Merge nested JSON dictionaries

mege_json.py

#!/usr/bin/env python
import collections
import json
import sys


MISSING = object()


class ConflictError(Exception): pass


def merge_files(filenames):
    trees = []
    for filename in filenames:
        with open(filename, 'rb') as f:
            tree = json.load(f, object_pairs_hook=collections.OrderedDict)
            trees.append(tree)
    merged = collections.OrderedDict()
    for tree in trees:
        for path, value in get_items(tree):
            set_value(merged, path, value)
    print json.dumps(merged, indent=2)


def set_value(tree, path, value):
    key = path[-1]
    parent_path = path[:-1]
    for parent_key in parent_path:
        try:
            tree = tree[parent_key]
        except KeyError:
            new_tree = collections.OrderedDict()
            tree[parent_key] = new_tree
            tree = new_tree
    old_value = tree.get(key, MISSING)
    try:
        new_value = resolve(value, old_value)
    except ConflictError as exc:
        raise RuntimeError('Conflict at {!r}: {}'.format(path, exc))
    else:
        tree[key] = new_value


def resolve(new_value, old_value):
    if old_value == new_value:
        return old_value
    if old_value is MISSING:
        return new_value
    if old_value is None:
        return new_value
    if new_value is None:
        return old_value
    raise ConflictError(
            "Can't resolve new value {!r} and old value {!r}".format(
                new_value, old_value))



def get_items(tree, parent_path=()):
    for key, value in tree.items():
        path = parent_path + (key,)
        if isinstance(value, collections.Mapping):
            for child_path, child_value in get_items(value, parent_path=path):
                yield child_path, child_value
        else:
            yield path, value


if __name__ == '__main__':
    merge_files(sys.argv[1:])