Debugging
=========

The testrunner module supports post-mortem debugging and debugging
using `pdb.set_trace`.  Let's look first at using `pdb.set_trace`.
To demonstrate this, we'll provide input via helper Input objects:

    >>> class Input:
    ...     def __init__(self, src):
    ...         self.lines = src.split('\n')
    ...     def readline(self):
    ...         line = self.lines.pop(0)
    ...         print line
    ...         return line+'\n'

If a test or code called by a test calls pdb.set_trace, then the
runner will enter pdb at that point:

    >>> import os.path, sys
    >>> directory_with_tests = os.path.join(this_directory, 'testrunner-ex')
    >>> from zope.testing import testrunner
    >>> defaults = [
    ...     '--path', directory_with_tests,
    ...     '--tests-pattern', '^sampletestsf?$',
    ...     ]

    >>> real_stdin = sys.stdin
    >>> if sys.version_info[:2] == (2, 3):
    ...     sys.stdin = Input('n\np x\nc')
    ... else:
    ...     sys.stdin = Input('p x\nc')

    >>> sys.argv = ('test -ssample3 --tests-pattern ^sampletests_d$'
    ...             ' -t set_trace1').split()
    >>> try: testrunner.run_internal(defaults)
    ... finally: sys.stdin = real_stdin
    ... # doctest: +ELLIPSIS
    Running zope.testing.testrunner.layer.UnitTests tests:
    ...
    > testrunner-ex/sample3/sampletests_d.py(27)test_set_trace1()
    -> y = x
    (Pdb) p x
    1
    (Pdb) c
      Ran 1 tests with 0 failures and 0 errors in 0.001 seconds.
    ...
    False

Note that, prior to Python 2.4, calling pdb.set_trace caused pdb to
break in the pdb.set_trace function.  It was necessary to use 'next'
or 'up' to get to the application code that called pdb.set_trace.  In
Python 2.4, pdb.set_trace causes pdb to stop right after the call to
pdb.set_trace.

You can also do post-mortem debugging, using the --post-mortem (-D)
option:

    >>> sys.stdin = Input('p x\nc')
    >>> sys.argv = ('test -ssample3 --tests-pattern ^sampletests_d$'
    ...             ' -t post_mortem1 -D').split()
    >>> try: testrunner.run_internal(defaults)
    ... finally: sys.stdin = real_stdin
    ... # doctest: +NORMALIZE_WHITESPACE +REPORT_NDIFF
    Running zope.testing.testrunner.layer.UnitTests tests:
    ...
    Error in test test_post_mortem1 (sample3.sampletests_d.TestSomething)
    Traceback (most recent call last):
      File "testrunner-ex/sample3/sampletests_d.py",
              line 34, in test_post_mortem1
        raise ValueError
    ValueError
    <BLANKLINE>
    exceptions.ValueError:
    <BLANKLINE>
    > testrunner-ex/sample3/sampletests_d.py(34)test_post_mortem1()
    -> raise ValueError
    (Pdb) p x
    1
    (Pdb) c
    True

Note that the test runner exits after post-mortem debugging.

In the example above, we debugged an error.  Failures are actually
converted to errors and can be debugged the same way:

    >>> sys.stdin = Input('up\np x\np y\nc')
    >>> sys.argv = ('test -ssample3 --tests-pattern ^sampletests_d$'
    ...             ' -t post_mortem_failure1 -D').split()
    >>> try: testrunner.run_internal(defaults)
    ... finally: sys.stdin = real_stdin
    ... # doctest: +NORMALIZE_WHITESPACE +REPORT_NDIFF
    Running zope.testing.testrunner.layer.UnitTests tests:
    ...
    Error in test test_post_mortem_failure1 (sample3.sampletests_d.TestSomething)
    Traceback (most recent call last):
      File ".../unittest.py",  line 252, in debug
        getattr(self, self.__testMethodName)()
      File "testrunner-ex/sample3/sampletests_d.py",
        line 42, in test_post_mortem_failure1
        self.assertEqual(x, y)
      File ".../unittest.py", line 302, in failUnlessEqual
        raise self.failureException, \
    AssertionError: 1 != 2
    <BLANKLINE>
    exceptions.AssertionError:
    1 != 2
    > .../unittest.py(302)failUnlessEqual()
    -> ...
    (Pdb) up
    > testrunner-ex/sample3/sampletests_d.py(42)test_post_mortem_failure1()
    -> self.assertEqual(x, y)
    (Pdb) p x
    1
    (Pdb) p y
    2
    (Pdb) c
    True