Python classes, objects, external functions and cleaning up.

I’ve been working in some code to be able to use z/OS datasets, and DD statements. It took me a while to understand how some bits of Python work.

I also did things such as open a file, allocate a 1MB buffer, and wondered how to close the file, and release the buffer to prevent a storage leak.

The Python import

The Python import makes external functions and classes available to a program. The syntax is like

import abc as xyz

x = xyz…..

abc can be

  • a file
  • a directory abc
  • a load module

I’ll focus on the load module.

The load module

This can define a function based approach, so you would use it like

fileHandle = zos.fopen(“colin.c”,”rb”)
data = zos.fread(fileHandle)

You can provide many functions. Some may return a “handle” object, such as fileHandle which is passed to other functions.

It can also be object based and the C load module external function creates a new type.

file = zos.fopen(“colin.c”,”rb”)
data = file.fread()

The functions are associated with the object “file”, rather than the load module zos.

Internally the object is passed to the function.

Cleaning up

Within my code I had fileHandle = fopen(“datasetname”….), which allocated a 1MB buffer for the read function.

I also had fclose(fileHandle) where I closed the file and freed the buffer.

However I could also do

fileHandle = fopen(“datasetname1″….)
fileHandle = fopen(“datasetname2″….)
fileHandle = fopen(“datasetname3″….)

with no intermediate fclose(), which would lead to a storage leak as the file was fclose routine was not being called.

Using a class to call a function at exit

If you have a Python class for your data you can use

def cb(self,a,b):
     self.handle =  zconsole.acb(a,b)

def clean_up(self,handle):
    if handle != None:

When function cb is used, it registers with the “at exit” routine atexit, and says, “at exit” call my routine “clean_up”, and pass the handle.

At shutdown the clean_up routine is called once for every instance, and gives the cancel code a chance to clean up.

Using a C external function and “functions”.

Within the external functions C code, is PyModuleDef which defines the module to Python.

As such there is no way to automatically get your clean up function to be called (and free my 1MB buffer).

However you can exploit the Python module state data. For example

struct {
myparm * ...
} myStatic;

static struct PyModuleDef zos_module = {
  zos_methods, // the functions (methods)
  NULL, // Multi phase init. NULL -> single
  NULL, // Garbage collection traversal
  zos_clear, // Garbage collection clear
  zos_free // Garbage collection free

The block of state data is allocated for you, and you can issue the PyModule_GetState(PythonModule) function to get access this block.

You chain could chain your data from the state data, perhaps in a linked list.

When the clean up occurs, your “zos_free” routine will be called, and you can free all the storage you allocated and clean up.

For example

PyMODINIT_FUNC PyInit_zos(void) { 
  PyObject *d; 
  /* Create the module  */ 
  mzos = PyModule_Create(&zos_module); 
  // get the state data and initialise it
  state * pState = (state * )  PyModule_GetState(mzos); 
  memcpy(pState -> eyec,"state   ",8);
  PyDict_SetItemString(d, "__doc__", Py23Text_FromString(zos_doc)); 
  PyDict_SetItemString(d,"__version__", Py23Text_FromString(__version__)); 
return mzos;

Using a C external function and “objects” or types.

With a “function based” function, you have Python code like

fileHandle = zos.fopen("myfilename"....)
data = zos.fread(fileHande)

With “object based” functions you have Python code like

fileHandle = zos.fopen("myfilename"...)
data = fileHandle.fread()

In this case the object is a Python type. There is a good description here.

As with function based code you define the attributes of the object, including the tp_dealloc function. This gets control when the object is deallocated. In the Custom_dealloc, function you can close the file and free the buffer etc.

static PyTypeObject CustomType = {
    PyVarObject_HEAD_INIT(NULL, 0)
    .tp_name = "custom.Custom",
    .tp_doc = PyDoc_STR("Custom objects"),
    .tp_basicsize = sizeof(CustomObject),
    .tp_itemsize = 0,
    .tp_dealloc = (destructor) Custom_dealloc,
    .tp_flags = Py_TPFLAGS_DEFAULT,
    .tp_new = PyType_GenericNew,

static void
Custom_dealloc(CustomObject *self)
   ... // put your code here

static PyModuleDef custommodule = {
    .m_name = "custom",
    .m_doc = "Example module that creates an extension type.",
    .m_size = -1,

    PyObject *m;

    m = PyModule_Create(&custommodule);
    if (m == NULL)
        return NULL;
    if (PyModule_AddObject(m, "Custom", (PyObject *) &CustomType) <  0) {
        return NULL;
    return m;

Note: The list of available .tp… definitions is available here.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s