Adding upstream version 2.2.0.

Signed-off-by: Daniel Baumann <daniel@debian.org>

tests/clientlib_test.py

@@ -0,0 +1,313 @@
+import logging
+
+import cfgv
+import pytest
+
+import pre_commit.constants as C
+from pre_commit.clientlib import check_type_tag
+from pre_commit.clientlib import CONFIG_HOOK_DICT
+from pre_commit.clientlib import CONFIG_REPO_DICT
+from pre_commit.clientlib import CONFIG_SCHEMA
+from pre_commit.clientlib import DEFAULT_LANGUAGE_VERSION
+from pre_commit.clientlib import MANIFEST_SCHEMA
+from pre_commit.clientlib import MigrateShaToRev
+from pre_commit.clientlib import validate_config_main
+from pre_commit.clientlib import validate_manifest_main
+from testing.fixtures import sample_local_config
+
+
+def is_valid_according_to_schema(obj, obj_schema):
+    try:
+        cfgv.validate(obj, obj_schema)
+        return True
+    except cfgv.ValidationError:
+        return False
+
+
+@pytest.mark.parametrize('value', ('definitely-not-a-tag', 'fiel'))
+def test_check_type_tag_failures(value):
+    with pytest.raises(cfgv.ValidationError):
+        check_type_tag(value)
+
+
+@pytest.mark.parametrize(
+    ('config_obj', 'expected'), (
+        (
+            {
+                'repos': [{
+                    'repo': 'git@github.com:pre-commit/pre-commit-hooks',
+                    'rev': 'cd74dc150c142c3be70b24eaf0b02cae9d235f37',
+                    'hooks': [{'id': 'pyflakes', 'files': '\\.py$'}],
+                }],
+            },
+            True,
+        ),
+        (
+            {
+                'repos': [{
+                    'repo': 'git@github.com:pre-commit/pre-commit-hooks',
+                    'rev': 'cd74dc150c142c3be70b24eaf0b02cae9d235f37',
+                    'hooks': [
+                        {
+                            'id': 'pyflakes',
+                            'files': '\\.py$',
+                            'args': ['foo', 'bar', 'baz'],
+                        },
+                    ],
+                }],
+            },
+            True,
+        ),
+        (
+            {
+                'repos': [{
+                    'repo': 'git@github.com:pre-commit/pre-commit-hooks',
+                    'rev': 'cd74dc150c142c3be70b24eaf0b02cae9d235f37',
+                    'hooks': [
+                        {
+                            'id': 'pyflakes',
+                            'files': '\\.py$',
+                            # Exclude pattern must be a string
+                            'exclude': 0,
+                            'args': ['foo', 'bar', 'baz'],
+                        },
+                    ],
+                }],
+            },
+            False,
+        ),
+    ),
+)
+def test_config_valid(config_obj, expected):
+    ret = is_valid_according_to_schema(config_obj, CONFIG_SCHEMA)
+    assert ret is expected
+
+
+def test_local_hooks_with_rev_fails():
+    config_obj = {'repos': [dict(sample_local_config(), rev='foo')]}
+    with pytest.raises(cfgv.ValidationError):
+        cfgv.validate(config_obj, CONFIG_SCHEMA)
+
+
+def test_config_with_local_hooks_definition_passes():
+    config_obj = {'repos': [sample_local_config()]}
+    cfgv.validate(config_obj, CONFIG_SCHEMA)
+
+
+def test_config_schema_does_not_contain_defaults():
+    """Due to the way our merging works, if this schema has any defaults they
+    will clobber potentially useful values in the backing manifest. #227
+    """
+    for item in CONFIG_HOOK_DICT.items:
+        assert not isinstance(item, cfgv.Optional)
+
+
+def test_validate_manifest_main_ok():
+    assert not validate_manifest_main(('.pre-commit-hooks.yaml',))
+
+
+def test_validate_config_main_ok():
+    assert not validate_config_main(('.pre-commit-config.yaml',))
+
+
+def test_validate_config_old_list_format_ok(tmpdir):
+    f = tmpdir.join('cfg.yaml')
+    f.write('-  {repo: meta, hooks: [{id: identity}]}')
+    assert not validate_config_main((f.strpath,))
+
+
+def test_validate_warn_on_unknown_keys_at_repo_level(tmpdir, caplog):
+    f = tmpdir.join('cfg.yaml')
+    f.write(
+        '-   repo: https://gitlab.com/pycqa/flake8\n'
+        '    rev: 3.7.7\n'
+        '    hooks:\n'
+        '    -   id: flake8\n'
+        '    args: [--some-args]\n',
+    )
+    ret_val = validate_config_main((f.strpath,))
+    assert not ret_val
+    assert caplog.record_tuples == [
+        (
+            'pre_commit',
+            logging.WARNING,
+            'Unexpected key(s) present on https://gitlab.com/pycqa/flake8: '
+            'args',
+        ),
+    ]
+
+
+def test_validate_warn_on_unknown_keys_at_top_level(tmpdir, caplog):
+    f = tmpdir.join('cfg.yaml')
+    f.write(
+        'repos:\n'
+        '-   repo: https://gitlab.com/pycqa/flake8\n'
+        '    rev: 3.7.7\n'
+        '    hooks:\n'
+        '    -   id: flake8\n'
+        'foo:\n'
+        '    id: 1.0.0\n',
+    )
+    ret_val = validate_config_main((f.strpath,))
+    assert not ret_val
+    assert caplog.record_tuples == [
+        (
+            'pre_commit',
+            logging.WARNING,
+            'Unexpected key(s) present at root: foo',
+        ),
+    ]
+
+
+@pytest.mark.parametrize('fn', (validate_config_main, validate_manifest_main))
+def test_mains_not_ok(tmpdir, fn):
+    not_yaml = tmpdir.join('f.notyaml')
+    not_yaml.write('{')
+    not_schema = tmpdir.join('notconfig.yaml')
+    not_schema.write('{}')
+
+    assert fn(('does-not-exist',))
+    assert fn((not_yaml.strpath,))
+    assert fn((not_schema.strpath,))
+
+
+@pytest.mark.parametrize(
+    ('manifest_obj', 'expected'),
+    (
+        (
+            [{
+                'id': 'a',
+                'name': 'b',
+                'entry': 'c',
+                'language': 'python',
+                'files': r'\.py$',
+            }],
+            True,
+        ),
+        (
+            [{
+                'id': 'a',
+                'name': 'b',
+                'entry': 'c',
+                'language': 'python',
+                'language_version': 'python3.4',
+                'files': r'\.py$',
+            }],
+            True,
+        ),
+        (
+            # A regression in 0.13.5: always_run and files are permissible
+            [{
+                'id': 'a',
+                'name': 'b',
+                'entry': 'c',
+                'language': 'python',
+                'files': '',
+                'always_run': True,
+            }],
+            True,
+        ),
+    ),
+)
+def test_valid_manifests(manifest_obj, expected):
+    ret = is_valid_according_to_schema(manifest_obj, MANIFEST_SCHEMA)
+    assert ret is expected
+
+
+@pytest.mark.parametrize(
+    'dct',
+    (
+        {'repo': 'local'}, {'repo': 'meta'},
+        {'repo': 'wat', 'sha': 'wat'}, {'repo': 'wat', 'rev': 'wat'},
+    ),
+)
+def test_migrate_sha_to_rev_ok(dct):
+    MigrateShaToRev().check(dct)
+
+
+def test_migrate_sha_to_rev_dont_specify_both():
+    with pytest.raises(cfgv.ValidationError) as excinfo:
+        MigrateShaToRev().check({'repo': 'a', 'sha': 'b', 'rev': 'c'})
+    msg, = excinfo.value.args
+    assert msg == 'Cannot specify both sha and rev'
+
+
+@pytest.mark.parametrize(
+    'dct',
+    (
+        {'repo': 'a'},
+        {'repo': 'meta', 'sha': 'a'}, {'repo': 'meta', 'rev': 'a'},
+    ),
+)
+def test_migrate_sha_to_rev_conditional_check_failures(dct):
+    with pytest.raises(cfgv.ValidationError):
+        MigrateShaToRev().check(dct)
+
+
+def test_migrate_to_sha_apply_default():
+    dct = {'repo': 'a', 'sha': 'b'}
+    MigrateShaToRev().apply_default(dct)
+    assert dct == {'repo': 'a', 'rev': 'b'}
+
+
+def test_migrate_to_sha_ok():
+    dct = {'repo': 'a', 'rev': 'b'}
+    MigrateShaToRev().apply_default(dct)
+    assert dct == {'repo': 'a', 'rev': 'b'}
+
+
+@pytest.mark.parametrize(
+    'config_repo',
+    (
+        # i-dont-exist isn't a valid hook
+        {'repo': 'meta', 'hooks': [{'id': 'i-dont-exist'}]},
+        # invalid to set a language for a meta hook
+        {'repo': 'meta', 'hooks': [{'id': 'identity', 'language': 'python'}]},
+        # name override must be string
+        {'repo': 'meta', 'hooks': [{'id': 'identity', 'name': False}]},
+    ),
+)
+def test_meta_hook_invalid(config_repo):
+    with pytest.raises(cfgv.ValidationError):
+        cfgv.validate(config_repo, CONFIG_REPO_DICT)
+
+
+@pytest.mark.parametrize(
+    'mapping',
+    (
+        # invalid language key
+        {'pony': '1.0'},
+        # not a string for version
+        {'python': 3},
+    ),
+)
+def test_default_language_version_invalid(mapping):
+    with pytest.raises(cfgv.ValidationError):
+        cfgv.validate(mapping, DEFAULT_LANGUAGE_VERSION)
+
+
+def test_minimum_pre_commit_version_failing():
+    with pytest.raises(cfgv.ValidationError) as excinfo:
+        cfg = {'repos': [], 'minimum_pre_commit_version': '999'}
+        cfgv.validate(cfg, CONFIG_SCHEMA)
+    assert str(excinfo.value) == (
+        f'\n'
+        f'==> At Config()\n'
+        f'==> At key: minimum_pre_commit_version\n'
+        f'=====> pre-commit version 999 is required but version {C.VERSION} '
+        f'is installed.  Perhaps run `pip install --upgrade pre-commit`.'
+    )
+
+
+def test_minimum_pre_commit_version_passing():
+    cfg = {'repos': [], 'minimum_pre_commit_version': '0'}
+    cfgv.validate(cfg, CONFIG_SCHEMA)
+
+
+@pytest.mark.parametrize('schema', (CONFIG_SCHEMA, CONFIG_REPO_DICT))
+def test_warn_additional(schema):
+    allowed_keys = {item.key for item in schema.items if hasattr(item, 'key')}
+    warn_additional, = [
+        x for x in schema.items if isinstance(x, cfgv.WarnAdditionalKeys)
+    ]
+    assert allowed_keys == set(warn_additional.keys)

tests/color_test.py

@@ -0,0 +1,59 @@
+import sys
+from unittest import mock
+
+import pytest
+
+from pre_commit import envcontext
+from pre_commit.color import format_color
+from pre_commit.color import GREEN
+from pre_commit.color import use_color
+
+
+@pytest.mark.parametrize(
+    ('in_text', 'in_color', 'in_use_color', 'expected'), (
+        ('foo', GREEN, True, f'{GREEN}foo\033[m'),
+        ('foo', GREEN, False, 'foo'),
+    ),
+)
+def test_format_color(in_text, in_color, in_use_color, expected):
+    ret = format_color(in_text, in_color, in_use_color)
+    assert ret == expected
+
+
+def test_use_color_never():
+    assert use_color('never') is False
+
+
+def test_use_color_always():
+    assert use_color('always') is True
+
+
+def test_use_color_no_tty():
+    with mock.patch.object(sys.stdout, 'isatty', return_value=False):
+        assert use_color('auto') is False
+
+
+def test_use_color_tty_with_color_support():
+    with mock.patch.object(sys.stdout, 'isatty', return_value=True):
+        with mock.patch('pre_commit.color.terminal_supports_color', True):
+            with envcontext.envcontext((('TERM', envcontext.UNSET),)):
+                assert use_color('auto') is True
+
+
+def test_use_color_tty_without_color_support():
+    with mock.patch.object(sys.stdout, 'isatty', return_value=True):
+        with mock.patch('pre_commit.color.terminal_supports_color', False):
+            with envcontext.envcontext((('TERM', envcontext.UNSET),)):
+                assert use_color('auto') is False
+
+
+def test_use_color_dumb_term():
+    with mock.patch.object(sys.stdout, 'isatty', return_value=True):
+        with mock.patch('pre_commit.color.terminal_supports_color', True):
+            with envcontext.envcontext((('TERM', 'dumb'),)):
+                assert use_color('auto') is False
+
+
+def test_use_color_raises_if_given_shenanigans():
+    with pytest.raises(ValueError):
+        use_color('herpaderp')

tests/commands/autoupdate_test.py

@@ -0,0 +1,437 @@
+import shlex
+
+import pytest
+
+import pre_commit.constants as C
+from pre_commit import git
+from pre_commit.commands.autoupdate import _check_hooks_still_exist_at_rev
+from pre_commit.commands.autoupdate import autoupdate
+from pre_commit.commands.autoupdate import RepositoryCannotBeUpdatedError
+from pre_commit.commands.autoupdate import RevInfo
+from pre_commit.util import cmd_output
+from testing.auto_namedtuple import auto_namedtuple
+from testing.fixtures import add_config_to_repo
+from testing.fixtures import make_config_from_repo
+from testing.fixtures import make_repo
+from testing.fixtures import modify_manifest
+from testing.fixtures import read_config
+from testing.fixtures import sample_local_config
+from testing.fixtures import write_config
+from testing.util import git_commit
+
+
+@pytest.fixture
+def up_to_date(tempdir_factory):
+    yield make_repo(tempdir_factory, 'python_hooks_repo')
+
+
+@pytest.fixture
+def out_of_date(tempdir_factory):
+    path = make_repo(tempdir_factory, 'python_hooks_repo')
+    original_rev = git.head_rev(path)
+
+    git_commit(cwd=path)
+    head_rev = git.head_rev(path)
+
+    yield auto_namedtuple(
+        path=path, original_rev=original_rev, head_rev=head_rev,
+    )
+
+
+@pytest.fixture
+def tagged(out_of_date):
+    cmd_output('git', 'tag', 'v1.2.3', cwd=out_of_date.path)
+    yield out_of_date
+
+
+@pytest.fixture
+def hook_disappearing(tempdir_factory):
+    path = make_repo(tempdir_factory, 'python_hooks_repo')
+    original_rev = git.head_rev(path)
+
+    with modify_manifest(path) as manifest:
+        manifest[0]['id'] = 'bar'
+
+    yield auto_namedtuple(path=path, original_rev=original_rev)
+
+
+def test_rev_info_from_config():
+    info = RevInfo.from_config({'repo': 'repo/path', 'rev': 'v1.2.3'})
+    assert info == RevInfo('repo/path', 'v1.2.3', None)
+
+
+def test_rev_info_update_up_to_date_repo(up_to_date):
+    config = make_config_from_repo(up_to_date)
+    info = RevInfo.from_config(config)
+    new_info = info.update(tags_only=False, freeze=False)
+    assert info == new_info
+
+
+def test_rev_info_update_out_of_date_repo(out_of_date):
+    config = make_config_from_repo(
+        out_of_date.path, rev=out_of_date.original_rev,
+    )
+    info = RevInfo.from_config(config)
+    new_info = info.update(tags_only=False, freeze=False)
+    assert new_info.rev == out_of_date.head_rev
+
+
+def test_rev_info_update_non_master_default_branch(out_of_date):
+    # change the default branch to be not-master
+    cmd_output('git', '-C', out_of_date.path, 'branch', '-m', 'dev')
+    test_rev_info_update_out_of_date_repo(out_of_date)
+
+
+def test_rev_info_update_tags_even_if_not_tags_only(tagged):
+    config = make_config_from_repo(tagged.path, rev=tagged.original_rev)
+    info = RevInfo.from_config(config)
+    new_info = info.update(tags_only=False, freeze=False)
+    assert new_info.rev == 'v1.2.3'
+
+
+def test_rev_info_update_tags_only_does_not_pick_tip(tagged):
+    git_commit(cwd=tagged.path)
+    config = make_config_from_repo(tagged.path, rev=tagged.original_rev)
+    info = RevInfo.from_config(config)
+    new_info = info.update(tags_only=True, freeze=False)
+    assert new_info.rev == 'v1.2.3'
+
+
+def test_rev_info_update_freeze_tag(tagged):
+    git_commit(cwd=tagged.path)
+    config = make_config_from_repo(tagged.path, rev=tagged.original_rev)
+    info = RevInfo.from_config(config)
+    new_info = info.update(tags_only=True, freeze=True)
+    assert new_info.rev == tagged.head_rev
+    assert new_info.frozen == 'v1.2.3'
+
+
+def test_rev_info_update_does_not_freeze_if_already_sha(out_of_date):
+    config = make_config_from_repo(
+        out_of_date.path, rev=out_of_date.original_rev,
+    )
+    info = RevInfo.from_config(config)
+    new_info = info.update(tags_only=True, freeze=True)
+    assert new_info.rev == out_of_date.head_rev
+    assert new_info.frozen is None
+
+
+def test_autoupdate_up_to_date_repo(up_to_date, tmpdir, store):
+    contents = (
+        f'repos:\n'
+        f'-   repo: {up_to_date}\n'
+        f'    rev: {git.head_rev(up_to_date)}\n'
+        f'    hooks:\n'
+        f'    -   id: foo\n'
+    )
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(contents)
+
+    assert autoupdate(str(cfg), store, freeze=False, tags_only=False) == 0
+    assert cfg.read() == contents
+
+
+def test_autoupdate_old_revision_broken(tempdir_factory, in_tmpdir, store):
+    """In $FUTURE_VERSION, hooks.yaml will no longer be supported.  This
+    asserts that when that day comes, pre-commit will be able to autoupdate
+    despite not being able to read hooks.yaml in that repository.
+    """
+    path = make_repo(tempdir_factory, 'python_hooks_repo')
+    config = make_config_from_repo(path, check=False)
+
+    cmd_output('git', 'mv', C.MANIFEST_FILE, 'nope.yaml', cwd=path)
+    git_commit(cwd=path)
+    # Assume this is the revision the user's old repository was at
+    rev = git.head_rev(path)
+    cmd_output('git', 'mv', 'nope.yaml', C.MANIFEST_FILE, cwd=path)
+    git_commit(cwd=path)
+    update_rev = git.head_rev(path)
+
+    config['rev'] = rev
+    write_config('.', config)
+    with open(C.CONFIG_FILE) as f:
+        before = f.read()
+    assert autoupdate(C.CONFIG_FILE, store, freeze=False, tags_only=False) == 0
+    with open(C.CONFIG_FILE) as f:
+        after = f.read()
+    assert before != after
+    assert update_rev in after
+
+
+def test_autoupdate_out_of_date_repo(out_of_date, tmpdir, store):
+    fmt = (
+        'repos:\n'
+        '-   repo: {}\n'
+        '    rev: {}\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+    )
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(fmt.format(out_of_date.path, out_of_date.original_rev))
+
+    assert autoupdate(str(cfg), store, freeze=False, tags_only=False) == 0
+    assert cfg.read() == fmt.format(out_of_date.path, out_of_date.head_rev)
+
+
+def test_autoupdate_only_one_to_update(up_to_date, out_of_date, tmpdir, store):
+    fmt = (
+        'repos:\n'
+        '-   repo: {}\n'
+        '    rev: {}\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '-   repo: {}\n'
+        '    rev: {}\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+    )
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    before = fmt.format(
+        up_to_date, git.head_rev(up_to_date),
+        out_of_date.path, out_of_date.original_rev,
+    )
+    cfg.write(before)
+
+    assert autoupdate(str(cfg), store, freeze=False, tags_only=False) == 0
+    assert cfg.read() == fmt.format(
+        up_to_date, git.head_rev(up_to_date),
+        out_of_date.path, out_of_date.head_rev,
+    )
+
+
+def test_autoupdate_out_of_date_repo_with_correct_repo_name(
+        out_of_date, in_tmpdir, store,
+):
+    stale_config = make_config_from_repo(
+        out_of_date.path, rev=out_of_date.original_rev, check=False,
+    )
+    local_config = sample_local_config()
+    config = {'repos': [stale_config, local_config]}
+    write_config('.', config)
+
+    with open(C.CONFIG_FILE) as f:
+        before = f.read()
+    repo_name = f'file://{out_of_date.path}'
+    ret = autoupdate(
+        C.CONFIG_FILE, store, freeze=False, tags_only=False,
+        repos=(repo_name,),
+    )
+    with open(C.CONFIG_FILE) as f:
+        after = f.read()
+    assert ret == 0
+    assert before != after
+    assert out_of_date.head_rev in after
+    assert 'local' in after
+
+
+def test_autoupdate_out_of_date_repo_with_wrong_repo_name(
+        out_of_date, in_tmpdir, store,
+):
+    config = make_config_from_repo(
+        out_of_date.path, rev=out_of_date.original_rev, check=False,
+    )
+    write_config('.', config)
+
+    with open(C.CONFIG_FILE) as f:
+        before = f.read()
+    # It will not update it, because the name doesn't match
+    ret = autoupdate(
+        C.CONFIG_FILE, store, freeze=False, tags_only=False,
+        repos=('dne',),
+    )
+    with open(C.CONFIG_FILE) as f:
+        after = f.read()
+    assert ret == 0
+    assert before == after
+
+
+def test_does_not_reformat(tmpdir, out_of_date, store):
+    fmt = (
+        'repos:\n'
+        '-   repo: {}\n'
+        '    rev: {}  # definitely the version I want!\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '        # These args are because reasons!\n'
+        '        args: [foo, bar, baz]\n'
+    )
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(fmt.format(out_of_date.path, out_of_date.original_rev))
+
+    assert autoupdate(str(cfg), store, freeze=False, tags_only=False) == 0
+    expected = fmt.format(out_of_date.path, out_of_date.head_rev)
+    assert cfg.read() == expected
+
+
+def test_loses_formatting_when_not_detectable(out_of_date, store, tmpdir):
+    """A best-effort attempt is made at updating rev without rewriting
+    formatting.  When the original formatting cannot be detected, this
+    is abandoned.
+    """
+    config = (
+        'repos: [\n'
+        '    {{\n'
+        '        repo: {}, rev: {},\n'
+        '        hooks: [\n'
+        '            # A comment!\n'
+        '            {{id: foo}},\n'
+        '        ],\n'
+        '    }}\n'
+        ']\n'.format(
+            shlex.quote(out_of_date.path), out_of_date.original_rev,
+        )
+    )
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(config)
+
+    assert autoupdate(str(cfg), store, freeze=False, tags_only=False) == 0
+    expected = (
+        f'repos:\n'
+        f'-   repo: {out_of_date.path}\n'
+        f'    rev: {out_of_date.head_rev}\n'
+        f'    hooks:\n'
+        f'    -   id: foo\n'
+    )
+    assert cfg.read() == expected
+
+
+def test_autoupdate_tagged_repo(tagged, in_tmpdir, store):
+    config = make_config_from_repo(tagged.path, rev=tagged.original_rev)
+    write_config('.', config)
+
+    assert autoupdate(C.CONFIG_FILE, store, freeze=False, tags_only=False) == 0
+    with open(C.CONFIG_FILE) as f:
+        assert 'v1.2.3' in f.read()
+
+
+def test_autoupdate_freeze(tagged, in_tmpdir, store):
+    config = make_config_from_repo(tagged.path, rev=tagged.original_rev)
+    write_config('.', config)
+
+    assert autoupdate(C.CONFIG_FILE, store, freeze=True, tags_only=False) == 0
+    with open(C.CONFIG_FILE) as f:
+        expected = f'rev: {tagged.head_rev}  # frozen: v1.2.3'
+        assert expected in f.read()
+
+    # if we un-freeze it should remove the frozen comment
+    assert autoupdate(C.CONFIG_FILE, store, freeze=False, tags_only=False) == 0
+    with open(C.CONFIG_FILE) as f:
+        assert 'rev: v1.2.3\n' in f.read()
+
+
+def test_autoupdate_tags_only(tagged, in_tmpdir, store):
+    # add some commits after the tag
+    git_commit(cwd=tagged.path)
+
+    config = make_config_from_repo(tagged.path, rev=tagged.original_rev)
+    write_config('.', config)
+
+    assert autoupdate(C.CONFIG_FILE, store, freeze=False, tags_only=True) == 0
+    with open(C.CONFIG_FILE) as f:
+        assert 'v1.2.3' in f.read()
+
+
+def test_autoupdate_latest_no_config(out_of_date, in_tmpdir, store):
+    config = make_config_from_repo(
+        out_of_date.path, rev=out_of_date.original_rev,
+    )
+    write_config('.', config)
+
+    cmd_output('git', 'rm', '-r', ':/', cwd=out_of_date.path)
+    git_commit(cwd=out_of_date.path)
+
+    assert autoupdate(C.CONFIG_FILE, store, freeze=False, tags_only=False) == 1
+    with open(C.CONFIG_FILE) as f:
+        assert out_of_date.original_rev in f.read()
+
+
+def test_hook_disppearing_repo_raises(hook_disappearing, store):
+    config = make_config_from_repo(
+        hook_disappearing.path,
+        rev=hook_disappearing.original_rev,
+        hooks=[{'id': 'foo'}],
+    )
+    info = RevInfo.from_config(config).update(tags_only=False, freeze=False)
+    with pytest.raises(RepositoryCannotBeUpdatedError):
+        _check_hooks_still_exist_at_rev(config, info, store)
+
+
+def test_autoupdate_hook_disappearing_repo(hook_disappearing, tmpdir, store):
+    contents = (
+        f'repos:\n'
+        f'-   repo: {hook_disappearing.path}\n'
+        f'    rev: {hook_disappearing.original_rev}\n'
+        f'    hooks:\n'
+        f'    -   id: foo\n'
+    )
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(contents)
+
+    assert autoupdate(str(cfg), store, freeze=False, tags_only=False) == 1
+    assert cfg.read() == contents
+
+
+def test_autoupdate_local_hooks(in_git_dir, store):
+    config = sample_local_config()
+    add_config_to_repo('.', config)
+    assert autoupdate(C.CONFIG_FILE, store, freeze=False, tags_only=False) == 0
+    new_config_writen = read_config('.')
+    assert len(new_config_writen['repos']) == 1
+    assert new_config_writen['repos'][0] == config
+
+
+def test_autoupdate_local_hooks_with_out_of_date_repo(
+        out_of_date, in_tmpdir, store,
+):
+    stale_config = make_config_from_repo(
+        out_of_date.path, rev=out_of_date.original_rev, check=False,
+    )
+    local_config = sample_local_config()
+    config = {'repos': [local_config, stale_config]}
+    write_config('.', config)
+    assert autoupdate(C.CONFIG_FILE, store, freeze=False, tags_only=False) == 0
+    new_config_writen = read_config('.')
+    assert len(new_config_writen['repos']) == 2
+    assert new_config_writen['repos'][0] == local_config
+
+
+def test_autoupdate_meta_hooks(tmpdir, store):
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(
+        'repos:\n'
+        '-   repo: meta\n'
+        '    hooks:\n'
+        '    -   id: check-useless-excludes\n',
+    )
+    assert autoupdate(str(cfg), store, freeze=False, tags_only=True) == 0
+    assert cfg.read() == (
+        'repos:\n'
+        '-   repo: meta\n'
+        '    hooks:\n'
+        '    -   id: check-useless-excludes\n'
+    )
+
+
+def test_updates_old_format_to_new_format(tmpdir, capsys, store):
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(
+        '-   repo: local\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '        name: foo\n'
+        '        entry: ./bin/foo.sh\n'
+        '        language: script\n',
+    )
+    assert autoupdate(str(cfg), store, freeze=False, tags_only=True) == 0
+    contents = cfg.read()
+    assert contents == (
+        'repos:\n'
+        '-   repo: local\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '        name: foo\n'
+        '        entry: ./bin/foo.sh\n'
+        '        language: script\n'
+    )
+    out, _ = capsys.readouterr()
+    assert out == 'Configuration has been migrated.\n'

tests/commands/clean_test.py

@@ -0,0 +1,33 @@
+import os.path
+from unittest import mock
+
+import pytest
+
+from pre_commit.commands.clean import clean
+
+
+@pytest.fixture(autouse=True)
+def fake_old_dir(tempdir_factory):
+    fake_old_dir = tempdir_factory.get()
+
+    def _expanduser(path, *args, **kwargs):
+        assert path == '~/.pre-commit'
+        return fake_old_dir
+
+    with mock.patch.object(os.path, 'expanduser', side_effect=_expanduser):
+        yield fake_old_dir
+
+
+def test_clean(store, fake_old_dir):
+    assert os.path.exists(fake_old_dir)
+    assert os.path.exists(store.directory)
+    clean(store)
+    assert not os.path.exists(fake_old_dir)
+    assert not os.path.exists(store.directory)
+
+
+def test_clean_idempotent(store):
+    clean(store)
+    assert not os.path.exists(store.directory)
+    clean(store)
+    assert not os.path.exists(store.directory)

tests/commands/gc_test.py

@@ -0,0 +1,161 @@
+import os
+
+import pre_commit.constants as C
+from pre_commit import git
+from pre_commit.clientlib import load_config
+from pre_commit.commands.autoupdate import autoupdate
+from pre_commit.commands.gc import gc
+from pre_commit.commands.install_uninstall import install_hooks
+from pre_commit.repository import all_hooks
+from testing.fixtures import make_config_from_repo
+from testing.fixtures import make_repo
+from testing.fixtures import modify_config
+from testing.fixtures import sample_local_config
+from testing.fixtures import sample_meta_config
+from testing.fixtures import write_config
+from testing.util import git_commit
+
+
+def _repo_count(store):
+    return len(store.select_all_repos())
+
+
+def _config_count(store):
+    return len(store.select_all_configs())
+
+
+def _remove_config_assert_cleared(store, cap_out):
+    os.remove(C.CONFIG_FILE)
+    assert not gc(store)
+    assert _config_count(store) == 0
+    assert _repo_count(store) == 0
+    assert cap_out.get().splitlines()[-1] == '1 repo(s) removed.'
+
+
+def test_gc(tempdir_factory, store, in_git_dir, cap_out):
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    old_rev = git.head_rev(path)
+    git_commit(cwd=path)
+
+    write_config('.', make_config_from_repo(path, rev=old_rev))
+    store.mark_config_used(C.CONFIG_FILE)
+
+    # update will clone both the old and new repo, making the old one gc-able
+    install_hooks(C.CONFIG_FILE, store)
+    assert not autoupdate(C.CONFIG_FILE, store, freeze=False, tags_only=False)
+
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 2
+    assert not gc(store)
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 1
+    assert cap_out.get().splitlines()[-1] == '1 repo(s) removed.'
+
+    _remove_config_assert_cleared(store, cap_out)
+
+
+def test_gc_repo_not_cloned(tempdir_factory, store, in_git_dir, cap_out):
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    write_config('.', make_config_from_repo(path))
+    store.mark_config_used(C.CONFIG_FILE)
+
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 0
+    assert not gc(store)
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 0
+    assert cap_out.get().splitlines()[-1] == '0 repo(s) removed.'
+
+
+def test_gc_meta_repo_does_not_crash(store, in_git_dir, cap_out):
+    write_config('.', sample_meta_config())
+    store.mark_config_used(C.CONFIG_FILE)
+    assert not gc(store)
+    assert cap_out.get().splitlines()[-1] == '0 repo(s) removed.'
+
+
+def test_gc_local_repo_does_not_crash(store, in_git_dir, cap_out):
+    write_config('.', sample_local_config())
+    store.mark_config_used(C.CONFIG_FILE)
+    assert not gc(store)
+    assert cap_out.get().splitlines()[-1] == '0 repo(s) removed.'
+
+
+def test_gc_unused_local_repo_with_env(store, in_git_dir, cap_out):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'flake8', 'name': 'flake8', 'entry': 'flake8',
+            # a `language: python` local hook will create an environment
+            'types': ['python'], 'language': 'python',
+        }],
+    }
+    write_config('.', config)
+    store.mark_config_used(C.CONFIG_FILE)
+
+    # this causes the repositories to be created
+    all_hooks(load_config(C.CONFIG_FILE), store)
+
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 1
+    assert not gc(store)
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 1
+    assert cap_out.get().splitlines()[-1] == '0 repo(s) removed.'
+
+    _remove_config_assert_cleared(store, cap_out)
+
+
+def test_gc_config_with_missing_hook(
+        tempdir_factory, store, in_git_dir, cap_out,
+):
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    write_config('.', make_config_from_repo(path))
+    store.mark_config_used(C.CONFIG_FILE)
+    # to trigger a clone
+    all_hooks(load_config(C.CONFIG_FILE), store)
+
+    with modify_config() as config:
+        # add a hook which does not exist, make sure we don't crash
+        config['repos'][0]['hooks'].append({'id': 'does-not-exist'})
+
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 1
+    assert not gc(store)
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 1
+    assert cap_out.get().splitlines()[-1] == '0 repo(s) removed.'
+
+    _remove_config_assert_cleared(store, cap_out)
+
+
+def test_gc_deletes_invalid_configs(store, in_git_dir, cap_out):
+    config = {'i am': 'invalid'}
+    write_config('.', config)
+    store.mark_config_used(C.CONFIG_FILE)
+
+    assert _config_count(store) == 1
+    assert not gc(store)
+    assert _config_count(store) == 0
+    assert cap_out.get().splitlines()[-1] == '0 repo(s) removed.'
+
+
+def test_invalid_manifest_gcd(tempdir_factory, store, in_git_dir, cap_out):
+    # clean up repos from old pre-commit versions
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    write_config('.', make_config_from_repo(path))
+    store.mark_config_used(C.CONFIG_FILE)
+
+    # trigger a clone
+    install_hooks(C.CONFIG_FILE, store)
+
+    # we'll "break" the manifest to simulate an old version clone
+    (_, _, path), = store.select_all_repos()
+    os.remove(os.path.join(path, C.MANIFEST_FILE))
+
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 1
+    assert not gc(store)
+    assert _config_count(store) == 1
+    assert _repo_count(store) == 0
+    assert cap_out.get().splitlines()[-1] == '1 repo(s) removed.'

tests/commands/hook_impl_test.py

@@ -0,0 +1,235 @@
+import subprocess
+import sys
+from unittest import mock
+
+import pytest
+
+import pre_commit.constants as C
+from pre_commit import git
+from pre_commit.commands import hook_impl
+from pre_commit.envcontext import envcontext
+from pre_commit.util import cmd_output
+from pre_commit.util import make_executable
+from testing.fixtures import git_dir
+from testing.fixtures import sample_local_config
+from testing.fixtures import write_config
+from testing.util import cwd
+from testing.util import git_commit
+
+
+def test_validate_config_file_exists(tmpdir):
+    cfg = tmpdir.join(C.CONFIG_FILE).ensure()
+    hook_impl._validate_config(0, cfg, True)
+
+
+def test_validate_config_missing(capsys):
+    with pytest.raises(SystemExit) as excinfo:
+        hook_impl._validate_config(123, 'DNE.yaml', False)
+    ret, = excinfo.value.args
+    assert ret == 1
+    assert capsys.readouterr().out == (
+        'No DNE.yaml file was found\n'
+        '- To temporarily silence this, run '
+        '`PRE_COMMIT_ALLOW_NO_CONFIG=1 git ...`\n'
+        '- To permanently silence this, install pre-commit with the '
+        '--allow-missing-config option\n'
+        '- To uninstall pre-commit run `pre-commit uninstall`\n'
+    )
+
+
+def test_validate_config_skip_missing_config(capsys):
+    with pytest.raises(SystemExit) as excinfo:
+        hook_impl._validate_config(123, 'DNE.yaml', True)
+    ret, = excinfo.value.args
+    assert ret == 123
+    expected = '`DNE.yaml` config file not found. Skipping `pre-commit`.\n'
+    assert capsys.readouterr().out == expected
+
+
+def test_validate_config_skip_via_env_variable(capsys):
+    with pytest.raises(SystemExit) as excinfo:
+        with envcontext((('PRE_COMMIT_ALLOW_NO_CONFIG', '1'),)):
+            hook_impl._validate_config(0, 'DNE.yaml', False)
+    ret, = excinfo.value.args
+    assert ret == 0
+    expected = '`DNE.yaml` config file not found. Skipping `pre-commit`.\n'
+    assert capsys.readouterr().out == expected
+
+
+def test_run_legacy_does_not_exist(tmpdir):
+    retv, stdin = hook_impl._run_legacy('pre-commit', tmpdir, ())
+    assert (retv, stdin) == (0, b'')
+
+
+def test_run_legacy_executes_legacy_script(tmpdir, capfd):
+    hook = tmpdir.join('pre-commit.legacy')
+    hook.write('#!/usr/bin/env bash\necho hi "$@"\nexit 1\n')
+    make_executable(hook)
+    retv, stdin = hook_impl._run_legacy('pre-commit', tmpdir, ('arg1', 'arg2'))
+    assert capfd.readouterr().out.strip() == 'hi arg1 arg2'
+    assert (retv, stdin) == (1, b'')
+
+
+def test_run_legacy_pre_push_returns_stdin(tmpdir):
+    with mock.patch.object(sys.stdin.buffer, 'read', return_value=b'stdin'):
+        retv, stdin = hook_impl._run_legacy('pre-push', tmpdir, ())
+    assert (retv, stdin) == (0, b'stdin')
+
+
+def test_run_legacy_recursive(tmpdir):
+    hook = tmpdir.join('pre-commit.legacy').ensure()
+    make_executable(hook)
+
+    # simulate a call being recursive
+    def call(*_, **__):
+        return hook_impl._run_legacy('pre-commit', tmpdir, ())
+
+    with mock.patch.object(subprocess, 'run', call):
+        with pytest.raises(SystemExit):
+            call()
+
+
+def test_run_ns_pre_commit():
+    ns = hook_impl._run_ns('pre-commit', True, (), b'')
+    assert ns is not None
+    assert ns.hook_stage == 'commit'
+    assert ns.color is True
+
+
+def test_run_ns_commit_msg():
+    ns = hook_impl._run_ns('commit-msg', False, ('.git/COMMIT_MSG',), b'')
+    assert ns is not None
+    assert ns.hook_stage == 'commit-msg'
+    assert ns.color is False
+    assert ns.commit_msg_filename == '.git/COMMIT_MSG'
+
+
+def test_run_ns_post_checkout():
+    ns = hook_impl._run_ns('post-checkout', True, ('a', 'b', 'c'), b'')
+    assert ns is not None
+    assert ns.hook_stage == 'post-checkout'
+    assert ns.color is True
+    assert ns.from_ref == 'a'
+    assert ns.to_ref == 'b'
+    assert ns.checkout_type == 'c'
+
+
+@pytest.fixture
+def push_example(tempdir_factory):
+    src = git_dir(tempdir_factory)
+    git_commit(cwd=src)
+    src_head = git.head_rev(src)
+
+    clone = tempdir_factory.get()
+    cmd_output('git', 'clone', src, clone)
+    git_commit(cwd=clone)
+    clone_head = git.head_rev(clone)
+    return (src, src_head, clone, clone_head)
+
+
+def test_run_ns_pre_push_updating_branch(push_example):
+    src, src_head, clone, clone_head = push_example
+
+    with cwd(clone):
+        args = ('origin', src)
+        stdin = f'HEAD {clone_head} refs/heads/b {src_head}\n'.encode()
+        ns = hook_impl._run_ns('pre-push', False, args, stdin)
+
+    assert ns is not None
+    assert ns.hook_stage == 'push'
+    assert ns.color is False
+    assert ns.remote_name == 'origin'
+    assert ns.remote_url == src
+    assert ns.from_ref == src_head
+    assert ns.to_ref == clone_head
+    assert ns.all_files is False
+
+
+def test_run_ns_pre_push_new_branch(push_example):
+    src, src_head, clone, clone_head = push_example
+
+    with cwd(clone):
+        args = ('origin', src)
+        stdin = f'HEAD {clone_head} refs/heads/b {hook_impl.Z40}\n'.encode()
+        ns = hook_impl._run_ns('pre-push', False, args, stdin)
+
+    assert ns is not None
+    assert ns.from_ref == src_head
+    assert ns.to_ref == clone_head
+
+
+def test_run_ns_pre_push_new_branch_existing_rev(push_example):
+    src, src_head, clone, _ = push_example
+
+    with cwd(clone):
+        args = ('origin', src)
+        stdin = f'HEAD {src_head} refs/heads/b2 {hook_impl.Z40}\n'.encode()
+        ns = hook_impl._run_ns('pre-push', False, args, stdin)
+
+    assert ns is None
+
+
+def test_pushing_orphan_branch(push_example):
+    src, src_head, clone, _ = push_example
+
+    cmd_output('git', 'checkout', '--orphan', 'b2', cwd=clone)
+    git_commit(cwd=clone, msg='something else to get unique hash')
+    clone_rev = git.head_rev(clone)
+
+    with cwd(clone):
+        args = ('origin', src)
+        stdin = f'HEAD {clone_rev} refs/heads/b2 {hook_impl.Z40}\n'.encode()
+        ns = hook_impl._run_ns('pre-push', False, args, stdin)
+
+    assert ns is not None
+    assert ns.all_files is True
+
+
+def test_run_ns_pre_push_deleting_branch(push_example):
+    src, src_head, clone, _ = push_example
+
+    with cwd(clone):
+        args = ('origin', src)
+        stdin = f'(delete) {hook_impl.Z40} refs/heads/b {src_head}'.encode()
+        ns = hook_impl._run_ns('pre-push', False, args, stdin)
+
+    assert ns is None
+
+
+def test_hook_impl_main_noop_pre_push(cap_out, store, push_example):
+    src, src_head, clone, _ = push_example
+
+    stdin = f'(delete) {hook_impl.Z40} refs/heads/b {src_head}'.encode()
+    with mock.patch.object(sys.stdin.buffer, 'read', return_value=stdin):
+        with cwd(clone):
+            write_config('.', sample_local_config())
+            ret = hook_impl.hook_impl(
+                store,
+                config=C.CONFIG_FILE,
+                color=False,
+                hook_type='pre-push',
+                hook_dir='.git/hooks',
+                skip_on_missing_config=False,
+                args=('origin', src),
+            )
+    assert ret == 0
+    assert cap_out.get() == ''
+
+
+def test_hook_impl_main_runs_hooks(cap_out, tempdir_factory, store):
+    with cwd(git_dir(tempdir_factory)):
+        write_config('.', sample_local_config())
+        ret = hook_impl.hook_impl(
+            store,
+            config=C.CONFIG_FILE,
+            color=False,
+            hook_type='pre-commit',
+            hook_dir='.git/hooks',
+            skip_on_missing_config=False,
+            args=(),
+        )
+    assert ret == 0
+    expected = '''\
+Block if "DO NOT COMMIT" is found....................(no files to check)Skipped
+'''
+    assert cap_out.get() == expected

tests/commands/init_templatedir_test.py

@@ -0,0 +1,92 @@
+import os.path
+from unittest import mock
+
+import pre_commit.constants as C
+from pre_commit.commands.init_templatedir import init_templatedir
+from pre_commit.envcontext import envcontext
+from pre_commit.util import cmd_output
+from testing.fixtures import git_dir
+from testing.fixtures import make_consuming_repo
+from testing.util import cmd_output_mocked_pre_commit_home
+from testing.util import cwd
+from testing.util import git_commit
+
+
+def test_init_templatedir(tmpdir, tempdir_factory, store, cap_out):
+    target = str(tmpdir.join('tmpl'))
+    init_templatedir(C.CONFIG_FILE, store, target, hook_types=['pre-commit'])
+    lines = cap_out.get().splitlines()
+    assert lines[0].startswith('pre-commit installed at ')
+    assert lines[1] == (
+        '[WARNING] `init.templateDir` not set to the target directory'
+    )
+    assert lines[2].startswith(
+        '[WARNING] maybe `git config --global init.templateDir',
+    )
+
+    with envcontext((('GIT_TEMPLATE_DIR', target),)):
+        path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+
+        with cwd(path):
+            retcode, output = git_commit(
+                fn=cmd_output_mocked_pre_commit_home,
+                tempdir_factory=tempdir_factory,
+            )
+            assert retcode == 0
+            assert 'Bash hook....' in output
+
+
+def test_init_templatedir_already_set(tmpdir, tempdir_factory, store, cap_out):
+    target = str(tmpdir.join('tmpl'))
+    tmp_git_dir = git_dir(tempdir_factory)
+    with cwd(tmp_git_dir):
+        cmd_output('git', 'config', 'init.templateDir', target)
+        init_templatedir(
+            C.CONFIG_FILE, store, target, hook_types=['pre-commit'],
+        )
+
+    lines = cap_out.get().splitlines()
+    assert len(lines) == 1
+    assert lines[0].startswith('pre-commit installed at')
+
+
+def test_init_templatedir_not_set(tmpdir, store, cap_out):
+    # set HOME to ignore the current `.gitconfig`
+    with envcontext((('HOME', str(tmpdir)),)):
+        with tmpdir.join('tmpl').ensure_dir().as_cwd():
+            # we have not set init.templateDir so this should produce a warning
+            init_templatedir(
+                C.CONFIG_FILE, store, '.', hook_types=['pre-commit'],
+            )
+
+    lines = cap_out.get().splitlines()
+    assert len(lines) == 3
+    assert lines[1] == (
+        '[WARNING] `init.templateDir` not set to the target directory'
+    )
+
+
+def test_init_templatedir_expanduser(tmpdir, tempdir_factory, store, cap_out):
+    target = str(tmpdir.join('tmpl'))
+    tmp_git_dir = git_dir(tempdir_factory)
+    with cwd(tmp_git_dir):
+        cmd_output('git', 'config', 'init.templateDir', '~/templatedir')
+        with mock.patch.object(os.path, 'expanduser', return_value=target):
+            init_templatedir(
+                C.CONFIG_FILE, store, target, hook_types=['pre-commit'],
+            )
+
+    lines = cap_out.get().splitlines()
+    assert len(lines) == 1
+    assert lines[0].startswith('pre-commit installed at')
+
+
+def test_init_templatedir_hookspath_set(tmpdir, tempdir_factory, store):
+    target = tmpdir.join('tmpl')
+    tmp_git_dir = git_dir(tempdir_factory)
+    with cwd(tmp_git_dir):
+        cmd_output('git', 'config', '--local', 'core.hooksPath', 'hooks')
+        init_templatedir(
+            C.CONFIG_FILE, store, target, hook_types=['pre-commit'],
+        )
+    assert target.join('hooks/pre-commit').exists()

tests/commands/install_uninstall_test.py

@@ -0,0 +1,901 @@
+import os.path
+import re
+import sys
+from unittest import mock
+
+import pre_commit.constants as C
+from pre_commit import git
+from pre_commit.commands import install_uninstall
+from pre_commit.commands.install_uninstall import CURRENT_HASH
+from pre_commit.commands.install_uninstall import install
+from pre_commit.commands.install_uninstall import install_hooks
+from pre_commit.commands.install_uninstall import is_our_script
+from pre_commit.commands.install_uninstall import PRIOR_HASHES
+from pre_commit.commands.install_uninstall import shebang
+from pre_commit.commands.install_uninstall import uninstall
+from pre_commit.parse_shebang import find_executable
+from pre_commit.util import cmd_output
+from pre_commit.util import make_executable
+from pre_commit.util import resource_text
+from testing.fixtures import add_config_to_repo
+from testing.fixtures import git_dir
+from testing.fixtures import make_consuming_repo
+from testing.fixtures import remove_config_from_repo
+from testing.fixtures import write_config
+from testing.util import cmd_output_mocked_pre_commit_home
+from testing.util import cwd
+from testing.util import git_commit
+
+
+def test_is_not_script():
+    assert is_our_script('setup.py') is False
+
+
+def test_is_script():
+    assert is_our_script('pre_commit/resources/hook-tmpl')
+
+
+def test_is_previous_pre_commit(tmpdir):
+    f = tmpdir.join('foo')
+    f.write(f'{PRIOR_HASHES[0]}\n')
+    assert is_our_script(f.strpath)
+
+
+def patch_platform(platform):
+    return mock.patch.object(sys, 'platform', platform)
+
+
+def patch_lookup_path(path):
+    return mock.patch.object(install_uninstall, 'POSIX_SEARCH_PATH', path)
+
+
+def patch_sys_exe(exe):
+    return mock.patch.object(install_uninstall, 'SYS_EXE', exe)
+
+
+def test_shebang_windows():
+    with patch_platform('win32'), patch_sys_exe('python.exe'):
+        assert shebang() == '#!/usr/bin/env python.exe'
+
+
+def test_shebang_posix_not_on_path():
+    with patch_platform('posix'), patch_lookup_path(()):
+        with patch_sys_exe('python3.6'):
+            assert shebang() == '#!/usr/bin/env python3.6'
+
+
+def test_shebang_posix_on_path(tmpdir):
+    exe = tmpdir.join(f'python{sys.version_info[0]}').ensure()
+    make_executable(exe)
+
+    with patch_platform('posix'), patch_lookup_path((tmpdir.strpath,)):
+        with patch_sys_exe('python'):
+            assert shebang() == f'#!/usr/bin/env python{sys.version_info[0]}'
+
+
+def test_install_pre_commit(in_git_dir, store):
+    assert not install(C.CONFIG_FILE, store, hook_types=['pre-commit'])
+    assert os.access(in_git_dir.join('.git/hooks/pre-commit').strpath, os.X_OK)
+
+    assert not install(C.CONFIG_FILE, store, hook_types=['pre-push'])
+    assert os.access(in_git_dir.join('.git/hooks/pre-push').strpath, os.X_OK)
+
+
+def test_install_hooks_directory_not_present(in_git_dir, store):
+    # Simulate some git clients which don't make .git/hooks #234
+    if in_git_dir.join('.git/hooks').exists():  # pragma: no cover (odd git)
+        in_git_dir.join('.git/hooks').remove()
+    install(C.CONFIG_FILE, store, hook_types=['pre-commit'])
+    assert in_git_dir.join('.git/hooks/pre-commit').exists()
+
+
+def test_install_multiple_hooks_at_once(in_git_dir, store):
+    install(C.CONFIG_FILE, store, hook_types=['pre-commit', 'pre-push'])
+    assert in_git_dir.join('.git/hooks/pre-commit').exists()
+    assert in_git_dir.join('.git/hooks/pre-push').exists()
+    uninstall(hook_types=['pre-commit', 'pre-push'])
+    assert not in_git_dir.join('.git/hooks/pre-commit').exists()
+    assert not in_git_dir.join('.git/hooks/pre-push').exists()
+
+
+def test_install_refuses_core_hookspath(in_git_dir, store):
+    cmd_output('git', 'config', '--local', 'core.hooksPath', 'hooks')
+    assert install(C.CONFIG_FILE, store, hook_types=['pre-commit'])
+
+
+def test_install_hooks_dead_symlink(in_git_dir, store):
+    hook = in_git_dir.join('.git/hooks').ensure_dir().join('pre-commit')
+    os.symlink('/fake/baz', hook.strpath)
+    install(C.CONFIG_FILE, store, hook_types=['pre-commit'])
+    assert hook.exists()
+
+
+def test_uninstall_does_not_blow_up_when_not_there(in_git_dir):
+    assert uninstall(hook_types=['pre-commit']) == 0
+
+
+def test_uninstall(in_git_dir, store):
+    assert not in_git_dir.join('.git/hooks/pre-commit').exists()
+    install(C.CONFIG_FILE, store, hook_types=['pre-commit'])
+    assert in_git_dir.join('.git/hooks/pre-commit').exists()
+    uninstall(hook_types=['pre-commit'])
+    assert not in_git_dir.join('.git/hooks/pre-commit').exists()
+
+
+def _get_commit_output(tempdir_factory, touch_file='foo', **kwargs):
+    open(touch_file, 'a').close()
+    cmd_output('git', 'add', touch_file)
+    return git_commit(
+        fn=cmd_output_mocked_pre_commit_home,
+        retcode=None,
+        tempdir_factory=tempdir_factory,
+        **kwargs,
+    )
+
+
+# osx does this different :(
+FILES_CHANGED = (
+    r'('
+    r' 1 file changed, 0 insertions\(\+\), 0 deletions\(-\)\n'
+    r'|'
+    r' 0 files changed\n'
+    r')'
+)
+
+
+NORMAL_PRE_COMMIT_RUN = re.compile(
+    fr'^\[INFO\] Initializing environment for .+\.\n'
+    fr'Bash hook\.+Passed\n'
+    fr'\[master [a-f0-9]{{7}}\] commit!\n'
+    fr'{FILES_CHANGED}'
+    fr' create mode 100644 foo\n$',
+)
+
+
+def test_install_pre_commit_and_run(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def test_install_pre_commit_and_run_custom_path(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        cmd_output('git', 'mv', C.CONFIG_FILE, 'custom.yaml')
+        git_commit(cwd=path)
+        assert install('custom.yaml', store, hook_types=['pre-commit']) == 0
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def test_install_in_submodule_and_run(tempdir_factory, store):
+    src_path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    parent_path = git_dir(tempdir_factory)
+    cmd_output('git', 'submodule', 'add', src_path, 'sub', cwd=parent_path)
+    git_commit(cwd=parent_path)
+
+    sub_pth = os.path.join(parent_path, 'sub')
+    with cwd(sub_pth):
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def test_install_in_worktree_and_run(tempdir_factory, store):
+    src_path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    path = tempdir_factory.get()
+    cmd_output('git', '-C', src_path, 'branch', '-m', 'notmaster')
+    cmd_output('git', '-C', src_path, 'worktree', 'add', path, '-b', 'master')
+
+    with cwd(path):
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def test_commit_am(tempdir_factory, store):
+    """Regression test for #322."""
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        # Make an unstaged change
+        open('unstaged', 'w').close()
+        cmd_output('git', 'add', '.')
+        git_commit(cwd=path)
+        with open('unstaged', 'w') as foo_file:
+            foo_file.write('Oh hai')
+
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+
+
+def test_unicode_merge_commit_message(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+        cmd_output('git', 'checkout', 'master', '-b', 'foo')
+        git_commit('-n', cwd=path)
+        cmd_output('git', 'checkout', 'master')
+        cmd_output('git', 'merge', 'foo', '--no-ff', '--no-commit', '-m', '☃')
+        # Used to crash
+        git_commit(
+            '--no-edit',
+            msg=None,
+            fn=cmd_output_mocked_pre_commit_home,
+            tempdir_factory=tempdir_factory,
+        )
+
+
+def test_install_idempotent(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def _path_without_us():
+    # Choose a path which *probably* doesn't include us
+    env = dict(os.environ)
+    exe = find_executable('pre-commit', _environ=env)
+    while exe:
+        parts = env['PATH'].split(os.pathsep)
+        after = [x for x in parts if x.lower() != os.path.dirname(exe).lower()]
+        if parts == after:
+            raise AssertionError(exe, parts)
+        env['PATH'] = os.pathsep.join(after)
+        exe = find_executable('pre-commit', _environ=env)
+    return env['PATH']
+
+
+def test_environment_not_sourced(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        assert not install(C.CONFIG_FILE, store, hook_types=['pre-commit'])
+        # simulate deleting the virtualenv by rewriting the exe
+        hook = os.path.join(path, '.git/hooks/pre-commit')
+        with open(hook) as f:
+            src = f.read()
+        src = re.sub(
+            '\nINSTALL_PYTHON =.*\n',
+            '\nINSTALL_PYTHON = "/dne"\n',
+            src,
+        )
+        with open(hook, 'w') as f:
+            f.write(src)
+
+        # Use a specific homedir to ignore --user installs
+        homedir = tempdir_factory.get()
+        ret, out = git_commit(
+            env={
+                'HOME': homedir,
+                'PATH': _path_without_us(),
+                # Git needs this to make a commit
+                'GIT_AUTHOR_NAME': os.environ['GIT_AUTHOR_NAME'],
+                'GIT_COMMITTER_NAME': os.environ['GIT_COMMITTER_NAME'],
+                'GIT_AUTHOR_EMAIL': os.environ['GIT_AUTHOR_EMAIL'],
+                'GIT_COMMITTER_EMAIL': os.environ['GIT_COMMITTER_EMAIL'],
+            },
+            retcode=None,
+        )
+        assert ret == 1
+        assert out == (
+            '`pre-commit` not found.  '
+            'Did you forget to activate your virtualenv?\n'
+        )
+
+
+FAILING_PRE_COMMIT_RUN = re.compile(
+    r'^\[INFO\] Initializing environment for .+\.\n'
+    r'Failing hook\.+Failed\n'
+    r'- hook id: failing_hook\n'
+    r'- exit code: 1\n'
+    r'\n'
+    r'Fail\n'
+    r'foo\n'
+    r'\n$',
+)
+
+
+def test_failing_hooks_returns_nonzero(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'failing_hook_repo')
+    with cwd(path):
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 1
+        assert FAILING_PRE_COMMIT_RUN.match(output)
+
+
+EXISTING_COMMIT_RUN = re.compile(
+    fr'^legacy hook\n'
+    fr'\[master [a-f0-9]{{7}}\] commit!\n'
+    fr'{FILES_CHANGED}'
+    fr' create mode 100644 baz\n$',
+)
+
+
+def _write_legacy_hook(path):
+    os.makedirs(os.path.join(path, '.git/hooks'), exist_ok=True)
+    with open(os.path.join(path, '.git/hooks/pre-commit'), 'w') as f:
+        f.write(f'{shebang()}\nprint("legacy hook")\n')
+    make_executable(f.name)
+
+
+def test_install_existing_hooks_no_overwrite(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        _write_legacy_hook(path)
+
+        # Make sure we installed the "old" hook correctly
+        ret, output = _get_commit_output(tempdir_factory, touch_file='baz')
+        assert ret == 0
+        assert EXISTING_COMMIT_RUN.match(output)
+
+        # Now install pre-commit (no-overwrite)
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+        # We should run both the legacy and pre-commit hooks
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert output.startswith('legacy hook\n')
+        assert NORMAL_PRE_COMMIT_RUN.match(output[len('legacy hook\n'):])
+
+
+def test_legacy_overwriting_legacy_hook(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        _write_legacy_hook(path)
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+        _write_legacy_hook(path)
+        # this previously crashed on windows.  See #1010
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+
+def test_install_existing_hook_no_overwrite_idempotent(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        _write_legacy_hook(path)
+
+        # Install twice
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+        # We should run both the legacy and pre-commit hooks
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert output.startswith('legacy hook\n')
+        assert NORMAL_PRE_COMMIT_RUN.match(output[len('legacy hook\n'):])
+
+
+FAIL_OLD_HOOK = re.compile(
+    r'fail!\n'
+    r'\[INFO\] Initializing environment for .+\.\n'
+    r'Bash hook\.+Passed\n',
+)
+
+
+def test_failing_existing_hook_returns_1(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        # Write out a failing "old" hook
+        os.makedirs(os.path.join(path, '.git/hooks'), exist_ok=True)
+        with open(os.path.join(path, '.git/hooks/pre-commit'), 'w') as f:
+            f.write('#!/usr/bin/env bash\necho "fail!"\nexit 1\n')
+        make_executable(f.name)
+
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+        # We should get a failure from the legacy hook
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 1
+        assert FAIL_OLD_HOOK.match(output)
+
+
+def test_install_overwrite_no_existing_hooks(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        assert not install(
+            C.CONFIG_FILE, store, hook_types=['pre-commit'], overwrite=True,
+        )
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def test_install_overwrite(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        _write_legacy_hook(path)
+        assert not install(
+            C.CONFIG_FILE, store, hook_types=['pre-commit'], overwrite=True,
+        )
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def test_uninstall_restores_legacy_hooks(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        _write_legacy_hook(path)
+
+        # Now install and uninstall pre-commit
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+        assert uninstall(hook_types=['pre-commit']) == 0
+
+        # Make sure we installed the "old" hook correctly
+        ret, output = _get_commit_output(tempdir_factory, touch_file='baz')
+        assert ret == 0
+        assert EXISTING_COMMIT_RUN.match(output)
+
+
+def test_replace_old_commit_script(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        # Install a script that looks like our old script
+        pre_commit_contents = resource_text('hook-tmpl')
+        new_contents = pre_commit_contents.replace(
+            CURRENT_HASH, PRIOR_HASHES[-1],
+        )
+
+        os.makedirs(os.path.join(path, '.git/hooks'), exist_ok=True)
+        with open(os.path.join(path, '.git/hooks/pre-commit'), 'w') as f:
+            f.write(new_contents)
+        make_executable(f.name)
+
+        # Install normally
+        assert install(C.CONFIG_FILE, store, hook_types=['pre-commit']) == 0
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def test_uninstall_doesnt_remove_not_our_hooks(in_git_dir):
+    pre_commit = in_git_dir.join('.git/hooks').ensure_dir().join('pre-commit')
+    pre_commit.write('#!/usr/bin/env bash\necho 1\n')
+    make_executable(pre_commit.strpath)
+
+    assert uninstall(hook_types=['pre-commit']) == 0
+
+    assert pre_commit.exists()
+
+
+PRE_INSTALLED = re.compile(
+    fr'Bash hook\.+Passed\n'
+    fr'\[master [a-f0-9]{{7}}\] commit!\n'
+    fr'{FILES_CHANGED}'
+    fr' create mode 100644 foo\n$',
+)
+
+
+def test_installs_hooks_with_hooks_True(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        install(C.CONFIG_FILE, store, hook_types=['pre-commit'], hooks=True)
+        ret, output = _get_commit_output(
+            tempdir_factory, pre_commit_home=store.directory,
+        )
+
+        assert ret == 0
+        assert PRE_INSTALLED.match(output)
+
+
+def test_install_hooks_command(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        install(C.CONFIG_FILE, store, hook_types=['pre-commit'])
+        install_hooks(C.CONFIG_FILE, store)
+        ret, output = _get_commit_output(
+            tempdir_factory, pre_commit_home=store.directory,
+        )
+
+        assert ret == 0
+        assert PRE_INSTALLED.match(output)
+
+
+def test_installed_from_venv(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        install(C.CONFIG_FILE, store, hook_types=['pre-commit'])
+        # No environment so pre-commit is not on the path when running!
+        # Should still pick up the python from when we installed
+        ret, output = _get_commit_output(
+            tempdir_factory,
+            env={
+                'HOME': os.path.expanduser('~'),
+                'PATH': _path_without_us(),
+                'TERM': os.environ.get('TERM', ''),
+                # Windows needs this to import `random`
+                'SYSTEMROOT': os.environ.get('SYSTEMROOT', ''),
+                # Windows needs this to resolve executables
+                'PATHEXT': os.environ.get('PATHEXT', ''),
+                # Git needs this to make a commit
+                'GIT_AUTHOR_NAME': os.environ['GIT_AUTHOR_NAME'],
+                'GIT_COMMITTER_NAME': os.environ['GIT_COMMITTER_NAME'],
+                'GIT_AUTHOR_EMAIL': os.environ['GIT_AUTHOR_EMAIL'],
+                'GIT_COMMITTER_EMAIL': os.environ['GIT_COMMITTER_EMAIL'],
+            },
+        )
+        assert ret == 0
+        assert NORMAL_PRE_COMMIT_RUN.match(output)
+
+
+def _get_push_output(tempdir_factory, remote='origin', opts=()):
+    return cmd_output_mocked_pre_commit_home(
+        'git', 'push', remote, 'HEAD:new_branch', *opts,
+        tempdir_factory=tempdir_factory,
+        retcode=None,
+    )[:2]
+
+
+def test_pre_push_integration_failing(tempdir_factory, store):
+    upstream = make_consuming_repo(tempdir_factory, 'failing_hook_repo')
+    path = tempdir_factory.get()
+    cmd_output('git', 'clone', upstream, path)
+    with cwd(path):
+        install(C.CONFIG_FILE, store, hook_types=['pre-push'])
+        # commit succeeds because pre-commit is only installed for pre-push
+        assert _get_commit_output(tempdir_factory)[0] == 0
+        assert _get_commit_output(tempdir_factory, touch_file='zzz')[0] == 0
+
+        retc, output = _get_push_output(tempdir_factory)
+        assert retc == 1
+        assert 'Failing hook' in output
+        assert 'Failed' in output
+        assert 'foo zzz' in output  # both filenames should be printed
+        assert 'hook id: failing_hook' in output
+
+
+def test_pre_push_integration_accepted(tempdir_factory, store):
+    upstream = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    path = tempdir_factory.get()
+    cmd_output('git', 'clone', upstream, path)
+    with cwd(path):
+        install(C.CONFIG_FILE, store, hook_types=['pre-push'])
+        assert _get_commit_output(tempdir_factory)[0] == 0
+
+        retc, output = _get_push_output(tempdir_factory)
+        assert retc == 0
+        assert 'Bash hook' in output
+        assert 'Passed' in output
+
+
+def test_pre_push_force_push_without_fetch(tempdir_factory, store):
+    upstream = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    path1 = tempdir_factory.get()
+    path2 = tempdir_factory.get()
+    cmd_output('git', 'clone', upstream, path1)
+    cmd_output('git', 'clone', upstream, path2)
+    with cwd(path1):
+        assert _get_commit_output(tempdir_factory)[0] == 0
+        assert _get_push_output(tempdir_factory)[0] == 0
+
+    with cwd(path2):
+        install(C.CONFIG_FILE, store, hook_types=['pre-push'])
+        assert _get_commit_output(tempdir_factory, msg='force!')[0] == 0
+
+        retc, output = _get_push_output(tempdir_factory, opts=('--force',))
+        assert retc == 0
+        assert 'Bash hook' in output
+        assert 'Passed' in output
+
+
+def test_pre_push_new_upstream(tempdir_factory, store):
+    upstream = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    upstream2 = git_dir(tempdir_factory)
+    path = tempdir_factory.get()
+    cmd_output('git', 'clone', upstream, path)
+    with cwd(path):
+        install(C.CONFIG_FILE, store, hook_types=['pre-push'])
+        assert _get_commit_output(tempdir_factory)[0] == 0
+
+        cmd_output('git', 'remote', 'rename', 'origin', 'upstream')
+        cmd_output('git', 'remote', 'add', 'origin', upstream2)
+        retc, output = _get_push_output(tempdir_factory)
+        assert retc == 0
+        assert 'Bash hook' in output
+        assert 'Passed' in output
+
+
+def test_pre_push_environment_variables(tempdir_factory, store):
+    config = {
+        'repo': 'local',
+        'hooks': [
+            {
+                'id': 'print-remote-info',
+                'name': 'print remote info',
+                'entry': 'bash -c "echo remote: $PRE_COMMIT_REMOTE_NAME"',
+                'language': 'system',
+                'verbose': True,
+            },
+        ],
+    }
+
+    upstream = git_dir(tempdir_factory)
+    clone = tempdir_factory.get()
+    cmd_output('git', 'clone', upstream, clone)
+    add_config_to_repo(clone, config)
+    with cwd(clone):
+        install(C.CONFIG_FILE, store, hook_types=['pre-push'])
+
+        cmd_output('git', 'remote', 'rename', 'origin', 'origin2')
+        retc, output = _get_push_output(tempdir_factory, remote='origin2')
+        assert retc == 0
+        assert '\nremote: origin2\n' in output
+
+
+def test_pre_push_integration_empty_push(tempdir_factory, store):
+    upstream = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    path = tempdir_factory.get()
+    cmd_output('git', 'clone', upstream, path)
+    with cwd(path):
+        install(C.CONFIG_FILE, store, hook_types=['pre-push'])
+        _get_push_output(tempdir_factory)
+        retc, output = _get_push_output(tempdir_factory)
+        assert output == 'Everything up-to-date\n'
+        assert retc == 0
+
+
+def test_pre_push_legacy(tempdir_factory, store):
+    upstream = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    path = tempdir_factory.get()
+    cmd_output('git', 'clone', upstream, path)
+    with cwd(path):
+        os.makedirs(os.path.join(path, '.git/hooks'), exist_ok=True)
+        with open(os.path.join(path, '.git/hooks/pre-push'), 'w') as f:
+            f.write(
+                '#!/usr/bin/env bash\n'
+                'set -eu\n'
+                'read lr ls rr rs\n'
+                'test -n "$lr" -a -n "$ls" -a -n "$rr" -a -n "$rs"\n'
+                'echo legacy\n',
+            )
+        make_executable(f.name)
+
+        install(C.CONFIG_FILE, store, hook_types=['pre-push'])
+        assert _get_commit_output(tempdir_factory)[0] == 0
+
+        retc, output = _get_push_output(tempdir_factory)
+        assert retc == 0
+        first_line, _, third_line = output.splitlines()[:3]
+        assert first_line == 'legacy'
+        assert third_line.startswith('Bash hook')
+        assert third_line.endswith('Passed')
+
+
+def test_commit_msg_integration_failing(
+        commit_msg_repo, tempdir_factory, store,
+):
+    install(C.CONFIG_FILE, store, hook_types=['commit-msg'])
+    retc, out = _get_commit_output(tempdir_factory)
+    assert retc == 1
+    assert out == '''\
+Must have "Signed off by:"...............................................Failed
+- hook id: must-have-signoff
+- exit code: 1
+'''
+
+
+def test_commit_msg_integration_passing(
+        commit_msg_repo, tempdir_factory, store,
+):
+    install(C.CONFIG_FILE, store, hook_types=['commit-msg'])
+    msg = 'Hi\nSigned off by: me, lol'
+    retc, out = _get_commit_output(tempdir_factory, msg=msg)
+    assert retc == 0
+    first_line = out.splitlines()[0]
+    assert first_line.startswith('Must have "Signed off by:"...')
+    assert first_line.endswith('...Passed')
+
+
+def test_commit_msg_legacy(commit_msg_repo, tempdir_factory, store):
+    hook_path = os.path.join(commit_msg_repo, '.git/hooks/commit-msg')
+    os.makedirs(os.path.dirname(hook_path), exist_ok=True)
+    with open(hook_path, 'w') as hook_file:
+        hook_file.write(
+            '#!/usr/bin/env bash\n'
+            'set -eu\n'
+            'test -e "$1"\n'
+            'echo legacy\n',
+        )
+    make_executable(hook_path)
+
+    install(C.CONFIG_FILE, store, hook_types=['commit-msg'])
+
+    msg = 'Hi\nSigned off by: asottile'
+    retc, out = _get_commit_output(tempdir_factory, msg=msg)
+    assert retc == 0
+    first_line, second_line = out.splitlines()[:2]
+    assert first_line == 'legacy'
+    assert second_line.startswith('Must have "Signed off by:"...')
+
+
+def test_post_checkout_integration(tempdir_factory, store):
+    path = git_dir(tempdir_factory)
+    config = [
+        {
+            'repo': 'local',
+            'hooks': [{
+                'id': 'post-checkout',
+                'name': 'Post checkout',
+                'entry': 'bash -c "echo ${PRE_COMMIT_TO_REF}"',
+                'language': 'system',
+                'always_run': True,
+                'verbose': True,
+                'stages': ['post-checkout'],
+            }],
+        },
+        {'repo': 'meta', 'hooks': [{'id': 'identity'}]},
+    ]
+    write_config(path, config)
+    with cwd(path):
+        cmd_output('git', 'add', '.')
+        git_commit()
+
+        # add a file only on `feature`, it should not be passed to hooks
+        cmd_output('git', 'checkout', '-b', 'feature')
+        open('some_file', 'a').close()
+        cmd_output('git', 'add', '.')
+        git_commit()
+        cmd_output('git', 'checkout', 'master')
+
+        install(C.CONFIG_FILE, store, hook_types=['post-checkout'])
+        retc, _, stderr = cmd_output('git', 'checkout', 'feature')
+        assert stderr is not None
+        assert retc == 0
+        assert git.head_rev(path) in stderr
+        assert 'some_file' not in stderr
+
+
+def test_prepare_commit_msg_integration_failing(
+        failing_prepare_commit_msg_repo, tempdir_factory, store,
+):
+    install(C.CONFIG_FILE, store, hook_types=['prepare-commit-msg'])
+    retc, out = _get_commit_output(tempdir_factory)
+    assert retc == 1
+    assert out == '''\
+Add "Signed off by:".....................................................Failed
+- hook id: add-signoff
+- exit code: 1
+'''
+
+
+def test_prepare_commit_msg_integration_passing(
+        prepare_commit_msg_repo, tempdir_factory, store,
+):
+    install(C.CONFIG_FILE, store, hook_types=['prepare-commit-msg'])
+    retc, out = _get_commit_output(tempdir_factory, msg='Hi')
+    assert retc == 0
+    first_line = out.splitlines()[0]
+    assert first_line.startswith('Add "Signed off by:"...')
+    assert first_line.endswith('...Passed')
+    commit_msg_path = os.path.join(
+        prepare_commit_msg_repo, '.git/COMMIT_EDITMSG',
+    )
+    with open(commit_msg_path) as f:
+        assert 'Signed off by: ' in f.read()
+
+
+def test_prepare_commit_msg_legacy(
+    prepare_commit_msg_repo, tempdir_factory, store,
+):
+    hook_path = os.path.join(
+        prepare_commit_msg_repo, '.git/hooks/prepare-commit-msg',
+    )
+    os.makedirs(os.path.dirname(hook_path), exist_ok=True)
+    with open(hook_path, 'w') as hook_file:
+        hook_file.write(
+            '#!/usr/bin/env bash\n'
+            'set -eu\n'
+            'test -e "$1"\n'
+            'echo legacy\n',
+        )
+    make_executable(hook_path)
+
+    install(C.CONFIG_FILE, store, hook_types=['prepare-commit-msg'])
+
+    retc, out = _get_commit_output(tempdir_factory, msg='Hi')
+    assert retc == 0
+    first_line, second_line = out.splitlines()[:2]
+    assert first_line == 'legacy'
+    assert second_line.startswith('Add "Signed off by:"...')
+    commit_msg_path = os.path.join(
+        prepare_commit_msg_repo, '.git/COMMIT_EDITMSG',
+    )
+    with open(commit_msg_path) as f:
+        assert 'Signed off by: ' in f.read()
+
+
+def test_pre_merge_commit_integration(tempdir_factory, store):
+    expected = re.compile(
+        r'^\[INFO\] Initializing environment for .+\n'
+        r'Bash hook\.+Passed\n'
+        r"Merge made by the 'recursive' strategy.\n"
+        r' foo \| 0\n'
+        r' 1 file changed, 0 insertions\(\+\), 0 deletions\(-\)\n'
+        r' create mode 100644 foo\n$',
+    )
+
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        ret = install(C.CONFIG_FILE, store, hook_types=['pre-merge-commit'])
+        assert ret == 0
+
+        cmd_output('git', 'checkout', 'master', '-b', 'feature')
+        _get_commit_output(tempdir_factory)
+        cmd_output('git', 'checkout', 'master')
+        ret, output, _ = cmd_output_mocked_pre_commit_home(
+            'git', 'merge', '--no-ff', '--no-edit', 'feature',
+            tempdir_factory=tempdir_factory,
+        )
+        assert ret == 0
+        assert expected.match(output)
+
+
+def test_install_disallow_missing_config(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        remove_config_from_repo(path)
+        ret = install(
+            C.CONFIG_FILE, store, hook_types=['pre-commit'],
+            overwrite=True, skip_on_missing_config=False,
+        )
+        assert ret == 0
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 1
+
+
+def test_install_allow_missing_config(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        remove_config_from_repo(path)
+        ret = install(
+            C.CONFIG_FILE, store, hook_types=['pre-commit'],
+            overwrite=True, skip_on_missing_config=True,
+        )
+        assert ret == 0
+
+        ret, output = _get_commit_output(tempdir_factory)
+        assert ret == 0
+        expected = (
+            '`.pre-commit-config.yaml` config file not found. '
+            'Skipping `pre-commit`.'
+        )
+        assert expected in output
+
+
+def test_install_temporarily_allow_mising_config(tempdir_factory, store):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    with cwd(path):
+        remove_config_from_repo(path)
+        ret = install(
+            C.CONFIG_FILE, store, hook_types=['pre-commit'],
+            overwrite=True, skip_on_missing_config=False,
+        )
+        assert ret == 0
+
+        env = dict(os.environ, PRE_COMMIT_ALLOW_NO_CONFIG='1')
+        ret, output = _get_commit_output(tempdir_factory, env=env)
+        assert ret == 0
+        expected = (
+            '`.pre-commit-config.yaml` config file not found. '
+            'Skipping `pre-commit`.'
+        )
+        assert expected in output

tests/commands/migrate_config_test.py

@@ -0,0 +1,156 @@
+import pytest
+
+import pre_commit.constants as C
+from pre_commit.commands.migrate_config import _indent
+from pre_commit.commands.migrate_config import migrate_config
+
+
+@pytest.mark.parametrize(
+    ('s', 'expected'),
+    (
+        ('', ''),
+        ('a', '    a'),
+        ('foo\nbar', '    foo\n    bar'),
+        ('foo\n\nbar\n', '    foo\n\n    bar\n'),
+        ('\n\n\n', '\n\n\n'),
+    ),
+)
+def test_indent(s, expected):
+    assert _indent(s) == expected
+
+
+def test_migrate_config_normal_format(tmpdir, capsys):
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(
+        '-   repo: local\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '        name: foo\n'
+        '        entry: ./bin/foo.sh\n'
+        '        language: script\n',
+    )
+    with tmpdir.as_cwd():
+        assert not migrate_config(C.CONFIG_FILE)
+    out, _ = capsys.readouterr()
+    assert out == 'Configuration has been migrated.\n'
+    contents = cfg.read()
+    assert contents == (
+        'repos:\n'
+        '-   repo: local\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '        name: foo\n'
+        '        entry: ./bin/foo.sh\n'
+        '        language: script\n'
+    )
+
+
+def test_migrate_config_document_marker(tmpdir):
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(
+        '# comment\n'
+        '\n'
+        '---\n'
+        '-   repo: local\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '        name: foo\n'
+        '        entry: ./bin/foo.sh\n'
+        '        language: script\n',
+    )
+    with tmpdir.as_cwd():
+        assert not migrate_config(C.CONFIG_FILE)
+    contents = cfg.read()
+    assert contents == (
+        '# comment\n'
+        '\n'
+        '---\n'
+        'repos:\n'
+        '-   repo: local\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '        name: foo\n'
+        '        entry: ./bin/foo.sh\n'
+        '        language: script\n'
+    )
+
+
+def test_migrate_config_list_literal(tmpdir):
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(
+        '[{\n'
+        '    repo: local,\n'
+        '    hooks: [{\n'
+        '        id: foo, name: foo, entry: ./bin/foo.sh,\n'
+        '        language: script,\n'
+        '    }]\n'
+        '}]',
+    )
+    with tmpdir.as_cwd():
+        assert not migrate_config(C.CONFIG_FILE)
+    contents = cfg.read()
+    assert contents == (
+        'repos:\n'
+        '    [{\n'
+        '        repo: local,\n'
+        '        hooks: [{\n'
+        '            id: foo, name: foo, entry: ./bin/foo.sh,\n'
+        '            language: script,\n'
+        '        }]\n'
+        '    }]'
+    )
+
+
+def test_already_migrated_configuration_noop(tmpdir, capsys):
+    contents = (
+        'repos:\n'
+        '-   repo: local\n'
+        '    hooks:\n'
+        '    -   id: foo\n'
+        '        name: foo\n'
+        '        entry: ./bin/foo.sh\n'
+        '        language: script\n'
+    )
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(contents)
+    with tmpdir.as_cwd():
+        assert not migrate_config(C.CONFIG_FILE)
+    out, _ = capsys.readouterr()
+    assert out == 'Configuration is already migrated.\n'
+    assert cfg.read() == contents
+
+
+def test_migrate_config_sha_to_rev(tmpdir):
+    contents = (
+        'repos:\n'
+        '-   repo: https://github.com/pre-commit/pre-commit-hooks\n'
+        '    sha: v1.2.0\n'
+        '    hooks: []\n'
+        '-   repo: https://github.com/pre-commit/pre-commit-hooks\n'
+        '    sha: v1.2.0\n'
+        '    hooks: []\n'
+    )
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(contents)
+    with tmpdir.as_cwd():
+        assert not migrate_config(C.CONFIG_FILE)
+    contents = cfg.read()
+    assert contents == (
+        'repos:\n'
+        '-   repo: https://github.com/pre-commit/pre-commit-hooks\n'
+        '    rev: v1.2.0\n'
+        '    hooks: []\n'
+        '-   repo: https://github.com/pre-commit/pre-commit-hooks\n'
+        '    rev: v1.2.0\n'
+        '    hooks: []\n'
+    )
+
+
+@pytest.mark.parametrize('contents', ('', '\n'))
+def test_empty_configuration_file_user_error(tmpdir, contents):
+    cfg = tmpdir.join(C.CONFIG_FILE)
+    cfg.write(contents)
+    with tmpdir.as_cwd():
+        assert not migrate_config(C.CONFIG_FILE)
+    # even though the config is invalid, this should be a noop
+    assert cfg.read() == contents

tests/commands/sample_config_test.py

@@ -0,0 +1,19 @@
+from pre_commit.commands.sample_config import sample_config
+
+
+def test_sample_config(capsys):
+    ret = sample_config()
+    assert ret == 0
+    out, _ = capsys.readouterr()
+    assert out == '''\
+# See https://pre-commit.com for more information
+# See https://pre-commit.com/hooks.html for more hooks
+repos:
+-   repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v2.4.0
+    hooks:
+    -   id: trailing-whitespace
+    -   id: end-of-file-fixer
+    -   id: check-yaml
+    -   id: check-added-large-files
+'''

tests/commands/try_repo_test.py

@@ -0,0 +1,151 @@
+import os.path
+import re
+import time
+from unittest import mock
+
+from pre_commit import git
+from pre_commit.commands.try_repo import try_repo
+from pre_commit.util import cmd_output
+from testing.auto_namedtuple import auto_namedtuple
+from testing.fixtures import git_dir
+from testing.fixtures import make_repo
+from testing.fixtures import modify_manifest
+from testing.util import cwd
+from testing.util import git_commit
+from testing.util import run_opts
+
+
+def try_repo_opts(repo, ref=None, **kwargs):
+    return auto_namedtuple(repo=repo, ref=ref, **run_opts(**kwargs)._asdict())
+
+
+def _get_out(cap_out):
+    out = re.sub(r'\[INFO\].+\n', '', cap_out.get())
+    start, using_config, config, rest = out.split(f'{"=" * 79}\n')
+    assert using_config == 'Using config:\n'
+    return start, config, rest
+
+
+def _add_test_file():
+    open('test-file', 'a').close()
+    cmd_output('git', 'add', '.')
+
+
+def _run_try_repo(tempdir_factory, **kwargs):
+    repo = make_repo(tempdir_factory, 'modified_file_returns_zero_repo')
+    with cwd(git_dir(tempdir_factory)):
+        _add_test_file()
+        assert not try_repo(try_repo_opts(repo, **kwargs))
+
+
+def test_try_repo_repo_only(cap_out, tempdir_factory):
+    with mock.patch.object(time, 'time', return_value=0.0):
+        _run_try_repo(tempdir_factory, verbose=True)
+    start, config, rest = _get_out(cap_out)
+    assert start == ''
+    assert re.match(
+        '^repos:\n'
+        '-   repo: .+\n'
+        '    rev: .+\n'
+        '    hooks:\n'
+        '    -   id: bash_hook\n'
+        '    -   id: bash_hook2\n'
+        '    -   id: bash_hook3\n$',
+        config,
+    )
+    assert rest == '''\
+Bash hook............................................(no files to check)Skipped
+- hook id: bash_hook
+Bash hook................................................................Passed
+- hook id: bash_hook2
+- duration: 0s
+
+test-file
+
+Bash hook............................................(no files to check)Skipped
+- hook id: bash_hook3
+'''
+
+
+def test_try_repo_with_specific_hook(cap_out, tempdir_factory):
+    _run_try_repo(tempdir_factory, hook='bash_hook', verbose=True)
+    start, config, rest = _get_out(cap_out)
+    assert start == ''
+    assert re.match(
+        '^repos:\n'
+        '-   repo: .+\n'
+        '    rev: .+\n'
+        '    hooks:\n'
+        '    -   id: bash_hook\n$',
+        config,
+    )
+    assert rest == '''\
+Bash hook............................................(no files to check)Skipped
+- hook id: bash_hook
+'''
+
+
+def test_try_repo_relative_path(cap_out, tempdir_factory):
+    repo = make_repo(tempdir_factory, 'modified_file_returns_zero_repo')
+    with cwd(git_dir(tempdir_factory)):
+        _add_test_file()
+        relative_repo = os.path.relpath(repo, '.')
+        # previously crashed on cloning a relative path
+        assert not try_repo(try_repo_opts(relative_repo, hook='bash_hook'))
+
+
+def test_try_repo_bare_repo(cap_out, tempdir_factory):
+    repo = make_repo(tempdir_factory, 'modified_file_returns_zero_repo')
+    with cwd(git_dir(tempdir_factory)):
+        _add_test_file()
+        bare_repo = os.path.join(repo, '.git')
+        # previously crashed attempting modification changes
+        assert not try_repo(try_repo_opts(bare_repo, hook='bash_hook'))
+
+
+def test_try_repo_specific_revision(cap_out, tempdir_factory):
+    repo = make_repo(tempdir_factory, 'script_hooks_repo')
+    ref = git.head_rev(repo)
+    git_commit(cwd=repo)
+    with cwd(git_dir(tempdir_factory)):
+        _add_test_file()
+        assert not try_repo(try_repo_opts(repo, ref=ref))
+
+    _, config, _ = _get_out(cap_out)
+    assert ref in config
+
+
+def test_try_repo_uncommitted_changes(cap_out, tempdir_factory):
+    repo = make_repo(tempdir_factory, 'script_hooks_repo')
+    # make an uncommitted change
+    with modify_manifest(repo, commit=False) as manifest:
+        manifest[0]['name'] = 'modified name!'
+
+    with cwd(git_dir(tempdir_factory)):
+        open('test-fie', 'a').close()
+        cmd_output('git', 'add', '.')
+        assert not try_repo(try_repo_opts(repo))
+
+    start, config, rest = _get_out(cap_out)
+    assert start == '[WARNING] Creating temporary repo with uncommitted changes...\n'  # noqa: E501
+    assert re.match(
+        '^repos:\n'
+        '-   repo: .+shadow-repo\n'
+        '    rev: .+\n'
+        '    hooks:\n'
+        '    -   id: bash_hook\n$',
+        config,
+    )
+    assert rest == 'modified name!...........................................................Passed\n'  # noqa: E501
+
+
+def test_try_repo_staged_changes(tempdir_factory):
+    repo = make_repo(tempdir_factory, 'modified_file_returns_zero_repo')
+
+    with cwd(repo):
+        open('staged-file', 'a').close()
+        open('second-staged-file', 'a').close()
+        cmd_output('git', 'add', '.')
+
+    with cwd(git_dir(tempdir_factory)):
+        assert not try_repo(try_repo_opts(repo, hook='bash_hook'))

tests/conftest.py

@@ -0,0 +1,277 @@
+import functools
+import io
+import logging
+import os.path
+from unittest import mock
+
+import pytest
+
+from pre_commit import output
+from pre_commit.envcontext import envcontext
+from pre_commit.logging_handler import logging_handler
+from pre_commit.store import Store
+from pre_commit.util import cmd_output
+from pre_commit.util import make_executable
+from testing.fixtures import git_dir
+from testing.fixtures import make_consuming_repo
+from testing.fixtures import write_config
+from testing.util import cwd
+from testing.util import git_commit
+
+
+@pytest.fixture(autouse=True)
+def no_warnings(recwarn):
+    yield
+    warnings = []
+    for warning in recwarn:  # pragma: no cover
+        message = str(warning.message)
+        # ImportWarning: Not importing directory '...' missing __init__(.py)
+        if not (
+            isinstance(warning.message, ImportWarning) and
+            message.startswith('Not importing directory ') and
+            ' missing __init__' in message
+        ):
+            warnings.append(
+                f'{warning.filename}:{warning.lineno} {message}',
+            )
+    assert not warnings
+
+
+@pytest.fixture
+def tempdir_factory(tmpdir):
+    class TmpdirFactory:
+        def __init__(self):
+            self.tmpdir_count = 0
+
+        def get(self):
+            path = tmpdir.join(str(self.tmpdir_count)).strpath
+            self.tmpdir_count += 1
+            os.mkdir(path)
+            return path
+
+    yield TmpdirFactory()
+
+
+@pytest.fixture
+def in_tmpdir(tempdir_factory):
+    path = tempdir_factory.get()
+    with cwd(path):
+        yield path
+
+
+@pytest.fixture
+def in_git_dir(tmpdir):
+    repo = tmpdir.join('repo').ensure_dir()
+    with repo.as_cwd():
+        cmd_output('git', 'init')
+        yield repo
+
+
+def _make_conflict():
+    cmd_output('git', 'checkout', 'origin/master', '-b', 'foo')
+    with open('conflict_file', 'w') as conflict_file:
+        conflict_file.write('herp\nderp\n')
+    cmd_output('git', 'add', 'conflict_file')
+    with open('foo_only_file', 'w') as foo_only_file:
+        foo_only_file.write('foo')
+    cmd_output('git', 'add', 'foo_only_file')
+    git_commit(msg=_make_conflict.__name__)
+    cmd_output('git', 'checkout', 'origin/master', '-b', 'bar')
+    with open('conflict_file', 'w') as conflict_file:
+        conflict_file.write('harp\nddrp\n')
+    cmd_output('git', 'add', 'conflict_file')
+    with open('bar_only_file', 'w') as bar_only_file:
+        bar_only_file.write('bar')
+    cmd_output('git', 'add', 'bar_only_file')
+    git_commit(msg=_make_conflict.__name__)
+    cmd_output('git', 'merge', 'foo', retcode=None)
+
+
+@pytest.fixture
+def in_merge_conflict(tempdir_factory):
+    path = make_consuming_repo(tempdir_factory, 'script_hooks_repo')
+    open(os.path.join(path, 'dummy'), 'a').close()
+    cmd_output('git', 'add', 'dummy', cwd=path)
+    git_commit(msg=in_merge_conflict.__name__, cwd=path)
+
+    conflict_path = tempdir_factory.get()
+    cmd_output('git', 'clone', path, conflict_path)
+    with cwd(conflict_path):
+        _make_conflict()
+        yield os.path.join(conflict_path)
+
+
+@pytest.fixture
+def in_conflicting_submodule(tempdir_factory):
+    git_dir_1 = git_dir(tempdir_factory)
+    git_dir_2 = git_dir(tempdir_factory)
+    git_commit(msg=in_conflicting_submodule.__name__, cwd=git_dir_2)
+    cmd_output('git', 'submodule', 'add', git_dir_2, 'sub', cwd=git_dir_1)
+    with cwd(os.path.join(git_dir_1, 'sub')):
+        _make_conflict()
+        yield
+
+
+@pytest.fixture
+def commit_msg_repo(tempdir_factory):
+    path = git_dir(tempdir_factory)
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'must-have-signoff',
+            'name': 'Must have "Signed off by:"',
+            'entry': 'grep -q "Signed off by:"',
+            'language': 'system',
+            'stages': ['commit-msg'],
+        }],
+    }
+    write_config(path, config)
+    with cwd(path):
+        cmd_output('git', 'add', '.')
+        git_commit(msg=commit_msg_repo.__name__)
+        yield path
+
+
+@pytest.fixture
+def prepare_commit_msg_repo(tempdir_factory):
+    path = git_dir(tempdir_factory)
+    script_name = 'add_sign_off.sh'
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'add-signoff',
+            'name': 'Add "Signed off by:"',
+            'entry': f'./{script_name}',
+            'language': 'script',
+            'stages': ['prepare-commit-msg'],
+        }],
+    }
+    write_config(path, config)
+    with cwd(path):
+        with open(script_name, 'w') as script_file:
+            script_file.write(
+                '#!/usr/bin/env bash\n'
+                'set -eu\n'
+                'echo "\nSigned off by: " >> "$1"\n',
+            )
+            make_executable(script_name)
+        cmd_output('git', 'add', '.')
+        git_commit(msg=prepare_commit_msg_repo.__name__)
+        yield path
+
+
+@pytest.fixture
+def failing_prepare_commit_msg_repo(tempdir_factory):
+    path = git_dir(tempdir_factory)
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'add-signoff',
+            'name': 'Add "Signed off by:"',
+            'entry': 'bash -c "exit 1"',
+            'language': 'system',
+            'stages': ['prepare-commit-msg'],
+        }],
+    }
+    write_config(path, config)
+    with cwd(path):
+        cmd_output('git', 'add', '.')
+        git_commit(msg=failing_prepare_commit_msg_repo.__name__)
+        yield path
+
+
+@pytest.fixture(autouse=True, scope='session')
+def dont_write_to_home_directory():
+    """pre_commit.store.Store will by default write to the home directory
+    We'll mock out `Store.get_default_directory` to raise invariantly so we
+    don't construct a `Store` object that writes to our home directory.
+    """
+    class YouForgotToExplicitlyChooseAStoreDirectory(AssertionError):
+        pass
+
+    with mock.patch.object(
+        Store,
+        'get_default_directory',
+        side_effect=YouForgotToExplicitlyChooseAStoreDirectory,
+    ):
+        yield
+
+
+@pytest.fixture(autouse=True, scope='session')
+def configure_logging():
+    with logging_handler(use_color=False):
+        yield
+
+
+@pytest.fixture
+def mock_store_dir(tempdir_factory):
+    tmpdir = tempdir_factory.get()
+    with mock.patch.object(
+        Store,
+        'get_default_directory',
+        return_value=tmpdir,
+    ):
+        yield tmpdir
+
+
+@pytest.fixture
+def store(tempdir_factory):
+    yield Store(os.path.join(tempdir_factory.get(), '.pre-commit'))
+
+
+@pytest.fixture
+def log_info_mock():
+    with mock.patch.object(logging.getLogger('pre_commit'), 'info') as mck:
+        yield mck
+
+
+class FakeStream:
+    def __init__(self):
+        self.data = io.BytesIO()
+
+    def write(self, s):
+        self.data.write(s)
+
+    def flush(self):
+        pass
+
+
+class Fixture:
+    def __init__(self, stream):
+        self._stream = stream
+
+    def get_bytes(self):
+        """Get the output as-if no encoding occurred"""
+        data = self._stream.data.getvalue()
+        self._stream.data.seek(0)
+        self._stream.data.truncate()
+        return data.replace(b'\r\n', b'\n')
+
+    def get(self):
+        """Get the output assuming it was written as UTF-8 bytes"""
+        return self.get_bytes().decode()
+
+
+@pytest.fixture
+def cap_out():
+    stream = FakeStream()
+    write = functools.partial(output.write, stream=stream)
+    write_line_b = functools.partial(output.write_line_b, stream=stream)
+    with mock.patch.multiple(output, write=write, write_line_b=write_line_b):
+        yield Fixture(stream)
+
+
+@pytest.fixture
+def fake_log_handler():
+    handler = mock.Mock(level=logging.INFO)
+    logger = logging.getLogger('pre_commit')
+    logger.addHandler(handler)
+    yield handler
+    logger.removeHandler(handler)
+
+
+@pytest.fixture(scope='session', autouse=True)
+def set_git_templatedir(tmpdir_factory):
+    tdir = str(tmpdir_factory.mktemp('git_template_dir'))
+    with envcontext((('GIT_TEMPLATE_DIR', tdir),)):
+        yield

tests/envcontext_test.py

@@ -0,0 +1,101 @@
+import os
+from unittest import mock
+
+import pytest
+
+from pre_commit.envcontext import envcontext
+from pre_commit.envcontext import UNSET
+from pre_commit.envcontext import Var
+
+
+def _test(*, before, patch, expected):
+    env = before.copy()
+    with envcontext(patch, _env=env):
+        assert env == expected
+    assert env == before
+
+
+def test_trivial():
+    _test(before={}, patch={}, expected={})
+
+
+def test_noop():
+    _test(before={'foo': 'bar'}, patch=(), expected={'foo': 'bar'})
+
+
+def test_adds():
+    _test(before={}, patch=[('foo', 'bar')], expected={'foo': 'bar'})
+
+
+def test_overrides():
+    _test(
+        before={'foo': 'baz'},
+        patch=[('foo', 'bar')],
+        expected={'foo': 'bar'},
+    )
+
+
+def test_unset_but_nothing_to_unset():
+    _test(before={}, patch=[('foo', UNSET)], expected={})
+
+
+def test_unset_things_to_remove():
+    _test(
+        before={'PYTHONHOME': ''},
+        patch=[('PYTHONHOME', UNSET)],
+        expected={},
+    )
+
+
+def test_templated_environment_variable_missing():
+    _test(
+        before={},
+        patch=[('PATH', ('~/bin:', Var('PATH')))],
+        expected={'PATH': '~/bin:'},
+    )
+
+
+def test_templated_environment_variable_defaults():
+    _test(
+        before={},
+        patch=[('PATH', ('~/bin:', Var('PATH', default='/bin')))],
+        expected={'PATH': '~/bin:/bin'},
+    )
+
+
+def test_templated_environment_variable_there():
+    _test(
+        before={'PATH': '/usr/local/bin:/usr/bin'},
+        patch=[('PATH', ('~/bin:', Var('PATH')))],
+        expected={'PATH': '~/bin:/usr/local/bin:/usr/bin'},
+    )
+
+
+def test_templated_environ_sources_from_previous():
+    _test(
+        before={'foo': 'bar'},
+        patch=(
+            ('foo', 'baz'),
+            ('herp', ('foo: ', Var('foo'))),
+        ),
+        expected={'foo': 'baz', 'herp': 'foo: bar'},
+    )
+
+
+def test_exception_safety():
+    class MyError(RuntimeError):
+        pass
+
+    env = {'hello': 'world'}
+    with pytest.raises(MyError):
+        with envcontext((('foo', 'bar'),), _env=env):
+            raise MyError()
+    assert env == {'hello': 'world'}
+
+
+def test_integration_os_environ():
+    with mock.patch.dict(os.environ, {'FOO': 'bar'}, clear=True):
+        assert os.environ == {'FOO': 'bar'}
+        with envcontext((('HERP', 'derp'),)):
+            assert os.environ == {'FOO': 'bar', 'HERP': 'derp'}
+        assert os.environ == {'FOO': 'bar'}

tests/error_handler_test.py

@@ -0,0 +1,170 @@
+import os.path
+import re
+import sys
+from unittest import mock
+
+import pytest
+
+from pre_commit import error_handler
+from pre_commit.util import CalledProcessError
+from testing.util import cmd_output_mocked_pre_commit_home
+
+
+@pytest.fixture
+def mocked_log_and_exit():
+    with mock.patch.object(error_handler, '_log_and_exit') as log_and_exit:
+        yield log_and_exit
+
+
+def test_error_handler_no_exception(mocked_log_and_exit):
+    with error_handler.error_handler():
+        pass
+    assert mocked_log_and_exit.call_count == 0
+
+
+def test_error_handler_fatal_error(mocked_log_and_exit):
+    exc = error_handler.FatalError('just a test')
+    with error_handler.error_handler():
+        raise exc
+
+    mocked_log_and_exit.assert_called_once_with(
+        'An error has occurred',
+        exc,
+        # Tested below
+        mock.ANY,
+    )
+
+    assert re.match(
+        r'Traceback \(most recent call last\):\n'
+        r'  File ".+pre_commit.error_handler.py", line \d+, in error_handler\n'
+        r'    yield\n'
+        r'  File ".+tests.error_handler_test.py", line \d+, '
+        r'in test_error_handler_fatal_error\n'
+        r'    raise exc\n'
+        r'(pre_commit\.error_handler\.)?FatalError: just a test\n',
+        mocked_log_and_exit.call_args[0][2],
+    )
+
+
+def test_error_handler_uncaught_error(mocked_log_and_exit):
+    exc = ValueError('another test')
+    with error_handler.error_handler():
+        raise exc
+
+    mocked_log_and_exit.assert_called_once_with(
+        'An unexpected error has occurred',
+        exc,
+        # Tested below
+        mock.ANY,
+    )
+    assert re.match(
+        r'Traceback \(most recent call last\):\n'
+        r'  File ".+pre_commit.error_handler.py", line \d+, in error_handler\n'
+        r'    yield\n'
+        r'  File ".+tests.error_handler_test.py", line \d+, '
+        r'in test_error_handler_uncaught_error\n'
+        r'    raise exc\n'
+        r'ValueError: another test\n',
+        mocked_log_and_exit.call_args[0][2],
+    )
+
+
+def test_error_handler_keyboardinterrupt(mocked_log_and_exit):
+    exc = KeyboardInterrupt()
+    with error_handler.error_handler():
+        raise exc
+
+    mocked_log_and_exit.assert_called_once_with(
+        'Interrupted (^C)',
+        exc,
+        # Tested below
+        mock.ANY,
+    )
+    assert re.match(
+        r'Traceback \(most recent call last\):\n'
+        r'  File ".+pre_commit.error_handler.py", line \d+, in error_handler\n'
+        r'    yield\n'
+        r'  File ".+tests.error_handler_test.py", line \d+, '
+        r'in test_error_handler_keyboardinterrupt\n'
+        r'    raise exc\n'
+        r'KeyboardInterrupt\n',
+        mocked_log_and_exit.call_args[0][2],
+    )
+
+
+def test_log_and_exit(cap_out, mock_store_dir):
+    with pytest.raises(SystemExit):
+        error_handler._log_and_exit(
+            'msg', error_handler.FatalError('hai'), "I'm a stacktrace",
+        )
+
+    printed = cap_out.get()
+    log_file = os.path.join(mock_store_dir, 'pre-commit.log')
+    assert printed == f'msg: FatalError: hai\nCheck the log at {log_file}\n'
+
+    assert os.path.exists(log_file)
+    with open(log_file) as f:
+        logged = f.read()
+        expected = (
+            r'^### version information\n'
+            r'\n'
+            r'```\n'
+            r'pre-commit version: \d+\.\d+\.\d+\n'
+            r'sys.version:\n'
+            r'(    .*\n)*'
+            r'sys.executable: .*\n'
+            r'os.name: .*\n'
+            r'sys.platform: .*\n'
+            r'```\n'
+            r'\n'
+            r'### error information\n'
+            r'\n'
+            r'```\n'
+            r'msg: FatalError: hai\n'
+            r'```\n'
+            r'\n'
+            r'```\n'
+            r"I'm a stacktrace\n"
+            r'```\n'
+        )
+        assert re.match(expected, logged)
+
+
+def test_error_handler_non_ascii_exception(mock_store_dir):
+    with pytest.raises(SystemExit):
+        with error_handler.error_handler():
+            raise ValueError('☃')
+
+
+def test_error_handler_non_utf8_exception(mock_store_dir):
+    with pytest.raises(SystemExit):
+        with error_handler.error_handler():
+            raise CalledProcessError(1, ('exe',), 0, b'error: \xa0\xe1', b'')
+
+
+def test_error_handler_non_stringable_exception(mock_store_dir):
+    class C(Exception):
+        def __str__(self):
+            raise RuntimeError('not today!')
+
+    with pytest.raises(SystemExit):
+        with error_handler.error_handler():
+            raise C()
+
+
+def test_error_handler_no_tty(tempdir_factory):
+    pre_commit_home = tempdir_factory.get()
+    ret, out, _ = cmd_output_mocked_pre_commit_home(
+        sys.executable,
+        '-c',
+        'from pre_commit.error_handler import error_handler\n'
+        'with error_handler():\n'
+        '    raise ValueError("\\u2603")\n',
+        retcode=1,
+        tempdir_factory=tempdir_factory,
+        pre_commit_home=pre_commit_home,
+    )
+    log_file = os.path.join(pre_commit_home, 'pre-commit.log')
+    out_lines = out.splitlines()
+    assert out_lines[-2] == 'An unexpected error has occurred: ValueError: ☃'
+    assert out_lines[-1] == f'Check the log at {log_file}'

tests/git_test.py

@@ -0,0 +1,188 @@
+import os.path
+
+import pytest
+
+from pre_commit import git
+from pre_commit.util import cmd_output
+from testing.util import git_commit
+
+
+def test_get_root_at_root(in_git_dir):
+    expected = os.path.normcase(in_git_dir.strpath)
+    assert os.path.normcase(git.get_root()) == expected
+
+
+def test_get_root_deeper(in_git_dir):
+    expected = os.path.normcase(in_git_dir.strpath)
+    with in_git_dir.join('foo').ensure_dir().as_cwd():
+        assert os.path.normcase(git.get_root()) == expected
+
+
+def test_get_staged_files_deleted(in_git_dir):
+    in_git_dir.join('test').ensure()
+    cmd_output('git', 'add', 'test')
+    git_commit()
+    cmd_output('git', 'rm', '--cached', 'test')
+    assert git.get_staged_files() == []
+
+
+def test_is_not_in_merge_conflict(in_git_dir):
+    assert git.is_in_merge_conflict() is False
+
+
+def test_is_in_merge_conflict(in_merge_conflict):
+    assert git.is_in_merge_conflict() is True
+
+
+def test_is_in_merge_conflict_submodule(in_conflicting_submodule):
+    assert git.is_in_merge_conflict() is True
+
+
+def test_cherry_pick_conflict(in_merge_conflict):
+    cmd_output('git', 'merge', '--abort')
+    foo_ref = cmd_output('git', 'rev-parse', 'foo')[1].strip()
+    cmd_output('git', 'cherry-pick', foo_ref, retcode=None)
+    assert git.is_in_merge_conflict() is False
+
+
+def resolve_conflict():
+    with open('conflict_file', 'w') as conflicted_file:
+        conflicted_file.write('herp\nderp\n')
+    cmd_output('git', 'add', 'conflict_file')
+
+
+def test_get_conflicted_files(in_merge_conflict):
+    resolve_conflict()
+    with open('other_file', 'w') as other_file:
+        other_file.write('oh hai')
+    cmd_output('git', 'add', 'other_file')
+
+    ret = set(git.get_conflicted_files())
+    assert ret == {'conflict_file', 'other_file'}
+
+
+def test_get_conflicted_files_in_submodule(in_conflicting_submodule):
+    resolve_conflict()
+    assert set(git.get_conflicted_files()) == {'conflict_file'}
+
+
+def test_get_conflicted_files_unstaged_files(in_merge_conflict):
+    """This case no longer occurs, but it is a useful test nonetheless"""
+    resolve_conflict()
+
+    # Make unstaged file.
+    with open('bar_only_file', 'w') as bar_only_file:
+        bar_only_file.write('new contents!\n')
+
+    ret = set(git.get_conflicted_files())
+    assert ret == {'conflict_file'}
+
+
+MERGE_MSG = b"Merge branch 'foo' into bar\n\nConflicts:\n\tconflict_file\n"
+OTHER_MERGE_MSG = MERGE_MSG + b'\tother_conflict_file\n'
+
+
+@pytest.mark.parametrize(
+    ('input', 'expected_output'),
+    (
+        (MERGE_MSG, ['conflict_file']),
+        (OTHER_MERGE_MSG, ['conflict_file', 'other_conflict_file']),
+    ),
+)
+def test_parse_merge_msg_for_conflicts(input, expected_output):
+    ret = git.parse_merge_msg_for_conflicts(input)
+    assert ret == expected_output
+
+
+def test_get_changed_files(in_git_dir):
+    git_commit()
+    in_git_dir.join('a.txt').ensure()
+    in_git_dir.join('b.txt').ensure()
+    cmd_output('git', 'add', '.')
+    git_commit()
+    files = git.get_changed_files('HEAD^', 'HEAD')
+    assert files == ['a.txt', 'b.txt']
+
+    # files changed in source but not in origin should not be returned
+    files = git.get_changed_files('HEAD', 'HEAD^')
+    assert files == []
+
+
+@pytest.mark.parametrize(
+    ('s', 'expected'),
+    (
+        ('foo\0bar\0', ['foo', 'bar']),
+        ('foo\0', ['foo']),
+        ('', []),
+        ('foo', ['foo']),
+    ),
+)
+def test_zsplit(s, expected):
+    assert git.zsplit(s) == expected
+
+
+@pytest.fixture
+def non_ascii_repo(in_git_dir):
+    git_commit()
+    in_git_dir.join('интервью').ensure()
+    cmd_output('git', 'add', '.')
+    git_commit()
+    yield in_git_dir
+
+
+def test_all_files_non_ascii(non_ascii_repo):
+    ret = git.get_all_files()
+    assert ret == ['интервью']
+
+
+def test_staged_files_non_ascii(non_ascii_repo):
+    non_ascii_repo.join('интервью').write('hi')
+    cmd_output('git', 'add', '.')
+    assert git.get_staged_files() == ['интервью']
+
+
+def test_changed_files_non_ascii(non_ascii_repo):
+    ret = git.get_changed_files('HEAD^', 'HEAD')
+    assert ret == ['интервью']
+
+
+def test_get_conflicted_files_non_ascii(in_merge_conflict):
+    open('интервью', 'a').close()
+    cmd_output('git', 'add', '.')
+    ret = git.get_conflicted_files()
+    assert ret == {'conflict_file', 'интервью'}
+
+
+def test_intent_to_add(in_git_dir):
+    in_git_dir.join('a').ensure()
+    cmd_output('git', 'add', '--intent-to-add', 'a')
+
+    assert git.intent_to_add_files() == ['a']
+
+
+def test_status_output_with_rename(in_git_dir):
+    in_git_dir.join('a').write('1\n2\n3\n4\n5\n6\n7\n8\n9\n10\n')
+    cmd_output('git', 'add', 'a')
+    git_commit()
+    cmd_output('git', 'mv', 'a', 'b')
+    in_git_dir.join('c').ensure()
+    cmd_output('git', 'add', '--intent-to-add', 'c')
+
+    assert git.intent_to_add_files() == ['c']
+
+
+def test_no_git_env():
+    env = {
+        'http_proxy': 'http://myproxy:80',
+        'GIT_EXEC_PATH': '/some/git/exec/path',
+        'GIT_SSH': '/usr/bin/ssh',
+        'GIT_SSH_COMMAND': 'ssh -o',
+        'GIT_DIR': '/none/shall/pass',
+    }
+    no_git_env = git.no_git_env(env)
+    assert no_git_env == {
+        'http_proxy': 'http://myproxy:80',
+        'GIT_EXEC_PATH': '/some/git/exec/path',
+        'GIT_SSH': '/usr/bin/ssh',
+        'GIT_SSH_COMMAND': 'ssh -o',
+    }

tests/languages/docker_test.py

@@ -0,0 +1,23 @@
+from unittest import mock
+
+from pre_commit.languages import docker
+from pre_commit.util import CalledProcessError
+
+
+def test_docker_is_running_process_error():
+    with mock.patch(
+        'pre_commit.languages.docker.cmd_output_b',
+        side_effect=CalledProcessError(1, (), 0, b'', None),
+    ):
+        assert docker.docker_is_running() is False
+
+
+def test_docker_fallback_user():
+    def invalid_attribute():
+        raise AttributeError
+    with mock.patch.multiple(
+        'os', create=True,
+        getuid=invalid_attribute,
+        getgid=invalid_attribute,
+    ):
+        assert docker.get_docker_user() == '1000:1000'

tests/languages/golang_test.py

@@ -0,0 +1,20 @@
+import pytest
+
+from pre_commit.languages.golang import guess_go_dir
+
+
+@pytest.mark.parametrize(
+    ('url', 'expected'),
+    (
+        ('/im/a/path/on/disk', 'unknown_src_dir'),
+        ('file:///im/a/path/on/disk', 'unknown_src_dir'),
+        ('git@github.com:golang/lint', 'github.com/golang/lint'),
+        ('git://github.com/golang/lint', 'github.com/golang/lint'),
+        ('http://github.com/golang/lint', 'github.com/golang/lint'),
+        ('https://github.com/golang/lint', 'github.com/golang/lint'),
+        ('ssh://git@github.com/golang/lint', 'github.com/golang/lint'),
+        ('git@github.com:golang/lint.git', 'github.com/golang/lint'),
+    ),
+)
+def test_guess_go_dir(url, expected):
+    assert guess_go_dir(url) == expected

tests/languages/helpers_test.py

@@ -0,0 +1,82 @@
+import multiprocessing
+import os
+import sys
+from unittest import mock
+
+import pytest
+
+import pre_commit.constants as C
+from pre_commit.languages import helpers
+from pre_commit.prefix import Prefix
+from pre_commit.util import CalledProcessError
+from testing.auto_namedtuple import auto_namedtuple
+
+
+def test_basic_get_default_version():
+    assert helpers.basic_get_default_version() == C.DEFAULT
+
+
+def test_basic_healthy():
+    assert helpers.basic_healthy(Prefix('.'), 'default') is True
+
+
+def test_failed_setup_command_does_not_unicode_error():
+    script = (
+        'import sys\n'
+        "getattr(sys.stderr, 'buffer', sys.stderr).write(b'\\x81\\xfe')\n"
+        'exit(1)\n'
+    )
+
+    # an assertion that this does not raise `UnicodeError`
+    with pytest.raises(CalledProcessError):
+        helpers.run_setup_cmd(Prefix('.'), (sys.executable, '-c', script))
+
+
+def test_assert_no_additional_deps():
+    with pytest.raises(AssertionError) as excinfo:
+        helpers.assert_no_additional_deps('lang', ['hmmm'])
+    msg, = excinfo.value.args
+    assert msg == (
+        'For now, pre-commit does not support additional_dependencies for lang'
+    )
+
+
+SERIAL_FALSE = auto_namedtuple(require_serial=False)
+SERIAL_TRUE = auto_namedtuple(require_serial=True)
+
+
+def test_target_concurrency_normal():
+    with mock.patch.object(multiprocessing, 'cpu_count', return_value=123):
+        with mock.patch.dict(os.environ, {}, clear=True):
+            assert helpers.target_concurrency(SERIAL_FALSE) == 123
+
+
+def test_target_concurrency_cpu_count_require_serial_true():
+    with mock.patch.dict(os.environ, {}, clear=True):
+        assert helpers.target_concurrency(SERIAL_TRUE) == 1
+
+
+def test_target_concurrency_testing_env_var():
+    with mock.patch.dict(
+            os.environ, {'PRE_COMMIT_NO_CONCURRENCY': '1'}, clear=True,
+    ):
+        assert helpers.target_concurrency(SERIAL_FALSE) == 1
+
+
+def test_target_concurrency_on_travis():
+    with mock.patch.dict(os.environ, {'TRAVIS': '1'}, clear=True):
+        assert helpers.target_concurrency(SERIAL_FALSE) == 2
+
+
+def test_target_concurrency_cpu_count_not_implemented():
+    with mock.patch.object(
+            multiprocessing, 'cpu_count', side_effect=NotImplementedError,
+    ):
+        with mock.patch.dict(os.environ, {}, clear=True):
+            assert helpers.target_concurrency(SERIAL_FALSE) == 1
+
+
+def test_shuffled_is_deterministic():
+    seq = [str(i) for i in range(10)]
+    expected = ['3', '7', '8', '2', '4', '6', '5', '1', '0', '9']
+    assert helpers._shuffled(seq) == expected

tests/languages/pygrep_test.py

@@ -0,0 +1,65 @@
+import pytest
+
+from pre_commit.languages import pygrep
+
+
+@pytest.fixture
+def some_files(tmpdir):
+    tmpdir.join('f1').write_binary(b'foo\nbar\n')
+    tmpdir.join('f2').write_binary(b'[INFO] hi\n')
+    tmpdir.join('f3').write_binary(b"with'quotes\n")
+    with tmpdir.as_cwd():
+        yield
+
+
+@pytest.mark.usefixtures('some_files')
+@pytest.mark.parametrize(
+    ('pattern', 'expected_retcode', 'expected_out'),
+    (
+        ('baz', 0, ''),
+        ('foo', 1, 'f1:1:foo\n'),
+        ('bar', 1, 'f1:2:bar\n'),
+        (r'(?i)\[info\]', 1, 'f2:1:[INFO] hi\n'),
+        ("h'q", 1, "f3:1:with'quotes\n"),
+    ),
+)
+def test_main(some_files, cap_out, pattern, expected_retcode, expected_out):
+    ret = pygrep.main((pattern, 'f1', 'f2', 'f3'))
+    out = cap_out.get()
+    assert ret == expected_retcode
+    assert out == expected_out
+
+
+def test_ignore_case(some_files, cap_out):
+    ret = pygrep.main(('--ignore-case', 'info', 'f1', 'f2', 'f3'))
+    out = cap_out.get()
+    assert ret == 1
+    assert out == 'f2:1:[INFO] hi\n'
+
+
+def test_multiline(some_files, cap_out):
+    ret = pygrep.main(('--multiline', r'foo\nbar', 'f1', 'f2', 'f3'))
+    out = cap_out.get()
+    assert ret == 1
+    assert out == 'f1:1:foo\nbar\n'
+
+
+def test_multiline_line_number(some_files, cap_out):
+    ret = pygrep.main(('--multiline', r'ar', 'f1', 'f2', 'f3'))
+    out = cap_out.get()
+    assert ret == 1
+    assert out == 'f1:2:bar\n'
+
+
+def test_multiline_dotall_flag_is_enabled(some_files, cap_out):
+    ret = pygrep.main(('--multiline', r'o.*bar', 'f1', 'f2', 'f3'))
+    out = cap_out.get()
+    assert ret == 1
+    assert out == 'f1:1:foo\nbar\n'
+
+
+def test_multiline_multiline_flag_is_enabled(some_files, cap_out):
+    ret = pygrep.main(('--multiline', r'foo$.*bar', 'f1', 'f2', 'f3'))
+    out = cap_out.get()
+    assert ret == 1
+    assert out == 'f1:1:foo\nbar\n'

tests/languages/python_test.py

@@ -0,0 +1,75 @@
+import os.path
+import sys
+from unittest import mock
+
+import pytest
+
+import pre_commit.constants as C
+from pre_commit.languages import python
+from pre_commit.prefix import Prefix
+
+
+def test_norm_version_expanduser():
+    home = os.path.expanduser('~')
+    if os.name == 'nt':  # pragma: nt cover
+        path = r'~\python343'
+        expected_path = fr'{home}\python343'
+    else:  # pragma: nt no cover
+        path = '~/.pyenv/versions/3.4.3/bin/python'
+        expected_path = f'{home}/.pyenv/versions/3.4.3/bin/python'
+    result = python.norm_version(path)
+    assert result == expected_path
+
+
+@pytest.mark.parametrize('v', ('python3.6', 'python3', 'python'))
+def test_sys_executable_matches(v):
+    with mock.patch.object(sys, 'version_info', (3, 6, 7)):
+        assert python._sys_executable_matches(v)
+
+
+@pytest.mark.parametrize('v', ('notpython', 'python3.x'))
+def test_sys_executable_matches_does_not_match(v):
+    with mock.patch.object(sys, 'version_info', (3, 6, 7)):
+        assert not python._sys_executable_matches(v)
+
+
+@pytest.mark.parametrize(
+    ('exe', 'realpath', 'expected'), (
+        ('/usr/bin/python3', '/usr/bin/python3.7', 'python3'),
+        ('/usr/bin/python', '/usr/bin/python3.7', 'python3.7'),
+        ('/usr/bin/python', '/usr/bin/python', None),
+        ('/usr/bin/python3.6m', '/usr/bin/python3.6m', 'python3.6m'),
+        ('v/bin/python', 'v/bin/pypy', 'pypy'),
+    ),
+)
+def test_find_by_sys_executable(exe, realpath, expected):
+    with mock.patch.object(sys, 'executable', exe):
+        with mock.patch.object(os.path, 'realpath', return_value=realpath):
+            with mock.patch.object(python, 'find_executable', lambda x: x):
+                assert python._find_by_sys_executable() == expected
+
+
+def test_healthy_types_py_in_cwd(tmpdir):
+    with tmpdir.as_cwd():
+        prefix = tmpdir.join('prefix').ensure_dir()
+        prefix.join('setup.py').write('import setuptools; setuptools.setup()')
+        prefix = Prefix(str(prefix))
+        python.install_environment(prefix, C.DEFAULT, ())
+
+        # even if a `types.py` file exists, should still be healthy
+        tmpdir.join('types.py').ensure()
+        assert python.healthy(prefix, C.DEFAULT) is True
+
+
+def test_healthy_python_goes_missing(tmpdir):
+    with tmpdir.as_cwd():
+        prefix = tmpdir.join('prefix').ensure_dir()
+        prefix.join('setup.py').write('import setuptools; setuptools.setup()')
+        prefix = Prefix(str(prefix))
+        python.install_environment(prefix, C.DEFAULT, ())
+
+        exe_name = 'python' if sys.platform != 'win32' else 'python.exe'
+        py_exe = prefix.path(python.bin_dir('py_env-default'), exe_name)
+        os.remove(py_exe)
+
+        assert python.healthy(prefix, C.DEFAULT) is False

tests/languages/ruby_test.py

@@ -0,0 +1,28 @@
+import os.path
+
+from pre_commit.languages import ruby
+from pre_commit.prefix import Prefix
+from pre_commit.util import cmd_output
+from testing.util import xfailif_windows_no_ruby
+
+
+@xfailif_windows_no_ruby
+def test_install_rbenv(tempdir_factory):
+    prefix = Prefix(tempdir_factory.get())
+    ruby._install_rbenv(prefix)
+    # Should have created rbenv directory
+    assert os.path.exists(prefix.path('rbenv-default'))
+
+    # Should be able to activate using our script and access rbenv
+    with ruby.in_env(prefix, 'default'):
+        cmd_output('rbenv', '--help')
+
+
+@xfailif_windows_no_ruby
+def test_install_rbenv_with_version(tempdir_factory):
+    prefix = Prefix(tempdir_factory.get())
+    ruby._install_rbenv(prefix, version='1.9.3p547')
+
+    # Should be able to activate and use rbenv install
+    with ruby.in_env(prefix, '1.9.3p547'):
+        cmd_output('rbenv', 'install', '--help')

tests/logging_handler_test.py

@@ -0,0 +1,21 @@
+import logging
+
+from pre_commit import color
+from pre_commit.logging_handler import LoggingHandler
+
+
+def _log_record(message, level):
+    return logging.LogRecord('name', level, '', 1, message, {}, None)
+
+
+def test_logging_handler_color(cap_out):
+    handler = LoggingHandler(True)
+    handler.emit(_log_record('hi', logging.WARNING))
+    ret = cap_out.get()
+    assert ret == f'{color.YELLOW}[WARNING]{color.NORMAL} hi\n'
+
+
+def test_logging_handler_no_color(cap_out):
+    handler = LoggingHandler(False)
+    handler.emit(_log_record('hi', logging.WARNING))
+    assert cap_out.get() == '[WARNING] hi\n'

tests/main_test.py

@@ -0,0 +1,189 @@
+import argparse
+import os.path
+from unittest import mock
+
+import pytest
+
+import pre_commit.constants as C
+from pre_commit import main
+from pre_commit.error_handler import FatalError
+from testing.auto_namedtuple import auto_namedtuple
+
+
+@pytest.mark.parametrize(
+    ('argv', 'expected'),
+    (
+        ((), ['f']),
+        (('--f', 'x'), ['x']),
+        (('--f', 'x', '--f', 'y'), ['x', 'y']),
+    ),
+)
+def test_append_replace_default(argv, expected):
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--f', action=main.AppendReplaceDefault, default=['f'])
+    assert parser.parse_args(argv).f == expected
+
+
+def _args(**kwargs):
+    kwargs.setdefault('command', 'help')
+    kwargs.setdefault('config', C.CONFIG_FILE)
+    return argparse.Namespace(**kwargs)
+
+
+def test_adjust_args_and_chdir_not_in_git_dir(in_tmpdir):
+    with pytest.raises(FatalError):
+        main._adjust_args_and_chdir(_args())
+
+
+def test_adjust_args_and_chdir_in_dot_git_dir(in_git_dir):
+    with in_git_dir.join('.git').as_cwd(), pytest.raises(FatalError):
+        main._adjust_args_and_chdir(_args())
+
+
+def test_adjust_args_and_chdir_noop(in_git_dir):
+    args = _args(command='run', files=['f1', 'f2'])
+    main._adjust_args_and_chdir(args)
+    assert os.getcwd() == in_git_dir
+    assert args.config == C.CONFIG_FILE
+    assert args.files == ['f1', 'f2']
+
+
+def test_adjust_args_and_chdir_relative_things(in_git_dir):
+    in_git_dir.join('foo/cfg.yaml').ensure()
+    in_git_dir.join('foo').chdir()
+
+    args = _args(command='run', files=['f1', 'f2'], config='cfg.yaml')
+    main._adjust_args_and_chdir(args)
+    assert os.getcwd() == in_git_dir
+    assert args.config == os.path.join('foo', 'cfg.yaml')
+    assert args.files == [os.path.join('foo', 'f1'), os.path.join('foo', 'f2')]
+
+
+def test_adjust_args_and_chdir_non_relative_config(in_git_dir):
+    in_git_dir.join('foo').ensure_dir().chdir()
+
+    args = _args()
+    main._adjust_args_and_chdir(args)
+    assert os.getcwd() == in_git_dir
+    assert args.config == C.CONFIG_FILE
+
+
+def test_adjust_args_try_repo_repo_relative(in_git_dir):
+    in_git_dir.join('foo').ensure_dir().chdir()
+
+    args = _args(command='try-repo', repo='../foo', files=[])
+    assert args.repo is not None
+    assert os.path.exists(args.repo)
+    main._adjust_args_and_chdir(args)
+    assert os.getcwd() == in_git_dir
+    assert os.path.exists(args.repo)
+    assert args.repo == 'foo'
+
+
+FNS = (
+    'autoupdate', 'clean', 'gc', 'hook_impl', 'install', 'install_hooks',
+    'migrate_config', 'run', 'sample_config', 'uninstall',
+)
+CMDS = tuple(fn.replace('_', '-') for fn in FNS)
+
+
+@pytest.fixture
+def mock_commands():
+    mcks = {fn: mock.patch.object(main, fn).start() for fn in FNS}
+    ret = auto_namedtuple(**mcks)
+    yield ret
+    for mck in ret:
+        mck.stop()
+
+
+@pytest.fixture
+def argparse_parse_args_spy():
+    parse_args_mock = mock.Mock()
+
+    original_parse_args = argparse.ArgumentParser.parse_args
+
+    def fake_parse_args(self, args):
+        # call our spy object
+        parse_args_mock(args)
+        return original_parse_args(self, args)
+
+    with mock.patch.object(
+        argparse.ArgumentParser, 'parse_args', fake_parse_args,
+    ):
+        yield parse_args_mock
+
+
+def assert_only_one_mock_called(mock_objs):
+    total_call_count = sum(mock_obj.call_count for mock_obj in mock_objs)
+    assert total_call_count == 1
+
+
+def test_overall_help(mock_commands):
+    with pytest.raises(SystemExit):
+        main.main(['--help'])
+
+
+def test_help_command(mock_commands, argparse_parse_args_spy):
+    with pytest.raises(SystemExit):
+        main.main(['help'])
+
+    argparse_parse_args_spy.assert_has_calls([
+        mock.call(['help']),
+        mock.call(['--help']),
+    ])
+
+
+def test_help_other_command(mock_commands, argparse_parse_args_spy):
+    with pytest.raises(SystemExit):
+        main.main(['help', 'run'])
+
+    argparse_parse_args_spy.assert_has_calls([
+        mock.call(['help', 'run']),
+        mock.call(['run', '--help']),
+    ])
+
+
+@pytest.mark.parametrize('command', CMDS)
+def test_all_cmds(command, mock_commands, mock_store_dir):
+    main.main((command,))
+    assert getattr(mock_commands, command.replace('-', '_')).call_count == 1
+    assert_only_one_mock_called(mock_commands)
+
+
+def test_try_repo(mock_store_dir):
+    with mock.patch.object(main, 'try_repo') as patch:
+        main.main(('try-repo', '.'))
+    assert patch.call_count == 1
+
+
+def test_init_templatedir(mock_store_dir):
+    with mock.patch.object(main, 'init_templatedir') as patch:
+        main.main(('init-templatedir', 'tdir'))
+    assert patch.call_count == 1
+
+
+def test_help_cmd_in_empty_directory(
+        in_tmpdir,
+        mock_commands,
+        argparse_parse_args_spy,
+):
+    with pytest.raises(SystemExit):
+        main.main(['help', 'run'])
+
+    argparse_parse_args_spy.assert_has_calls([
+        mock.call(['help', 'run']),
+        mock.call(['run', '--help']),
+    ])
+
+
+def test_expected_fatal_error_no_git_repo(in_tmpdir, cap_out, mock_store_dir):
+    with pytest.raises(SystemExit):
+        main.main([])
+    log_file = os.path.join(mock_store_dir, 'pre-commit.log')
+    cap_out_lines = cap_out.get().splitlines()
+    assert (
+        cap_out_lines[-2] ==
+        'An error has occurred: FatalError: git failed. '
+        'Is it installed, and are you in a Git repository directory?'
+    )
+    assert cap_out_lines[-1] == f'Check the log at {log_file}'

tests/make_archives_test.py

@@ -0,0 +1,46 @@
+import tarfile
+
+from pre_commit import git
+from pre_commit import make_archives
+from pre_commit.util import cmd_output
+from testing.util import git_commit
+
+
+def test_make_archive(in_git_dir, tmpdir):
+    output_dir = tmpdir.join('output').ensure_dir()
+    # Add a files to the git directory
+    in_git_dir.join('foo').ensure()
+    cmd_output('git', 'add', '.')
+    git_commit()
+    # We'll use this rev
+    head_rev = git.head_rev('.')
+    # And check that this file doesn't exist
+    in_git_dir.join('bar').ensure()
+    cmd_output('git', 'add', '.')
+    git_commit()
+
+    # Do the thing
+    archive_path = make_archives.make_archive(
+        'foo', in_git_dir.strpath, head_rev, output_dir.strpath,
+    )
+
+    expected = output_dir.join('foo.tar.gz')
+    assert archive_path == expected.strpath
+    assert expected.exists()
+
+    extract_dir = tmpdir.join('extract').ensure_dir()
+    with tarfile.open(archive_path) as tf:
+        tf.extractall(extract_dir.strpath)
+
+    # Verify the contents of the tar
+    assert extract_dir.join('foo').isdir()
+    assert extract_dir.join('foo/foo').exists()
+    assert not extract_dir.join('foo/.git').exists()
+    assert not extract_dir.join('foo/bar').exists()
+
+
+def test_main(tmpdir):
+    make_archives.main(('--dest', tmpdir.strpath))
+
+    for archive, _, _ in make_archives.REPOS:
+        assert tmpdir.join(f'{archive}.tar.gz').exists()

tests/meta_hooks/check_hooks_apply_test.py

@@ -0,0 +1,138 @@
+from pre_commit.meta_hooks import check_hooks_apply
+from testing.fixtures import add_config_to_repo
+
+
+def test_hook_excludes_everything(capsys, in_git_dir, mock_store_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [
+                    {
+                        'id': 'check-useless-excludes',
+                        'exclude': '.pre-commit-config.yaml',
+                    },
+                ],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_hooks_apply.main(()) == 1
+
+    out, _ = capsys.readouterr()
+    assert 'check-useless-excludes does not apply to this repository' in out
+
+
+def test_hook_includes_nothing(capsys, in_git_dir, mock_store_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [
+                    {
+                        'id': 'check-useless-excludes',
+                        'files': 'foo',
+                    },
+                ],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_hooks_apply.main(()) == 1
+
+    out, _ = capsys.readouterr()
+    assert 'check-useless-excludes does not apply to this repository' in out
+
+
+def test_hook_types_not_matched(capsys, in_git_dir, mock_store_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [
+                    {
+                        'id': 'check-useless-excludes',
+                        'types': ['python'],
+                    },
+                ],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_hooks_apply.main(()) == 1
+
+    out, _ = capsys.readouterr()
+    assert 'check-useless-excludes does not apply to this repository' in out
+
+
+def test_hook_types_excludes_everything(capsys, in_git_dir, mock_store_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [
+                    {
+                        'id': 'check-useless-excludes',
+                        'exclude_types': ['yaml'],
+                    },
+                ],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_hooks_apply.main(()) == 1
+
+    out, _ = capsys.readouterr()
+    assert 'check-useless-excludes does not apply to this repository' in out
+
+
+def test_valid_exceptions(capsys, in_git_dir, mock_store_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'local',
+                'hooks': [
+                    # applies to a file
+                    {
+                        'id': 'check-yaml',
+                        'name': 'check yaml',
+                        'entry': './check-yaml',
+                        'language': 'script',
+                        'files': r'\.yaml$',
+                    },
+                    # Should not be reported as an error due to language: fail
+                    {
+                        'id': 'changelogs-rst',
+                        'name': 'changelogs must be rst',
+                        'entry': 'changelog filenames must end in .rst',
+                        'language': 'fail',
+                        'files': r'changelog/.*(?<!\.rst)$',
+                    },
+                    # Should not be reported as an error due to always_run
+                    {
+                        'id': 'i-always-run',
+                        'name': 'make check',
+                        'entry': 'make check',
+                        'language': 'system',
+                        'files': '^$',
+                        'always_run': True,
+                    },
+                ],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_hooks_apply.main(()) == 0
+
+    out, _ = capsys.readouterr()
+    assert out == ''

tests/meta_hooks/check_useless_excludes_test.py

@@ -0,0 +1,115 @@
+from pre_commit.meta_hooks import check_useless_excludes
+from testing.fixtures import add_config_to_repo
+
+
+def test_useless_exclude_global(capsys, in_git_dir):
+    config = {
+        'exclude': 'foo',
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [{'id': 'check-useless-excludes'}],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_useless_excludes.main(()) == 1
+
+    out, _ = capsys.readouterr()
+    out = out.strip()
+    assert "The global exclude pattern 'foo' does not match any files" == out
+
+
+def test_useless_exclude_for_hook(capsys, in_git_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [{'id': 'check-useless-excludes', 'exclude': 'foo'}],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_useless_excludes.main(()) == 1
+
+    out, _ = capsys.readouterr()
+    out = out.strip()
+    expected = (
+        "The exclude pattern 'foo' for check-useless-excludes "
+        'does not match any files'
+    )
+    assert expected == out
+
+
+def test_useless_exclude_with_types_filter(capsys, in_git_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [
+                    {
+                        'id': 'check-useless-excludes',
+                        'exclude': '.pre-commit-config.yaml',
+                        'types': ['python'],
+                    },
+                ],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_useless_excludes.main(()) == 1
+
+    out, _ = capsys.readouterr()
+    out = out.strip()
+    expected = (
+        "The exclude pattern '.pre-commit-config.yaml' for "
+        'check-useless-excludes does not match any files'
+    )
+    assert expected == out
+
+
+def test_no_excludes(capsys, in_git_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [{'id': 'check-useless-excludes'}],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_useless_excludes.main(()) == 0
+
+    out, _ = capsys.readouterr()
+    assert out == ''
+
+
+def test_valid_exclude(capsys, in_git_dir):
+    config = {
+        'repos': [
+            {
+                'repo': 'meta',
+                'hooks': [
+                    {
+                        'id': 'check-useless-excludes',
+                        'exclude': '.pre-commit-config.yaml',
+                    },
+                ],
+            },
+        ],
+    }
+
+    add_config_to_repo(in_git_dir.strpath, config)
+
+    assert check_useless_excludes.main(()) == 0
+
+    out, _ = capsys.readouterr()
+    assert out == ''

tests/meta_hooks/identity_test.py

@@ -0,0 +1,6 @@
+from pre_commit.meta_hooks import identity
+
+
+def test_identity(cap_out):
+    assert not identity.main(('a', 'b', 'c'))
+    assert cap_out.get() == 'a\nb\nc\n'

tests/output_test.py

@@ -0,0 +1,9 @@
+import io
+
+from pre_commit import output
+
+
+def test_output_write_writes():
+    stream = io.BytesIO()
+    output.write('hello world', stream)
+    assert stream.getvalue() == b'hello world'

tests/parse_shebang_test.py

@@ -0,0 +1,152 @@
+import contextlib
+import os.path
+import shutil
+import sys
+
+import pytest
+
+from pre_commit import parse_shebang
+from pre_commit.envcontext import envcontext
+from pre_commit.envcontext import Var
+from pre_commit.util import make_executable
+
+
+def _echo_exe() -> str:
+    exe = shutil.which('echo')
+    assert exe is not None
+    return exe
+
+
+def test_file_doesnt_exist():
+    assert parse_shebang.parse_filename('herp derp derp') == ()
+
+
+def test_simple_case(tmpdir):
+    x = tmpdir.join('f')
+    x.write('#!/usr/bin/env echo')
+    make_executable(x.strpath)
+    assert parse_shebang.parse_filename(x.strpath) == ('echo',)
+
+
+def test_find_executable_full_path():
+    assert parse_shebang.find_executable(sys.executable) == sys.executable
+
+
+def test_find_executable_on_path():
+    assert parse_shebang.find_executable('echo') == _echo_exe()
+
+
+def test_find_executable_not_found_none():
+    assert parse_shebang.find_executable('not-a-real-executable') is None
+
+
+def write_executable(shebang, filename='run'):
+    os.mkdir('bin')
+    path = os.path.join('bin', filename)
+    with open(path, 'w') as f:
+        f.write(f'#!{shebang}')
+    make_executable(path)
+    return path
+
+
+@contextlib.contextmanager
+def bin_on_path():
+    bindir = os.path.join(os.getcwd(), 'bin')
+    with envcontext((('PATH', (bindir, os.pathsep, Var('PATH'))),)):
+        yield
+
+
+def test_find_executable_path_added(in_tmpdir):
+    path = os.path.abspath(write_executable('/usr/bin/env sh'))
+    assert parse_shebang.find_executable('run') is None
+    with bin_on_path():
+        assert parse_shebang.find_executable('run') == path
+
+
+def test_find_executable_path_ext(in_tmpdir):
+    """Windows exports PATHEXT as a list of extensions to automatically add
+    to executables when doing PATH searching.
+    """
+    exe_path = os.path.abspath(
+        write_executable('/usr/bin/env sh', filename='run.myext'),
+    )
+    env_path = {'PATH': os.path.dirname(exe_path)}
+    env_path_ext = dict(env_path, PATHEXT=os.pathsep.join(('.exe', '.myext')))
+    assert parse_shebang.find_executable('run') is None
+    assert parse_shebang.find_executable('run', _environ=env_path) is None
+    ret = parse_shebang.find_executable('run.myext', _environ=env_path)
+    assert ret == exe_path
+    ret = parse_shebang.find_executable('run', _environ=env_path_ext)
+    assert ret == exe_path
+
+
+def test_normexe_does_not_exist():
+    with pytest.raises(OSError) as excinfo:
+        parse_shebang.normexe('i-dont-exist-lol')
+    assert excinfo.value.args == ('Executable `i-dont-exist-lol` not found',)
+
+
+def test_normexe_does_not_exist_sep():
+    with pytest.raises(OSError) as excinfo:
+        parse_shebang.normexe('./i-dont-exist-lol')
+    assert excinfo.value.args == ('Executable `./i-dont-exist-lol` not found',)
+
+
+@pytest.mark.xfail(os.name == 'nt', reason='posix only')
+def test_normexe_not_executable(tmpdir):  # pragma: win32 no cover
+    tmpdir.join('exe').ensure()
+    with tmpdir.as_cwd(), pytest.raises(OSError) as excinfo:
+        parse_shebang.normexe('./exe')
+    assert excinfo.value.args == ('Executable `./exe` is not executable',)
+
+
+def test_normexe_is_a_directory(tmpdir):
+    with tmpdir.as_cwd():
+        tmpdir.join('exe').ensure_dir()
+        exe = os.path.join('.', 'exe')
+        with pytest.raises(OSError) as excinfo:
+            parse_shebang.normexe(exe)
+        msg, = excinfo.value.args
+        assert msg == f'Executable `{exe}` is a directory'
+
+
+def test_normexe_already_full_path():
+    assert parse_shebang.normexe(sys.executable) == sys.executable
+
+
+def test_normexe_gives_full_path():
+    assert parse_shebang.normexe('echo') == _echo_exe()
+    assert os.sep in _echo_exe()
+
+
+def test_normalize_cmd_trivial():
+    cmd = (_echo_exe(), 'hi')
+    assert parse_shebang.normalize_cmd(cmd) == cmd
+
+
+def test_normalize_cmd_PATH():
+    cmd = ('echo', '--version')
+    expected = (_echo_exe(), '--version')
+    assert parse_shebang.normalize_cmd(cmd) == expected
+
+
+def test_normalize_cmd_shebang(in_tmpdir):
+    echo = _echo_exe().replace(os.sep, '/')
+    path = write_executable(echo)
+    assert parse_shebang.normalize_cmd((path,)) == (echo, path)
+
+
+def test_normalize_cmd_PATH_shebang_full_path(in_tmpdir):
+    echo = _echo_exe().replace(os.sep, '/')
+    path = write_executable(echo)
+    with bin_on_path():
+        ret = parse_shebang.normalize_cmd(('run',))
+        assert ret == (echo, os.path.abspath(path))
+
+
+def test_normalize_cmd_PATH_shebang_PATH(in_tmpdir):
+    echo = _echo_exe()
+    path = write_executable('/usr/bin/env echo')
+    with bin_on_path():
+        ret = parse_shebang.normalize_cmd(('run',))
+        assert ret == (echo, os.path.abspath(path))

tests/prefix_test.py

@@ -0,0 +1,44 @@
+import os.path
+
+import pytest
+
+from pre_commit.prefix import Prefix
+
+
+def norm_slash(*args):
+    return tuple(x.replace('/', os.sep) for x in args)
+
+
+@pytest.mark.parametrize(
+    ('prefix', 'path_end', 'expected_output'),
+    (
+        norm_slash('foo', '', 'foo'),
+        norm_slash('foo', 'bar', 'foo/bar'),
+        norm_slash('foo/bar', '../baz', 'foo/baz'),
+        norm_slash('./', 'bar', 'bar'),
+        norm_slash('./', '', '.'),
+        norm_slash('/tmp/foo', '/tmp/bar', '/tmp/bar'),
+    ),
+)
+def test_path(prefix, path_end, expected_output):
+    instance = Prefix(prefix)
+    ret = instance.path(path_end)
+    assert ret == expected_output
+
+
+def test_path_multiple_args():
+    instance = Prefix('foo')
+    ret = instance.path('bar', 'baz')
+    assert ret == os.path.join('foo', 'bar', 'baz')
+
+
+def test_exists(tmpdir):
+    assert not Prefix(str(tmpdir)).exists('foo')
+    tmpdir.ensure('foo')
+    assert Prefix(str(tmpdir)).exists('foo')
+
+
+def test_star(tmpdir):
+    for f in ('a.txt', 'b.txt', 'c.py'):
+        tmpdir.join(f).ensure()
+    assert set(Prefix(str(tmpdir)).star('.txt')) == {'a.txt', 'b.txt'}

tests/repository_test.py

@@ -0,0 +1,902 @@
+import os.path
+import re
+import shutil
+import sys
+from typing import Any
+from typing import Dict
+from unittest import mock
+
+import cfgv
+import pytest
+
+import pre_commit.constants as C
+from pre_commit.clientlib import CONFIG_SCHEMA
+from pre_commit.clientlib import load_manifest
+from pre_commit.envcontext import envcontext
+from pre_commit.hook import Hook
+from pre_commit.languages import golang
+from pre_commit.languages import helpers
+from pre_commit.languages import node
+from pre_commit.languages import python
+from pre_commit.languages import ruby
+from pre_commit.languages import rust
+from pre_commit.languages.all import languages
+from pre_commit.prefix import Prefix
+from pre_commit.repository import all_hooks
+from pre_commit.repository import install_hook_envs
+from pre_commit.util import cmd_output
+from pre_commit.util import cmd_output_b
+from testing.fixtures import make_config_from_repo
+from testing.fixtures import make_repo
+from testing.fixtures import modify_manifest
+from testing.util import cwd
+from testing.util import get_resource_path
+from testing.util import skipif_cant_run_docker
+from testing.util import skipif_cant_run_swift
+from testing.util import xfailif_no_venv
+from testing.util import xfailif_windows_no_ruby
+
+
+def _norm_out(b):
+    return b.replace(b'\r\n', b'\n')
+
+
+def _hook_run(hook, filenames, color):
+    return languages[hook.language].run_hook(hook, filenames, color)
+
+
+def _get_hook_no_install(repo_config, store, hook_id):
+    config = {'repos': [repo_config]}
+    config = cfgv.validate(config, CONFIG_SCHEMA)
+    config = cfgv.apply_defaults(config, CONFIG_SCHEMA)
+    hooks = all_hooks(config, store)
+    hook, = [hook for hook in hooks if hook.id == hook_id]
+    return hook
+
+
+def _get_hook(repo_config, store, hook_id):
+    hook = _get_hook_no_install(repo_config, store, hook_id)
+    install_hook_envs([hook], store)
+    return hook
+
+
+def _test_hook_repo(
+        tempdir_factory,
+        store,
+        repo_path,
+        hook_id,
+        args,
+        expected,
+        expected_return_code=0,
+        config_kwargs=None,
+        color=False,
+):
+    path = make_repo(tempdir_factory, repo_path)
+    config = make_config_from_repo(path, **(config_kwargs or {}))
+    hook = _get_hook(config, store, hook_id)
+    ret, out = _hook_run(hook, args, color=color)
+    assert ret == expected_return_code
+    assert _norm_out(out) == expected
+
+
+def test_conda_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'conda_hooks_repo',
+        'sys-exec', [os.devnull],
+        b'conda-default\n',
+    )
+
+
+def test_conda_with_additional_dependencies_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'conda_hooks_repo',
+        'additional-deps', [os.devnull],
+        b'OK\n',
+        config_kwargs={
+            'hooks': [{
+                'id': 'additional-deps',
+                'args': ['-c', 'import mccabe; print("OK")'],
+                'additional_dependencies': ['mccabe'],
+            }],
+        },
+    )
+
+
+def test_local_conda_additional_dependencies(store):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'local-conda',
+            'name': 'local-conda',
+            'entry': 'python',
+            'language': 'conda',
+            'args': ['-c', 'import mccabe; print("OK")'],
+            'additional_dependencies': ['mccabe'],
+        }],
+    }
+    hook = _get_hook(config, store, 'local-conda')
+    ret, out = _hook_run(hook, (), color=False)
+    assert ret == 0
+    assert _norm_out(out) == b'OK\n'
+
+
+def test_python_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'python_hooks_repo',
+        'foo', [os.devnull],
+        f'[{os.devnull!r}]\nHello World\n'.encode(),
+    )
+
+
+def test_python_hook_default_version(tempdir_factory, store):
+    # make sure that this continues to work for platforms where default
+    # language detection does not work
+    with mock.patch.object(
+            python, 'get_default_version', return_value=C.DEFAULT,
+    ):
+        test_python_hook(tempdir_factory, store)
+
+
+def test_python_hook_args_with_spaces(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'python_hooks_repo',
+        'foo',
+        [],
+        b"['i have spaces', 'and\"\\'quotes', '$and !this']\n"
+        b'Hello World\n',
+        config_kwargs={
+            'hooks': [{
+                'id': 'foo',
+                'args': ['i have spaces', 'and"\'quotes', '$and !this'],
+            }],
+        },
+    )
+
+
+def test_python_hook_weird_setup_cfg(in_git_dir, tempdir_factory, store):
+    in_git_dir.join('setup.cfg').write('[install]\ninstall_scripts=/usr/sbin')
+
+    _test_hook_repo(
+        tempdir_factory, store, 'python_hooks_repo',
+        'foo', [os.devnull],
+        f'[{os.devnull!r}]\nHello World\n'.encode(),
+    )
+
+
+@xfailif_no_venv
+def test_python_venv(tempdir_factory, store):  # pragma: no cover (no venv)
+    _test_hook_repo(
+        tempdir_factory, store, 'python_venv_hooks_repo',
+        'foo', [os.devnull],
+        f'[{os.devnull!r}]\nHello World\n'.encode(),
+    )
+
+
+def test_switch_language_versions_doesnt_clobber(tempdir_factory, store):
+    # We're using the python3 repo because it prints the python version
+    path = make_repo(tempdir_factory, 'python3_hooks_repo')
+
+    def run_on_version(version, expected_output):
+        config = make_config_from_repo(path)
+        config['hooks'][0]['language_version'] = version
+        hook = _get_hook(config, store, 'python3-hook')
+        ret, out = _hook_run(hook, [], color=False)
+        assert ret == 0
+        assert _norm_out(out) == expected_output
+
+    run_on_version('python2', b'2\n[]\nHello World\n')
+    run_on_version('python3', b'3\n[]\nHello World\n')
+
+
+def test_versioned_python_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'python3_hooks_repo',
+        'python3-hook',
+        [os.devnull],
+        f'3\n[{os.devnull!r}]\nHello World\n'.encode(),
+    )
+
+
+@skipif_cant_run_docker  # pragma: win32 no cover
+def test_run_a_docker_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'docker_hooks_repo',
+        'docker-hook',
+        ['Hello World from docker'], b'Hello World from docker\n',
+    )
+
+
+@skipif_cant_run_docker  # pragma: win32 no cover
+def test_run_a_docker_hook_with_entry_args(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'docker_hooks_repo',
+        'docker-hook-arg',
+        ['Hello World from docker'], b'Hello World from docker',
+    )
+
+
+@skipif_cant_run_docker  # pragma: win32 no cover
+def test_run_a_failing_docker_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'docker_hooks_repo',
+        'docker-hook-failing',
+        ['Hello World from docker'],
+        mock.ANY,  # an error message about `bork` not existing
+        expected_return_code=127,
+    )
+
+
+@skipif_cant_run_docker  # pragma: win32 no cover
+@pytest.mark.parametrize('hook_id', ('echo-entrypoint', 'echo-cmd'))
+def test_run_a_docker_image_hook(tempdir_factory, store, hook_id):
+    _test_hook_repo(
+        tempdir_factory, store, 'docker_image_hooks_repo',
+        hook_id,
+        ['Hello World from docker'], b'Hello World from docker\n',
+    )
+
+
+def test_run_a_node_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'node_hooks_repo',
+        'foo', [os.devnull], b'Hello World\n',
+    )
+
+
+def test_run_versioned_node_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'node_versioned_hooks_repo',
+        'versioned-node-hook', [os.devnull], b'v9.3.0\nHello World\n',
+    )
+
+
+@xfailif_windows_no_ruby
+def test_run_a_ruby_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'ruby_hooks_repo',
+        'ruby_hook', [os.devnull], b'Hello world from a ruby hook\n',
+    )
+
+
+@xfailif_windows_no_ruby
+def test_run_versioned_ruby_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'ruby_versioned_hooks_repo',
+        'ruby_hook',
+        [os.devnull],
+        b'2.5.1\nHello world from a ruby hook\n',
+    )
+
+
+@xfailif_windows_no_ruby
+def test_run_ruby_hook_with_disable_shared_gems(
+        tempdir_factory,
+        store,
+        tmpdir,
+):
+    """Make sure a Gemfile in the project doesn't interfere."""
+    tmpdir.join('Gemfile').write('gem "lol_hai"')
+    tmpdir.join('.bundle').mkdir()
+    tmpdir.join('.bundle', 'config').write(
+        'BUNDLE_DISABLE_SHARED_GEMS: true\n'
+        'BUNDLE_PATH: vendor/gem\n',
+    )
+    with cwd(tmpdir.strpath):
+        _test_hook_repo(
+            tempdir_factory, store, 'ruby_versioned_hooks_repo',
+            'ruby_hook',
+            [os.devnull],
+            b'2.5.1\nHello world from a ruby hook\n',
+        )
+
+
+def test_system_hook_with_spaces(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'system_hook_with_spaces_repo',
+        'system-hook-with-spaces', [os.devnull], b'Hello World\n',
+    )
+
+
+@skipif_cant_run_swift  # pragma: win32 no cover
+def test_swift_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'swift_hooks_repo',
+        'swift-hooks-repo', [], b'Hello, world!\n',
+    )
+
+
+def test_golang_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'golang_hooks_repo',
+        'golang-hook', [], b'hello world\n',
+    )
+
+
+def test_golang_hook_still_works_when_gobin_is_set(tempdir_factory, store):
+    gobin_dir = tempdir_factory.get()
+    with envcontext((('GOBIN', gobin_dir),)):
+        test_golang_hook(tempdir_factory, store)
+    assert os.listdir(gobin_dir) == []
+
+
+def test_rust_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'rust_hooks_repo',
+        'rust-hook', [], b'hello world\n',
+    )
+
+
+@pytest.mark.parametrize('dep', ('cli:shellharden:3.1.0', 'cli:shellharden'))
+def test_additional_rust_cli_dependencies_installed(
+        tempdir_factory, store, dep,
+):
+    path = make_repo(tempdir_factory, 'rust_hooks_repo')
+    config = make_config_from_repo(path)
+    # A small rust package with no dependencies.
+    config['hooks'][0]['additional_dependencies'] = [dep]
+    hook = _get_hook(config, store, 'rust-hook')
+    binaries = os.listdir(
+        hook.prefix.path(
+            helpers.environment_dir(rust.ENVIRONMENT_DIR, C.DEFAULT), 'bin',
+        ),
+    )
+    # normalize for windows
+    binaries = [os.path.splitext(binary)[0] for binary in binaries]
+    assert 'shellharden' in binaries
+
+
+def test_additional_rust_lib_dependencies_installed(
+        tempdir_factory, store,
+):
+    path = make_repo(tempdir_factory, 'rust_hooks_repo')
+    config = make_config_from_repo(path)
+    # A small rust package with no dependencies.
+    deps = ['shellharden:3.1.0']
+    config['hooks'][0]['additional_dependencies'] = deps
+    hook = _get_hook(config, store, 'rust-hook')
+    binaries = os.listdir(
+        hook.prefix.path(
+            helpers.environment_dir(rust.ENVIRONMENT_DIR, C.DEFAULT), 'bin',
+        ),
+    )
+    # normalize for windows
+    binaries = [os.path.splitext(binary)[0] for binary in binaries]
+    assert 'rust-hello-world' in binaries
+    assert 'shellharden' not in binaries
+
+
+def test_missing_executable(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'not_found_exe',
+        'not-found-exe', [os.devnull],
+        b'Executable `i-dont-exist-lol` not found',
+        expected_return_code=1,
+    )
+
+
+def test_run_a_script_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'script_hooks_repo',
+        'bash_hook', ['bar'], b'bar\nHello World\n',
+    )
+
+
+def test_run_hook_with_spaced_args(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'arg_per_line_hooks_repo',
+        'arg-per-line',
+        ['foo bar', 'baz'],
+        b'arg: hello\narg: world\narg: foo bar\narg: baz\n',
+    )
+
+
+def test_run_hook_with_curly_braced_arguments(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'arg_per_line_hooks_repo',
+        'arg-per-line',
+        [],
+        b"arg: hi {1}\narg: I'm {a} problem\n",
+        config_kwargs={
+            'hooks': [{
+                'id': 'arg-per-line',
+                'args': ['hi {1}', "I'm {a} problem"],
+            }],
+        },
+    )
+
+
+def test_intermixed_stdout_stderr(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'stdout_stderr_repo',
+        'stdout-stderr',
+        [],
+        b'0\n1\n2\n3\n4\n5\n',
+    )
+
+
+@pytest.mark.xfail(os.name == 'nt', reason='ptys are posix-only')
+def test_output_isatty(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'stdout_stderr_repo',
+        'tty-check',
+        [],
+        b'stdin: False\nstdout: True\nstderr: True\n',
+        color=True,
+    )
+
+
+def _make_grep_repo(entry, store, args=()):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'grep-hook',
+            'name': 'grep-hook',
+            'language': 'pygrep',
+            'entry': entry,
+            'args': args,
+            'types': ['text'],
+        }],
+    }
+    return _get_hook(config, store, 'grep-hook')
+
+
+@pytest.fixture
+def greppable_files(tmpdir):
+    with tmpdir.as_cwd():
+        cmd_output_b('git', 'init', '.')
+        tmpdir.join('f1').write_binary(b"hello'hi\nworld\n")
+        tmpdir.join('f2').write_binary(b'foo\nbar\nbaz\n')
+        tmpdir.join('f3').write_binary(b'[WARN] hi\n')
+        yield tmpdir
+
+
+def test_grep_hook_matching(greppable_files, store):
+    hook = _make_grep_repo('ello', store)
+    ret, out = _hook_run(hook, ('f1', 'f2', 'f3'), color=False)
+    assert ret == 1
+    assert _norm_out(out) == b"f1:1:hello'hi\n"
+
+
+def test_grep_hook_case_insensitive(greppable_files, store):
+    hook = _make_grep_repo('ELLO', store, args=['-i'])
+    ret, out = _hook_run(hook, ('f1', 'f2', 'f3'), color=False)
+    assert ret == 1
+    assert _norm_out(out) == b"f1:1:hello'hi\n"
+
+
+@pytest.mark.parametrize('regex', ('nope', "foo'bar", r'^\[INFO\]'))
+def test_grep_hook_not_matching(regex, greppable_files, store):
+    hook = _make_grep_repo(regex, store)
+    ret, out = _hook_run(hook, ('f1', 'f2', 'f3'), color=False)
+    assert (ret, out) == (0, b'')
+
+
+def _norm_pwd(path):
+    # Under windows bash's temp and windows temp is different.
+    # This normalizes to the bash /tmp
+    return cmd_output_b(
+        'bash', '-c', f"cd '{path}' && pwd",
+    )[1].strip()
+
+
+def test_cwd_of_hook(in_git_dir, tempdir_factory, store):
+    # Note: this doubles as a test for `system` hooks
+    _test_hook_repo(
+        tempdir_factory, store, 'prints_cwd_repo',
+        'prints_cwd', ['-L'], _norm_pwd(in_git_dir.strpath) + b'\n',
+    )
+
+
+def test_lots_of_files(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'script_hooks_repo',
+        'bash_hook', [os.devnull] * 15000, mock.ANY,
+    )
+
+
+def test_additional_dependencies_roll_forward(tempdir_factory, store):
+    path = make_repo(tempdir_factory, 'python_hooks_repo')
+
+    config1 = make_config_from_repo(path)
+    hook1 = _get_hook(config1, store, 'foo')
+    with python.in_env(hook1.prefix, hook1.language_version):
+        assert 'mccabe' not in cmd_output('pip', 'freeze', '-l')[1]
+
+    # Make another repo with additional dependencies
+    config2 = make_config_from_repo(path)
+    config2['hooks'][0]['additional_dependencies'] = ['mccabe']
+    hook2 = _get_hook(config2, store, 'foo')
+    with python.in_env(hook2.prefix, hook2.language_version):
+        assert 'mccabe' in cmd_output('pip', 'freeze', '-l')[1]
+
+    # should not have affected original
+    with python.in_env(hook1.prefix, hook1.language_version):
+        assert 'mccabe' not in cmd_output('pip', 'freeze', '-l')[1]
+
+
+@xfailif_windows_no_ruby  # pragma: win32 no cover
+def test_additional_ruby_dependencies_installed(tempdir_factory, store):
+    path = make_repo(tempdir_factory, 'ruby_hooks_repo')
+    config = make_config_from_repo(path)
+    config['hooks'][0]['additional_dependencies'] = ['tins']
+    hook = _get_hook(config, store, 'ruby_hook')
+    with ruby.in_env(hook.prefix, hook.language_version):
+        output = cmd_output('gem', 'list', '--local')[1]
+        assert 'tins' in output
+
+
+def test_additional_node_dependencies_installed(tempdir_factory, store):
+    path = make_repo(tempdir_factory, 'node_hooks_repo')
+    config = make_config_from_repo(path)
+    # Careful to choose a small package that's not depped by npm
+    config['hooks'][0]['additional_dependencies'] = ['lodash']
+    hook = _get_hook(config, store, 'foo')
+    with node.in_env(hook.prefix, hook.language_version):
+        output = cmd_output('npm', 'ls', '-g')[1]
+        assert 'lodash' in output
+
+
+def test_additional_golang_dependencies_installed(
+        tempdir_factory, store,
+):
+    path = make_repo(tempdir_factory, 'golang_hooks_repo')
+    config = make_config_from_repo(path)
+    # A small go package
+    deps = ['github.com/golang/example/hello']
+    config['hooks'][0]['additional_dependencies'] = deps
+    hook = _get_hook(config, store, 'golang-hook')
+    binaries = os.listdir(
+        hook.prefix.path(
+            helpers.environment_dir(golang.ENVIRONMENT_DIR, C.DEFAULT), 'bin',
+        ),
+    )
+    # normalize for windows
+    binaries = [os.path.splitext(binary)[0] for binary in binaries]
+    assert 'hello' in binaries
+
+
+def test_local_golang_additional_dependencies(store):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'hello',
+            'name': 'hello',
+            'entry': 'hello',
+            'language': 'golang',
+            'additional_dependencies': ['github.com/golang/example/hello'],
+        }],
+    }
+    hook = _get_hook(config, store, 'hello')
+    ret, out = _hook_run(hook, (), color=False)
+    assert ret == 0
+    assert _norm_out(out) == b'Hello, Go examples!\n'
+
+
+def test_local_rust_additional_dependencies(store):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'hello',
+            'name': 'hello',
+            'entry': 'hello',
+            'language': 'rust',
+            'additional_dependencies': ['cli:hello-cli:0.2.2'],
+        }],
+    }
+    hook = _get_hook(config, store, 'hello')
+    ret, out = _hook_run(hook, (), color=False)
+    assert ret == 0
+    assert _norm_out(out) == b'Hello World!\n'
+
+
+def test_fail_hooks(store):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'fail',
+            'name': 'fail',
+            'language': 'fail',
+            'entry': 'make sure to name changelogs as .rst!',
+            'files': r'changelog/.*(?<!\.rst)$',
+        }],
+    }
+    hook = _get_hook(config, store, 'fail')
+    ret, out = _hook_run(
+        hook, ('changelog/123.bugfix', 'changelog/wat'), color=False,
+    )
+    assert ret == 1
+    assert out == (
+        b'make sure to name changelogs as .rst!\n'
+        b'\n'
+        b'changelog/123.bugfix\n'
+        b'changelog/wat\n'
+    )
+
+
+def test_unknown_keys(store, fake_log_handler):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'too-much',
+            'name': 'too much',
+            'hello': 'world',
+            'foo': 'bar',
+            'language': 'system',
+            'entry': 'true',
+        }],
+    }
+    _get_hook(config, store, 'too-much')
+    expected = 'Unexpected key(s) present on local => too-much: foo, hello'
+    assert fake_log_handler.handle.call_args[0][0].msg == expected
+
+
+def test_reinstall(tempdir_factory, store, log_info_mock):
+    path = make_repo(tempdir_factory, 'python_hooks_repo')
+    config = make_config_from_repo(path)
+    _get_hook(config, store, 'foo')
+    # We print some logging during clone (1) + install (3)
+    assert log_info_mock.call_count == 4
+    log_info_mock.reset_mock()
+    # Reinstall on another run should not trigger another install
+    _get_hook(config, store, 'foo')
+    assert log_info_mock.call_count == 0
+
+
+def test_control_c_control_c_on_install(tempdir_factory, store):
+    """Regression test for #186."""
+    path = make_repo(tempdir_factory, 'python_hooks_repo')
+    config = make_config_from_repo(path)
+    hooks = [_get_hook_no_install(config, store, 'foo')]
+
+    class MyKeyboardInterrupt(KeyboardInterrupt):
+        pass
+
+    # To simulate a killed install, we'll make PythonEnv.run raise ^C
+    # and then to simulate a second ^C during cleanup, we'll make shutil.rmtree
+    # raise as well.
+    with pytest.raises(MyKeyboardInterrupt):
+        with mock.patch.object(
+            helpers, 'run_setup_cmd', side_effect=MyKeyboardInterrupt,
+        ):
+            with mock.patch.object(
+                shutil, 'rmtree', side_effect=MyKeyboardInterrupt,
+            ):
+                install_hook_envs(hooks, store)
+
+    # Should have made an environment, however this environment is broken!
+    hook, = hooks
+    assert hook.prefix.exists(
+        helpers.environment_dir(python.ENVIRONMENT_DIR, hook.language_version),
+    )
+
+    # However, it should be perfectly runnable (reinstall after botched
+    # install)
+    install_hook_envs(hooks, store)
+    ret, out = _hook_run(hook, (), color=False)
+    assert ret == 0
+
+
+def test_invalidated_virtualenv(tempdir_factory, store):
+    # A cached virtualenv may become invalidated if the system python upgrades
+    # This should not cause every hook in that virtualenv to fail.
+    path = make_repo(tempdir_factory, 'python_hooks_repo')
+    config = make_config_from_repo(path)
+    hook = _get_hook(config, store, 'foo')
+
+    # Simulate breaking of the virtualenv
+    libdir = hook.prefix.path(
+        helpers.environment_dir(python.ENVIRONMENT_DIR, hook.language_version),
+        'lib', hook.language_version,
+    )
+    paths = [
+        os.path.join(libdir, p) for p in ('site.py', 'site.pyc', '__pycache__')
+    ]
+    cmd_output_b('rm', '-rf', *paths)
+
+    # pre-commit should rebuild the virtualenv and it should be runnable
+    hook = _get_hook(config, store, 'foo')
+    ret, out = _hook_run(hook, (), color=False)
+    assert ret == 0
+
+
+def test_really_long_file_paths(tempdir_factory, store):
+    base_path = tempdir_factory.get()
+    really_long_path = os.path.join(base_path, 'really_long' * 10)
+    cmd_output_b('git', 'init', really_long_path)
+
+    path = make_repo(tempdir_factory, 'python_hooks_repo')
+    config = make_config_from_repo(path)
+
+    with cwd(really_long_path):
+        _get_hook(config, store, 'foo')
+
+
+def test_config_overrides_repo_specifics(tempdir_factory, store):
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    config = make_config_from_repo(path)
+
+    hook = _get_hook(config, store, 'bash_hook')
+    assert hook.files == ''
+    # Set the file regex to something else
+    config['hooks'][0]['files'] = '\\.sh$'
+    hook = _get_hook(config, store, 'bash_hook')
+    assert hook.files == '\\.sh$'
+
+
+def _create_repo_with_tags(tempdir_factory, src, tag):
+    path = make_repo(tempdir_factory, src)
+    cmd_output_b('git', 'tag', tag, cwd=path)
+    return path
+
+
+def test_tags_on_repositories(in_tmpdir, tempdir_factory, store):
+    tag = 'v1.1'
+    git1 = _create_repo_with_tags(tempdir_factory, 'prints_cwd_repo', tag)
+    git2 = _create_repo_with_tags(tempdir_factory, 'script_hooks_repo', tag)
+
+    config1 = make_config_from_repo(git1, rev=tag)
+    hook1 = _get_hook(config1, store, 'prints_cwd')
+    ret1, out1 = _hook_run(hook1, ('-L',), color=False)
+    assert ret1 == 0
+    assert out1.strip() == _norm_pwd(in_tmpdir)
+
+    config2 = make_config_from_repo(git2, rev=tag)
+    hook2 = _get_hook(config2, store, 'bash_hook')
+    ret2, out2 = _hook_run(hook2, ('bar',), color=False)
+    assert ret2 == 0
+    assert out2 == b'bar\nHello World\n'
+
+
+@pytest.fixture
+def local_python_config():
+    # Make a "local" hooks repo that just installs our other hooks repo
+    repo_path = get_resource_path('python_hooks_repo')
+    manifest = load_manifest(os.path.join(repo_path, C.MANIFEST_FILE))
+    hooks = [
+        dict(hook, additional_dependencies=[repo_path]) for hook in manifest
+    ]
+    return {'repo': 'local', 'hooks': hooks}
+
+
+@pytest.mark.xfail(  # pragma: win32 no cover
+    sys.platform == 'win32',
+    reason='microsoft/azure-pipelines-image-generation#989',
+)
+def test_local_python_repo(store, local_python_config):
+    hook = _get_hook(local_python_config, store, 'foo')
+    # language_version should have been adjusted to the interpreter version
+    assert hook.language_version != C.DEFAULT
+    ret, out = _hook_run(hook, ('filename',), color=False)
+    assert ret == 0
+    assert _norm_out(out) == b"['filename']\nHello World\n"
+
+
+def test_default_language_version(store, local_python_config):
+    config: Dict[str, Any] = {
+        'default_language_version': {'python': 'fake'},
+        'default_stages': ['commit'],
+        'repos': [local_python_config],
+    }
+
+    # `language_version` was not set, should default
+    hook, = all_hooks(config, store)
+    assert hook.language_version == 'fake'
+
+    # `language_version` is set, should not default
+    config['repos'][0]['hooks'][0]['language_version'] = 'fake2'
+    hook, = all_hooks(config, store)
+    assert hook.language_version == 'fake2'
+
+
+def test_default_stages(store, local_python_config):
+    config: Dict[str, Any] = {
+        'default_language_version': {'python': C.DEFAULT},
+        'default_stages': ['commit'],
+        'repos': [local_python_config],
+    }
+
+    # `stages` was not set, should default
+    hook, = all_hooks(config, store)
+    assert hook.stages == ['commit']
+
+    # `stages` is set, should not default
+    config['repos'][0]['hooks'][0]['stages'] = ['push']
+    hook, = all_hooks(config, store)
+    assert hook.stages == ['push']
+
+
+def test_hook_id_not_present(tempdir_factory, store, fake_log_handler):
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    config = make_config_from_repo(path)
+    config['hooks'][0]['id'] = 'i-dont-exist'
+    with pytest.raises(SystemExit):
+        _get_hook(config, store, 'i-dont-exist')
+    assert fake_log_handler.handle.call_args[0][0].msg == (
+        f'`i-dont-exist` is not present in repository file://{path}.  '
+        f'Typo? Perhaps it is introduced in a newer version?  '
+        f'Often `pre-commit autoupdate` fixes this.'
+    )
+
+
+def test_too_new_version(tempdir_factory, store, fake_log_handler):
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    with modify_manifest(path) as manifest:
+        manifest[0]['minimum_pre_commit_version'] = '999.0.0'
+    config = make_config_from_repo(path)
+    with pytest.raises(SystemExit):
+        _get_hook(config, store, 'bash_hook')
+    msg = fake_log_handler.handle.call_args[0][0].msg
+    assert re.match(
+        r'^The hook `bash_hook` requires pre-commit version 999\.0\.0 but '
+        r'version \d+\.\d+\.\d+ is installed.  '
+        r'Perhaps run `pip install --upgrade pre-commit`\.$',
+        msg,
+    )
+
+
+@pytest.mark.parametrize('version', ('0.1.0', C.VERSION))
+def test_versions_ok(tempdir_factory, store, version):
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    with modify_manifest(path) as manifest:
+        manifest[0]['minimum_pre_commit_version'] = version
+    config = make_config_from_repo(path)
+    # Should succeed
+    _get_hook(config, store, 'bash_hook')
+
+
+def test_manifest_hooks(tempdir_factory, store):
+    path = make_repo(tempdir_factory, 'script_hooks_repo')
+    config = make_config_from_repo(path)
+    hook = _get_hook(config, store, 'bash_hook')
+
+    assert hook == Hook(
+        src=f'file://{path}',
+        prefix=Prefix(mock.ANY),
+        additional_dependencies=[],
+        alias='',
+        always_run=False,
+        args=[],
+        description='',
+        entry='bin/hook.sh',
+        exclude='^$',
+        exclude_types=[],
+        files='',
+        id='bash_hook',
+        language='script',
+        language_version='default',
+        log_file='',
+        minimum_pre_commit_version='0',
+        name='Bash hook',
+        pass_filenames=True,
+        require_serial=False,
+        stages=(
+            'commit', 'merge-commit', 'prepare-commit-msg', 'commit-msg',
+            'manual', 'post-checkout', 'push',
+        ),
+        types=['file'],
+        verbose=False,
+    )
+
+
+def test_perl_hook(tempdir_factory, store):
+    _test_hook_repo(
+        tempdir_factory, store, 'perl_hooks_repo',
+        'perl-hook', [], b'Hello from perl-commit Perl!\n',
+    )
+
+
+def test_local_perl_additional_dependencies(store):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'hello',
+            'name': 'hello',
+            'entry': 'perltidy --version',
+            'language': 'perl',
+            'additional_dependencies': ['SHANCOCK/Perl-Tidy-20200110.tar.gz'],
+        }],
+    }
+    hook = _get_hook(config, store, 'hello')
+    ret, out = _hook_run(hook, (), color=False)
+    assert ret == 0
+    assert _norm_out(out).startswith(b'This is perltidy, v20200110')

tests/staged_files_only_test.py

@@ -0,0 +1,349 @@
+import itertools
+import os.path
+import shutil
+
+import pytest
+
+from pre_commit import git
+from pre_commit.staged_files_only import staged_files_only
+from pre_commit.util import cmd_output
+from testing.auto_namedtuple import auto_namedtuple
+from testing.fixtures import git_dir
+from testing.util import cwd
+from testing.util import get_resource_path
+from testing.util import git_commit
+
+
+FOO_CONTENTS = '\n'.join(('1', '2', '3', '4', '5', '6', '7', '8', ''))
+
+
+@pytest.fixture
+def patch_dir(tempdir_factory):
+    return tempdir_factory.get()
+
+
+def get_short_git_status():
+    git_status = cmd_output('git', 'status', '-s')[1]
+    line_parts = [line.split() for line in git_status.splitlines()]
+    return {v: k for k, v in line_parts}
+
+
+@pytest.fixture
+def foo_staged(in_git_dir):
+    foo = in_git_dir.join('foo')
+    foo.write(FOO_CONTENTS)
+    cmd_output('git', 'add', 'foo')
+    yield auto_namedtuple(path=in_git_dir.strpath, foo_filename=foo.strpath)
+
+
+def _test_foo_state(
+        path,
+        foo_contents=FOO_CONTENTS,
+        status='A',
+        encoding='UTF-8',
+):
+    assert os.path.exists(path.foo_filename)
+    with open(path.foo_filename, encoding=encoding) as f:
+        assert f.read() == foo_contents
+    actual_status = get_short_git_status()['foo']
+    assert status == actual_status
+
+
+def test_foo_staged(foo_staged):
+    _test_foo_state(foo_staged)
+
+
+def test_foo_nothing_unstaged(foo_staged, patch_dir):
+    with staged_files_only(patch_dir):
+        _test_foo_state(foo_staged)
+    _test_foo_state(foo_staged)
+
+
+def test_foo_something_unstaged(foo_staged, patch_dir):
+    with open(foo_staged.foo_filename, 'w') as foo_file:
+        foo_file.write('herp\nderp\n')
+
+    _test_foo_state(foo_staged, 'herp\nderp\n', 'AM')
+
+    with staged_files_only(patch_dir):
+        _test_foo_state(foo_staged)
+
+    _test_foo_state(foo_staged, 'herp\nderp\n', 'AM')
+
+
+def test_does_not_crash_patch_dir_does_not_exist(foo_staged, patch_dir):
+    with open(foo_staged.foo_filename, 'w') as foo_file:
+        foo_file.write('hello\nworld\n')
+
+    shutil.rmtree(patch_dir)
+    with staged_files_only(patch_dir):
+        pass
+
+
+def test_something_unstaged_ext_diff_tool(foo_staged, patch_dir, tmpdir):
+    diff_tool = tmpdir.join('diff-tool.sh')
+    diff_tool.write('#!/usr/bin/env bash\necho "$@"\n')
+    cmd_output('git', 'config', 'diff.external', diff_tool.strpath)
+    test_foo_something_unstaged(foo_staged, patch_dir)
+
+
+def test_foo_something_unstaged_diff_color_always(foo_staged, patch_dir):
+    cmd_output('git', 'config', '--local', 'color.diff', 'always')
+    test_foo_something_unstaged(foo_staged, patch_dir)
+
+
+def test_foo_both_modify_non_conflicting(foo_staged, patch_dir):
+    with open(foo_staged.foo_filename, 'w') as foo_file:
+        foo_file.write(f'{FOO_CONTENTS}9\n')
+
+    _test_foo_state(foo_staged, f'{FOO_CONTENTS}9\n', 'AM')
+
+    with staged_files_only(patch_dir):
+        _test_foo_state(foo_staged)
+
+        # Modify the file as part of the "pre-commit"
+        with open(foo_staged.foo_filename, 'w') as foo_file:
+            foo_file.write(FOO_CONTENTS.replace('1', 'a'))
+
+        _test_foo_state(foo_staged, FOO_CONTENTS.replace('1', 'a'), 'AM')
+
+    _test_foo_state(foo_staged, f'{FOO_CONTENTS.replace("1", "a")}9\n', 'AM')
+
+
+def test_foo_both_modify_conflicting(foo_staged, patch_dir):
+    with open(foo_staged.foo_filename, 'w') as foo_file:
+        foo_file.write(FOO_CONTENTS.replace('1', 'a'))
+
+    _test_foo_state(foo_staged, FOO_CONTENTS.replace('1', 'a'), 'AM')
+
+    with staged_files_only(patch_dir):
+        _test_foo_state(foo_staged)
+
+        # Modify in the same place as the stashed diff
+        with open(foo_staged.foo_filename, 'w') as foo_file:
+            foo_file.write(FOO_CONTENTS.replace('1', 'b'))
+
+        _test_foo_state(foo_staged, FOO_CONTENTS.replace('1', 'b'), 'AM')
+
+    _test_foo_state(foo_staged, FOO_CONTENTS.replace('1', 'a'), 'AM')
+
+
+@pytest.fixture
+def img_staged(in_git_dir):
+    img = in_git_dir.join('img.jpg')
+    shutil.copy(get_resource_path('img1.jpg'), img.strpath)
+    cmd_output('git', 'add', 'img.jpg')
+    yield auto_namedtuple(path=in_git_dir.strpath, img_filename=img.strpath)
+
+
+def _test_img_state(path, expected_file='img1.jpg', status='A'):
+    assert os.path.exists(path.img_filename)
+    with open(path.img_filename, 'rb') as f1:
+        with open(get_resource_path(expected_file), 'rb') as f2:
+            assert f1.read() == f2.read()
+    actual_status = get_short_git_status()['img.jpg']
+    assert status == actual_status
+
+
+def test_img_staged(img_staged):
+    _test_img_state(img_staged)
+
+
+def test_img_nothing_unstaged(img_staged, patch_dir):
+    with staged_files_only(patch_dir):
+        _test_img_state(img_staged)
+    _test_img_state(img_staged)
+
+
+def test_img_something_unstaged(img_staged, patch_dir):
+    shutil.copy(get_resource_path('img2.jpg'), img_staged.img_filename)
+
+    _test_img_state(img_staged, 'img2.jpg', 'AM')
+
+    with staged_files_only(patch_dir):
+        _test_img_state(img_staged)
+
+    _test_img_state(img_staged, 'img2.jpg', 'AM')
+
+
+def test_img_conflict(img_staged, patch_dir):
+    """Admittedly, this shouldn't happen, but just in case."""
+    shutil.copy(get_resource_path('img2.jpg'), img_staged.img_filename)
+
+    _test_img_state(img_staged, 'img2.jpg', 'AM')
+
+    with staged_files_only(patch_dir):
+        _test_img_state(img_staged)
+        shutil.copy(get_resource_path('img3.jpg'), img_staged.img_filename)
+        _test_img_state(img_staged, 'img3.jpg', 'AM')
+
+    _test_img_state(img_staged, 'img2.jpg', 'AM')
+
+
+@pytest.fixture
+def submodule_with_commits(tempdir_factory):
+    path = git_dir(tempdir_factory)
+    with cwd(path):
+        git_commit()
+        rev1 = cmd_output('git', 'rev-parse', 'HEAD')[1].strip()
+        git_commit()
+        rev2 = cmd_output('git', 'rev-parse', 'HEAD')[1].strip()
+        yield auto_namedtuple(path=path, rev1=rev1, rev2=rev2)
+
+
+def checkout_submodule(rev):
+    cmd_output('git', 'checkout', rev, cwd='sub')
+
+
+@pytest.fixture
+def sub_staged(submodule_with_commits, tempdir_factory):
+    path = git_dir(tempdir_factory)
+    with cwd(path):
+        cmd_output(
+            'git', 'submodule', 'add', submodule_with_commits.path, 'sub',
+        )
+        checkout_submodule(submodule_with_commits.rev1)
+        cmd_output('git', 'add', 'sub')
+        yield auto_namedtuple(
+            path=path,
+            sub_path=os.path.join(path, 'sub'),
+            submodule=submodule_with_commits,
+        )
+
+
+def _test_sub_state(path, rev='rev1', status='A'):
+    assert os.path.exists(path.sub_path)
+    with cwd(path.sub_path):
+        actual_rev = cmd_output('git', 'rev-parse', 'HEAD')[1].strip()
+    assert actual_rev == getattr(path.submodule, rev)
+    actual_status = get_short_git_status()['sub']
+    assert actual_status == status
+
+
+def test_sub_staged(sub_staged):
+    _test_sub_state(sub_staged)
+
+
+def test_sub_nothing_unstaged(sub_staged, patch_dir):
+    with staged_files_only(patch_dir):
+        _test_sub_state(sub_staged)
+    _test_sub_state(sub_staged)
+
+
+def test_sub_something_unstaged(sub_staged, patch_dir):
+    checkout_submodule(sub_staged.submodule.rev2)
+
+    _test_sub_state(sub_staged, 'rev2', 'AM')
+
+    with staged_files_only(patch_dir):
+        # This is different from others, we don't want to touch subs
+        _test_sub_state(sub_staged, 'rev2', 'AM')
+
+    _test_sub_state(sub_staged, 'rev2', 'AM')
+
+
+def test_stage_utf8_changes(foo_staged, patch_dir):
+    contents = '\u2603'
+    with open('foo', 'w', encoding='UTF-8') as foo_file:
+        foo_file.write(contents)
+
+    _test_foo_state(foo_staged, contents, 'AM')
+    with staged_files_only(patch_dir):
+        _test_foo_state(foo_staged)
+    _test_foo_state(foo_staged, contents, 'AM')
+
+
+def test_stage_non_utf8_changes(foo_staged, patch_dir):
+    contents = 'ú'
+    # Produce a latin-1 diff
+    with open('foo', 'w', encoding='latin-1') as foo_file:
+        foo_file.write(contents)
+
+    _test_foo_state(foo_staged, contents, 'AM', encoding='latin-1')
+    with staged_files_only(patch_dir):
+        _test_foo_state(foo_staged)
+    _test_foo_state(foo_staged, contents, 'AM', encoding='latin-1')
+
+
+def test_non_utf8_conflicting_diff(foo_staged, patch_dir):
+    """Regression test for #397"""
+    # The trailing whitespace is important here, this triggers git to produce
+    # an error message which looks like:
+    #
+    # ...patch1471530032:14: trailing whitespace.
+    # [[unprintable character]][[space character]]
+    # error: patch failed: foo:1
+    # error: foo: patch does not apply
+    #
+    # Previously, the error message (though discarded immediately) was being
+    # decoded with the UTF-8 codec (causing a crash)
+    contents = 'ú \n'
+    with open('foo', 'w', encoding='latin-1') as foo_file:
+        foo_file.write(contents)
+
+    _test_foo_state(foo_staged, contents, 'AM', encoding='latin-1')
+    with staged_files_only(patch_dir):
+        _test_foo_state(foo_staged)
+        # Create a conflicting diff that will need to be rolled back
+        with open('foo', 'w') as foo_file:
+            foo_file.write('')
+    _test_foo_state(foo_staged, contents, 'AM', encoding='latin-1')
+
+
+def _write(b):
+    with open('foo', 'wb') as f:
+        f.write(b)
+
+
+def assert_no_diff():
+    tree = cmd_output('git', 'write-tree')[1].strip()
+    cmd_output('git', 'diff-index', tree, '--exit-code')
+
+
+bool_product = tuple(itertools.product((True, False), repeat=2))
+
+
+@pytest.mark.parametrize(('crlf_before', 'crlf_after'), bool_product)
+@pytest.mark.parametrize('autocrlf', ('true', 'false', 'input'))
+def test_crlf(in_git_dir, patch_dir, crlf_before, crlf_after, autocrlf):
+    cmd_output('git', 'config', '--local', 'core.autocrlf', autocrlf)
+
+    before, after = b'1\n2\n', b'3\n4\n\n'
+    before = before.replace(b'\n', b'\r\n') if crlf_before else before
+    after = after.replace(b'\n', b'\r\n') if crlf_after else after
+
+    _write(before)
+    cmd_output('git', 'add', 'foo')
+    _write(after)
+    with staged_files_only(patch_dir):
+        assert_no_diff()
+
+
+def test_whitespace_errors(in_git_dir, patch_dir):
+    cmd_output('git', 'config', '--local', 'apply.whitespace', 'error')
+    test_crlf(in_git_dir, patch_dir, True, True, 'true')
+
+
+def test_autocrlf_committed_crlf(in_git_dir, patch_dir):
+    """Regression test for #570"""
+    cmd_output('git', 'config', '--local', 'core.autocrlf', 'false')
+    _write(b'1\r\n2\r\n')
+    cmd_output('git', 'add', 'foo')
+    git_commit()
+
+    cmd_output('git', 'config', '--local', 'core.autocrlf', 'true')
+    _write(b'1\r\n2\r\n\r\n\r\n\r\n')
+
+    with staged_files_only(patch_dir):
+        assert_no_diff()
+
+
+def test_intent_to_add(in_git_dir, patch_dir):
+    """Regression test for #881"""
+    _write(b'hello\nworld\n')
+    cmd_output('git', 'add', '--intent-to-add', 'foo')
+
+    assert git.intent_to_add_files() == ['foo']
+    with staged_files_only(patch_dir):
+        assert_no_diff()
+    assert git.intent_to_add_files() == ['foo']

tests/store_test.py

@@ -0,0 +1,216 @@
+import os.path
+import sqlite3
+from unittest import mock
+
+import pytest
+
+from pre_commit import git
+from pre_commit.store import _get_default_directory
+from pre_commit.store import Store
+from pre_commit.util import CalledProcessError
+from pre_commit.util import cmd_output
+from testing.fixtures import git_dir
+from testing.util import cwd
+from testing.util import git_commit
+
+
+def test_our_session_fixture_works():
+    """There's a session fixture which makes `Store` invariantly raise to
+    prevent writing to the home directory.
+    """
+    with pytest.raises(AssertionError):
+        Store()
+
+
+def test_get_default_directory_defaults_to_home():
+    # Not we use the module level one which is not mocked
+    ret = _get_default_directory()
+    assert ret == os.path.join(os.path.expanduser('~/.cache'), 'pre-commit')
+
+
+def test_adheres_to_xdg_specification():
+    with mock.patch.dict(
+        os.environ, {'XDG_CACHE_HOME': '/tmp/fakehome'},
+    ):
+        ret = _get_default_directory()
+        assert ret == os.path.join('/tmp/fakehome', 'pre-commit')
+
+
+def test_uses_environment_variable_when_present():
+    with mock.patch.dict(
+        os.environ, {'PRE_COMMIT_HOME': '/tmp/pre_commit_home'},
+    ):
+        ret = _get_default_directory()
+        assert ret == '/tmp/pre_commit_home'
+
+
+def test_store_init(store):
+    # Should create the store directory
+    assert os.path.exists(store.directory)
+    # Should create a README file indicating what the directory is about
+    with open(os.path.join(store.directory, 'README')) as readme_file:
+        readme_contents = readme_file.read()
+        for text_line in (
+            'This directory is maintained by the pre-commit project.',
+            'Learn more: https://github.com/pre-commit/pre-commit',
+        ):
+            assert text_line in readme_contents
+
+
+def test_clone(store, tempdir_factory, log_info_mock):
+    path = git_dir(tempdir_factory)
+    with cwd(path):
+        git_commit()
+        rev = git.head_rev(path)
+        git_commit()
+
+    ret = store.clone(path, rev)
+    # Should have printed some stuff
+    assert log_info_mock.call_args_list[0][0][0].startswith(
+        'Initializing environment for ',
+    )
+
+    # Should return a directory inside of the store
+    assert os.path.exists(ret)
+    assert ret.startswith(store.directory)
+    # Directory should start with `repo`
+    _, dirname = os.path.split(ret)
+    assert dirname.startswith('repo')
+    # Should be checked out to the rev we specified
+    assert git.head_rev(ret) == rev
+
+    # Assert there's an entry in the sqlite db for this
+    assert store.select_all_repos() == [(path, rev, ret)]
+
+
+def test_clone_cleans_up_on_checkout_failure(store):
+    with pytest.raises(Exception) as excinfo:
+        # This raises an exception because you can't clone something that
+        # doesn't exist!
+        store.clone('/i_dont_exist_lol', 'fake_rev')
+    assert '/i_dont_exist_lol' in str(excinfo.value)
+
+    repo_dirs = [
+        d for d in os.listdir(store.directory) if d.startswith('repo')
+    ]
+    assert repo_dirs == []
+
+
+def test_clone_when_repo_already_exists(store):
+    # Create an entry in the sqlite db that makes it look like the repo has
+    # been cloned.
+    with sqlite3.connect(store.db_path) as db:
+        db.execute(
+            'INSERT INTO repos (repo, ref, path) '
+            'VALUES ("fake_repo", "fake_ref", "fake_path")',
+        )
+
+    assert store.clone('fake_repo', 'fake_ref') == 'fake_path'
+
+
+def test_clone_shallow_failure_fallback_to_complete(
+    store, tempdir_factory,
+    log_info_mock,
+):
+    path = git_dir(tempdir_factory)
+    with cwd(path):
+        git_commit()
+        rev = git.head_rev(path)
+        git_commit()
+
+    # Force shallow clone failure
+    def fake_shallow_clone(self, *args, **kwargs):
+        raise CalledProcessError(1, (), 0, b'', None)
+    store._shallow_clone = fake_shallow_clone
+
+    ret = store.clone(path, rev)
+
+    # Should have printed some stuff
+    assert log_info_mock.call_args_list[0][0][0].startswith(
+        'Initializing environment for ',
+    )
+
+    # Should return a directory inside of the store
+    assert os.path.exists(ret)
+    assert ret.startswith(store.directory)
+    # Directory should start with `repo`
+    _, dirname = os.path.split(ret)
+    assert dirname.startswith('repo')
+    # Should be checked out to the rev we specified
+    assert git.head_rev(ret) == rev
+
+    # Assert there's an entry in the sqlite db for this
+    assert store.select_all_repos() == [(path, rev, ret)]
+
+
+def test_clone_tag_not_on_mainline(store, tempdir_factory):
+    path = git_dir(tempdir_factory)
+    with cwd(path):
+        git_commit()
+        cmd_output('git', 'checkout', 'master', '-b', 'branch')
+        git_commit()
+        cmd_output('git', 'tag', 'v1')
+        cmd_output('git', 'checkout', 'master')
+        cmd_output('git', 'branch', '-D', 'branch')
+
+    # previously crashed on unreachable refs
+    store.clone(path, 'v1')
+
+
+def test_create_when_directory_exists_but_not_db(store):
+    # In versions <= 0.3.5, there was no sqlite db causing a need for
+    # backward compatibility
+    os.remove(store.db_path)
+    store = Store(store.directory)
+    assert os.path.exists(store.db_path)
+
+
+def test_create_when_store_already_exists(store):
+    # an assertion that this is idempotent and does not crash
+    Store(store.directory)
+
+
+def test_db_repo_name(store):
+    assert store.db_repo_name('repo', ()) == 'repo'
+    assert store.db_repo_name('repo', ('b', 'a', 'c')) == 'repo:a,b,c'
+
+
+def test_local_resources_reflects_reality():
+    on_disk = {
+        res[len('empty_template_'):]
+        for res in os.listdir('pre_commit/resources')
+        if res.startswith('empty_template_')
+    }
+    assert on_disk == set(Store.LOCAL_RESOURCES)
+
+
+def test_mark_config_as_used(store, tmpdir):
+    with tmpdir.as_cwd():
+        f = tmpdir.join('f').ensure()
+        store.mark_config_used('f')
+        assert store.select_all_configs() == [f.strpath]
+
+
+def test_mark_config_as_used_idempotent(store, tmpdir):
+    test_mark_config_as_used(store, tmpdir)
+    test_mark_config_as_used(store, tmpdir)
+
+
+def test_mark_config_as_used_does_not_exist(store):
+    store.mark_config_used('f')
+    assert store.select_all_configs() == []
+
+
+def _simulate_pre_1_14_0(store):
+    with store.connect() as db:
+        db.executescript('DROP TABLE configs')
+
+
+def test_select_all_configs_roll_forward(store):
+    _simulate_pre_1_14_0(store)
+    assert store.select_all_configs() == []
+
+
+def test_mark_config_as_used_roll_forward(store, tmpdir):
+    _simulate_pre_1_14_0(store)
+    test_mark_config_as_used(store, tmpdir)

tests/util_test.py

@@ -0,0 +1,122 @@
+import os.path
+import stat
+import subprocess
+
+import pytest
+
+from pre_commit.util import CalledProcessError
+from pre_commit.util import clean_path_on_failure
+from pre_commit.util import cmd_output
+from pre_commit.util import cmd_output_b
+from pre_commit.util import cmd_output_p
+from pre_commit.util import make_executable
+from pre_commit.util import parse_version
+from pre_commit.util import rmtree
+from pre_commit.util import tmpdir
+
+
+def test_CalledProcessError_str():
+    error = CalledProcessError(1, ('exe',), 0, b'output', b'errors')
+    assert str(error) == (
+        "command: ('exe',)\n"
+        'return code: 1\n'
+        'expected return code: 0\n'
+        'stdout:\n'
+        '    output\n'
+        'stderr:\n'
+        '    errors'
+    )
+
+
+def test_CalledProcessError_str_nooutput():
+    error = CalledProcessError(1, ('exe',), 0, b'', b'')
+    assert str(error) == (
+        "command: ('exe',)\n"
+        'return code: 1\n'
+        'expected return code: 0\n'
+        'stdout: (none)\n'
+        'stderr: (none)'
+    )
+
+
+def test_clean_on_failure_noop(in_tmpdir):
+    with clean_path_on_failure('foo'):
+        pass
+
+
+def test_clean_path_on_failure_does_nothing_when_not_raising(in_tmpdir):
+    with clean_path_on_failure('foo'):
+        os.mkdir('foo')
+    assert os.path.exists('foo')
+
+
+def test_clean_path_on_failure_cleans_for_normal_exception(in_tmpdir):
+    class MyException(Exception):
+        pass
+
+    with pytest.raises(MyException):
+        with clean_path_on_failure('foo'):
+            os.mkdir('foo')
+            raise MyException
+
+    assert not os.path.exists('foo')
+
+
+def test_clean_path_on_failure_cleans_for_system_exit(in_tmpdir):
+    class MySystemExit(SystemExit):
+        pass
+
+    with pytest.raises(MySystemExit):
+        with clean_path_on_failure('foo'):
+            os.mkdir('foo')
+            raise MySystemExit
+
+    assert not os.path.exists('foo')
+
+
+def test_tmpdir():
+    with tmpdir() as tempdir:
+        assert os.path.exists(tempdir)
+    assert not os.path.exists(tempdir)
+
+
+def test_cmd_output_exe_not_found():
+    ret, out, _ = cmd_output('dne', retcode=None)
+    assert ret == 1
+    assert out == 'Executable `dne` not found'
+
+
+@pytest.mark.parametrize('fn', (cmd_output_b, cmd_output_p))
+def test_cmd_output_exe_not_found_bytes(fn):
+    ret, out, _ = fn('dne', retcode=None, stderr=subprocess.STDOUT)
+    assert ret == 1
+    assert out == b'Executable `dne` not found'
+
+
+@pytest.mark.parametrize('fn', (cmd_output_b, cmd_output_p))
+def test_cmd_output_no_shebang(tmpdir, fn):
+    f = tmpdir.join('f').ensure()
+    make_executable(f)
+
+    # previously this raised `OSError` -- the output is platform specific
+    ret, out, _ = fn(str(f), retcode=None, stderr=subprocess.STDOUT)
+    assert ret == 1
+    assert isinstance(out, bytes)
+    assert out.endswith(b'\n')
+
+
+def test_parse_version():
+    assert parse_version('0.0') == parse_version('0.0')
+    assert parse_version('0.1') > parse_version('0.0')
+    assert parse_version('2.1') >= parse_version('2')
+
+
+def test_rmtree_read_only_directories(tmpdir):
+    """Simulates the go module tree.  See #1042"""
+    tmpdir.join('x/y/z').ensure_dir().join('a').ensure()
+    mode = os.stat(str(tmpdir.join('x'))).st_mode
+    mode_no_w = mode & ~(stat.S_IWUSR | stat.S_IWGRP | stat.S_IWOTH)
+    tmpdir.join('x/y/z').chmod(mode_no_w)
+    tmpdir.join('x/y/z').chmod(mode_no_w)
+    tmpdir.join('x/y/z').chmod(mode_no_w)
+    rmtree(str(tmpdir.join('x')))

tests/xargs_test.py

@@ -0,0 +1,197 @@
+import concurrent.futures
+import os
+import sys
+import time
+from typing import Tuple
+from unittest import mock
+
+import pytest
+
+from pre_commit import parse_shebang
+from pre_commit import xargs
+
+
+@pytest.mark.parametrize(
+    ('env', 'expected'),
+    (
+        ({}, 0),
+        ({b'x': b'1'}, 12),
+        ({b'x': b'12'}, 13),
+        ({b'x': b'1', b'y': b'2'}, 24),
+    ),
+)
+def test_environ_size(env, expected):
+    # normalize integer sizing
+    assert xargs._environ_size(_env=env) == expected
+
+
+@pytest.fixture
+def win32_mock():
+    with mock.patch.object(sys, 'getfilesystemencoding', return_value='utf-8'):
+        with mock.patch.object(sys, 'platform', 'win32'):
+            yield
+
+
+@pytest.fixture
+def linux_mock():
+    with mock.patch.object(sys, 'getfilesystemencoding', return_value='utf-8'):
+        with mock.patch.object(sys, 'platform', 'linux'):
+            yield
+
+
+def test_partition_trivial():
+    assert xargs.partition(('cmd',), (), 1) == (('cmd',),)
+
+
+def test_partition_simple():
+    assert xargs.partition(('cmd',), ('foo',), 1) == (('cmd', 'foo'),)
+
+
+def test_partition_limits():
+    ret = xargs.partition(
+        ('ninechars',), (
+            # Just match the end (with spaces)
+            '.' * 5, '.' * 4,
+            # Just match the end (single arg)
+            '.' * 10,
+            # Goes over the end
+            '.' * 5,
+            '.' * 6,
+        ),
+        1,
+        _max_length=21,
+    )
+    assert ret == (
+        ('ninechars', '.' * 5, '.' * 4),
+        ('ninechars', '.' * 10),
+        ('ninechars', '.' * 5),
+        ('ninechars', '.' * 6),
+    )
+
+
+def test_partition_limit_win32(win32_mock):
+    cmd = ('ninechars',)
+    # counted as half because of utf-16 encode
+    varargs = ('😑' * 5,)
+    ret = xargs.partition(cmd, varargs, 1, _max_length=21)
+    assert ret == (cmd + varargs,)
+
+
+def test_partition_limit_linux(linux_mock):
+    cmd = ('ninechars',)
+    varargs = ('😑' * 5,)
+    ret = xargs.partition(cmd, varargs, 1, _max_length=31)
+    assert ret == (cmd + varargs,)
+
+
+def test_argument_too_long_with_large_unicode(linux_mock):
+    cmd = ('ninechars',)
+    varargs = ('😑' * 10,)  # 4 bytes * 10
+    with pytest.raises(xargs.ArgumentTooLongError):
+        xargs.partition(cmd, varargs, 1, _max_length=20)
+
+
+def test_partition_target_concurrency():
+    ret = xargs.partition(
+        ('foo',), ('A',) * 22,
+        4,
+        _max_length=50,
+    )
+    assert ret == (
+        ('foo',) + ('A',) * 6,
+        ('foo',) + ('A',) * 6,
+        ('foo',) + ('A',) * 6,
+        ('foo',) + ('A',) * 4,
+    )
+
+
+def test_partition_target_concurrency_wont_make_tiny_partitions():
+    ret = xargs.partition(
+        ('foo',), ('A',) * 10,
+        4,
+        _max_length=50,
+    )
+    assert ret == (
+        ('foo',) + ('A',) * 4,
+        ('foo',) + ('A',) * 4,
+        ('foo',) + ('A',) * 2,
+    )
+
+
+def test_argument_too_long():
+    with pytest.raises(xargs.ArgumentTooLongError):
+        xargs.partition(('a' * 5,), ('a' * 5,), 1, _max_length=10)
+
+
+def test_xargs_smoke():
+    ret, out = xargs.xargs(('echo',), ('hello', 'world'))
+    assert ret == 0
+    assert out.replace(b'\r\n', b'\n') == b'hello world\n'
+
+
+exit_cmd = parse_shebang.normalize_cmd(('bash', '-c', 'exit $1', '--'))
+# Abuse max_length to control the exit code
+max_length = len(' '.join(exit_cmd)) + 3
+
+
+def test_xargs_retcode_normal():
+    ret, _ = xargs.xargs(exit_cmd, ('0',), _max_length=max_length)
+    assert ret == 0
+
+    ret, _ = xargs.xargs(exit_cmd, ('0', '1'), _max_length=max_length)
+    assert ret == 1
+
+    # takes the maximum return code
+    ret, _ = xargs.xargs(exit_cmd, ('0', '5', '1'), _max_length=max_length)
+    assert ret == 5
+
+
+def test_xargs_concurrency():
+    bash_cmd = parse_shebang.normalize_cmd(('bash', '-c'))
+    print_pid = ('sleep 0.5 && echo $$',)
+
+    start = time.time()
+    ret, stdout = xargs.xargs(
+        bash_cmd, print_pid * 5,
+        target_concurrency=5,
+        _max_length=len(' '.join(bash_cmd + print_pid)) + 1,
+    )
+    elapsed = time.time() - start
+    assert ret == 0
+    pids = stdout.splitlines()
+    assert len(pids) == 5
+    # It would take 0.5*5=2.5 seconds ot run all of these in serial, so if it
+    # takes less, they must have run concurrently.
+    assert elapsed < 2.5
+
+
+def test_thread_mapper_concurrency_uses_threadpoolexecutor_map():
+    with xargs._thread_mapper(10) as thread_map:
+        _self = thread_map.__self__  # type: ignore
+        assert isinstance(_self, concurrent.futures.ThreadPoolExecutor)
+
+
+def test_thread_mapper_concurrency_uses_regular_map():
+    with xargs._thread_mapper(1) as thread_map:
+        assert thread_map is map
+
+
+def test_xargs_propagate_kwargs_to_cmd():
+    env = {'PRE_COMMIT_TEST_VAR': 'Pre commit is awesome'}
+    cmd: Tuple[str, ...] = ('bash', '-c', 'echo $PRE_COMMIT_TEST_VAR', '--')
+    cmd = parse_shebang.normalize_cmd(cmd)
+
+    ret, stdout = xargs.xargs(cmd, ('1',), env=env)
+    assert ret == 0
+    assert b'Pre commit is awesome' in stdout
+
+
+@pytest.mark.xfail(os.name == 'nt', reason='posix only')
+def test_xargs_color_true_makes_tty():
+    retcode, out = xargs.xargs(
+        (sys.executable, '-c', 'import sys; print(sys.stdout.isatty())'),
+        ('1',),
+        color=True,
+    )
+    assert retcode == 0
+    assert out == b'True\n'