Parallel asyncio Python setup.(cfg|py) Test Runner

pip install ptr==19.2.12


🏃‍♀️ ptr - Python Test Runner 🏃‍♂️

Code style: black Build Status

Python Test Runner (ptr) was born to run tests in an opinionated way, within arbitrary code repositories. ptr supports many Python projects with unit tests defined in their setup.(cfg|py) files per repository. ptr allows developers to test multiple projects/modules in one Python environment through the use of a single test virtual environment.

  • ptr requires >= python 3.6
  • ptr itself uses ptr to run its tests 👌🏼

By adding ptr configuration to your setup.cfg or you can have ptr perform the following, per test suite, in parallel:

  • run your test suite
  • check and enforce coverage requirements (via coverage),
  • format code (via black)
  • perform static type analysis (via mypy)


  • Install ptr into you virtualenv
    • pip install ptr
  • Ensure your tests have a base file that can be executed directly
    • i.e. python3 (possibly using unittest.main())
  • After adding ptr_params to (see example below), run:
cd repo

How does ptr perform this magic? 🎩

I'm glad you ask. Under the covers ptr performs:

  • Recursively searches for setup.(cfg|py) files from BASE_DIR (defaults to your "current working directory" (CWD))
    • AST parses out the config for each test requirements
    • If a setup.cfg exists, load via configparser and prefer if a [ptr] section exists
  • Creates a Python Virtual Environment (OPTIONALLY pointed at an internal PyPI mirror)
  • Runs ATONCE tests suites in parallel (i.e. per setup.(cfg|ptr))
  • All steps will be run for each suite and ONLY FAILED runs will have output written to stdout

Usage 🤓

To use ptr all you need to do is cd to your project or set the base dir via -b and execute:

$ ptr [-dk] [-b some/path] [--venv /tmp/existing_venv]

For faster runs when testing, it is recommended to reuse a Virtual Environment:

  • -k - To keep the virtualenv created by ptr.
  • Use --venv VENV_PATH to reuse to an existing virtualenv created by the user.

Help Output 🙋‍♀️ 🙋‍♂️

usage: [-h] [-a ATONCE] [-b BASE_DIR] [-d] [-k] [-m MIRROR]
              [--progress-interval PROGRESS_INTERVAL] [--venv VENV]
optional arguments:
  -h, --help            show this help message and exit
  -a ATONCE, --atonce ATONCE
                        How many tests to run at once [Default: 24]
  -b BASE_DIR, --base-dir BASE_DIR
                        Path to recursively look for files [Default:
  -d, --debug           Verbose debug output
  -k, --keep-venv       Do not remove created venv
  -m MIRROR, --mirror MIRROR
                        URL for pip to use for Simple API [Default:
  --progress-interval PROGRESS_INTERVAL
                        Seconds between status update on test running
                        [Default: Disabled]
  --venv VENV           Path to venv to reuse

Configuration 🧰

ptr is configured by placing directives in one or more of the following files. .ptrconfig provides base configuration and default values for all projects in the repository, while each setup.(cfg|py) overrides the base configuration for the respective packages they define.


ptr supports a general config in ini (ConfigParser) format. A .ptrconfig file can be placed at the root of any repository or in any directory within your repository. The first .ptrconfig file found via a recursive walk to the root ("/" in POSIX systems) will be used.

Please refer to ptrconfig.sample for the options available.

This is per project in your repository. A simple example, based on ptr itself:

# Specific Python Test Runner (ptr) params for Unit Testing Enforcement
ptr_params = {
    # Where mypy will run to type check your program
    "entry_point_module": "ptr",
    # Base Unittest file
    "test_suite": "ptr_tests",
    "test_suite_timeout": 300,
    # Relative path from to module (e.g. ptr ==
    "required_coverage": {"": 99, "TOTAL": 99},
    # Run `black --check` or not
    "run_black": False,
    # Run mypy or not
    "run_mypy": True,


This is per project in your repository and if exists is preferred over

Please refer to setup.cfg.sample for the options available + format.

mypy Specifics

When enabled, (in setup.(cfg|py)) mypy can support using a custom mypy.ini for each (module) defined.

To have ptr run mypy using you config:

  • create a mypy.ini in the same directory as your
  • OR add [mypy] section to your setup.cfg

mypy Configuration Documentation can be found here

python_version = 3.5
check_untyped_defs = True
disallow_incomplete_defs = True
disallow_untyped_decorators = True
disallow_untyped_defs = True
no_implicit_optional = True
show_error_context = True
warn_unused_ignores = True
warn_return_any = True

Example Output 📝

Here are some example runs.

Successful ptr Run:

Here is what you want to see in your CI logs!

[2019-02-06 21:51:45,442] INFO: Starting (
[2019-02-06 21:51:59,471] INFO: Successfully created venv @ /var/folders/tc/hbwxh76j1hn6gqjd2n2sjn4j9k1glp/T/ptr_venv_24397 to run tests (14s) (
[2019-02-06 21:51:59,472] INFO: Installing /Users/cooper/repos/ptr/ + deps (
[2019-02-06 21:52:00,726] INFO: Running /Users/cooper/repos/ptr/ tests via coverage (
[2019-02-06 21:52:04,153] INFO: Analyzing coverage report for /Users/cooper/repos/ptr/ (
[2019-02-06 21:52:04,368] INFO: Running mypy for /Users/cooper/repos/ptr/ (
[2019-02-06 21:52:07,733] INFO: /Users/cooper/repos/ptr/ has passed all configured tests (
-- Summary (total time 22s):

✅ PASS: 1
❌ FAIL: 0
💩 TOTAL: 1

-- 1 / 1 (100%) ``'s have `ptr` tests running

Unsuccessful ptr Run Examples:

Here are some examples of runs failing. Any "step" can fail. All output is predominately the underlying tool.

Unit Test Failure

[2019-02-06 21:53:58,121] INFO: Starting (
[2019-02-06 21:53:58,143] INFO: Installing /Users/cooper/repos/ptr/ + deps (
[2019-02-06 21:53:59,698] INFO: Running /Users/cooper/repos/ptr/ tests via coverage (
-- Summary (total time 5s):

✅ PASS: 0
❌ FAIL: 1
💩 TOTAL: 1

-- 1 / 1 (100%) ``'s have `ptr` tests running

-- Failure Output --

FAIL: test_config (__main__.TestPtr)
Traceback (most recent call last):
  File "/Users/cooper/repos/ptr/", line 125, in test_config
    self.assertEqual(len(sc["ptr"]["venv_pkgs"].split()), 4)
AssertionError: 5 != 4

Ran 24 tests in 3.221s

FAILED (failures=1)

Coverage Failure

[2019-02-06 21:55:42,947] INFO: Starting (
[2019-02-06 21:55:42,969] INFO: Installing /Users/cooper/repos/ptr/ + deps (
[2019-02-06 21:55:44,920] INFO: Running /Users/cooper/repos/ptr/ tests via coverage (
[2019-02-06 21:55:49,628] INFO: Analyzing coverage report for /Users/cooper/repos/ptr/ (
-- Summary (total time 7s):

✅ PASS: 0
❌ FAIL: 1
💩 TOTAL: 1

-- 1 / 1 (100%) ``'s have `ptr` tests running

-- Failure Output --

The following files did not meet coverage requirements: 84 < 99 - Missing: 146-147, 175, 209, 245, 269, 288-291, 334-336, 414-415, 425-446, 466, 497, 506, 541-543, 562, 611-614, 639-688


[2019-02-06 22:34:20,029] INFO: Starting (
[2019-02-06 22:34:20,060] INFO: Installing /Users/cooper/repos/ptr/ + deps (
[2019-02-06 22:34:21,614] INFO: Running /Users/cooper/repos/ptr/ tests via coverage (
[2019-02-06 22:34:25,208] INFO: Analyzing coverage report for /Users/cooper/repos/ptr/ (
[2019-02-06 22:34:25,450] INFO: Running mypy for /Users/cooper/repos/ptr/ (
[2019-02-06 22:34:26,422] INFO: Running black for /Users/cooper/repos/ptr/ (
-- Summary (total time 7s):

✅ PASS: 0
❌ FAIL: 1
💩 TOTAL: 1

-- 1 / 1 (100%) ``'s have `ptr` tests running

-- Failure Output --

would reformat /Users/cooper/repos/ptr/
All done! 💥 💔 💥
1 file would be reformatted, 4 files would be left unchanged.


[2019-02-06 22:35:39,480] INFO: Starting (
[2019-02-06 22:35:39,531] INFO: Installing /Users/cooper/repos/ptr/ + deps (
[2019-02-06 22:35:41,203] INFO: Running /Users/cooper/repos/ptr/ tests via coverage (
[2019-02-06 22:35:45,156] INFO: Analyzing coverage report for /Users/cooper/repos/ptr/ (
[2019-02-06 22:35:45,413] INFO: Running mypy for /Users/cooper/repos/ptr/ (
-- Summary (total time 6s):

✅ PASS: 0
❌ FAIL: 1
💩 TOTAL: 1

-- 1 / 1 (100%) ``'s have `ptr` tests running

-- Failure Output --

/Users/cooper/repos/ptr/ note: In function "_write_stats_file":
/Users/cooper/repos/ptr/ error: Argument 1 to "open" has incompatible type "Path"; expected "Union[str, bytes, int]"
/Users/cooper/repos/ptr/ note: In function "run_tests":
/Users/cooper/repos/ptr/ error: Argument 1 to "_write_stats_file" has incompatible type "str"; expected "Path"

FAQ ⁉️

Q. How do I debug? I need output!

  • ptr developers recommend that if you want output, please cause a test to fail
    • e.g. raise ZeroDivisionError
  • Another recommended way is to run your tests with the default test using a ptr created venv:
    • cd to/my/code
    • /tmp/venv/bin/python test

Q. How do I get specific version of black, coverage, mypy etc.?

Q. Why is the venv creation so slow?

  • ptr attempts to update from a PyPI compatible mirror (PEP 381) or PyPI itself
  • Running a package cache or local mirror can greatly increase speed. Example software to do this:
    • bandersnatch: Can do selected or FULL PyPI mirrors. The maintainer is also devilishly good looking.
    • devpi: Can be ran and used to proxy packages locally when pip goes out to grab your dependencies.
  • Please ensure you're using the -k or --venv option to no recreate a virtualenv each run when debugging your tests!

Contact or join the ptr community 💬

To chat in real time, hit us up on IRC. Otherwise, GitHub issues are always welcome! IRC: #pythontestrunner on FreeNode

See the CONTRIBUTING file for how to help out.


ptr is MIT licensed, as found in the LICENSE file.