Often times, when parsing a user-inputted JSON converted to dict object for a RESTful web service, I have a need to verify the high-level structure of the JSON. In the REST service, we may have many resources with various needed input dicts. Thus I want

dict_format.py

TYPE_PARSERS = {
    'http://www.w3.org/2001/XMLSchema.xsd#string': str,
    'http://www.w3.org/2001/XMLSchema.xsd#int': int,
    'http://www.w3.org/2001/XMLSchema.xsd#float': float,
    'http://www.w3.org/2001/XMLSchema.xsd#date': { 'module': 'datetime',
                                                       'klass': 'datetime',
                                                       'method': 'strptime',
                                                       'args': '%d-%m-%y'},
    'http://www.w3.org/2001/XMLSchema.xsd#datetime': { 'module': 'datetime',
                                                       'klass': 'datetime',
                                                       'method': 'strptime',
                                                       'args': '%d-%m-%y %H:%M'},
    'http://www.w3.org/2001/XMLSchema.xsd#hexBinary': { 'module': 'uuid',
                                                       'klass': 'uuid',
                                                       'method': 'UUID'},
    'http://www.w3.org/2001/XMLSchema.xsd#localElement': dict
}


def convert_value_to_def(func_name, field_value):
    """
    Converts a value from any possible input type to the defined type from above
    """
    if isinstance(func_name, dict):
        # Setup for more complex conversions, like Dates
        m_imp = __import__(func_name.get('module'),
            globals(), locals(),
            [func_name.get('klass')], -1)
        try:
            klass = getattr(m_imp, func_name.get('klass'))
        except AttributeError:
            klass = m_imp
        func = getattr(klass, func_name.get('method'))
        convert = func(field_value, func_name.get('args'))
    elif func_name is not None:
        # For simple conversions, like string
        convert = func_name(field_value)
    return convert


class JSON_Struct(object):
    """
    Type checking class that behaves like a callable after instantiation
    """
    def __init__(self, sub_types=None, addl_funks=None):
        """
        Create the checking object
        :arg: sub_types - dict defining the keys to search for in the input dict
              The keys of sub_types are the keys that are available for the input
              The values of sub_types are a dict with keys - 'required' and 'types'
                  required - boolean, Is the tag required in the input dict
                  types - list, predefined types from TYPE_PARSERS or a user-defined type
        :arg: addl_funks - dict defining the conversion callable for the user-defined types
              The keys of addl_funks are the user-defined types (normally nested dict structs)
              The values of addl_funks are the conversion callables (normally an instance of JSON_Struct)
        """
        self.sub_types = sub_types or {}
        self.addl_funks = addl_funks or {}

    def __call__(self, args):
        """
        Allows the instance to be called as a function to validate the structure of the input dict
        :arg: args - input dict from the user that should be validated
        """
        missed_flds = [tag for tag, specs in self.sub_types.iteritems() 
                      if specs.get('required', False) and tag not in args.keys()]
        if missed_flds:
            raise KeyError
        for key, value in args.iteritems():
            if self.sub_types.has_key('_'):
                key = '_'
            elif not self.sub_types.has_key(key):
                raise KeyError
            avail_types = self.sub_types[key]['types']
            raised_errors = list()
            converted = None
            for sub_type in avail_types:
                func_name = TYPE_PARSERS.get(sub_type, 
                                             self.addl_funks.get(sub_type))
                try:
                    converted = convert_value_to_def(func_name, value)
                except Exception, e:
                    raised_errors.append(e)
            if converted is None and raised_errors:
                raise ValueError


if __name__ == '__main__':
    """
    Test for handling a basic structure of a 1-level dict
    """
    simple = JSON_Struct({'a': {'types': ['http://www.w3.org/2001/XMLSchema.xsd#string']}, 
                          'b': {'types': ['http://www.w3.org/2001/XMLSchema.xsd#int']}})
    conv = simple({'a': 'some string', 'b': '100'})
    """
    Test for handling unvalidated 1-level dict
    """
    try:
        conv = simple({'a': 'some string', 'b': 'bad string'})
        raise Exception
    except (KeyError, ValueError):
        pass
    """
    Test for nested 2-level dict
    """ 
    simple = JSON_Struct({'c': {'types': ['http://www.w3.org/2001/XMLSchema.xsd#string']}, 
                          'd': {'types': ['http://www.w3.org/2001/XMLSchema.xsd#string']}})
    complex = JSON_Struct({'a': {'types': ['http://www.w3.org/2001/XMLSchema.xsd#int', 
                                 'http://www.w3.org/2001/XMLSchema.xsd#float']}, 
                           'b': {'types': ['simple_def']}},
                           {'simple_def': simple})
    conv = complex({'a': '45', 'b': {'c': 'some string', 'd': 'another'}})
    
    """
    Test for recursive nested dict with nested child requirements
    The 'b' tag accepts another a and b dict, and for the children, the 'a' tag is required
    """
    simple = JSON_Struct({'a': {'types': ['http://www.w3.org/2001/XMLSchema.xsd#string']}})
    import copy
    copied = copy.deepcopy(simple)
    copied.sub_types['a'].update({'required': True})
    copied.sub_types.update({'b': {'types': ['child_req']}})
    copied.addl_funks.update({'child_req': copied})
    simple.sub_types.update({'b': {'types': ['child_req']}})
    simple.addl_funks.update({'child_req': copied})
    conv = simple({'a': 'some', 'b': {'a': 'foo'}})
    conv = simple({'a': 'foo', 'b': {'a': 'bar', 'b': {'a': 'baz', 'b': {'a': 'bay'}}}})
    conv = simple({'b': {'a': 'foobar'}})