microsoft/onnxruntime-extensions
Publicmirrored from https://github.com/microsoft/onnxruntime-extensionsAvailable
operators/tokenizer/blingfire_sentencebreaker.cc
95lines · modecode
| 1 | // Copyright (c) Microsoft Corporation. All rights reserved. |
| 2 | // Licensed under the MIT License. |
| 3 | |
| 4 | #include "blingfire_sentencebreaker.hpp" |
| 5 | #include "string_tensor.h" |
| 6 | #include <vector> |
| 7 | #include <locale> |
| 8 | #include <codecvt> |
| 9 | #include <algorithm> |
| 10 | #include <memory> |
| 11 | |
| 12 | KernelBlingFireSentenceBreaker::KernelBlingFireSentenceBreaker(OrtApi api, const OrtKernelInfo* info) : BaseKernel(api, info), max_sentence(-1) { |
| 13 | model_data_ = ort_.KernelInfoGetAttribute<std::string>(info, "model"); |
| 14 | if (model_data_.empty()) { |
| 15 | ORT_CXX_API_THROW("vocabulary shouldn't be empty.", ORT_INVALID_ARGUMENT); |
| 16 | } |
| 17 | |
| 18 | void* model_ptr = SetModel(reinterpret_cast<unsigned char*>(model_data_.data()), model_data_.size()); |
| 19 | |
| 20 | if (model_ptr == nullptr) { |
| 21 | ORT_CXX_API_THROW("Invalid model", ORT_INVALID_ARGUMENT); |
| 22 | } |
| 23 | |
| 24 | model_ = std::shared_ptr<void>(model_ptr, FreeModel); |
| 25 | |
| 26 | if (HasAttribute("max_sentence")) { |
| 27 | max_sentence = ort_.KernelInfoGetAttribute<int64_t>(info, "max_sentence"); |
| 28 | } |
| 29 | } |
| 30 | |
| 31 | void KernelBlingFireSentenceBreaker::Compute(OrtKernelContext* context) { |
| 32 | // Setup inputs |
| 33 | const OrtValue* input = ort_.KernelContext_GetInput(context, 0); |
| 34 | OrtTensorDimensions dimensions(ort_, input); |
| 35 | |
| 36 | // TODO: fix this scalar check. |
| 37 | if (dimensions.Size() != 1 && dimensions[0] != 1) { |
| 38 | ORT_CXX_API_THROW("We only support string scalar.", ORT_INVALID_ARGUMENT); |
| 39 | } |
| 40 | |
| 41 | std::vector<std::string> input_data; |
| 42 | GetTensorMutableDataString(api_, ort_, context, input, input_data); |
| 43 | |
| 44 | std::string& input_string = input_data[0]; |
| 45 | int max_length = 2 * input_string.size() + 1; |
| 46 | std::unique_ptr<char[]> output_str = std::make_unique<char[]>(max_length); |
| 47 | |
| 48 | int output_length = TextToSentencesWithOffsetsWithModel(input_string.data(), input_string.size(), output_str.get(), nullptr, nullptr, max_length, model_.get()); |
| 49 | if (output_length < 0) { |
| 50 | ORT_CXX_API_THROW(MakeString("splitting input:\"", input_string, "\" failed"), ORT_INVALID_ARGUMENT); |
| 51 | } |
| 52 | |
| 53 | // inline split output_str by newline '\n' |
| 54 | std::vector<char*> output_sentences; |
| 55 | bool head_flag = true; |
| 56 | for (int i = 0; i < output_length; i++) { |
| 57 | if (head_flag) { |
| 58 | output_sentences.push_back(&output_str[i]); |
| 59 | head_flag = false; |
| 60 | } |
| 61 | |
| 62 | if (output_str[i] == '\n') { |
| 63 | head_flag = true; |
| 64 | output_str[i] = '\0'; |
| 65 | } |
| 66 | } |
| 67 | |
| 68 | std::vector<int64_t> output_dimensions(1); |
| 69 | output_dimensions[0] = output_sentences.size(); |
| 70 | |
| 71 | OrtValue* output = ort_.KernelContext_GetOutput(context, 0, output_dimensions.data(), output_dimensions.size()); |
| 72 | Ort::ThrowOnError(api_, api_.FillStringTensor(output, output_sentences.data(), output_sentences.size())); |
| 73 | } |
| 74 | |
| 75 | void* CustomOpBlingFireSentenceBreaker::CreateKernel(OrtApi api, const OrtKernelInfo* info) const { |
| 76 | return new KernelBlingFireSentenceBreaker(api, info); |
| 77 | }; |
| 78 | |
| 79 | const char* CustomOpBlingFireSentenceBreaker::GetName() const { return "BlingFireSentenceBreaker"; }; |
| 80 | |
| 81 | size_t CustomOpBlingFireSentenceBreaker::GetInputTypeCount() const { |
| 82 | return 1; |
| 83 | }; |
| 84 | |
| 85 | ONNXTensorElementDataType CustomOpBlingFireSentenceBreaker::GetInputType(size_t /*index*/) const { |
| 86 | return ONNX_TENSOR_ELEMENT_DATA_TYPE_STRING; |
| 87 | }; |
| 88 | |
| 89 | size_t CustomOpBlingFireSentenceBreaker::GetOutputTypeCount() const { |
| 90 | return 1; |
| 91 | }; |
| 92 | |
| 93 | ONNXTensorElementDataType CustomOpBlingFireSentenceBreaker::GetOutputType(size_t /*index*/) const { |
| 94 | return ONNX_TENSOR_ELEMENT_DATA_TYPE_STRING; |
| 95 | }; |
| 96 | |