pytorch/caffe2/operators/string_ops.h

78 lines
2.0 KiB
C++

#ifndef CAFFE2_OPERATORS_STRING_OPS_H_
#define CAFFE2_OPERATORS_STRING_OPS_H_
#include "caffe2/core/operator.h"
#include "caffe2/operators/elementwise_ops.h"
namespace caffe2 {
/**
* ForEach is a unary functor that forwards each element of the input array
* into the elementwise Functor provided, and gathers the results of each
* call into the resulting array. Use it as an adaptor if you want to create
* a UnaryElementwiseOp that acts on each element of the tensor per function
* call -- this is reasonable for complex types where vectorization wouldn't
* be much of a gain, performance-wise.
*/
template <typename Functor>
struct ForEach {
explicit ForEach(OperatorBase& op) : functor(op) {}
template <typename In, typename Out, typename Context>
bool operator()(int n, const In* in, Out* out, Context* /*c*/) {
for (const auto i : c10::irange(n)) {
out[i] = functor(in[i]);
}
return true;
}
Functor functor;
};
template <typename ScalarFunctor, typename TypeMap = FixedType<std::string>>
using StringElementwiseOp = UnaryElementwiseWithArgsOp<
TensorTypes<std::string>,
CPUContext,
ForEach<ScalarFunctor>,
TypeMap>;
template <class Context>
class StringJoinOp final : public Operator<Context> {
public:
USE_OPERATOR_CONTEXT_FUNCTIONS;
template <class... Args>
explicit StringJoinOp(Args&&... args)
: Operator<Context>(std::forward<Args>(args)...),
delimiter_(
this->template GetSingleArgument<std::string>("delimiter", ",")),
axis_(this->template GetSingleArgument<int>("axis", 0)) {
CAFFE_ENFORCE(axis_ == 0 || axis_ == 1);
}
bool RunOnDevice() override {
return DispatchHelper<TensorTypes<
float,
double,
int8_t,
uint8_t,
int16_t,
uint16_t,
int32_t,
int64_t,
std::string,
bool>>::call(this, Input(0));
}
template <typename T>
bool DoRunWithType();
protected:
std::string delimiter_;
int axis_;
};
} // namespace caffe2
#endif // CAFFE2_OPERATORS_STRING_OPS_H_