ansible/test/units/module_utils/common/test_collections.py
Matt Martz 48f12c14e9
AnsibleVaultEncryptedUnicode should be considered a string (#71609)
* AnsibleVaultEncryptedUnicode should be considered a string

* linting fix

* clog frag
2020-09-03 14:54:00 -05:00

175 lines
5.2 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# -*- coding: utf-8 -*-
# Copyright (c) 20182019, Sviatoslav Sydorenko <webknjaz@redhat.com>
# Simplified BSD License (see licenses/simplified_bsd.txt or https://opensource.org/licenses/BSD-2-Clause)
"""Test low-level utility functions from ``module_utils.common.collections``."""
from __future__ import absolute_import, division, print_function
__metaclass__ = type
import pytest
from ansible.module_utils.six import Iterator
from ansible.module_utils.common._collections_compat import Sequence
from ansible.module_utils.common.collections import ImmutableDict, is_iterable, is_sequence
class SeqStub:
"""Stub emulating a sequence type.
>>> from collections.abc import Sequence
>>> assert issubclass(SeqStub, Sequence)
>>> assert isinstance(SeqStub(), Sequence)
"""
Sequence.register(SeqStub)
class IteratorStub(Iterator):
def __next__(self):
raise StopIteration
class IterableStub:
def __iter__(self):
return IteratorStub()
class FakeAnsibleVaultEncryptedUnicode(Sequence):
__ENCRYPTED__ = True
def __init__(self, data):
self.data = data
def __getitem__(self, index):
return self.data[index]
def __len__(self):
return len(self.data)
TEST_STRINGS = u'he', u'Україна', u'Česká republika'
TEST_STRINGS = TEST_STRINGS + tuple(s.encode('utf-8') for s in TEST_STRINGS) + (FakeAnsibleVaultEncryptedUnicode(u'foo'),)
TEST_ITEMS_NON_SEQUENCES = (
{}, object(), frozenset(),
4, 0.,
) + TEST_STRINGS
TEST_ITEMS_SEQUENCES = (
[], (),
SeqStub(),
)
TEST_ITEMS_SEQUENCES = TEST_ITEMS_SEQUENCES + (
# Iterable effectively containing nested random data:
TEST_ITEMS_NON_SEQUENCES,
)
@pytest.mark.parametrize('sequence_input', TEST_ITEMS_SEQUENCES)
def test_sequence_positive(sequence_input):
"""Test that non-string item sequences are identified correctly."""
assert is_sequence(sequence_input)
assert is_sequence(sequence_input, include_strings=False)
@pytest.mark.parametrize('non_sequence_input', TEST_ITEMS_NON_SEQUENCES)
def test_sequence_negative(non_sequence_input):
"""Test that non-sequences are identified correctly."""
assert not is_sequence(non_sequence_input)
@pytest.mark.parametrize('string_input', TEST_STRINGS)
def test_sequence_string_types_with_strings(string_input):
"""Test that ``is_sequence`` can separate string and non-string."""
assert is_sequence(string_input, include_strings=True)
@pytest.mark.parametrize('string_input', TEST_STRINGS)
def test_sequence_string_types_without_strings(string_input):
"""Test that ``is_sequence`` can separate string and non-string."""
assert not is_sequence(string_input, include_strings=False)
@pytest.mark.parametrize(
'seq',
([], (), {}, set(), frozenset(), IterableStub()),
)
def test_iterable_positive(seq):
assert is_iterable(seq)
@pytest.mark.parametrize(
'seq', (IteratorStub(), object(), 5, 9.)
)
def test_iterable_negative(seq):
assert not is_iterable(seq)
@pytest.mark.parametrize('string_input', TEST_STRINGS)
def test_iterable_including_strings(string_input):
assert is_iterable(string_input, include_strings=True)
@pytest.mark.parametrize('string_input', TEST_STRINGS)
def test_iterable_excluding_strings(string_input):
assert not is_iterable(string_input, include_strings=False)
class TestImmutableDict:
def test_scalar(self):
imdict = ImmutableDict({1: 2})
assert imdict[1] == 2
def test_string(self):
imdict = ImmutableDict({u'café': u'くらとみ'})
assert imdict[u'café'] == u'くらとみ'
def test_container(self):
imdict = ImmutableDict({(1, 2): ['1', '2']})
assert imdict[(1, 2)] == ['1', '2']
def test_from_tuples(self):
imdict = ImmutableDict((('a', 1), ('b', 2)))
assert frozenset(imdict.items()) == frozenset((('a', 1), ('b', 2)))
def test_from_kwargs(self):
imdict = ImmutableDict(a=1, b=2)
assert frozenset(imdict.items()) == frozenset((('a', 1), ('b', 2)))
def test_immutable(self):
imdict = ImmutableDict({1: 2})
expected_reason = r"^'ImmutableDict' object does not support item assignment$"
with pytest.raises(TypeError, match=expected_reason):
imdict[1] = 3
with pytest.raises(TypeError, match=expected_reason):
imdict[5] = 3
def test_hashable(self):
# ImmutableDict is hashable when all of its values are hashable
imdict = ImmutableDict({u'café': u'くらとみ'})
assert hash(imdict)
def test_nonhashable(self):
# ImmutableDict is unhashable when one of its values is unhashable
imdict = ImmutableDict({u'café': u'くらとみ', 1: [1, 2]})
expected_reason = r"^unhashable type: 'list'$"
with pytest.raises(TypeError, match=expected_reason):
hash(imdict)
def test_len(self):
imdict = ImmutableDict({1: 2, 'a': 'b'})
assert len(imdict) == 2
def test_repr(self):
initial_data = {1: 2, 'a': 'b'}
initial_data_repr = repr(initial_data)
imdict = ImmutableDict(initial_data)
actual_repr = repr(imdict)
expected_repr = "ImmutableDict({0})".format(initial_data_repr)
assert actual_repr == expected_repr