pydeepdiff

Computes deep differences between objects


License
Other
Install
pip install pydeepdiff==1.0.3

Documentation

pydeepdiff

Description

This package allows to compute deep differences between two python dictionnaries

Differences are stored in the following format :

lhs stands for Left Hand Side rhs stands for Right Hand Side

  • Creation : {'path':path_to_object,'kind':'N','rhs':value}
  • Deletion : {'path':path_to_object,'kind':'D','lhs':value}
  • Edition : {'path':path_to_object,'kind':'E','lhs':value,'rhs':value}
  • Lists
    • Creation : {'path':path_to_object,'kind':'N','rhs_idx':index,'rhs':value}
    • Deletion : {'path':path_to_object,'kind':'D','lhs_idx':index,'lhs':value}
    • Move : {'path':path_to_object,'kind':'M','lhs_idx':index,'rhs_idx':index}

Requirements

This lib requires python 3.2+

Install

    pip install pydeepdiff

Example of use

    from pydeepdiff.diff import get_diff

        dict_a = {'field_1': 'id1', 'field_2': 'vala1'}
        dict_b = {'field_1': 'id3', 'field_2': 'valb1'}

        differences = get_diff(dict_a, dict_b)

        # Differences contains one difference :
            [
                {
                    'path': [
                        'field_1'
                    ],
                    'kind': 'E',
                    'lhs': 'id1',
                    'rhs': 'id3'
                },
                {
                    'path': [
                        'field_2'
                    ],
                    'kind': 'E',
                    'lhs': 'vala1',
                    'rhs': 'valb1'
                }
            ]


        dict_a = {'id': '1', 'bloc': {'act': '1'}}
        dict_b = {'id': '1', 'bloc': {'act': '2'}}

        differences = get_diff(dict_a, dict_b))

        # Differences contains one difference :
            [
                {
                    'path': [
                        'bloc',
                        'act'
                    ],
                    'kind': 'E',
                    'lhs': '1',
                    'rhs': '2'
                }
            ]

Ignored fields

It is possible to ignore some fields when comparing objects. For this, pass a list of these fields to the comparison function.

Example : don't compare the field 'act' of the nested 'bloc' object :

    dict_a = {'id': '1', 'bloc': {'act': '1'}}
    dict_b = {'id': '1', 'bloc': {'act': '2'}}

    diff = get_diff(dict_a, dict_b, 'root', {}, ['root.bloc.act'])

In this example the result is an empty list of differences.

Specific mapping

When comparing two list of dict, we have to "associate" each item of the left side list to an item of the right side list. For this, we have to know "HOW" making this association : often a dict will have a field that represents its id, and we want to use it. In pydeepdiff, this case is resolved with a mapping file.

In the following example, we explicitly use the 'field_1' to identify an object of the list.

    list_a = [{'field_1': 'id1', 'field_2': 'vala1'}, {'field_1': 'id2', 'field_2': 'vala2'}]
    list_b = [{'field_1': 'id3', 'field_2': 'valb1'}, {'field_1': 'id1', 'field_2': 'valb2'}]

    mapping = [{'path': '', 'id': 'field_1'}]

    diff = _get_list_dict_diff(list_a, list_b, 'root', mapping, p_complex_details=True)

The mapping is represented by a list of object. Each object has a :

  • a "path" field that contains the json path to the object
  • an "id" field that contains the name of the field representing the id of the object (pointed by "path")

Note that a path to the root object is the empty string

The result is the following list of differences :

    [
        {'path_to_object': '', 'filter': '', 'rhs_idx': 1, 'lhs_idx': 0, 'kind': 'M'},
        {'rhs': 'valb2', 'lhs': 'vala1', 'kind': 'E', 'path_to_object': '[0].field_2', 'filter': 'field_2'},
        {'lhs_idx': 1, 'kind': 'D', 'lhs': {'field_2': 'vala2', 'field_1': 'id2'}, 'path_to_object': '', 'filter': ''},
        {'rhs_idx': 0, 'kind': 'N', 'rhs': {'field_2': 'valb1', 'field_1': 'id3'}, 'path_to_object': '', 'filter': ''}
    ]

Tests

To launch unit tests, just run this command from the project home directory (you will need py.test installed)

py.test test