 
   for (auto& str : unk_strs) {
     int64_t value;
     auto [end, ec] = std::from_chars(str.data(), str.data() + str.size(), value);
     if (end != str.data() + str.size()) {
-      throw std::runtime_error(MakeString("Failed to parse unknown_value when processing the number: ", str));
+      ORT_CXX_API_THROW(MakeString("Failed to parse unknown_value when processing the number: ", str), ORT_INVALID_ARGUMENT);
     }
 
     unk_value_.push_back(value);
@@ -80,7 +80,7 @@ size_t StringToVectorImpl::ParseVectorLen(const std::string_view& line) {
   auto kv = SplitString(line, "\t", true);
 
   if (kv.size() != 2) {
-    throw std::runtime_error(MakeString("Failed to parse mapping_table when processing the line: ", line));
+    ORT_CXX_API_THROW(MakeString("Failed to parse mapping_table when processing the line: ", line), ORT_INVALID_ARGUMENT);
   }
 
   auto value_strs = SplitString(kv[1], " ", true);
@@ -94,7 +94,7 @@ void StringToVectorImpl::ParseValues(const std::string_view& v, std::vector<int6
   for (int i = 0; i < value_strs.size(); i++) {
     auto [end, ec] = std::from_chars(value_strs[i].data(), value_strs[i].data() + value_strs[i].size(), value);
     if (end != value_strs[i].data() + value_strs[i].size()) {
-      throw std::runtime_error(MakeString("Failed to parse map when processing the number: ", value_strs[i]));
+      ORT_CXX_API_THROW(MakeString("Failed to parse map when processing the number: ", value_strs[i]), ORT_INVALID_ARGUMENT);
     }
     values[i] = value;
   }
diff --git a/operators/text/text.cc b/operators/text/text.cc
new file mode 100644
index 00000000..027c8b96
--- /dev/null
+++ b/operators/text/text.cc
@@ -0,0 +1,43 @@
+#include "text/op_equal.hpp"
+#include "text/op_ragged_tensor.hpp"
+#include "text/string_hash.hpp"
+#include "text/string_join.hpp"
+#include "text/string_lower.hpp"
+#include "text/string_split.hpp"
+#include "text/string_to_vector.hpp"
+#include "text/string_upper.hpp"
+#include "text/vector_to_string.hpp"
+#include "text/string_length.hpp"
+#include "text/string_concat.hpp"
+#include "text/string_ecmaregex_replace.hpp"
+#include "text/string_ecmaregex_split.hpp"
+
+#if defined(ENABLE_RE2_REGEX)
+#include "text/re2_strings/string_regex_replace.hpp"
+#include "text/re2_strings/string_regex_split.hpp"
+#endif // ENABLE_RE2_REGEX
+
+
+FxLoadCustomOpFactory LoadCustomOpClasses_Text = 
+    LoadCustomOpClasses<CustomOpClassBegin, 
+#if defined(ENABLE_RE2_REGEX)
+                        CustomOpStringRegexReplace,
+                        CustomOpStringRegexSplitWithOffsets,
+#endif // ENABLE_RE2_REGEX
+                        CustomOpRaggedTensorToDense,
+                        CustomOpRaggedTensorToSparse,
+                        CustomOpStringRaggedTensorToDense,
+                        CustomOpStringEqual,
+                        CustomOpStringHash,
+                        CustomOpStringHashFast,
+                        CustomOpStringJoin,
+                        CustomOpStringLower,
+                        CustomOpStringUpper,
+                        CustomOpStringSplit,
+                        CustomOpStringToVector,
+                        CustomOpVectorToString,
+                        CustomOpStringLength,
+                        CustomOpStringConcat,
+                        CustomOpStringECMARegexReplace,
+                        CustomOpStringECMARegexSplitWithOffsets
+                        >;
diff --git a/operators/text/vector_to_string.cc b/operators/text/vector_to_string.cc
index 6cc8696d..4725e920 100644
--- a/operators/text/vector_to_string.cc
+++ b/operators/text/vector_to_string.cc
@@ -29,7 +29,7 @@ std::vector<std::string> VectorToStringImpl::Compute(const void* input, const Or
     output_dim = input_dim;
   } else {
     if (input_dim[input_dim.size() - 1] != vector_len_) {
-      throw std::runtime_error(MakeString("Incompatible dimension: required vector length should be ", vector_len_));
+      ORT_CXX_API_THROW(MakeString("Incompatible dimension: required vector length should be ", vector_len_), ORT_INVALID_ARGUMENT);
     }
 
     output_dim = input_dim;
@@ -70,7 +70,7 @@ void VectorToStringImpl::ParseMappingTable(std::string& map) {
     auto kv = SplitString(line, "\t", true);
 
     if (kv.size() != 2) {
-      throw std::runtime_error(MakeString("Failed to parse mapping_table when processing the line: ", line));
+      ORT_CXX_API_THROW(MakeString("Failed to parse mapping_table when processing the line: ", line), ORT_INVALID_ARGUMENT);
     }
 
     ParseValues(kv[1], values);
@@ -83,7 +83,7 @@ size_t VectorToStringImpl::ParseVectorLen(const std::string_view& line) {
   auto kv = SplitString(line, "\t", true);
 
   if (kv.size() != 2) {
-    throw std::runtime_error(MakeString("Failed to parse mapping_table when processing the line: ", line));
+    ORT_CXX_API_THROW(MakeString("Failed to parse mapping_table when processing the line: ", line), ORT_INVALID_ARGUMENT);
   }
 
   auto value_strs = SplitString(kv[1], " ", true);
@@ -97,7 +97,7 @@ void VectorToStringImpl::ParseValues(const std::string_view& v, std::vector<int6
   for (int i = 0; i < value_strs.size(); i++) {
     auto [end, ec] = std::from_chars(value_strs[i].data(), value_strs[i].data() + value_strs[i].size(), value);
     if (end != value_strs[i].data() + value_strs[i].size()) {
-      throw std::runtime_error(MakeString("Failed to parse map when processing the number: ", value_strs[i]));
+      ORT_CXX_API_THROW(MakeString("Failed to parse map when processing the number: ", value_strs[i]), ORT_INVALID_ARGUMENT);
     }
     values[i] = value;
   }
diff --git a/operators/tokenizer/blingfire_sentencebreaker.cc b/operators/tokenizer/blingfire_sentencebreaker.cc
index a485e088..66d91f84 100644
--- a/operators/tokenizer/blingfire_sentencebreaker.cc
+++ b/operators/tokenizer/blingfire_sentencebreaker.cc
@@ -11,13 +11,13 @@
 KernelBlingFireSentenceBreaker::KernelBlingFireSentenceBreaker(OrtApi api, const OrtKernelInfo* info) : BaseKernel(api, info), max_sentence(-1) {
   model_data_ = ort_.KernelInfoGetAttribute<std::string>(info, "model");
   if (model_data_.empty()) {
-    throw std::runtime_error("vocabulary shouldn't be empty.");
+    ORT_CXX_API_THROW("vocabulary shouldn't be empty.", ORT_INVALID_ARGUMENT);
   }
 
   void* model_ptr = SetModel(reinterpret_cast<unsigned char*>(model_data_.data()), model_data_.size());
 
   if (model_ptr == nullptr) {
-    throw std::runtime_error("Invalid model");
+    ORT_CXX_API_THROW("Invalid model", ORT_INVALID_ARGUMENT);
   }
 
   model_ = std::shared_ptr<void>(model_ptr, FreeModel);
@@ -33,7 +33,7 @@ void KernelBlingFireSentenceBreaker::Compute(OrtKernelContext* context) {
   OrtTensorDimensions dimensions(ort_, input);
 
   if (dimensions.Size() != 1 && dimensions[0] != 1) {
-    throw std::runtime_error("We only support string scalar.");
+    ORT_CXX_API_THROW("We only support string scalar.", ORT_INVALID_ARGUMENT);
   }
 
   std::vector<std::string> input_data;
@@ -46,7 +46,7 @@ void KernelBlingFireSentenceBreaker::Compute(OrtKernelContext* context) {
 
   int output_length = TextToSentencesWithOffsetsWithModel(input_string.data(), input_string.size(), output_str.data(), nullptr, nullptr, max_length, model_.get());
   if (output_length < 0) {
-    throw std::runtime_error(MakeString("splitting input:\"", input_string, "\"  failed"));
+    ORT_CXX_API_THROW(MakeString("splitting input:\"", input_string, "\"  failed"), ORT_INVALID_ARGUMENT);
   }
 
   // inline split output_str by newline '\n'
diff --git a/operators/tokenizer/wordpiece_tokenizer.cc b/operators/tokenizer/wordpiece_tokenizer.cc
index 1e2bd2e9..ae0eba0e 100644
--- a/operators/tokenizer/wordpiece_tokenizer.cc
+++ b/operators/tokenizer/wordpiece_tokenizer.cc
@@ -72,8 +72,8 @@ void KernelWordpieceTokenizer_Tokenizer(const std::unordered_map<std::u32string,
       rows.push_back(indices.size());
     } else if (text_index == existing_rows[row_index]) {
       if (row_index >= n_existing_rows)
-        throw std::runtime_error(MakeString(
-            "row_index=", row_index, " is out of range=", n_existing_rows, "."));
+        ORT_CXX_API_THROW(MakeString(
+            "row_index=", row_index, " is out of range=", n_existing_rows, "."), ORT_INVALID_ARGUMENT);
       rows.push_back(indices.size());
       ++row_index;
     }
@@ -181,7 +181,7 @@ ONNXTensorElementDataType CustomOpWordpieceTokenizer::GetInputType(size_t index)
     case 1:
       return ONNX_TENSOR_ELEMENT_DATA_TYPE_INT64;
     default:
-      throw std::runtime_error(MakeString("Unexpected input index ", index));
+      ORT_CXX_API_THROW(MakeString("Unexpected input index ", index), ORT_INVALID_ARGUMENT);
   }
 };
 
@@ -198,6 +198,6 @@ ONNXTensorElementDataType CustomOpWordpieceTokenizer::GetOutputType(size_t index
     case 3:
       return ONNX_TENSOR_ELEMENT_DATA_TYPE_INT64;
     default:
-      throw std::runtime_error(MakeString("[WordpieceTokenizer] Unexpected output index ", index));
+      ORT_CXX_API_THROW(MakeString("[WordpieceTokenizer] Unexpected output index ", index), ORT_INVALID_ARGUMENT);
   }
 };
\ No newline at end of file
diff --git a/shared/ortcustomops.cc b/shared/ortcustomops.cc
index f77309ba..143b6b69 100644
--- a/shared/ortcustomops.cc
+++ b/shared/ortcustomops.cc
@@ -4,89 +4,7 @@
 #include <set>
 
 #include "onnxruntime_extensions.h"
-#include "string_utils.h"
-
-#include "text/op_equal.hpp"
-#include "text/op_segment_sum.hpp"
-#include "text/op_segment_extraction.hpp"
-#include "text/op_ragged_tensor.hpp"
-#include "text/string_hash.hpp"
-#include "text/string_join.hpp"
-#include "text/string_lower.hpp"
-#include "text/string_regex_replace.hpp"
-#include "text/string_regex_split.hpp"
-#include "text/string_split.hpp"
-#include "text/string_to_vector.hpp"
-#include "text/string_upper.hpp"
-#include "text/vector_to_string.hpp"
-#include "text/string_length.hpp"
-#include "text/string_concat.hpp"
-#include "text/string_ecmaregex_replace.hpp"
-#include "text/string_ecmaregex_split.hpp"
-
-
-#ifdef ENABLE_TF_STRING
-CustomOpSegmentExtraction c_CustomOpSegmentExtraction;
-CustomOpSegmentSum c_CustomOpSegmentSum;
-CustomOpRaggedTensorToDense c_CustomOpRaggedTensorToDense;
-CustomOpRaggedTensorToSparse c_CustomOpRaggedTensorToSparse;
-CustomOpStringEqual c_CustomOpStringEqual;
-CustomOpStringHash c_CustomOpStringHash;
-CustomOpStringHashFast c_CustomOpStringHashFast;
-CustomOpStringJoin c_CustomOpStringJoin;
-CustomOpStringLower c_CustomOpStringLower;
-CustomOpStringRaggedTensorToDense c_CustomOpStringRaggedTensorToDense;
-CustomOpStringECMARegexReplace c_CustomOpStringECMARegexReplace;
-CustomOpStringECMARegexSplitWithOffsets c_CustomOpStringECMARegexSplitWithOffsets;
-CustomOpStringSplit c_CustomOpStringSplit;
-CustomOpStringToVector c_CustomOpStringToVector;
-CustomOpStringUpper c_CustomOpStringUpper;
-CustomOpVectorToString c_CustomOpVectorToString;
-CustomOpStringLength c_CustomOpStringLength;
-CustomOpStringConcat c_CustomOpStringConcat;
-#endif
-
-#ifdef ENABLE_RE2_REGEX
-CustomOpStringRegexReplace c_CustomOpStringRegexReplace;
-CustomOpStringRegexSplitWithOffsets c_CustomOpStringRegexSplitWithOffsets;
-#endif
-
-OrtCustomOp* operator_lists[] = {
-#ifdef ENABLE_TF_STRING
-    &c_CustomOpRaggedTensorToDense,
-    &c_CustomOpRaggedTensorToSparse,
-    &c_CustomOpSegmentSum,
-    &c_CustomOpSegmentExtraction,
-    &c_CustomOpStringEqual,
-    &c_CustomOpStringHash,
-    &c_CustomOpStringHashFast,
-    &c_CustomOpStringJoin,
-    &c_CustomOpStringLower,
-    &c_CustomOpStringRaggedTensorToDense,
-    &c_CustomOpStringECMARegexReplace,
-    &c_CustomOpStringECMARegexSplitWithOffsets,
-    &c_CustomOpStringSplit,
-    &c_CustomOpStringToVector,
-    &c_CustomOpStringUpper,
-    &c_CustomOpVectorToString,
-    &c_CustomOpStringLength,
-    &c_CustomOpStringConcat,
-#endif
-
-#ifdef ENABLE_RE2_REGEX
-    &c_CustomOpStringRegexReplace,
-    &c_CustomOpStringRegexSplitWithOffsets,
-#endif
-
-    nullptr };
-
-#ifdef ENABLE_MATH
-extern FxLoadCustomOpFactory LoadCustomOpClasses_Math;
-#endif  // ENABLE_MATH
-
-#ifdef ENABLE_TOKENIZER
-extern FxLoadCustomOpFactory LoadCustomOpClasses_Tokenizer;
-#endif // ENABLE_TOKENIZER
+#include "ocos.h"
 
 
 class ExternalCustomOps {
@@ -154,7 +72,10 @@ extern "C" OrtStatus* ORT_API_CALL RegisterCustomOps(OrtSessionOptions* options,
 #endif
 
   static std::vector<FxLoadCustomOpFactory> c_factories = {
-    []() { return const_cast<const OrtCustomOp**>(operator_lists); }
+    LoadCustomOpClasses<CustomOpClassBegin>
+#if defined(ENABLE_TF_STRING)
+    , LoadCustomOpClasses_Text
+#endif // ENABLE_TF_STRING
 #if defined(ENABLE_MATH)
     , LoadCustomOpClasses_Math
 #endif
diff --git a/test/static_test/test_strings.cc b/test/static_test/test_strings.cc
index 3ac72225..9805f7dc 100644
--- a/test/static_test/test_strings.cc
+++ b/test/static_test/test_strings.cc
@@ -3,7 +3,7 @@
 
 #include "gtest/gtest.h"
 #include "string_utils.h"
-#include "text/string_regex_split_re.hpp"
+#include "text/re2_strings/string_regex_split_re.hpp"
 #include "text/string_ecmaregex_split.hpp"