sign_structure - sign nested Python structures (including dicts) while ignoring dict ordering

structsign.py

from hashlib import sha1
import unittest

__all__ = ['sign_structure']


def sign_structure(data):
    """Create a (non-cryptographic) signature of the data structure.

    This function makes sure that dict item ordering doesn't matter.
    """
    content = []

    def process_dict(x):
        content.append('{')
        for k in sorted(x.keys()):
            process(k)
            content.append('=')
            process(x[k])
            content.append(';')
        content.append('}')

    def process_sequence(x, typ):
        content.append('(' if typ == tuple else '[')
        for el in x:
            process(el)
            content.append(';')
        content.append(')' if typ == tuple else ']')

    def process(x):
        if isinstance(x, dict):
            process_dict(x)
        elif isinstance(x, list) or isinstance(x, tuple):
            process_sequence(x, type(x))
        elif isinstance(x, unicode):
            content.append(u"'%s'" % x.encode('utf-8'))
        elif isinstance(x, float):
            content.append('%.8f;' % x)
        elif isinstance(x, int):
            content.append('%d;' % x)
        else:
            content.append("'%s';" % str(x))

    process(data)
    return sha1(''.join(content)).hexdigest()


class SignTest(unittest.TestCase):

    def sha1(self, txt):
        return sha1(txt).hexdigest()

    def test_string_signing(self):
        self.assertEqual(sign_structure(''),
            'f3c3af94dfbad817c6a52c11882ba03102d322d6')
        self.assertEqual(sign_structure('abc'),
            'adab040b65c164a21f2cee29d068e37baef8cafd')

    def test_unicode_signing(self):
        self.assertEqual(
            sign_structure('\xc5\xa0\xc4\x90\xc4\x8c\xc4\x86\xc5\xbd'),
            '5d66fe7a5014f54d509aa23db1a2e8203ff9869d')

    def test_integer_signing(self):
        self.assertEqual(sign_structure(42),
            '3ea704a51effa29961c48dbda82a85bdad323602')

    def test_float_signing(self):
        self.assertEqual(sign_structure(3.14159265389),
            '07cab0e741f60c7a7051db9abd672b738b1420f6')

    def test_list_signing(self):
        self.assertEqual(sign_structure([1, 2, 3, 4, 5]),
            '7b16598465b45e23a7f8bb6d187ab1b282a45185')

    def test_tuple_signing(self):
        self.assertEqual(sign_structure((1, 2, 3, 4, 5)),
            '5549c793d5b1d44995156d20e4a97171b65adbdc')

    def test_list_and_tuple_signature_differ(self):
        self.assertNotEqual(sign_structure(1,), sign_structure([1]))

    def test_dict_signing(self):
        self.assertEqual(sign_structure({'a': 1, 'b': 2}),
            'a46171bee1eca62076215398ba0d4cbe69bde70f')

    def test_dict_ordering_doesnt_matter(self):
        from collections import OrderedDict
        a = OrderedDict()
        a['foo'] = 1
        a['bar'] = 2
        b = OrderedDict()
        b['bar'] = 2
        b['foo'] = 1
        self.assertNotEqual(a, b)
        self.assertEqual(dict(a), dict(b))
        self.assertEqual(sign_structure(a), sign_structure(b))


if __name__ == '__main__':
    unittest.main()