Skip to content

Latest commit

 

History

History
204 lines (141 loc) · 6.19 KB

README.md

File metadata and controls

204 lines (141 loc) · 6.19 KB
Raw

packse

Python packaging scenarios.

Installation

Only a local installation is supported at this time:

poetry install

Once installed, the packse command-line interface will be available.

Depending on your Poetry configuration, you may need to use poetry run packse instead or activate Poetry's virtual environment.

Usage

Scenarios

A scenario is a JSON description of a dependency tree.

See scenarios/examples/

Each scenario file can contain one or more scenarios.

Listing scenarios

A list of available scenarios can be printed with the list command:

packse list

By default, packse will search for scenarios in the current tree. You may also pass a file to read from:

packse list scenarios/examples/example.json

Each scenario will be listed with its unique identifier e.g. example-cd797223. This is the name of the package that can be installed to run the scenario once it is built and published.

Each packse command supports reading multiple scenario files. For example, with list:

packse list scenarios/examples/example.json scenarios/requires-does-not-exist.json

Viewing scenarios

The dependency tree of a scenario can be previewed using the view command:

$ packse view scenarios/examples/example.json
example-89cac9f1
├── root
│   └── requires a
│       └── satisfied by a-1.0.0
├── a
│   └── a-1.0.0
│       └── requires b>1.0.0
│           ├── satisfied by b-2.0.0
│           └── satisfied by b-3.0.0
└── b
    ├── b-1.0.0
    ├── b-2.0.0
    │   └── requires c
    │       └── unsatisfied: no versions for package
    └── b-3.0.0

Note the view command will view all scenarios in a file by default. A single scenario can be viewed by providing the --name option:

$ packse view scenarios/examples/example.json --name example
example

This is an example scenario, in which the user depends on a single package `a` which requires `b`

example-89cac9f1
├── root
│   └── requires a
│       └── satisfied by a-1.0.0
├── a
│   └── a-1.0.0
│       └── requires b>1.0.0
│           ├── satisfied by b-2.0.0
│           └── satisfied by b-3.0.0
└── b
    ├── b-1.0.0
    ├── b-2.0.0
    │   └── requires c
    │       └── unsatisfied: no versions for package
    └── b-3.0.0

Notice, when a specific scenario is specified, there is more information displayed.

Building scenarios

A scenario can be used to generate packages and build distributions:

packse build scenario/example.json

The build/ directory will contain sources for all of the packages in the scenario. The dist/ directory will contain built distributions for all of the packages in the scenario.

When a scenario is built, it is given a unique identifier based on a hash of the scenario contents and the project templates used to generate the packages. Each package and requirement in the scenario will be prefixed with the identifier. The unique identifier can be excluded using the --no-hash argument, however, this will prevent publishing multiple times to a registry that does not allow overwrites.

The PACKSE_VERSION_SEED environment variable can be used to override the seed used to generate the unique identifier, allowing versions to differ based on information outside of packse.

Publishing scenarios

Built scenarios can be published to a Python Package Index with the publish command:

packse publish dist/example-cd797223

By default, packages are published to the Test PyPI server.

Credentials must be provided via the PACKSE_PYPI_PASSWORD environment variable. PACKSE_PYPI_USERNAME can be used to set a username if not using an API token. If using a server which does not require authentication, the --anonymous flag can be passed.

Running a package index

Requires installation with the index extra

A local package index can be controlled with the index command. For example, to start a local package index:

packse index up

The --bg flag can be passed to run the index in the background. When running an index in the background, state will be stored in the ~/.packse directory. The PACKSE_STATE_PATH environment variable can be used to change the state path.

Packages can be installed by passing the --index-url flag to the installer e.g. with pip:

pip install --index-url http://127.0.0.1:3141 example-0611cb74

Packages can be published to the index by providing the --index-url and --anonymous flags to the publish command:

packse publish dist/example-cd797223 --index-url http://localhost:3141 --anonymous

Packages can also be published to the index by placing their distributions in the configured --dist-dir. This defaults to ./dist which is also the default location used in packse build.

By default, the index will fallback to PyPI for missing packages. To test in isolation, use the --offline flag.

To stop an index running in the background use packse index down:

packse index down

Serving scenarios

Requires installation with the serve extra

The serve command can be used to host, build, and publish scenarios in one step.

packse serve scenarios

Packse will watch for changes to the scenarios directory, and publish new versions on each change.

Note when developing, it is often useful to use the --no-hash flag to avoid having to determine the latest hash for the scenario.

Testing scenarios

Published scenarios can then be tested with your choice of package manager.

For example, with pip:

pip install -i https://test.pypi.org/simple/ example-cd797223

Exporting scenarios

Scenario information can be exported with the packse inspect. This creates a JSON representation of the scenarios with additional generated information such as the root package name and the tree displayed during packse view.

Writing new scenarios

Scenario files may contain one or more scenarios written in JSON. See the existing scenarios for examples and the Scenario type for details on the supported schema.