How To catch python stdout in c++ code

Question

I have a program which during it's run sometimes needs to call python in order to preform some tasks. I need a function that calls python and catches pythons stdout and puts it in some file. This is a declaration of the function

  pythonCallBackFunc(const char* pythonInput)

My problem is to catch all the python output for a given command (pythonInput). I have no experience with python API and I don't know what is the right technique to do this. First thing I've tried is to redirect python's sdtout and stderr using Py_run_SimpleString this is some example of the code i've written.

#include "boost\python.hpp"
#include <iostream>

void pythonCallBackFunc(const char* inputStr){   

    PyRun_SimpleString(inputStr); 
}


int main () {
    ...
   //S0me outside functions does this
   Py_Initialize();
   PyRun_SimpleString("import sys");
   PyRun_SimpleString("old_stdout = sys.stdout");
   PyRun_SimpleString("fsock = open('python_out.log','a')");
   PyRun_SimpleString("sys.stdout = fsock");
   ...

   //my func   
   pythonCallBackFunc("print 'HAHAHAHAHA'");
   pythonCallBackFunc("result = 5");
   pythonCallBackFunc("print result");

   pythonCallBackFunc("result = 'Hello '+'World!'");
   pythonCallBackFunc("print result");

   pythonCallBackFunc("'KUKU '+'KAKA'");
   pythonCallBackFunc("5**3");

   pythonCallBackFunc("prinhghult");

   pythonCallBackFunc("execfile('stdout_close.py')");
   ... 

   //Again anothers function code
   PyRun_SimpleString("sys.stdout = old_stdout");
   PyRun_SimpleString("fsock.close()");

   Py_Finalize();
   return 0;
}

Is there a better way to do this? Besides, for some reason PyRun_SimpleString does nothing when it gets some mathematical expression, for example PyRun_SimpleString("5**3") prints nothing (python conlsul prints the result: 125)

maybe it is important, i am using visual studio 2008. Thanks, Alex

---

Changes I've made according Mark's suggestion:

  #include <python.h>
  #include <string>

  using namespace std;

  void PythonPrinting(string inputStr){ 
     string stdOutErr =
    "import sys\n\
     class CatchOut:\n\
        def __init__(self):\n\
           self.value = ''\n\
        def write(self, txt):\n\
           self.value += txt\n\
     catchOut = CatchOut()\n\
     sys.stdout = catchOut\n\
     sys.stderr = catchOut\n\
    "; //this is python code to redirect stdouts/stderr

     PyObject *pModule = PyImport_AddModule("__main__"); //create main module
     PyRun_SimpleString(stdOutErr.c_str()); //invoke code to redirect

     PyRun_SimpleString(inputStr.c_str());
     PyObject *catcher = PyObject_GetAttrString(pModule,"catchOut");

     PyObject *output = PyObject_GetAttrString(catcher,"value");
     printf("Here's the output: %s\n", PyString_AsString(output)); 
     }

  int main(int argc, char** argv){
         Py_Initialize();

     PythonPrinting("print 123");
     PythonPrinting("1+5");
     PythonPrinting("result = 2");
         PythonPrinting("print result");

         Py_Finalize();
         return 0;
  }

The output i get after running main:

 Here's the output: 123

 Here's the output:
 Here's the output: 
 Here's the output: 2

It is good for me , but only one problem, it should be

 Here's the output: 123

 Here's the output: 6

 Here's the output: 
 Here's the output: 2

I dont know why but after running this command: PythonPrinting("1+5"), PyString_AsString(output) command returns an empty string (char*) instead of 6... :( Is there somthing i can do not to loose this output?

Thaks, Alex

Solution 1

If I'm reading your question correctly, you want to capture stdout/stderr into a variable within your C++? You can do this by redirecting stdout/stderr into a python variable and then querying this variable into your C++. Please not that I have not done the proper ref counting below:

#include <Python.h>
#include <string>

int main(int argc, char** argv)
{
    std::string stdOutErr =
"import sys\n\
class CatchOutErr:\n\
    def __init__(self):\n\
        self.value = ''\n\
    def write(self, txt):\n\
        self.value += txt\n\
catchOutErr = CatchOutErr()\n\
sys.stdout = catchOutErr\n\
sys.stderr = catchOutErr\n\
"; //this is python code to redirect stdouts/stderr

    Py_Initialize();
    PyObject *pModule = PyImport_AddModule("__main__"); //create main module
    PyRun_SimpleString(stdOutErr.c_str()); //invoke code to redirect
    PyRun_SimpleString("print(1+1)"); //this is ok stdout
    PyRun_SimpleString("1+a"); //this creates an error
    PyObject *catcher = PyObject_GetAttrString(pModule,"catchOutErr"); //get our catchOutErr created above
    PyErr_Print(); //make python print any errors

    PyObject *output = PyObject_GetAttrString(catcher,"value"); //get the stdout and stderr from our catchOutErr object

    printf("Here's the output:\n %s", PyString_AsString(output)); //it's not in our C++ portion

    Py_Finalize();


    return 0;

}

Solution 2

Here is a C++ friendly solution I have developed lately.

I explain a few details of it on my blog: Python sys.stdout redirection in C++ where I also point to repository at my GitHub where most recent version can be found. Here is complete example based on the current code at the time of posting this answer:

#include <functional>
#include <iostream>
#include <string>
#include <Python.h>

namespace emb
{

typedef std::function<void(std::string)> stdout_write_type;

struct Stdout
{
    PyObject_HEAD
    stdout_write_type write;
};

PyObject* Stdout_write(PyObject* self, PyObject* args)
{
    std::size_t written(0);
    Stdout* selfimpl = reinterpret_cast<Stdout*>(self);
    if (selfimpl->write)
    {
        char* data;
        if (!PyArg_ParseTuple(args, "s", &data))
            return 0;

        std::string str(data);
        selfimpl->write(str);
        written = str.size();
    }
    return PyLong_FromSize_t(written);
}

PyObject* Stdout_flush(PyObject* self, PyObject* args)
{
    // no-op
    return Py_BuildValue("");
}

PyMethodDef Stdout_methods[] =
{
    {"write", Stdout_write, METH_VARARGS, "sys.stdout.write"},
    {"flush", Stdout_flush, METH_VARARGS, "sys.stdout.flush"},
    {0, 0, 0, 0} // sentinel
};

PyTypeObject StdoutType =
{
    PyVarObject_HEAD_INIT(0, 0)
    "emb.StdoutType",     /* tp_name */
    sizeof(Stdout),       /* tp_basicsize */
    0,                    /* tp_itemsize */
    0,                    /* tp_dealloc */
    0,                    /* tp_print */
    0,                    /* tp_getattr */
    0,                    /* tp_setattr */
    0,                    /* tp_reserved */
    0,                    /* tp_repr */
    0,                    /* tp_as_number */
    0,                    /* tp_as_sequence */
    0,                    /* tp_as_mapping */
    0,                    /* tp_hash  */
    0,                    /* tp_call */
    0,                    /* tp_str */
    0,                    /* tp_getattro */
    0,                    /* tp_setattro */
    0,                    /* tp_as_buffer */
    Py_TPFLAGS_DEFAULT,   /* tp_flags */
    "emb.Stdout objects", /* tp_doc */
    0,                    /* tp_traverse */
    0,                    /* tp_clear */
    0,                    /* tp_richcompare */
    0,                    /* tp_weaklistoffset */
    0,                    /* tp_iter */
    0,                    /* tp_iternext */
    Stdout_methods,       /* tp_methods */
    0,                    /* tp_members */
    0,                    /* tp_getset */
    0,                    /* tp_base */
    0,                    /* tp_dict */
    0,                    /* tp_descr_get */
    0,                    /* tp_descr_set */
    0,                    /* tp_dictoffset */
    0,                    /* tp_init */
    0,                    /* tp_alloc */
    0,                    /* tp_new */
};

PyModuleDef embmodule =
{
    PyModuleDef_HEAD_INIT,
    "emb", 0, -1, 0,
};

// Internal state
PyObject* g_stdout;
PyObject* g_stdout_saved;

PyMODINIT_FUNC PyInit_emb(void) 
{
    g_stdout = 0;
    g_stdout_saved = 0;

    StdoutType.tp_new = PyType_GenericNew;
    if (PyType_Ready(&StdoutType) < 0)
        return 0;

    PyObject* m = PyModule_Create(&embmodule);
    if (m)
    {
        Py_INCREF(&StdoutType);
        PyModule_AddObject(m, "Stdout", reinterpret_cast<PyObject*>(&StdoutType));
    }
    return m;
}

void set_stdout(stdout_write_type write)
{
    if (!g_stdout)
    {
        g_stdout_saved = PySys_GetObject("stdout"); // borrowed
        g_stdout = StdoutType.tp_new(&StdoutType, 0, 0);
    }

    Stdout* impl = reinterpret_cast<Stdout*>(g_stdout);
    impl->write = write;
    PySys_SetObject("stdout", g_stdout);    
}

void reset_stdout()
{
    if (g_stdout_saved)
        PySys_SetObject("stdout", g_stdout_saved);

    Py_XDECREF(g_stdout);
    g_stdout = 0;
}

} // namespace emb

int main()
{
    PyImport_AppendInittab("emb", emb::PyInit_emb);
    Py_Initialize();
    PyImport_ImportModule("emb");

    PyRun_SimpleString("print(\'hello to console\')");

    // here comes the ***magic***
    std::string buffer;
    {
        // switch sys.stdout to custom handler
        emb::stdout_write_type write = [&buffer] (std::string s) { buffer += s; };
        emb::set_stdout(write);
        PyRun_SimpleString("print(\'hello to buffer\')");
        PyRun_SimpleString("print(3.14)");
        PyRun_SimpleString("print(\'still talking to buffer\')");
        emb::reset_stdout();
    }

    PyRun_SimpleString("print(\'hello to console again\')");
    Py_Finalize();

    // output what was written to buffer object
    std::clog << buffer << std::endl;
}

This allows to intercept sys.stdout.write output with any kind of callable C++ entity: free function, class member function, named function objects or even anonymous functions as in the example above where I use C++11 lambda.

Note, this is a minimal example to present the essential concept. In production-ready code, it certainly needs more attention around reference counting of PyObject, getting rid of global state, and so on.

Solution 3

I know this question is old, but one part of the question has not been answered yet:

"How to catch output of commands that don't directly write to the stdout of Python, like: 1+1 ?"

Here are the steps (for Python 3.4):

1. Redirect stdout/stderr into a Python variable using Mark's solution: https://stackoverflow.com/a/4307737/1046299

2. Copy function PyRun_InteractiveOneObject(FILE *fp, PyObject *filename, PyCompilerFlags *flags) from Python source code. It is located in file pythonrun.c

3. Modify the PyRun_InteractiveOneObject function name and signature so that the new function takes a const char* (your command) as first parameter instead of a FILE*. Then you will need to use PyParser_ASTFromStringObject instead of PyParser_ASTFromFileObject in the function implementation. Note that you will need to copy the function run_mod as is from Python since it is called within the function.

4. Call the new function with your command, for example 1+1. Stdout should now receive the output 2.

Comments

[alexpov]

Hello mark, thak you , it is very helpfull. Can you please explain several things. First of all, how does the cather works, second, i've posted on my question changes i've made according to your suggestion. When i run the second command in main (PythonPrinting("1+5");), the PyString_AsString(output) function return an empty string, meanin, i am loosing the original python output which is: 6. What changes can i do not to lose this ? Thank you again ...:)

[Mark]

@alexpov, the catcher works by simply redirecting the Pythons stdout and stderr to a variable. You do not see any output for "1+5" because python does not send anything to stdout for that case. You should use "print(1+5)". Also, refactor your code, you should not make multiple calls to the PyImport_AddModule

[alexpov]

hello, in my case i simply run python commands from c and all i need is a way to capture all the pythons output. I don't know which command it will be, a command that makes python to print to sdtout or stderr or a command "1+1". I can't wrap all my commands with print. Do you know a way to make a redirection (or something else) so i could catch those outputs to? (where does python send that output?) About PyImport_AddModule,when i call it once,catcher "value" holds all the previous outputs.How do i init this value to empty string after each call for PythonPrinting? Thank you again, Alex

[chown]

Don't forget to check to make sure it is a PyString object first with: if (PyObject_IsInstance(output, (PyObject *)&PyString_Type)) { and free up your memory with: Py_DecRef(output);.

[Leo]

update: you must also implement flush(). is there a way to override only the write method?

[G.Vanem]

I'm not very familiar with Python3.x. But for Python3, I had to replace PyString_AsString() with PyBytes_AsString(). Correct?

[jpo38]

Works like a charm. Would be even perfect if also redirecting stderr to a different function....;-)

[mloskot]

@jpo38 I'm glad to know. I kept the example as short as possible, I think, leaving any improvements to reader. It should be not so difficult to achieve what you suggest :)

[jpo38]

@mloskot: Absolutely, I extended it esaily to redirect hem both. That was just a free comment. Thank you for this code, it really helped!

[jpo38]

Arg, this actually fails when the message has utf-8 encoding. Then strings are truncated...

[soegaard]

Beware that self.value += txt can lead to quadratic behaviour if the program repeatedly writes small strings to the output.