Python / SWIG：按索引返回数组

Question

I am trying to output an array of values from a C++ function wrapped using SWIG for Python. I know how to do this for the simple case of returning a C++ array using the following typemap:

%apply (double* ARGOUT_ARRAY1, int DIM1) {(double* output_array, int length)};

with the associated C++ code looking like:

void MyClass::retrieveArray(double* output_array, int length) {
    for (int i=0; i<length; i++)
        output_array[i] = _array[i];
}

The usage from Python looks like:

output_array = my_class_instance.returnArray(length)

However, I want to take this a step further and return an array using an index into a multi-dimensional array. So the C++ code would look like:

void MyClass::retrieveArray(double* output_array, int length, int index) {
    for (int i=0; i<length; i++)
        output_array[i] = _multi_dimensional_array[index][i];
}

However, the typemap would need to be changed to accommodate the index. I have tried:

%apply (double* ARGOUT_ARRAY1, int DIM1, int index) {(double* output_array, 
        int length, int index)};

but that does not seem to work. I also know that the size of the returned array in this case will be 6 doubles, but I am not sure how to relay that information to SWIG.

Answer 1

You should be able to apply the same typemap as before:

%apply (double* ARGOUT_ARRAY1, int DIM1) {(double* output_array, int length)};

This will match the parameters double* output_array and int length of MyClass::retrieveArray . To call the function from python, use:

output_array = my_class_instance.retrieveArray(length, index)

or, since you already know that the length will be 6:

output_array = my_class_instance.retrieveArray(6, index)

You can also modify your C++ code to use the following function signature if you already know the size of the array at compile time:

void MyClass::retrieveArray(double output_array[6], int index) {...}

and then use a different typemap:

%apply (double ARGOUT_ARRAY1[6]) {(double output_array[6])};

See also the documentation for numpy.i , which I assume you are using.