[TIP] doctest with nested dictionaries

[Raphael Marvie]

hi,

Couldn't YAML (http://www.yaml.org) come to the rescue?

"""
   >>> import yaml
   >>> def foo(request):
   ...     return dict(name="bob", children=[dict(name="mary",  
children=[])])
   >>> print yaml.dump(foo(None))
   children:
   - children: []
     name: mary
   name: bob
   <BLANKLINE>
"""

HTH

r.

On Jul 5, 2007, at 12:35 PM, Graham Carlyle wrote:

> Hi
>
> I've been testing some code with doctest recently and am wondering  
> whats
> the best way to proceed with a deeply nested data structure.
>
> I've got some functions that return a nested data structure of lists &
> dictionaries (django-amf views)
>
>>>> def foo(request):
>     ...     return dict(name="bob", children=[dict(name="mary",
> children=[])])
>
> I could just do a simple comparison:
>
>>>> foo(None) == dict(name="bob", children=[dict(name="mary",
> children=[])])
>     True
>
> but if it fails the error message isn't very helpful about whats  
> wrong.
>
> so as an alternative:
>
>>>> def areEqual(a, b):
>     ...     if a == b:
>     ...        return True
>     ...     else:
>     ...        return False, "%r != %r" % (a, b)
>
>>>> areEqual(foo(None), dict(name="bob", children=[]))
>     (False, "{'children': [{'children': [], 'name': 'mary'}], 'name':
> 'bob'} != {'children': [], 'name': 'bob'}")
>
> but that seems a retrograde step :) as the testing infrastructure is
> getting in the way
>
> Another way of solving this would be to pformat the output to  
> normalise
> it so:
>
>>>> from pprint import pformat
>>>> pformat(foo(None))
>     "{'children': [{'children': [], 'name': 'mary'}], 'name': 'bob'}"
>
> and for other parts of the response that may be dynamic substitute
> placeholders...
>
>>>> bobs_child = "mary"
>>>> response_str = pformat(foo(None))
>>>> response_str.replace(bobs_child, "<bobs child>")
>     "{'children': [{'children': [], 'name': '<bobs child>'}], 'name':
> 'bob'}"
>
> however i was wondering whether the lack of the pformat  
> normalisation as
> a standard doctest option indicated that this is not considered a good
> practice in using doctest. Also this seems similar to the assertEquals
> solution in that the testing infrastructure is getting in the way.
>
> A purer doctest option seems to be to break the datastructure apart  
> and
> compare it bit by bit but that would seem less readable and more
> verbose.
>
>>>> response = foo(None)
>>>> type(response)
>     <type 'dict'>
>>>> response["name"]
>     'bob'
>>>> children = response["children"]
>
>     ...etc...
>
> Any opinions welcomed!
>
> thanks,
> Graham
>
>
>
> _______________________________________________
> testing-in-python mailing list
> testing-in-python at lists.idyll.org
> http://lists.idyll.org/listinfo/testing-in-python

--
Raphael Marvie, PhD                http://www.lifl.fr/~marvie/
Maître de Conférences / Associate Professor  @  LIFL -- IRCICA
Directeur du Master Informatique Professionnel spécialité IAGL
Head of Master's in Software Engineering     +33 3 28 77 85 83