tach mod

Tach comes bundled with a command to set up your initial boundaries - tach mod.

usage: tach mod [-h] [-d [DEPTH]] [-e file_or_path,...]

Configure module boundaries interactively

options:
  -h, --help            show this help message and exit
  -d [DEPTH], --depth [DEPTH]
                        The number of child directories to expand from the root
  -e file_or_path,..., --exclude file_or_path,...
                        Comma separated path list to exclude. tests/, ci/, etc.

Running tach mod will open an editor in your terminal where you can mark your module boundaries.

You can navigate with the arrow keys, mark individual modules with Enter, and mark all siblings as modules with Ctrl + a.

You can also mark your Python source roots by pressing s. This allows Tach to understand module paths and correctly identify first-party imports.

To save your modules, use Ctrl + s. Otherwise, to exit without saving, use Ctrl + c.

Any time you make changes with tach mod, run tach sync to automatically configure dependency rules.

tach sync

Tach can automatically sync your project configuration (tach.toml) with your project’s actual dependencies.

usage: tach sync [-h] [--add] [-e file_or_path,...]

Sync constraints with actual dependencies in your project.

options:
  -h, --help            show this help message and exit
  --add                 add all existing constraints and re-sync dependencies.
  -e file_or_path,..., --exclude file_or_path,...
                        Comma separated path list to exclude. tests/, ci/, etc.

When this command runs, Tach will analyze the imports in your Python project.

Any undeclared dependencies will be automatically resolved by adding the corresponding dependencies to your tach.toml file.

With --add, any missing dependencies in your tach.toml will be added, but does not remove unused dependencies.

When run without the --add flag, tach sync will remove modules from the tach.yml file that do not exist in the project’s source roots.

tach check

Tach will flag any unwanted imports between modules. We recommend you run tach check like a linter or test runner, e.g. in pre-commit hooks, on-save hooks, and in CI pipelines.

usage: tach check [-h] [--exact] [-e file_or_path,...]

Check existing boundaries against your dependencies and module interfaces

options:
  -h, --help            show this help message and exit
  --exact               Raise errors if any dependency constraints are unused.
  -e file_or_path,..., --exclude file_or_path,...
                        Comma separated path list to exclude. tests/, ci/, etc.

An error will indicate:

  • the file path in which the error was detected
  • the path associated with that file
  • the path associated with the attempted import

If --exact is provided, additional errors will be raised if a dependency exists in tach.toml that does not exist in the code.

Example:

> tach check
❌ tach/check.py[L8]: Cannot import 'tach.filesystem'. Module 'tach' cannot depend on 'tach.filesystem'.

NOTE: If your terminal supports hyperlinks, you can click on the failing file path to go directly to the error.

tach report

Tach can generate a report showing all the dependencies and usages of a given module.

usage: tach report [-h] [-d module_path,...] [--no-deps] [-u module_path,...] [--no-usages] [-e file_or_path,...] path

Create a report of dependencies and usages of the given path or directory.

positional arguments:
  path                  The path or directory path used to generate the report.

options:
  -h, --help            show this help message and exit
  -d module_path,..., --dependencies module_path,...
                        Comma separated module list of dependencies to include [includes everything by default]
  --no-deps             Do not include dependencies in the report.
  -u module_path,..., --usages module_path,...
                        Comma separated module list of usages to include [includes everything by default]
  --no-usages           Do not include usages in the report.
  -e file_or_path,..., --exclude file_or_path,...
                        Comma separated path list to exclude. tests/, ci/, etc.

This will generate a textual report showing the file and line number of each relevant import. By default, the report will include a section containing all external usages of a module, as well as all dependencies of the module. These can be toggled off with --no-usages and --no-deps respectively.

To filter the output, the -d and -u options allow specifying which module paths to include in the dependency section and usage section respectively.

tach show

Tach will generate a visual representation of your dependency graph!

usage: tach show [-h] [--web] [-o [OUT]]

Visualize the dependency graph of your project.

options:
  -h, --help            show this help message and exit
  --web                 Open your dependency graph in a remote web viewer.
  -o [OUT], --out [OUT]
                        Specify an output path for a locally generated module graph file.

These are the results of tach show --web on the Tach codebase itself: tach show

tach test

Tach also functions as an intelligent test runner.

usage: tach test [-h] [--base [BASE]] [--head [HEAD]] [--disable-cache] ...
Run tests on modules impacted by the current changes.
positional arguments:
  pytest_args      Arguments forwarded to pytest. Use '--' to separate
                   these arguments. Ex: 'tach test -- -v'
options:
  -h, --help       show this help message and exit
  --base [BASE]    The base commit to use when determining which modules
                   are impacted by changes. [default: 'main']
  --head [HEAD]    The head commit to use when determining which modules
                   are impacted by changes. [default: current filesystem]
  --disable-cache  Do not check cache for results, and
                   do not push results to cache.

Using pytest, running tach test will perform impact analysis on the changes between your current filesystem and your main branch to determine which test files need to be run. This can dramatically speed up your test suite in CI, particularly when you make a small change to a large codebase. This command also takes advantage of Tach’s computation cache.

tach check-external

Tach can validate that the external imports in your Python packages match your declared package dependencies in pyproject.toml.

usage: tach check-external [-h] [-e file_or_path,...]

Perform checks related to third-party dependencies

options:
  -h, --help  show this help message and exit
  -e file_or_path,..., --exclude file_or_path,...
                        Comma separated path list to exclude. tests/, ci/, etc.

For each pyproject.toml found from the project root, Tach will scan all imports in the associated Python source and compare them with the declared dependencies. Tach will report an error for any external import which is not satisfied by the declared dependencies - preventing your users from errors due to missing imports.

This is typically useful if you are developing more than one Python package from a single virtual environment. Although your local environment may contain the dependencies for all your packages, when an end-user installs each package they will only install the dependencies listed in the pyproject.toml.

This means that, although tests may pass in your shared environment, an invalid import can still cause errors at runtime for your users.

In case you would like to explicitly allow a certain external module, this can be configured in your tach.toml

It is recommended to run Tach within a virtual environment containing all of your dependencies across all packages. This is because Tach uses the distribution metadata to map module names like ‘git’ to their distributions (‘GitPython’).

tach report-external

Tach can determine what external packages are used in a given path in your project.

usage: tach report-external [-h] [--raw] path

Create a report of third-party dependencies.

positional arguments:
  path        The path or directory path used to generate the report.

options:
  -h, --help  show this help message and exit
  --raw       Print machine-readable raw output. Each line will contain a PEP 508 dependency string.

This is typically useful for ‘tree-shaking’ during a build process. For example, if you are building a deployable image with only a subset of your code, you would run:

> tach report-external --raw [source_path]
tomli-w
pydantic
stdlib-list
importlib-metadata
gitpython
prompt-toolkit
networkx
rich

to generate a minimal set of external dependencies for that source code.

It is recommended to run Tach within a virtual environment containing all of your dependencies across all packages. This is because Tach uses the distribution metadata to map module names like ‘git’ to their distributions (‘GitPython’).

tach install

Tach can be installed into your development workflow automatically as a pre-commit hook.

With pre-commit framework

If you use the pre-commit framework, you can add the following to your .pre-commit-hooks.yaml:

repos:
  - repo: https://github.com/gauge-sh/tach-pre-commit
    rev: v0.10.7 # change this to the latest tag!
    hooks:
      - id: tach

Note that you should specify the version you are using in the rev key.

Standard install

If you don’t already have pre-commit hooks set up, you can run:

tach install pre-commit

The command above will install tach check as a pre-commit hook, directly into .git/hooks/pre-commit.

If that file already exists, you will need to manually add tach check to your existing .git/hooks/pre-commit file.