python-snippet/nested_mapping.py

from collections import Iterable
from boltons.iterutils import remap

def unique(seq):
    """Return a new sequence made of unique elements from seq.
       Code borrowed from the Internet.
    """
    uniques = []
    [uniques.append(x) for x in seq if not uniques.count(x)]
    return type(seq)(uniques)


class StrType(str):
    """Helper class to print and compare object types (used in function 'get_structure').
       When instanciating this class, provide the object whose type you want to print.
       __hash__ and __eq__ implemented in order to filter duplicates and yet stay compatible with
       uniqueness detection (to fit in sets for instance).
    """
    def __init__(self, obj):
        self.value = type(obj).__name__
        
    def __hash__(self):
        return hash(self.value)
    
    def __eq__(self, other):
        if not isinstance(other, StrType):
            return False
        else:
            return self.value == other.value
    
    def __str__(self):
        return f"<{self.value}>"
    
    __repr__ = __str__


def get_structure(data, folding_keys=[], folding_tag=" (folded)"):
    """Return a kind of map of the could-be nested data.
       Duplicates are removed from sequences.
       If 'folding_keys' are provided then the corresponding values in the data set will be folded,
       that is to say replaced by their data type (thus transformed into leaves). The key will be
       appended with 'folding_tag'.
    """
    if folding_keys:
        def visit_fold(path, key, value):
            if key in folding_keys:
                # We only fold iterable types (except for strings and bytes).
                if (isinstance(data, Iterable)
                        and not isinstance(data, str)
                        and not isinstance(data, bytes)):
                    return "".join([key, folding_tag]), type(value)()
            return True
        
        data = remap(data, visit=visit_fold)
    
    def extract_structure(data):
        if data and isinstance(data, dict):
            return {key:get_structure(value) for key, value in data.items()}
        elif data and (isinstance(data, Iterable)
                 and not isinstance(data, str)
                 and not isinstance(data, bytes)):
            # The iterable is not empty, we map each of its elements
            return unique(type(data)([get_structure(item) for item in data]))
        # It's a leaf (or empty iterable),
        # we return the name of its type
        return StrType(data)
    
    return extract_structure(data)


if __name__ == "__main__":        
    from pprint import pprint
    from datetime import datetime

    class MaClasse:
        pass
    
    nested = {'some types': {'MaClasse': MaClasse(),
                             'datetime': datetime.now(),
                             'empty-list': [],
                             'empty-tuple': (),
                             'float': 3.14,
                             'int': 0,
                             'list': [False,],
                             'set': {'a',},
                             'str': 'Babylone',
                             'tuple': ('jah',)},
              'sequence types': {'lists': ['a', 'b', 'c',
                                           datetime.now(),
                                           [1, 2, 3],
                                           [1, 2, 3]],
                                 'tuples': ('a', 'b', 'c',
                                            datetime.now(),
                                            [1, 2, 3],
                                            [1, 2, 3]),
                                 'sets': {'hello', 'world', 42, True}},
              '2 same dict 1 other': [{'same':3, 'as':b'the other'},
                                      {'same':3, 'as':b'the other'},
                                      {'different':'from','the':'other'}],
              'play me': MaClasse,
              'Two folded keys':{'Fold me :)':[1,2,3,4,5],
                                 'Fold me too':"folded, can't see me!"}}
    
    structure = get_structure(nested, ['Fold me :)','Fold me too'], ' [+]')

    print("The nested data:")
    pprint(nested)
    
    print("")
    print("Its mapping:")
    pprint(structure)
    
    ## Output:
    ##
    ##{'2 same dict 1 other': [{'as': <bytes>, 'same': <int>},
    ##                         {'different': <str>, 'the': <str>}],
    ## 'Two folded keys': {'Fold me :) [+]': <list>, 'Fold me too [+]': <str>},
    ## 'play me': <type>,
    ## 'sequence types': {'lists': [<str>, <datetime>, [<int>]],
    ##                    'sets': {<int>, <bool>, <str>},
    ##                    'tuples': (<str>, <datetime>, [<int>])},
    ## 'some types': {'MaClasse': <MaClasse>,
    ##                'datetime': <datetime>,
    ##                'empty-list': <list>,
    ##                'empty-tuple': <tuple>,
    ##                'float': <float>,
    ##                'int': <int>,
    ##                'list': [<bool>],
    ##                'set': {<str>},
    ##                'str': <str>,
    ##                'tuple': (<str>,)}}