CODEWITHSUNDEEP
pyarrowapache-arrow
clery00
clery00

Reputation: 251

Converting Python seqence to arrow Array via C++ API

I'm attempting to investigate how Arrow converts a python list into an equivalent arrow::Array using the C++ API below.

#include <memory>
#include <Python.h>
#include <string>
#include <iostream>
#include <arrow/memory_pool.h>
#include <arrow/python/python_to_arrow.h>


PyObject* clist(void)
{
    PyObject* lst = PyList_New(0);
    PyList_Append(lst, PyLong_FromLong(1));
    PyList_Append(lst, PyLong_FromLong(2));
    PyList_Append(lst, PyLong_FromLong(5));

    return lst;
}

int main()
{
    Py_Initialize();

    PyObject* list = clist();

    std::shared_ptr<arrow::ChunkedArray> carr;

    arrow::py::PyConversionOptions ops;
    ops.from_pandas = false;
    ops.pool = arrow::default_memory_pool(); 
    ConvertPySequence(list, ops, &carr);

    Py_Finalize();

}

The file compiles fine, however I get a segmentation fault at arrow/cpp/src/arrow/python/iterators.h line 44 on PyCheck_Array.

The error in my debugger is EXC_BAD_ACCESS, however when I interrogate it in the debug console it appears to be there in memory:

enter image description here

Any help is appreciated.

Upvotes: 0

Views: 220

Answers (1)

Wes McKinney
Wes McKinney

Reputation: 105521

You need to initialize the NumPy C API by calling arrow_init_numpy(). See

https://github.com/apache/arrow/blob/master/cpp/src/arrow/python/util/test_main.cc#L24

Upvotes: 1

Related Questions