Merging upstream version 3.1.0.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

17 files changed, 576 insertions, 318 deletions

diff --git a/tests/all_languages_test.py b/tests/all_languages_test.py
new file mode 100644
index 0000000..98c9121
--- /dev/null
+++ b/tests/all_languages_test.py
@@ -0,0 +1,7 @@
+from __future__ import annotations
+
+from pre_commit.all_languages import languages
+
+
+def test_python_venv_is_an_alias_to_python():
+    assert languages['python_venv'] is languages['python']
diff --git a/tests/commands/migrate_config_test.py b/tests/commands/migrate_config_test.py
index fca1ad9..ba18463 100644
--- a/tests/commands/migrate_config_test.py
+++ b/tests/commands/migrate_config_test.py
@@ -134,6 +134,39 @@ def test_migrate_config_sha_to_rev(tmpdir):
     )
 
 
+def test_migrate_config_language_python_venv(tmp_path):
+    src = '''\
+repos:
+-   repo: local
+    hooks:
+    -   id: example
+        name: example
+        entry: example
+        language: python_venv
+    -   id: example
+        name: example
+        entry: example
+        language: system
+'''
+    expected = '''\
+repos:
+-   repo: local
+    hooks:
+    -   id: example
+        name: example
+        entry: example
+        language: python
+    -   id: example
+        name: example
+        entry: example
+        language: system
+'''
+    cfg = tmp_path.joinpath('cfg.yaml')
+    cfg.write_text(src)
+    assert migrate_config(str(cfg)) == 0
+    assert cfg.read_text() == expected
+
+
 def test_migrate_config_invalid_yaml(tmpdir):
     contents = '['
     cfg = tmpdir.join(C.CONFIG_FILE)
diff --git a/tests/languages/helpers_test.py b/tests/lang_base_test.py
index c209e7e..a532b6a 100644
--- a/tests/languages/helpers_test.py
+++ b/tests/lang_base_test.py
@@ -8,8 +8,8 @@ from unittest import mock
 import pytest
 
 import pre_commit.constants as C
+from pre_commit import lang_base
 from pre_commit import parse_shebang
-from pre_commit.languages import helpers
 from pre_commit.prefix import Prefix
 from pre_commit.util import CalledProcessError
 
@@ -32,42 +32,42 @@ def homedir_mck():
 
 def test_exe_exists_does_not_exist(find_exe_mck, homedir_mck):
     find_exe_mck.return_value = None
-    assert helpers.exe_exists('ruby') is False
+    assert lang_base.exe_exists('ruby') is False
 
 
 def test_exe_exists_exists(find_exe_mck, homedir_mck):
     find_exe_mck.return_value = os.path.normpath('/usr/bin/ruby')
-    assert helpers.exe_exists('ruby') is True
+    assert lang_base.exe_exists('ruby') is True
 
 
 def test_exe_exists_false_if_shim(find_exe_mck, homedir_mck):
     find_exe_mck.return_value = os.path.normpath('/foo/shims/ruby')
-    assert helpers.exe_exists('ruby') is False
+    assert lang_base.exe_exists('ruby') is False
 
 
 def test_exe_exists_false_if_homedir(find_exe_mck, homedir_mck):
     find_exe_mck.return_value = os.path.normpath('/home/me/somedir/ruby')
-    assert helpers.exe_exists('ruby') is False
+    assert lang_base.exe_exists('ruby') is False
 
 
 def test_exe_exists_commonpath_raises_ValueError(find_exe_mck, homedir_mck):
     find_exe_mck.return_value = os.path.normpath('/usr/bin/ruby')
     with mock.patch.object(os.path, 'commonpath', side_effect=ValueError):
-        assert helpers.exe_exists('ruby') is True
+        assert lang_base.exe_exists('ruby') is True
 
 
 def test_exe_exists_true_when_homedir_is_slash(find_exe_mck):
     find_exe_mck.return_value = os.path.normpath('/usr/bin/ruby')
     with mock.patch.object(os.path, 'expanduser', return_value=os.sep):
-        assert helpers.exe_exists('ruby') is True
+        assert lang_base.exe_exists('ruby') is True
 
 
 def test_basic_get_default_version():
-    assert helpers.basic_get_default_version() == C.DEFAULT
+    assert lang_base.basic_get_default_version() == C.DEFAULT
 
 
 def test_basic_health_check():
-    assert helpers.basic_health_check(Prefix('.'), 'default') is None
+    assert lang_base.basic_health_check(Prefix('.'), 'default') is None
 
 
 def test_failed_setup_command_does_not_unicode_error():
@@ -79,12 +79,27 @@ def test_failed_setup_command_does_not_unicode_error():
 
     # an assertion that this does not raise `UnicodeError`
     with pytest.raises(CalledProcessError):
-        helpers.run_setup_cmd(Prefix('.'), (sys.executable, '-c', script))
+        lang_base.setup_cmd(Prefix('.'), (sys.executable, '-c', script))
+
+
+def test_environment_dir(tmp_path):
+    ret = lang_base.environment_dir(Prefix(tmp_path), 'langenv', 'default')
+    assert ret == f'{tmp_path}{os.sep}langenv-default'
+
+
+def test_assert_version_default():
+    with pytest.raises(AssertionError) as excinfo:
+        lang_base.assert_version_default('lang', '1.2.3')
+    msg, = excinfo.value.args
+    assert msg == (
+        'for now, pre-commit requires system-installed lang -- '
+        'you selected `language_version: 1.2.3`'
+    )
 
 
 def test_assert_no_additional_deps():
     with pytest.raises(AssertionError) as excinfo:
-        helpers.assert_no_additional_deps('lang', ['hmmm'])
+        lang_base.assert_no_additional_deps('lang', ['hmmm'])
     msg, = excinfo.value.args
     assert msg == (
         'for now, pre-commit does not support additional_dependencies for '
@@ -93,22 +108,30 @@ def test_assert_no_additional_deps():
     )
 
 
+def test_no_env_noop(tmp_path):
+    before = os.environ.copy()
+    with lang_base.no_env(Prefix(tmp_path), '1.2.3'):
+        inside = os.environ.copy()
+    after = os.environ.copy()
+    assert before == inside == after
+
+
 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() == 123
+            assert lang_base.target_concurrency() == 123
 
 
 def test_target_concurrency_testing_env_var():
     with mock.patch.dict(
             os.environ, {'PRE_COMMIT_NO_CONCURRENCY': '1'}, clear=True,
     ):
-        assert helpers.target_concurrency() == 1
+        assert lang_base.target_concurrency() == 1
 
 
 def test_target_concurrency_on_travis():
     with mock.patch.dict(os.environ, {'TRAVIS': '1'}, clear=True):
-        assert helpers.target_concurrency() == 2
+        assert lang_base.target_concurrency() == 2
 
 
 def test_target_concurrency_cpu_count_not_implemented():
@@ -116,20 +139,35 @@ def test_target_concurrency_cpu_count_not_implemented():
             multiprocessing, 'cpu_count', side_effect=NotImplementedError,
     ):
         with mock.patch.dict(os.environ, {}, clear=True):
-            assert helpers.target_concurrency() == 1
+            assert lang_base.target_concurrency() == 1
 
 
 def test_shuffled_is_deterministic():
     seq = [str(i) for i in range(10)]
     expected = ['4', '0', '5', '1', '8', '6', '2', '3', '7', '9']
-    assert helpers._shuffled(seq) == expected
+    assert lang_base._shuffled(seq) == expected
 
 
 def test_xargs_require_serial_is_not_shuffled():
-    ret, out = helpers.run_xargs(
+    ret, out = lang_base.run_xargs(
         ('echo',), [str(i) for i in range(10)],
         require_serial=True,
         color=False,
     )
     assert ret == 0
     assert out.strip() == b'0 1 2 3 4 5 6 7 8 9'
+
+
+def test_basic_run_hook(tmp_path):
+    ret, out = lang_base.basic_run_hook(
+        Prefix(tmp_path),
+        'echo hi',
+        ['hello'],
+        ['file', 'file', 'file'],
+        is_local=False,
+        require_serial=False,
+        color=False,
+    )
+    assert ret == 0
+    out = out.replace(b'\r\n', b'\n')
+    assert out == b'hi hello file file file\n'
diff --git a/tests/languages/docker_image_test.py b/tests/languages/docker_image_test.py
new file mode 100644
index 0000000..7993c11
--- /dev/null
+++ b/tests/languages/docker_image_test.py
@@ -0,0 +1,27 @@
+from __future__ import annotations
+
+from pre_commit.languages import docker_image
+from testing.language_helpers import run_language
+from testing.util import xfailif_windows
+
+
+@xfailif_windows  # pragma: win32 no cover
+def test_docker_image_hook_via_entrypoint(tmp_path):
+    ret = run_language(
+        tmp_path,
+        docker_image,
+        '--entrypoint echo ubuntu:22.04',
+        args=('hello hello world',),
+    )
+    assert ret == (0, b'hello hello world\n')
+
+
+@xfailif_windows  # pragma: win32 no cover
+def test_docker_image_hook_via_args(tmp_path):
+    ret = run_language(
+        tmp_path,
+        docker_image,
+        'ubuntu:22.04 echo',
+        args=('hello hello world',),
+    )
+    assert ret == (0, b'hello hello world\n')
diff --git a/tests/languages/docker_test.py b/tests/languages/docker_test.py
index 5f7c85e..836382a 100644
--- a/tests/languages/docker_test.py
+++ b/tests/languages/docker_test.py
@@ -11,6 +11,8 @@ import pytest
 
 from pre_commit.languages import docker
 from pre_commit.util import CalledProcessError
+from testing.language_helpers import run_language
+from testing.util import xfailif_windows
 
 DOCKER_CGROUP_EXAMPLE = b'''\
 12:hugetlb:/docker/c33988ec7651ebc867cb24755eaf637a6734088bc7eef59d5799293a9e5450f7
@@ -181,3 +183,15 @@ def test_get_docker_path_in_docker_docker_in_docker(in_docker):
     err = CalledProcessError(1, (), b'', b'')
     with mock.patch.object(docker, 'cmd_output_b', side_effect=err):
         assert docker._get_docker_path('/project') == '/project'
+
+
+@xfailif_windows  # pragma: win32 no cover
+def test_docker_hook(tmp_path):
+    dockerfile = '''\
+FROM ubuntu:22.04
+CMD ["echo", "This is overwritten by the entry"']
+'''
+    tmp_path.joinpath('Dockerfile').write_text(dockerfile)
+
+    ret = run_language(tmp_path, docker, 'echo hello hello world')
+    assert ret == (0, b'hello hello world\n')
diff --git a/tests/languages/dotnet_test.py b/tests/languages/dotnet_test.py
index e69de29..470c03b 100644
--- a/tests/languages/dotnet_test.py
+++ b/tests/languages/dotnet_test.py
@@ -0,0 +1,154 @@
+from __future__ import annotations
+
+from pre_commit.languages import dotnet
+from testing.language_helpers import run_language
+
+
+def _write_program_cs(tmp_path):
+    program_cs = '''\
+using System;
+
+namespace dotnet_tests
+{
+    class Program
+    {
+        static void Main(string[] args)
+        {
+            Console.WriteLine("Hello from dotnet!");
+        }
+    }
+}
+'''
+    tmp_path.joinpath('Program.cs').write_text(program_cs)
+
+
+def _csproj(tool_name):
+    return f'''\
+<Project Sdk="Microsoft.NET.Sdk">
+  <PropertyGroup>
+    <OutputType>Exe</OutputType>
+    <TargetFramework>net6</TargetFramework>
+    <PackAsTool>true</PackAsTool>
+    <ToolCommandName>{tool_name}</ToolCommandName>
+    <PackageOutputPath>./nupkg</PackageOutputPath>
+  </PropertyGroup>
+</Project>
+'''
+
+
+def test_dotnet_csproj(tmp_path):
+    csproj = _csproj('testeroni')
+    _write_program_cs(tmp_path)
+    tmp_path.joinpath('dotnet_csproj.csproj').write_text(csproj)
+    ret = run_language(tmp_path, dotnet, 'testeroni')
+    assert ret == (0, b'Hello from dotnet!\n')
+
+
+def test_dotnet_csproj_prefix(tmp_path):
+    csproj = _csproj('testeroni.tool')
+    _write_program_cs(tmp_path)
+    tmp_path.joinpath('dotnet_hooks_csproj_prefix.csproj').write_text(csproj)
+    ret = run_language(tmp_path, dotnet, 'testeroni.tool')
+    assert ret == (0, b'Hello from dotnet!\n')
+
+
+def test_dotnet_sln(tmp_path):
+    csproj = _csproj('testeroni')
+    sln = '''\
+Microsoft Visual Studio Solution File, Format Version 12.00
+# Visual Studio 15
+VisualStudioVersion = 15.0.26124.0
+MinimumVisualStudioVersion = 15.0.26124.0
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "dotnet_hooks_sln_repo", "dotnet_hooks_sln_repo.csproj", "{6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}"
+EndProject
+Global
+        GlobalSection(SolutionConfigurationPlatforms) = preSolution
+                Debug|Any CPU = Debug|Any CPU
+                Debug|x64 = Debug|x64
+                Debug|x86 = Debug|x86
+                Release|Any CPU = Release|Any CPU
+                Release|x64 = Release|x64
+                Release|x86 = Release|x86
+        EndGlobalSection
+        GlobalSection(SolutionProperties) = preSolution
+                HideSolutionNode = FALSE
+        EndGlobalSection
+        GlobalSection(ProjectConfigurationPlatforms) = postSolution
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Debug|Any CPU.Build.0 = Debug|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Debug|x64.ActiveCfg = Debug|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Debug|x64.Build.0 = Debug|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Debug|x86.ActiveCfg = Debug|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Debug|x86.Build.0 = Debug|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Release|Any CPU.ActiveCfg = Release|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Release|Any CPU.Build.0 = Release|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Release|x64.ActiveCfg = Release|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Release|x64.Build.0 = Release|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Release|x86.ActiveCfg = Release|Any CPU
+                {6568CFDB-6F6F-45A9-932C-8C7DAABC8E56}.Release|x86.Build.0 = Release|Any CPU
+        EndGlobalSection
+EndGlobal
+'''  # noqa: E501
+    _write_program_cs(tmp_path)
+    tmp_path.joinpath('dotnet_hooks_sln_repo.csproj').write_text(csproj)
+    tmp_path.joinpath('dotnet_hooks_sln_repo.sln').write_text(sln)
+
+    ret = run_language(tmp_path, dotnet, 'testeroni')
+    assert ret == (0, b'Hello from dotnet!\n')
+
+
+def _setup_dotnet_combo(tmp_path):
+    sln = '''\
+Microsoft Visual Studio Solution File, Format Version 12.00
+# Visual Studio Version 16
+VisualStudioVersion = 16.0.30114.105
+MinimumVisualStudioVersion = 10.0.40219.1
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "proj1", "proj1\\proj1.csproj", "{38A939C3-DEA4-47D7-9B75-0418C4249662}"
+EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "proj2", "proj2\\proj2.csproj", "{4C9916CB-165C-4EF5-8A57-4CB6794C1EBF}"
+EndProject
+Global
+        GlobalSection(SolutionConfigurationPlatforms) = preSolution
+                Debug|Any CPU = Debug|Any CPU
+                Release|Any CPU = Release|Any CPU
+        EndGlobalSection
+        GlobalSection(SolutionProperties) = preSolution
+                HideSolutionNode = FALSE
+        EndGlobalSection
+        GlobalSection(ProjectConfigurationPlatforms) = postSolution
+                {38A939C3-DEA4-47D7-9B75-0418C4249662}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+                {38A939C3-DEA4-47D7-9B75-0418C4249662}.Debug|Any CPU.Build.0 = Debug|Any CPU
+                {38A939C3-DEA4-47D7-9B75-0418C4249662}.Release|Any CPU.ActiveCfg = Release|Any CPU
+                {38A939C3-DEA4-47D7-9B75-0418C4249662}.Release|Any CPU.Build.0 = Release|Any CPU
+                {4C9916CB-165C-4EF5-8A57-4CB6794C1EBF}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+                {4C9916CB-165C-4EF5-8A57-4CB6794C1EBF}.Debug|Any CPU.Build.0 = Debug|Any CPU
+                {4C9916CB-165C-4EF5-8A57-4CB6794C1EBF}.Release|Any CPU.ActiveCfg = Release|Any CPU
+                {4C9916CB-165C-4EF5-8A57-4CB6794C1EBF}.Release|Any CPU.Build.0 = Release|Any CPU
+        EndGlobalSection
+EndGlobal
+'''  # noqa: E501
+    tmp_path.joinpath('dotnet_hooks_combo_repo.sln').write_text(sln)
+
+    csproj1 = _csproj('proj1')
+    proj1 = tmp_path.joinpath('proj1')
+    proj1.mkdir()
+    proj1.joinpath('proj1.csproj').write_text(csproj1)
+    _write_program_cs(proj1)
+
+    csproj2 = _csproj('proj2')
+    proj2 = tmp_path.joinpath('proj2')
+    proj2.mkdir()
+    proj2.joinpath('proj2.csproj').write_text(csproj2)
+    _write_program_cs(proj2)
+
+
+def test_dotnet_combo_proj1(tmp_path):
+    _setup_dotnet_combo(tmp_path)
+    ret = run_language(tmp_path, dotnet, 'proj1')
+    assert ret == (0, b'Hello from dotnet!\n')
+
+
+def test_dotnet_combo_proj2(tmp_path):
+    _setup_dotnet_combo(tmp_path)
+    ret = run_language(tmp_path, dotnet, 'proj2')
+    assert ret == (0, b'Hello from dotnet!\n')
diff --git a/tests/languages/fail_test.py b/tests/languages/fail_test.py
new file mode 100644
index 0000000..7c74886
--- /dev/null
+++ b/tests/languages/fail_test.py
@@ -0,0 +1,14 @@
+from __future__ import annotations
+
+from pre_commit.languages import fail
+from testing.language_helpers import run_language
+
+
+def test_fail_hooks(tmp_path):
+    ret = run_language(
+        tmp_path,
+        fail,
+        'watch out for',
+        file_args=('bunnies',),
+    )
+    assert ret == (1, b'watch out for\n\nbunnies\n')
diff --git a/tests/languages/golang_test.py b/tests/languages/golang_test.py
index 7c04255..ec5a878 100644
--- a/tests/languages/golang_test.py
+++ b/tests/languages/golang_test.py
@@ -6,8 +6,11 @@ import pytest
 import re_assert
 
 import pre_commit.constants as C
+from pre_commit import lang_base
+from pre_commit.envcontext import envcontext
 from pre_commit.languages import golang
-from pre_commit.languages import helpers
+from pre_commit.store import _make_local_repo
+from testing.language_helpers import run_language
 
 
 ACTUAL_GET_DEFAULT_VERSION = golang.get_default_version.__wrapped__
@@ -15,7 +18,7 @@ ACTUAL_GET_DEFAULT_VERSION = golang.get_default_version.__wrapped__
 
 @pytest.fixture
 def exe_exists_mck():
-    with mock.patch.object(helpers, 'exe_exists') as mck:
+    with mock.patch.object(lang_base, 'exe_exists') as mck:
         yield mck
 
 
@@ -41,3 +44,93 @@ def test_golang_infer_go_version_default():
 
     assert version != C.DEFAULT
     re_assert.Matches(r'^\d+\.\d+(?:\.\d+)?$').assert_matches(version)
+
+
+def _make_hello_world(tmp_path):
+    go_mod = '''\
+module golang-hello-world
+
+go 1.18
+
+require github.com/BurntSushi/toml v1.1.0
+'''
+    go_sum = '''\
+github.com/BurntSushi/toml v1.1.0 h1:ksErzDEI1khOiGPgpwuI7x2ebx/uXQNw7xJpn9Eq1+I=
+github.com/BurntSushi/toml v1.1.0/go.mod h1:CxXYINrC8qIiEnFrOxCa7Jy5BFHlXnUU2pbicEuybxQ=
+'''  # noqa: E501
+    hello_world_go = '''\
+package main
+
+
+import (
+        "fmt"
+        "github.com/BurntSushi/toml"
+)
+
+type Config struct {
+        What string
+}
+
+func main() {
+        var conf Config
+        toml.Decode("What = 'world'\\n", &conf)
+        fmt.Printf("hello %v\\n", conf.What)
+}
+'''
+    tmp_path.joinpath('go.mod').write_text(go_mod)
+    tmp_path.joinpath('go.sum').write_text(go_sum)
+    mod_dir = tmp_path.joinpath('golang-hello-world')
+    mod_dir.mkdir()
+    main_file = mod_dir.joinpath('main.go')
+    main_file.write_text(hello_world_go)
+
+
+def test_golang_system(tmp_path):
+    _make_hello_world(tmp_path)
+
+    ret = run_language(tmp_path, golang, 'golang-hello-world')
+    assert ret == (0, b'hello world\n')
+
+
+def test_golang_default_version(tmp_path):
+    _make_hello_world(tmp_path)
+
+    ret = run_language(
+        tmp_path,
+        golang,
+        'golang-hello-world',
+        version=C.DEFAULT,
+    )
+    assert ret == (0, b'hello world\n')
+
+
+def test_golang_versioned(tmp_path):
+    _make_local_repo(str(tmp_path))
+
+    ret, out = run_language(
+        tmp_path,
+        golang,
+        'go version',
+        version='1.18.4',
+    )
+
+    assert ret == 0
+    assert out.startswith(b'go version go1.18.4')
+
+
+def test_local_golang_additional_deps(tmp_path):
+    _make_local_repo(str(tmp_path))
+
+    ret = run_language(
+        tmp_path,
+        golang,
+        'hello',
+        deps=('golang.org/x/example/hello@latest',),
+    )
+
+    assert ret == (0, b'Hello, Go examples!\n')
+
+
+def test_golang_hook_still_works_when_gobin_is_set(tmp_path):
+    with envcontext((('GOBIN', str(tmp_path.joinpath('gobin'))),)):
+        test_golang_system(tmp_path)
diff --git a/tests/languages/pygrep_test.py b/tests/languages/pygrep_test.py
index 8420046..c6271c8 100644
--- a/tests/languages/pygrep_test.py
+++ b/tests/languages/pygrep_test.py
@@ -3,6 +3,7 @@ from __future__ import annotations
 import pytest
 
 from pre_commit.languages import pygrep
+from testing.language_helpers import run_language
 
 
 @pytest.fixture
@@ -13,6 +14,9 @@ def some_files(tmpdir):
     tmpdir.join('f4').write_binary(b'foo\npattern\nbar\n')
     tmpdir.join('f5').write_binary(b'[INFO] hi\npattern\nbar')
     tmpdir.join('f6').write_binary(b"pattern\nbarwith'foo\n")
+    tmpdir.join('f7').write_binary(b"hello'hi\nworld\n")
+    tmpdir.join('f8').write_binary(b'foo\nbar\nbaz\n')
+    tmpdir.join('f9').write_binary(b'[WARN] hi\n')
     with tmpdir.as_cwd():
         yield
 
@@ -125,3 +129,16 @@ def test_multiline_multiline_flag_is_enabled(cap_out):
     out = cap_out.get()
     assert ret == 1
     assert out == 'f1:1:foo\nbar\n'
+
+
+def test_grep_hook_matching(some_files, tmp_path):
+    ret = run_language(
+        tmp_path, pygrep, 'ello', file_args=('f7', 'f8', 'f9'),
+    )
+    assert ret == (1, b"f7:1:hello'hi\n")
+
+
+@pytest.mark.parametrize('regex', ('nope', "foo'bar", r'^\[INFO\]'))
+def test_grep_hook_not_matching(regex, some_files, tmp_path):
+    ret = run_language(tmp_path, pygrep, regex, file_args=('f7', 'f8', 'f9'))
+    assert ret == (0, b'')
diff --git a/tests/languages/python_test.py b/tests/languages/python_test.py
index 54fb98f..8bb284e 100644
--- a/tests/languages/python_test.py
+++ b/tests/languages/python_test.py
@@ -12,6 +12,7 @@ from pre_commit.languages import python
 from pre_commit.prefix import Prefix
 from pre_commit.util import make_executable
 from pre_commit.util import win_exe
+from testing.language_helpers import run_language
 
 
 def test_read_pyvenv_cfg(tmpdir):
@@ -210,3 +211,25 @@ def test_unhealthy_then_replaced(python_dir):
     os.replace(f'{py_exe}.tmp', py_exe)
 
     assert python.health_check(prefix, C.DEFAULT) is None
+
+
+def test_language_versioned_python_hook(tmp_path):
+    setup_py = '''\
+from setuptools import setup
+setup(
+    name='example',
+    py_modules=['mod'],
+    entry_points={'console_scripts': ['myexe=mod:main']},
+)
+'''
+    tmp_path.joinpath('setup.py').write_text(setup_py)
+    tmp_path.joinpath('mod.py').write_text('def main(): print("ohai")')
+
+    # we patch this to force virtualenv executing with `-p` since we can't
+    # reliably have multiple pythons available in CI
+    with mock.patch.object(
+            python,
+            '_sys_executable_matches',
+            return_value=False,
+    ):
+        assert run_language(tmp_path, python, 'myexe') == (0, b'ohai\n')
diff --git a/tests/languages/ruby_test.py b/tests/languages/ruby_test.py
index 9cfaad5..6397a43 100644
--- a/tests/languages/ruby_test.py
+++ b/tests/languages/ruby_test.py
@@ -9,8 +9,8 @@ import pre_commit.constants as C
 from pre_commit import parse_shebang
 from pre_commit.envcontext import envcontext
 from pre_commit.languages import ruby
+from pre_commit.languages.ruby import _resource_bytesio
 from pre_commit.store import _make_local_repo
-from pre_commit.util import resource_bytesio
 from testing.language_helpers import run_language
 from testing.util import cwd
 from testing.util import xfailif_windows
@@ -40,7 +40,7 @@ def test_uses_system_if_both_gem_and_ruby_are_available(find_exe_mck):
     ('rbenv.tar.gz', 'ruby-build.tar.gz', 'ruby-download.tar.gz'),
 )
 def test_archive_root_stat(filename):
-    with resource_bytesio(filename) as f:
+    with _resource_bytesio(filename) as f:
         with tarfile.open(fileobj=f) as tarf:
             root, _, _ = filename.partition('.')
             assert oct(tarf.getmember(root).mode) == '0o755'
diff --git a/tests/languages/script_test.py b/tests/languages/script_test.py
new file mode 100644
index 0000000..a02f615
--- /dev/null
+++ b/tests/languages/script_test.py
@@ -0,0 +1,14 @@
+from __future__ import annotations
+
+from pre_commit.languages import script
+from pre_commit.util import make_executable
+from testing.language_helpers import run_language
+
+
+def test_script_language(tmp_path):
+    exe = tmp_path.joinpath('main')
+    exe.write_text('#!/usr/bin/env bash\necho hello hello world\n')
+    make_executable(exe)
+
+    expected = (0, b'hello hello world\n')
+    assert run_language(tmp_path, script, 'main') == expected
diff --git a/tests/languages/system_test.py b/tests/languages/system_test.py
new file mode 100644
index 0000000..dcd9cf1
--- /dev/null
+++ b/tests/languages/system_test.py
@@ -0,0 +1,9 @@
+from __future__ import annotations
+
+from pre_commit.languages import system
+from testing.language_helpers import run_language
+
+
+def test_system_language(tmp_path):
+    expected = (0, b'hello hello world\n')
+    assert run_language(tmp_path, system, 'echo hello hello world') == expected
diff --git a/tests/repository_test.py b/tests/repository_test.py
index b43b344..9e5d9d6 100644
--- a/tests/repository_test.py
+++ b/tests/repository_test.py
@@ -10,15 +10,12 @@ import pytest
 import re_assert
 
 import pre_commit.constants as C
-from pre_commit import git
+from pre_commit import lang_base
+from pre_commit.all_languages import languages
 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 python
-from pre_commit.languages.all import languages
 from pre_commit.prefix import Prefix
 from pre_commit.repository import _hook_installed
 from pre_commit.repository import all_hooks
@@ -28,26 +25,24 @@ 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.language_helpers import run_language
 from testing.util import cwd
 from testing.util import get_resource_path
-from testing.util import skipif_cant_run_docker
-
-
-def _norm_out(b):
-    return b.replace(b'\r\n', b'\n')
 
 
 def _hook_run(hook, filenames, color):
-    with languages[hook.language].in_env(hook.prefix, hook.language_version):
-        return languages[hook.language].run_hook(
-            hook.prefix,
-            hook.entry,
-            hook.args,
-            filenames,
-            is_local=hook.src == 'local',
-            require_serial=hook.require_serial,
-            color=color,
-        )
+    return run_language(
+        path=hook.prefix.prefix_dir,
+        language=languages[hook.language],
+        exe=hook.entry,
+        args=hook.args,
+        file_args=filenames,
+        version=hook.language_version,
+        deps=hook.additional_dependencies,
+        is_local=hook.src == 'local',
+        require_serial=hook.require_serial,
+        color=color,
+    )
 
 
 def _get_hook_no_install(repo_config, store, hook_id):
@@ -81,7 +76,7 @@ def _test_hook_repo(
     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
+    assert out == expected
 
 
 def test_python_hook(tempdir_factory, store):
@@ -129,66 +124,21 @@ def test_python_hook_weird_setup_cfg(in_git_dir, tempdir_factory, store):
     )
 
 
-def test_python_venv(tempdir_factory, store):
-    _test_hook_repo(
-        tempdir_factory, store, 'python_venv_hooks_repo',
-        'foo', [os.devnull],
-        f'[{os.devnull!r}]\nHello World\n'.encode(),
-    )
-
-
-def test_language_versioned_python_hook(tempdir_factory, store):
-    # we patch this force virtualenv executing with `-p` since we can't
-    # reliably have multiple pythons available in CI
-    with mock.patch.object(
-            python,
-            '_sys_executable_matches',
-            return_value=False,
-    ):
-        _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_python_venv_deprecation(store, caplog):
+    config = {
+        'repo': 'local',
+        'hooks': [{
+            'id': 'example',
+            'name': 'example',
+            'language': 'python_venv',
+            'entry': 'echo hi',
+        }],
+    }
+    _get_hook(config, store, 'example')
+    assert caplog.messages[-1] == (
+        '`repo: local` uses deprecated `language: python_venv`.  '
+        'This is an alias for `language: python`.  '
+        'Often `pre-commit autoupdate --repo local` will fix this.'
     )
 
 
@@ -199,92 +149,6 @@ def test_system_hook_with_spaces(tempdir_factory, store):
     )
 
 
-def test_golang_system_hook(tempdir_factory, store):
-    _test_hook_repo(
-        tempdir_factory, store, 'golang_hooks_repo',
-        'golang-hook', ['system'], b'hello world from system\n',
-        config_kwargs={
-            'hooks': [{
-                'id': 'golang-hook',
-                'language_version': 'system',
-            }],
-        },
-    )
-
-
-def test_golang_versioned_hook(tempdir_factory, store):
-    _test_hook_repo(
-        tempdir_factory, store, 'golang_hooks_repo',
-        'golang-hook', [], b'hello world from go1.18.4\n',
-        config_kwargs={
-            'hooks': [{
-                'id': 'golang-hook',
-                'language_version': '1.18.4',
-            }],
-        },
-    )
-
-
-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_system_hook(tempdir_factory, store)
-    assert os.listdir(gobin_dir) == []
-
-
-def test_golang_with_recursive_submodule(tmpdir, tempdir_factory, store):
-    sub_go = '''\
-package sub
-
-import "fmt"
-
-func Func() {
-    fmt.Println("hello hello world")
-}
-'''
-    sub = tmpdir.join('sub').ensure_dir()
-    sub.join('sub.go').write(sub_go)
-    cmd_output('git', '-C', str(sub), 'init', '.')
-    cmd_output('git', '-C', str(sub), 'add', '.')
-    git.commit(str(sub))
-
-    pre_commit_hooks = '''\
--   id: example
-    name: example
-    entry: example
-    language: golang
-    verbose: true
-'''
-    go_mod = '''\
-module github.com/asottile/example
-
-go 1.14
-'''
-    main_go = '''\
-package main
-
-import "github.com/asottile/example/sub"
-
-func main() {
-    sub.Func()
-}
-'''
-    repo = tmpdir.join('repo').ensure_dir()
-    repo.join('.pre-commit-hooks.yaml').write(pre_commit_hooks)
-    repo.join('go.mod').write(go_mod)
-    repo.join('main.go').write(main_go)
-    cmd_output('git', '-C', str(repo), 'init', '.')
-    cmd_output('git', '-C', str(repo), 'add', '.')
-    cmd_output('git', '-C', str(repo), 'submodule', 'add', str(sub), 'sub')
-    git.commit(str(repo))
-
-    config = make_config_from_repo(str(repo))
-    hook = _get_hook(config, store, 'example')
-    ret, out = _hook_run(hook, (), color=False)
-    assert ret == 0
-    assert _norm_out(out) == b'hello hello world\n'
-
-
 def test_missing_executable(tempdir_factory, store):
     _test_hook_repo(
         tempdir_factory, store, 'not_found_exe',
@@ -345,52 +209,6 @@ def test_output_isatty(tempdir_factory, store):
     )
 
 
-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
@@ -440,7 +258,7 @@ def test_repository_state_compatibility(tempdir_factory, store, v):
 
     config = make_config_from_repo(path)
     hook = _get_hook(config, store, 'foo')
-    envdir = helpers.environment_dir(
+    envdir = lang_base.environment_dir(
         hook.prefix,
         python.ENVIRONMENT_DIR,
         hook.language_version,
@@ -449,67 +267,6 @@ def test_repository_state_compatibility(tempdir_factory, store, v):
     assert _hook_installed(hook) is True
 
 
-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 = ['golang.org/x/example/hello@latest']
-    config['hooks'][0]['additional_dependencies'] = deps
-    hook = _get_hook(config, store, 'golang-hook')
-    envdir = helpers.environment_dir(
-        hook.prefix,
-        golang.ENVIRONMENT_DIR,
-        golang.get_default_version(),
-    )
-    binaries = os.listdir(os.path.join(envdir, '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': ['golang.org/x/example/hello@latest'],
-        }],
-    }
-    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_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, caplog):
     config = {
         'repo': 'local',
@@ -553,7 +310,7 @@ def test_control_c_control_c_on_install(tempdir_factory, store):
     # raise as well.
     with pytest.raises(MyKeyboardInterrupt):
         with mock.patch.object(
-            helpers, 'run_setup_cmd', side_effect=MyKeyboardInterrupt,
+            lang_base, 'setup_cmd', side_effect=MyKeyboardInterrupt,
         ):
             with mock.patch.object(
                 shutil, 'rmtree', side_effect=MyKeyboardInterrupt,
@@ -562,7 +319,7 @@ def test_control_c_control_c_on_install(tempdir_factory, store):
 
     # Should have made an environment, however this environment is broken!
     hook, = hooks
-    envdir = helpers.environment_dir(
+    envdir = lang_base.environment_dir(
         hook.prefix,
         python.ENVIRONMENT_DIR,
         hook.language_version,
@@ -585,7 +342,7 @@ def test_invalidated_virtualenv(tempdir_factory, store):
     hook = _get_hook(config, store, 'foo')
 
     # Simulate breaking of the virtualenv
-    envdir = helpers.environment_dir(
+    envdir = lang_base.environment_dir(
         hook.prefix,
         python.ENVIRONMENT_DIR,
         hook.language_version,
@@ -667,7 +424,7 @@ def test_local_python_repo(store, local_python_config):
     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"
+    assert out == b"['filename']\nHello World\n"
 
 
 def test_default_language_version(store, local_python_config):
@@ -781,22 +538,6 @@ def test_manifest_hooks(tempdir_factory, store):
     )
 
 
-@pytest.mark.parametrize(
-    'repo',
-    (
-        'dotnet_hooks_csproj_repo',
-        'dotnet_hooks_sln_repo',
-        'dotnet_hooks_combo_repo',
-        'dotnet_hooks_csproj_prefix_repo',
-    ),
-)
-def test_dotnet_hook(tempdir_factory, store, repo):
-    _test_hook_repo(
-        tempdir_factory, store, repo,
-        'dotnet-example-hook', [], b'Hello from dotnet!\n',
-    )
-
-
 def test_non_installable_hook_error_for_language_version(store, caplog):
     config = {
         'repo': 'local',
diff --git a/tests/staged_files_only_test.py b/tests/staged_files_only_test.py
index a91f315..58dbe5a 100644
--- a/tests/staged_files_only_test.py
+++ b/tests/staged_files_only_test.py
@@ -1,12 +1,15 @@
 from __future__ import annotations
 
+import contextlib
 import itertools
 import os.path
 import shutil
 
 import pytest
+import re_assert
 
 from pre_commit import git
+from pre_commit.errors import FatalError
 from pre_commit.staged_files_only import staged_files_only
 from pre_commit.util import cmd_output
 from testing.auto_namedtuple import auto_namedtuple
@@ -14,6 +17,7 @@ from testing.fixtures import git_dir
 from testing.util import cwd
 from testing.util import get_resource_path
 from testing.util import git_commit
+from testing.util import xfailif_windows
 
 
 FOO_CONTENTS = '\n'.join(('1', '2', '3', '4', '5', '6', '7', '8', ''))
@@ -382,3 +386,49 @@ def test_intent_to_add(in_git_dir, patch_dir):
     with staged_files_only(patch_dir):
         assert_no_diff()
     assert git.intent_to_add_files() == ['foo']
+
+
+@contextlib.contextmanager
+def _unreadable(f):
+    orig = os.stat(f).st_mode
+    os.chmod(f, 0o000)
+    try:
+        yield
+    finally:
+        os.chmod(f, orig)
+
+
+@xfailif_windows  # pragma: win32 no cover
+def test_failed_diff_does_not_discard_changes(in_git_dir, patch_dir):
+    # stage 3 files
+    for i in range(3):
+        with open(str(i), 'w') as f:
+            f.write(str(i))
+    cmd_output('git', 'add', '0', '1', '2')
+
+    # modify all of their contents
+    for i in range(3):
+        with open(str(i), 'w') as f:
+            f.write('new contents')
+
+    with _unreadable('1'):
+        with pytest.raises(FatalError) as excinfo:
+            with staged_files_only(patch_dir):
+                raise AssertionError('should have errored on enter')
+
+    # the diff command failed to produce a diff of `1`
+    msg, = excinfo.value.args
+    re_assert.Matches(
+        r'^pre-commit failed to diff -- perhaps due to permissions\?\n\n'
+        r'command: .*\n'
+        r'return code: 128\n'
+        r'stdout: \(none\)\n'
+        r'stderr:\n'
+        r'    error: open\("1"\): Permission denied\n'
+        r'    fatal: cannot hash 1$',
+    ).assert_matches(msg)
+
+    # even though it errored, the unstaged changes should still be present
+    for i in range(3):
+        with open(str(i)) as f:
+            assert f.read() == 'new contents'
diff --git a/tests/store_test.py b/tests/store_test.py
index c42ce65..eaab940 100644
--- a/tests/store_test.py
+++ b/tests/store_test.py
@@ -180,7 +180,7 @@ def test_create_when_store_already_exists(store):
 
 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'
+    assert store.db_repo_name('repo', ('b', 'a', 'c')) == 'repo:b,a,c'
 
 
 def test_local_resources_reflects_reality():
@@ -246,3 +246,27 @@ def test_mark_config_as_used_readonly(tmpdir):
     # should be skipped due to readonly
     store.mark_config_used(str(cfg))
     assert store.select_all_configs() == []
+
+
+def test_clone_with_recursive_submodules(store, tmp_path):
+    sub = tmp_path.joinpath('sub')
+    sub.mkdir()
+    sub.joinpath('submodule').write_text('i am a submodule')
+    cmd_output('git', '-C', str(sub), 'init', '.')
+    cmd_output('git', '-C', str(sub), 'add', '.')
+    git.commit(str(sub))
+
+    repo = tmp_path.joinpath('repo')
+    repo.mkdir()
+    repo.joinpath('repository').write_text('i am a repo')
+    cmd_output('git', '-C', str(repo), 'init', '.')
+    cmd_output('git', '-C', str(repo), 'add', '.')
+    cmd_output('git', '-C', str(repo), 'submodule', 'add', str(sub), 'sub')
+    git.commit(str(repo))
+
+    rev = git.head_rev(str(repo))
+    ret = store.clone(str(repo), rev)
+
+    assert os.path.exists(ret)
+    assert os.path.exists(os.path.join(ret, str(repo), 'repository'))
+    assert os.path.exists(os.path.join(ret, str(sub), 'submodule'))
diff --git a/tests/util_test.py b/tests/util_test.py
index 310f8f5..5b26211 100644
--- a/tests/util_test.py
+++ b/tests/util_test.py
@@ -16,7 +16,7 @@ from pre_commit.util import rmtree
 
 
 def test_CalledProcessError_str():
-    error = CalledProcessError(1, ('exe',), b'output', b'errors')
+    error = CalledProcessError(1, ('exe',), b'output\n', b'errors\n')
     assert str(error) == (
         "command: ('exe',)\n"
         'return code: 1\n'