fixtures

Collection of pytest fixtures used in conda tests.

Classes

CondaCLIFixture
PipCLIFixture	Fixture for calling pip in specific conda environments.
PathFactoryFixture
TmpEnvFixture
TmpChannelFixture
HttpTestServerFixture	Fixture providing HTTP test server for serving local files.

Functions

suppress_resource_warning()	Suppress Unclosed Socket Warning
tmpdir(tmpdir, request)
clear_subdir_cache()
reset_conda_context()	Resets the context object after each test function is run.
temp_package_cache(...)	Used to isolate package or index cache from other tests.
parametrized_solver_fixture(...)	A parameterized fixture that sets the solver backend to (1) libmamba
solver_classic(...)
solver_libmamba(...)
_solver_helper(→ collections.abc.Iterable[Solver])
session_capsys(...)
conda_cli(→ collections.abc.Iterator[CondaCLIFixture])	A function scoped fixture returning CondaCLIFixture instance.
session_conda_cli(...)	A session scoped fixture returning CondaCLIFixture instance.
pip_cli(→ collections.abc.Iterator[PipCLIFixture])	A function scoped fixture returning PipCLIFixture instance.
path_factory(...)	A function scoped fixture returning PathFactoryFixture instance.
tmp_env(→ collections.abc.Iterator[TmpEnvFixture])	A function scoped fixture returning TmpEnvFixture instance.
empty_env(→ pathlib.Path)	A function scoped fixture returning an empty environment.
session_tmp_env(→ collections.abc.Iterator[TmpEnvFixture])	A session scoped fixture returning TmpEnvFixture instance.
tmp_channel(→ collections.abc.Iterator[TmpChannelFixture])	A function scoped fixture returning TmpChannelFixture instance.
context_aware_monkeypatch(→ pytest.MonkeyPatch)	A monkeypatch fixture that resets context after each test.
tmp_pkgs_dir(→ collections.abc.Iterator[pathlib.Path])	A function scoped fixture returning a temporary package cache directory.
tmp_envs_dir(→ collections.abc.Iterator[pathlib.Path])	A function scoped fixture returning a temporary environment directory.
PYTHONPATH()	We need to set this so Python loads the dev version of 'conda', usually taken
context_testdata(→ None)
http_test_server(...)	Function-scoped HTTP test server for serving local files.

Attributes

TEST_CONDARC
Solver

TEST_CONDARC

suppress_resource_warning()

Suppress Unclosed Socket Warning

It seems urllib3 keeps a socket open to avoid costly recreation costs.

xref: kennethreitz/requests#1882

tmpdir(tmpdir, request)

clear_subdir_cache()

reset_conda_context()

Resets the context object after each test function is run.

temp_package_cache(tmp_path_factory, monkeypatch: pytest.MonkeyPatch) → collections.abc.Iterator[pathlib.Path]

Used to isolate package or index cache from other tests.

parametrized_solver_fixture(request: pytest.FixtureRequest) → collections.abc.Iterable[Literal[libmamba, classic]]

A parameterized fixture that sets the solver backend to (1) libmamba and (2) classic for each test. It's using autouse=True, so only import it in modules that actually need it.

Note that skips and xfails need to be done _inside_ the test body. Decorators can't be used because they are evaluated before the fixture has done its work!

So, instead of:

@pytest.mark.skipif(context.solver == "libmamba", reason="...") def test_foo():

...

Do:

def test_foo():

if context.solver == "libmamba":

pytest.skip("...")

...

solver_classic(request: pytest.FixtureRequest) → collections.abc.Iterable[Literal[classic]]

solver_libmamba(request: pytest.FixtureRequest) → collections.abc.Iterable[Literal[libmamba]]

Solver

_solver_helper(request: pytest.FixtureRequest, solver: Solver) → collections.abc.Iterable[Solver]

session_capsys(request) → collections.abc.Iterator[_pytest.capture.MultiCapture]

class CondaCLIFixture

capsys: pytest.CaptureFixture | None

__call__(*argv: conda.common.path.PathType, raises: type[Exception] | tuple[type[Exception], Ellipsis]) → tuple[str, str, pytest.ExceptionInfo]

__call__(*argv: conda.common.path.PathType) → tuple[str, str, int]

Test conda CLI. Mimic what is done in conda.cli.main.main.

conda ... == conda_cli(...)

Parameters:

• argv -- Arguments to parse.

• raises -- Expected exception to intercept. If provided, the raised exception will be returned instead of exit code (see pytest.raises and pytest.ExceptionInfo).

Returns:

Command results (stdout, stderr, exit code or pytest.ExceptionInfo).

static _cast_args(argv: tuple[conda.common.path.PathType, Ellipsis]) → collections.abc.Iterable[str]

Cast args to string.

conda_cli(capsys: pytest.CaptureFixture) → collections.abc.Iterator[CondaCLIFixture]

A function scoped fixture returning CondaCLIFixture instance.

Use this for any commands that are local to the current test (e.g., creating a conda environment only used in the test).

session_conda_cli() → collections.abc.Iterator[CondaCLIFixture]

A session scoped fixture returning CondaCLIFixture instance.

Use this for any commands that are global to the test session (e.g., creating a conda environment shared across tests, conda info, etc.).

class PipCLIFixture

Fixture for calling pip in specific conda environments.

__call__(*argv: conda.common.path.PathType, prefix: conda.common.path.PathType, raises: type[Exception] | tuple[type[Exception], Ellipsis]) → tuple[str, str, pytest.ExceptionInfo]

__call__(*argv: conda.common.path.PathType, prefix: conda.common.path.PathType) → tuple[str, str, int]

Test pip CLI in a specific conda environment.

pip ... in environment == pip_cli(..., prefix=env_path)

Parameters:

• argv -- Arguments to pass to pip.

• prefix -- Path to the conda environment containing pip.

• raises -- Expected exception to intercept. If provided, the raised exception will be returned instead of exit code (see pytest.raises and pytest.ExceptionInfo).

Returns:

Command results (stdout, stderr, exit code or pytest.ExceptionInfo).

pip_cli() → collections.abc.Iterator[PipCLIFixture]

A function scoped fixture returning PipCLIFixture instance.

Use this for calling pip commands in specific conda environments during tests. Uses python -m pip for reliable cross-platform execution.

Example

def test_pip_install(tmp_env, pip_cli):

with tmp_env("python=3.10", "pip") as prefix:

stdout, stderr, code = pip_cli("install", "requests", prefix=prefix) assert code == 0

class PathFactoryFixture

tmp_path: pathlib.Path

__call__(name: str | None = None, *, prefix: str | None = None, infix: str | None = None, suffix: str | None = None) → pathlib.Path

Unique, non-existent path factory.

Extends pytest's tmp_path fixture with a new unique, non-existent path for usage in cases where we need a temporary path that doesn't exist yet.

Default behavior (no arguments):

path_factory() → tmp_path/ab12cd34ef56 (12-char UUID)

Two modes of operation (mutually exclusive):

1. Name mode: Pass a complete path name. path_factory("myfile.txt") → tmp_path/myfile.txt

2. Parts mode: Pass prefix/infix/suffix; unspecified parts get UUID defaults. path_factory(infix="!") → tmp_path/ab12!ef56 path_factory(suffix=".yml") → tmp_path/ab12cd34.yml

Parameters:

• name -- Complete path name (mutually exclusive with prefix/infix/suffix)

• prefix -- Prefix for generated name (mutually exclusive with name param)

• infix -- Infix for generated name (mutually exclusive with name param)

• suffix -- Suffix for generated name (mutually exclusive with name param)

Returns:

A new unique path

path_factory(tmp_path: pathlib.Path) → collections.abc.Iterator[PathFactoryFixture]

A function scoped fixture returning PathFactoryFixture instance.

Use this to generate any number of temporary paths for the test that are unique and do not exist yet.

class TmpEnvFixture

path_factory: PathFactoryFixture | pytest.TempPathFactory

conda_cli: CondaCLIFixture

get_path(name: str | None = None, prefix: str | None = None, infix: str | None = None, suffix: str | None = None) → pathlib.Path

__call__(*args: str, prefix: str | os.PathLike | None = None, name: str | None = None, path_prefix: str | None = None, path_infix: str | None = None, path_suffix: str | None = None, shallow: bool | None = None) → collections.abc.Iterator[pathlib.Path]

Generate a conda environment with the provided packages.

Path customization (mutually exclusive options):

1. Auto-generated path (default): Unique path in tmp_path. tmp_env() → tmp_path/ab12cd34ef56 (12-char UUID)

2. Custom prefix: Specify exact location. tmp_env(prefix="/path/to/env") → /path/to/env

3. Name mode: Specify env name directly. tmp_env(name="my-test-env") → tmp_path/my-test-env

4. Parts mode: Customize path name generation (useful for special char testing). tmp_env(path_infix="!") → tmp_path/ab12!ef56

Parameters:

• args -- Arguments to pass to conda create (e.g., packages, flags)

• prefix -- Exact prefix path (mutually exclusive with name/path_* params)

• name -- Env name (mutually exclusive with prefix/path_* params)

• path_prefix -- Prefix for path name (mutually exclusive with prefix/name params)

• path_infix -- Infix for path name (mutually exclusive with prefix/name params)

• path_suffix -- Suffix for path name (mutually exclusive with prefix/name params)

• shallow -- If True, create env on disk only without conda create

Returns:

The conda environment's prefix

tmp_env(path_factory: PathFactoryFixture, conda_cli: CondaCLIFixture) → collections.abc.Iterator[TmpEnvFixture]

A function scoped fixture returning TmpEnvFixture instance.

Use this when creating a conda environment that is local to the current test.

empty_env(tmp_env: TmpEnvFixture) → pathlib.Path

A function scoped fixture returning an empty environment.

Use this when creating a conda environment that is empty.

session_tmp_env(tmp_path_factory: pytest.TempPathFactory, session_conda_cli: CondaCLIFixture) → collections.abc.Iterator[TmpEnvFixture]

A session scoped fixture returning TmpEnvFixture instance.

Use this when creating a conda environment that is shared across tests.

class TmpChannelFixture

path_factory: PathFactoryFixture

conda_cli: CondaCLIFixture

__call__(*specs: str) → collections.abc.Iterator[tuple[pathlib.Path, str]]

tmp_channel(path_factory: PathFactoryFixture, conda_cli: CondaCLIFixture) → collections.abc.Iterator[TmpChannelFixture]

A function scoped fixture returning TmpChannelFixture instance.

context_aware_monkeypatch(monkeypatch: pytest.MonkeyPatch) → pytest.MonkeyPatch

A monkeypatch fixture that resets context after each test.

tmp_pkgs_dir(path_factory: PathFactoryFixture, mocker: pytest_mock.MockerFixture) → collections.abc.Iterator[pathlib.Path]

A function scoped fixture returning a temporary package cache directory.

tmp_envs_dir(path_factory: PathFactoryFixture, mocker: pytest_mock.MockerFixture) → collections.abc.Iterator[pathlib.Path]

A function scoped fixture returning a temporary environment directory.

PYTHONPATH()

We need to set this so Python loads the dev version of 'conda', usually taken from conda/ in the root of the cloned repo. This root is usually the working directory when we run pytest. Otherwise, it will import the one installed in the base environment, which might have not been overwritten with pip install -e . --no-deps. This doesn't happen in other tests because they run with the equivalent of python -m conda. However, some tests directly run conda (shell function) which calls `conda (Python entry point). When a script is called this way, it bypasses the automatic "working directory is first on sys.path" behavior you find in python -m style calls. See https://docs.python.org/3/library/sys_path_init.html for details.

context_testdata() → None

class HttpTestServerFixture

Fixture providing HTTP test server for serving local files.

server: http.server.ThreadingHTTPServer

host: str

port: int

url: str

directory: pathlib.Path

__post_init__()

Log server startup for debugging.

get_url(path: str = '') → str

Get full URL for a given path on the server.

Parameters:

path -- Relative path on the server (e.g., "subdir/package.tar.bz2")

Returns:

Full URL

http_test_server(request: pytest.FixtureRequest, path_factory: PathFactoryFixture) → collections.abc.Iterator[HttpTestServerFixture]

Function-scoped HTTP test server for serving local files.

This fixture starts an HTTP server on a random port and serves files from a directory. The server supports both IPv4 and IPv6.

Usage without parametrize (dynamic content):

def test_dynamic(http_test_server):

# Server uses temporary directory automatically (http_test_server.directory / "file.txt").write_text("content") url = http_test_server.get_url("file.txt") response = requests.get(url) assert response.status_code == 200

Usage with parametrize (pre-existing directory):

@pytest.mark.parametrize("http_test_server", ["tests/data/mock-channel"], indirect=True) def test_existing(http_test_server):

url = http_test_server.get_url("file.txt") response = requests.get(url) assert response.status_code == 200

Use None in parametrize to mix pre-existing directories with dynamic content:

@pytest.mark.parametrize("http_test_server", ["tests/data", None], indirect=True)

Parameters:

• request -- pytest fixture request object

• path_factory -- path_factory fixture for creating unique temp directories

Returns:

HttpTestServerFixture with server, host, port, url, and directory attributes

Raises:

ValueError -- If parametrized directory is invalid