use doctest and logging in python program

Question

#!/usr/bin/python2.4
import logging
import sys
import doctest
def foo(x):
        """
    >>> foo (0)
    0
        """
        print ("%d" %(x))
        _logger.debug("%d" %(x))
def _test():
        doctest.testmod()
_logger = logging.getLogger()
_logger.setLevel(logging.DEBUG)
_formatter = logging.Formatter('%(message)s')
_handler = logging.StreamHandler(sys.stdout)
_handler.setFormatter(_formatter)
_logger.addHandler(_handler)
_test()

I would like to use logger module for all of my print statements. I have looked at the first 50 top google links for this, and they seem to agree that doctest uses it's own copy of the stdout. If print is used it works if logger is used it logs to the root console. Can someone please demonstrate a working example with a code snippet that will allow me to combine. Note running nose to test doctest will just append the log output at the end of the test, (assuming you set the switches) it does not treat them as a print statement.

Answer 1

OK, this is a simple one-liner that works well:

Within your doctest, before any logging capture is needed, do a addHandler(logging.Streamhandler(sys.stdout)) on your logger. So, assuming loggeris your logging object:

"""
This is a doctest that will capture output from the logging module.

>>> logger.addHandler(logging.StreamHandler(sys.stdout))

The rest of your doctest, now able to use output from the logging
module...
"""

Explanation: Once the doctest is running, doctest has already put its spoof in place, so sys.stdout is now set to DocTestRunner._fakeout. If you create a logging.StreamHandler for sys.stdout at this point, sys.stdout will point to doctest's spoof for sys.stdout rather than the real sys.stdout.

This solution also has the advantage not depending on internal private variables in doctest, such as _fakeout or _SpoofOut, in case they change in the future.

If you get:

error: [Errno 128] Transport endpoint is not connected

you might have forgotten to import sys.

Answer 2

I'm not sure why you want to do this, but if you really need to do it, then you can define your own subclass of DocTestRunner, and override the run method:

#imports left out, code not tested - just to point you in the right direction
class MyDocTestRunner(DocTestRunner):
    def run(self, test, compileflags=None, out=None, clear_globs=True):
        if out is None:
            handler = None
        else:
            handler = StreamHandler(out)
        logger = logging.getLogger() # root logger (say)
        if handler:
            logger.addHandler(handler)
        try:
            DocTestRunner.run(self, test, compileflags, out, clear_globs)
        finally:
            if handler:
                logger.removeHandler(handler)
                handler.close()

Then use this runner in place of DocTestRunner.

Answer 3

One simple and general-purpose approach looks like this:

LOGGER = logging.getLogger(__name__)
if hasattr(sys.modules['__main__'], '_SpoofOut'):
    LOGGER.setLevel(logging.DEBUG)
    LOGGER.addHandler(logger.addHandler(logging.StreamHandler()))

The _SpoofOut attribute is injected by the doctest module. If it's present, you can configure logging specifically for doctest. E.g. in my example, set a verbose debug mode and log to the console.