Python C API and C++ functions - c++

I'm trying to extend the Python interpreter in my C++ program, my problem is as follows.
When I'm trying to call a function, explained in the code below, I get a NameError, from the Python interpreter.
The error is
Traceback (most recent call last):
File "", line 3, in module
NameError: name 'func' is not defined
I used the following code to bind it, according to the Python wiki for version 3.3.2, which I'm using here
double func( int a )
{
return a*a-0.5;
}
static PyObject *TestError;
static PyObject * func_test(PyObject * self, PyObject *args)
{
const int * command;
double sts;
if( !PyArg_ParseTuple(args, "i", &command) )
return NULL;
sts = func( *command );
return PyFloat_FromDouble(sts);
}
static PyMethodDef TestMethods[] = {
{"func", func_test, METH_VARARGS,
"Thing."},
{NULL, NULL, 0, NULL} /* Sentinel */
};
static struct PyModuleDef testmodule = {
PyModuleDef_HEAD_INIT,
"test", /* name of module */
NULL, /* module documentation, may be NULL */
-1, /* size of per-interpreter state of the module,
or -1 if the module keeps state in global variables. */
TestMethods
};
PyMODINIT_FUNC PyInit_test()
{
PyObject *m;
m = PyModule_Create(&testmodule);
if (m == NULL)
return NULL;
TestError = PyErr_NewException("test.error", NULL, NULL);
Py_INCREF(TestError);
PyModule_AddObject(m, "error", TestError);
return m;
}
Then I'm calling PyImport_AppendInittab("test", PyInit_test);
Py_Initialize();, and then I'm trying to run a simple test, with
PyRun_SimpleString("import test\n"
"print('Hi!')\n"
"b = func(5)\n"
"print(b)\n");
Yet, I keep getting the error. Can someone please explain, what am I doing wrong here?

PyRun_SimpleString("import test\n"
"print('Hi!')\n"
"b = test.func(5)\n" # <--
"print(b)\n");
EDIT: Another problem:
int command; // not "int *"
double sts;
if( !PyArg_ParseTuple(args, "i", &command) )
Note that I'd recommend using CFFI if you're not familiar yet with how to write a CPython C extension module.

I agree with all the fixes by Armin Rigo and I would add this one:
PyImport_AppendInittab("test", &PyInit_test);
Pass the address of the function to PyImport_AppendInittab.

Related

Numpy C++: How to iterate over PyArrayObject without a segfault

For me, the following all result in a segfault:
my_array->descr->subarray->shape;
my_array->dimensions;
PyArray_SHAPE(my_array);
PyArray_DIMS(my_array);
PyArray_ITEMSIZE(my_array);
PyArray_NBYTES(my_array);
My function looks like this:
static PyObject* exterior(PyObject* self, PyArrayObject* old_simplices_array)
{//code here
The rest of my cpp file looks like this:
#include "Python.h"
#include "numpy/arrayobject.h"
/* function */
static PyMethodDef compiled_methods[] =
{
{"_exterior",(PyCFunction)exterior , METH_VARARGS},
{NULL, NULL} /* Sentinel */
};
PyMODINIT_FUNC init_alto(void)
{
(void) Py_InitModule("_alto", compiled_methods);
import_array();
}
The python code that passes the array to "exterior" just passes an NxM uint array. That part works. I can access the array's strides and data. I just cannot determine the bounds of iteration. I am working from within sage if that makes any difference.
How am I supposed to iterate over an array without segfaulting? If the answer is obvious, please idiotproof your answer.
For a better idea of what the function looks like, see here.
In the past I have done the following to iterate over a PyArrayObject:
static PyObject *func1(PyObject *self, PyObject *args) {
PyArrayObject *X;
int ndX;
npy_intp *shapeX;
PyArray_Descr *dtype;
NpyIter *iter;
NpyIter_IterNextFunc *iternext;
PyArg_ParseTuple(args, "O!", &PyArray_Type, &X);
ndX = PyArray_NDIM(X);
shapeX = PyArray_SHAPE(X);
dtype = PyArray_DescrFromType(NPY_DOUBLE);
iter = NpyIter_New(X, NPY_ITER_READONLY, NPY_KEEPORDER, NPY_NO_CASTING, dtype);
if (iter==NULL) {
return NULL;
}
iternext = NpyIter_GetIterNext(iter, NULL);
dataptr = (double **) NpyIter_GetDataPtrArray(iter);
do {
cout << **dataptr << endl;
} while (iternext(iter));
NpyIter_Deallocate(iter);
return Py_BuildValue(something);
}
To find out more information check out this link: http://docs.scipy.org/doc/numpy/reference/c-api.iterator.html

Calling a Python function within a C++ program

I have a real simple Python function:
def myfunc(x):
return 2.0 * x
I want to send this function to a C++ program and call it so I have done this:
#include "Python.h"
static PyObject *execMyPyFunc(PyObject *self, PyObject *args) {
PyObject *Fx, *pyresult;
double x;
PyArg_ParseTuple(args, "dO", &x, &Fx);
pyresult = PyObject_CallFunction(Fx, "d", x);
return pyresult;
}
static PyMethodDef C_API_TestMethods[] = {
{"execMyPyFunc", execMyPyFunc, METH_VARARGS, "Add documentation here.."},
{NULL, NULL}
};
PyMODINIT_FUNC initC_API_Test(void) {
Py_InitModule("C_API_Test", C_API_TestMethods);
}
My Python program works correctly:
from C_API_Test import execMyPyFunc
def myfunc(x):
return 2.0 * x
fx = execMyPyFunc(1.28,myfunc)
print fx
What I would like to do though is to somehow get the pointer from my Python function (PyObject *Fx) and pass this to a C++ function expecting: double(*fx)(double). Does anyone know how to do this (if possible)? I tried to initialize double(*cFx)(double) and cast my Python function as cFx = (double(*)(double))Fx but this does not work. Any ideas?
You aren't going to be able to simply cast a Python function to C like that.
Instead pass the PyObject function pointer, call the function, and convert to C double. This code will return -1 on failure.
static double cFx(PyObject *fx, double x){
PyObject *pyresult = PyObject_CallFunction(fx, "d", x);
if (pyresult == NULL) return -1;
double cppresult = PyFloat_AsDouble(pyresult);
Py_DECREF(pyresult);
if (PyErr_Occurred()) return -1;
return cppresult;
}
Important part of this is to decrement the reference count to the return value of PyObject_CallFunction since you aren't passing it off to the Python interpreter to deal with.

C/C++ embedding python module PIL, the python function want to return all pixel data, but i why can't?

add c/c++ code embedding python. all problem is in c code, I don't known how to get the pixel value.
python code:
import Image
format = ""
mode = ""
size = ""
data = list()
def getImage(file):
im = Image.open(file)
global format
global mode
global size
global data
format = im.format
mode = im.mode
size = im.size
width, height = im.size
for x in range(0, height):
for y in range(0, width):
data.append(im.getpixel((x,y)))
return None
in C/C++ code
My get the data length is 0. Any problems of the two for loop?
void loadImage(char *file) {
Image* img = NULL;
Py_Initialize();
if ( !Py_IsInitialized() ) {
std::cerr<<"Python initalize failed.\n";
return NULL;
}
PyRun_SimpleString("import sys");
PyRun_SimpleString("sys.path.append('./')");
PyObject *pName, *pModule, *pFunc, *pArgs;
pName = PyString_FromString("loadImage");
pModule = PyImport_Import(pName);
Py_DECREF(pName);
if ( !pModule ) {
PyErr_Print();
std::cerr<<"import loadImage module faild, please confirm where the file 'loadImage.py' is.\n";
return NULL;
}
pFunc = PyObject_GetAttrString(pModule, "getImage");
if ( !pFunc ) {
PyErr_Print();
std::cerr<<"Can't find method getImage in loadImage module.\n";
return NULL;
}
pArgs = PyTuple_New(1);
PyTuple_SetItem(pArgs, 0, Py_BuildValue("s", file));
PyObject_CallObject(pFunc, pArgs);
PyObject *pFormat, *pMode, *pSize, *pData, *pSeq, *pPixel;
pFormat = PyObject_GetAttrString(pModule, "format");
pMode = PyObject_GetAttrString(pModule, "mode");
pSize = PyObject_GetAttrString(pModule, "size");
if ( !PyTuple_Check(pSize) ) {
std::cerr<<"pSize is not tupple object.\n";
return NULL;
}
pData = PyObject_GetAttrString(pModule, "data");
if ( !pData ) {
std::cerr<<"pData is null.\n";
return NULL;
}
if ( !PyList_Check(pData) ) {
std::cerr<<"pData is not list object.\n";
return NULL;
}
int n = PyList_GET_SIZE(pData);
std::cerr<<n<<"\n";
Py_DECREF(pData);
Py_DECREF(pFunc);
Py_DECREF(pModule);
Py_DECREF(pArgs);
std::cerr<<"Py_DECREF over.\n";
}
I get the lenght of pData(is n) is 0.
Well, your code can literally be replaced with a single line:
data = list(Image.open(file).getdata())
The getdata method of Image objects returns the pixels packed into a list, line-by-line. If you need them in column-major order (as you are doing), apply .transpose with the appropriate arguments first. The result will still be substantially faster and simpler than iterating over all the pixels.
You aren't checking the return value from PyObject_CallObject. It might have returned NULL (Python exception), and consequently failed to modify the data list.
You should check every Python API call you make, since exception handling is critically important to making the interpreter work correctly.
If you do get a NULL, you can call PyErr_Print() to get the traceback printed out to stderr (cerr).
The Python code you run by calling PyObject_CallObject is probably throwing an exception.
NULL means an exception was thrown. You can use various python api functions to get at
the actual error: http://docs.python.org/c-api/exceptions.html. This is probably the simplest.
if( PyObject_CallObject(pFunc, pArgs ) == NULL )
{
// Print the exception (a traceback may or may not be available)
PyErr_Print();
// Do any cleanup...
// Always return NULL after a python api function returns NULL.
return NULL;
}
The problem may caused by the python version 2.7 in my computer. I changed to python 3.2, all code can got ideal return.

Trouble calling on a Java method from a native thread using JNI (C++)

I have a JNI problem which I hope someone can help me out with.
I'm trying to call on a constructor of a Java class called LUSOutputJNI from a native thread.
It keeps failing on FindClass(...) of this specific class.
Here is the code:
LOGE("1");
JNIEnv *env = NULL;
LOGE("2");
int res = -1;
res = g_vm->AttachCurrentThread(&env, NULL);
if(env == NULL)
{
LOGE("env is NULL, AttachCurrentThread failed");;
}
if(res >= 0)
LOGE("AttachCurrentThread was successful");
jclass clazz = NULL;
jmethodID cid;
jclass clazzESEngine;
jmethodID callbackid;
jobject jCoreOut;
static jfieldID fid_ActionState = NULL;
static jfieldID fid_nSpeed = NULL;
static jfieldID fid_nType = NULL;
static jfieldID fid_nInProcess = NULL;
static jfieldID fid_nX = NULL;
static jfieldID fid_nY = NULL;
LOGE("3");
static const char* const ECOClassName = "lus/android/sdk/LUSOutputJNI";
//static const char* const ECOClassName = "android/widget/TextView";
clazz = env->FindClass(ECOClassName);
if (clazz == NULL) {
LOGE("Can't find class LUSOutputJNI");
}
else
LOGE("lus/android/sdk/LUSOutputJNI was found, YEY!!");
LOGE("4");
cid = env->GetMethodID(clazz,"<init>", "()V");
LOGE("5");
jCoreOut = env->NewObject(clazz, cid);
LOGE("6");
Here is the logcat output from when it fails:
E/lusCore_JNI( 3040): 1
E/lusCore_JNI( 3040): 2
E/lusCore_JNI( 3040): AttachCurrentThread was successful
E/lusCore_JNI( 3040): 3
E/lusCore_JNI( 3040): Can't find class LUSOutputJNI
E/lusCore_JNI( 3040): 4
W/dalvikvm( 3040): JNI WARNING: JNI method called with exception raised
Observations:
I get a result from AttachCurrentThread which is 0, which means that this attachment was successful + the env pointer is no longer NULL.
I'm sure about the package name declaration of LUSOutputJNI (triple checked it...)
When I try to run FindClass(..) with a more popular class name such as android/widget/TextView , I get a positive match. It is there. Meaning the thread attachment and the env variables are ok. (Can I assume that?)
When I try to run the following code from a JNI method which has a JNI thread running it, it finds the LUSOutputJNI class without a problem.
What am I doing wrong?
Help will be much appreciated :)
Thanks for your time,
Ita
Found the answer \ work around in here. (Look for FAQ: "FindClass didn't find my class" in
JNI tips)
I basicaly saved a global ref to the needed jclass objects.
Did however had to overcome some evil JNI changes between C/JNI and C++/JNI in order for the code to compile.
This is how I got the NewGlobalRef to compile and work.
jclass localRefCls = env->FindClass(strLUSClassName);
if (localRefCls == NULL) {
LOGE("Can't find class %s",strLUSCoreClassName);
return result;
}
//cache the EyeSightCore ref as global
/* Create a global reference */
clazzLUSCore = (_jclass*)env->NewGlobalRef(localRefCls);
/* The local reference is no longer useful */
env->DeleteLocalRef(localRefCls);
/* Is the global reference created successfully? */
if (clazzLUSCore == NULL) {
LOGE("Error - clazzLUSCore is still null when it is suppose to be global");
return result; /* out of memory exception thrown */
}
I hope this helps anyone.
To get a bit more information about whats going wrong ( it will log to stderr, not LOGE, and I'm not sure how to change that) you can add some exception printing - you can change this:
//static const char* const ECOClassName = "android/widget/TextView";
clazz = env->FindClass(ECOClassName);
if (clazz == NULL) {
LOGE("Can't find class LUSOutputJNI");
}
else
LOGE("lus/android/sdk/LUSOutputJNI was found, YEY!!");
To this:
//static const char* const ECOClassName = "android/widget/TextView";
clazz = env->FindClass(ECOClassName);
if (clazz == NULL) {
LOGE("Can't find class LUSOutputJNI");
jthrowable exc = env->ExceptionOccurred();
if(exc)
{
env->ExceptionDescribe();
env->ExceptionClear();
}
}
else
LOGE("lus/android/sdk/LUSOutputJNI was found, YEY!!");

How do you extend python with C++?

I've successfully extended python with C, thanks to this handy skeleton module. But I can't find one for C++, and I have circular dependency trouble when trying to fix the errors that C++ gives when I compile this skeleton module.
How do you extend Python with C++?
I'd rather not depend on Boost (or SWIP or other libraries) if I don't have to. Dependencies are a pain in the butt. Best case scenario, I find a skeleton file that already compiles with C++.
Here's the edited skeleton I've made for C++:
#include <Python.h>
#include "Flp.h"
static PyObject * ErrorObject;
typedef struct {
PyObject_HEAD
PyObject * x_attr; // attributes dictionary
} FlpObject;
static void Flp_dealloc(FlpObject * self);
static PyObject * Flp_getattr(FlpObject * self, char * name);
static int Flp_setattr(FlpObject * self, char * name, PyObject * v);
DL_EXPORT(void) initflp();
static PyTypeObject Flp_Type = {
/* The ob_type field must be initialized in the module init function
* to be portable to Windows without using C++. */
PyObject_HEAD_INIT(NULL)
0, /*ob_size*/
"Flp", /*tp_name*/
sizeof(FlpObject), /*tp_basicsize*/
0, /*tp_itemsize*/
/* methods */
(destructor)Flp_dealloc, /*tp_dealloc*/
0, /*tp_print*/
(getattrfunc)Flp_getattr, /*tp_getattr*/
(setattrfunc)Flp_setattr, /*tp_setattr*/
0, /*tp_compare*/
0, /*tp_repr*/
0, /*tp_as_number*/
0, /*tp_as_sequence*/
0, /*tp_as_mapping*/
0, /*tp_hash*/
};
#define FlpObject_Check(v) ((v)->ob_type == &Flp_Type)
static FlpObject * newFlpObject(PyObject * arg)
{
FlpObject * self;
self = PyObject_NEW(FlpObject, &Flp_Type);
if (self == NULL)
return NULL;
self->x_attr = NULL;
return self;
}
// Flp methods
static void Flp_dealloc(FlpObject * self)
{
Py_XDECREF(self->x_attr);
PyMem_DEL(self);
}
static PyObject * Flp_demo(FlpObject * self, PyObject * args)
{
if (! PyArg_ParseTuple(args, ""))
return NULL;
Py_INCREF(Py_None);
return Py_None;
}
static PyMethodDef Flp_methods[] = {
{"demo", (PyCFunction)Flp_demo, 1},
{NULL, NULL} // sentinel
};
static PyObject * Flp_getattr(FlpObject * self, char * name)
{
if (self->x_attr != NULL) {
PyObject * v = PyDict_GetItemString(self->x_attr, name);
if (v != NULL) {
Py_INCREF(v);
return v;
}
}
return Py_FindMethod(Flp_methods, (PyObject *)self, name);
}
static int Flp_setattr(FlpObject * self, char * name, PyObject * v)
{
if (self->x_attr == NULL) {
self->x_attr = PyDict_New();
if (self->x_attr == NULL)
return -1;
}
if (v == NULL) {
int rv = PyDict_DelItemString(self->x_attr, name);
if (rv < 0)
PyErr_SetString(PyExc_AttributeError,
"delete non-existing Flp attribute");
return rv;
}
else
return PyDict_SetItemString(self->x_attr, name, v);
}
/* --------------------------------------------------------------------- */
/* Function of two integers returning integer */
static PyObject * flp_foo(PyObject * self, PyObject * args)
{
long i, j;
long res;
if (!PyArg_ParseTuple(args, "ll", &i, &j))
return NULL;
res = i+j; /* flpX Do something here */
return PyInt_FromLong(res);
}
/* Function of no arguments returning new Flp object */
static PyObject * flp_new(PyObject * self, PyObject * args)
{
FlpObject *rv;
if (!PyArg_ParseTuple(args, ""))
return NULL;
rv = newFlpObject(args);
if ( rv == NULL )
return NULL;
return (PyObject *)rv;
}
/* Example with subtle bug from extensions manual ("Thin Ice"). */
static PyObject * flp_bug(PyObject * self, PyObject * args)
{
PyObject *list, *item;
if (!PyArg_ParseTuple(args, "O", &list))
return NULL;
item = PyList_GetItem(list, 0);
/* Py_INCREF(item); */
PyList_SetItem(list, 1, PyInt_FromLong(0L));
PyObject_Print(item, stdout, 0);
printf("\n");
/* Py_DECREF(item); */
Py_INCREF(Py_None);
return Py_None;
}
/* Test bad format character */
static PyObject * flp_roj(PyObject * self, PyObject * args)
{
PyObject *a;
long b;
if (!PyArg_ParseTuple(args, "O#", &a, &b))
return NULL;
Py_INCREF(Py_None);
return Py_None;
}
/* List of functions defined in the module */
static PyMethodDef flp_methods[] = {
{"roj", flp_roj, 1},
{"foo", flp_foo, 1},
{"new", flp_new, 1},
{"bug", flp_bug, 1},
{NULL, NULL} /* sentinel */
};
/* Initialization function for the module (*must* be called initflp) */
DL_EXPORT(void) initflp()
{
PyObject *m, *d;
/* Initialize the type of the new type object here; doing it here
* is required for portability to Windows without requiring C++. */
Flp_Type.ob_type = &PyType_Type;
/* Create the module and add the functions */
m = Py_InitModule("flp", flp_methods);
/* Add some symbolic constants to the module */
d = PyModule_GetDict(m);
ErrorObject = PyErr_NewException("flp.error", NULL, NULL);
PyDict_SetItemString(d, "error", ErrorObject);
}
This compiles fine for me, but when I test it:
$ python
Python 2.6.5 (r265:79063, Apr 16 2010, 13:57:41)
[GCC 4.4.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import flp
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: dynamic module does not define init function (initflp)
>>>
First of all, even though you don't want to introduce an additional dependency, I suggest you to have a look at PyCXX. Quoting its webpage:
CXX/Objects is a set of C++ facilities to make it easier to write Python extensions. The chief way in which PyCXX makes it easier to write Python extensions is that it greatly increases the probability that your program will not make a reference-counting error and will not have to continually check error returns from the Python C API. CXX/Objects integrates Python with C++ in these ways:
C++ exception handling is relied on to detect errors and clean up. In a complicated function this is often a tremendous problem when writing in C. With PyCXX, we let the compiler keep track of what objects need to be dereferenced when an error occurs.
The Standard Template Library (STL) and its many algorithms plug and play with Python containers such as lists and tuples.
The optional CXX/Extensions facility allows you to replace the clumsy C tables with objects and method calls that define your modules and extension objects.
I think PyCXX is licensed under the BSD license, which means that you can just as well include the whole source code of PyCXX in the distributed tarball of your extension if your extension will be released under a similar license.
If you really and absolutely don't want to depend on PyCXX or any other third-party library, I think you only have to wrap functions that will be called by the Python interpreter in extern "C" { and } to avoid name mangling.
Here's the corrected code:
#include <Python.h>
#include "Flp.h"
static PyObject * ErrorObject;
typedef struct {
PyObject_HEAD
PyObject * x_attr; // attributes dictionary
} FlpObject;
extern "C" {
static void Flp_dealloc(FlpObject * self);
static PyObject * Flp_getattr(FlpObject * self, char * name);
static int Flp_setattr(FlpObject * self, char * name, PyObject * v);
DL_EXPORT(void) initflp();
}
static PyTypeObject Flp_Type = {
/* The ob_type field must be initialized in the module init function
* to be portable to Windows without using C++. */
PyObject_HEAD_INIT(NULL)
0, /*ob_size*/
"Flp", /*tp_name*/
sizeof(FlpObject), /*tp_basicsize*/
0, /*tp_itemsize*/
/* methods */
(destructor)Flp_dealloc, /*tp_dealloc*/
0, /*tp_print*/
(getattrfunc)Flp_getattr, /*tp_getattr*/
(setattrfunc)Flp_setattr, /*tp_setattr*/
0, /*tp_compare*/
0, /*tp_repr*/
0, /*tp_as_number*/
0, /*tp_as_sequence*/
0, /*tp_as_mapping*/
0, /*tp_hash*/
};
#define FlpObject_Check(v) ((v)->ob_type == &Flp_Type)
static FlpObject * newFlpObject(PyObject * arg)
{
FlpObject * self;
self = PyObject_NEW(FlpObject, &Flp_Type);
if (self == NULL)
return NULL;
self->x_attr = NULL;
return self;
}
// Flp methods
static void Flp_dealloc(FlpObject * self)
{
Py_XDECREF(self->x_attr);
PyMem_DEL(self);
}
static PyObject * Flp_demo(FlpObject * self, PyObject * args)
{
if (! PyArg_ParseTuple(args, ""))
return NULL;
Py_INCREF(Py_None);
return Py_None;
}
static PyMethodDef Flp_methods[] = {
{"demo", (PyCFunction)Flp_demo, 1},
{NULL, NULL} // sentinel
};
static PyObject * Flp_getattr(FlpObject * self, char * name)
{
if (self->x_attr != NULL) {
PyObject * v = PyDict_GetItemString(self->x_attr, name);
if (v != NULL) {
Py_INCREF(v);
return v;
}
}
return Py_FindMethod(Flp_methods, (PyObject *)self, name);
}
static int Flp_setattr(FlpObject * self, char * name, PyObject * v)
{
if (self->x_attr == NULL) {
self->x_attr = PyDict_New();
if (self->x_attr == NULL)
return -1;
}
if (v == NULL) {
int rv = PyDict_DelItemString(self->x_attr, name);
if (rv < 0)
PyErr_SetString(PyExc_AttributeError,
"delete non-existing Flp attribute");
return rv;
}
else
return PyDict_SetItemString(self->x_attr, name, v);
}
/* --------------------------------------------------------------------- */
/* Function of two integers returning integer */
static PyObject * flp_foo(PyObject * self, PyObject * args)
{
long i, j;
long res;
if (!PyArg_ParseTuple(args, "ll", &i, &j))
return NULL;
res = i+j; /* flpX Do something here */
return PyInt_FromLong(res);
}
/* Function of no arguments returning new Flp object */
static PyObject * flp_new(PyObject * self, PyObject * args)
{
FlpObject *rv;
if (!PyArg_ParseTuple(args, ""))
return NULL;
rv = newFlpObject(args);
if ( rv == NULL )
return NULL;
return (PyObject *)rv;
}
/* Example with subtle bug from extensions manual ("Thin Ice"). */
static PyObject * flp_bug(PyObject * self, PyObject * args)
{
PyObject *list, *item;
if (!PyArg_ParseTuple(args, "O", &list))
return NULL;
item = PyList_GetItem(list, 0);
/* Py_INCREF(item); */
PyList_SetItem(list, 1, PyInt_FromLong(0L));
PyObject_Print(item, stdout, 0);
printf("\n");
/* Py_DECREF(item); */
Py_INCREF(Py_None);
return Py_None;
}
/* Test bad format character */
static PyObject * flp_roj(PyObject * self, PyObject * args)
{
PyObject *a;
long b;
if (!PyArg_ParseTuple(args, "O#", &a, &b))
return NULL;
Py_INCREF(Py_None);
return Py_None;
}
/* List of functions defined in the module */
static PyMethodDef flp_methods[] = {
{"roj", flp_roj, 1},
{"foo", flp_foo, 1},
{"new", flp_new, 1},
{"bug", flp_bug, 1},
{NULL, NULL} /* sentinel */
};
/* Initialization function for the module (*must* be called initflp) */
DL_EXPORT(void) initflp()
{
PyObject *m, *d;
/* Initialize the type of the new type object here; doing it here
* is required for portability to Windows without requiring C++. */
Flp_Type.ob_type = &PyType_Type;
/* Create the module and add the functions */
m = Py_InitModule("flp", flp_methods);
/* Add some symbolic constants to the module */
d = PyModule_GetDict(m);
ErrorObject = PyErr_NewException("flp.error", NULL, NULL);
PyDict_SetItemString(d, "error", ErrorObject);
}
use extern C to wrap all the function names that get called from python. Because C++ compilers use something called 'name mangling' (necessary for dealing with overloading), python can't read c++ libraries. But extern C will solve your problems. Do it like this:
// most of your code can go whereever
void cpp_function() {}
extern "C" {
// all functions that python calls directly must go in here
void python_function() {}
}
Make extra sure you put every function python needs inside the extern block. You can still use c++ features inside the functions, it's just that the names will be exported without 'name mangling'.
What about Boost::Python?
EDIT: sorry, I did oversee that you don't want to depend on boost but I think it might still be one of the best options.