• Stars
    star
    346
  • Rank 117,854 (Top 3 %)
  • Language
    Python
  • License
    MIT License
  • Created almost 3 years ago
  • Updated about 1 month ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Poetry plugin for dynamically extracting the package version from a __version__ variable or a Git tag.

Poetry Version Plugin

Test Publish Coverage Package version

A Poetry plugin for dynamically extracting the package version.

It can read the version from a file __init__.py with:

# __init__.py

__version__ = "0.1.0"

Alternatively, it can read it from a git tag, set with a GitHub release or with:

$ git tag 0.1.0

🚨 Consider this in the alpha stage. Read the warning below.

When to use

This is useful mainly if you are building a package library for others to use and you want to set the version in a place different than pyproject.toml, but you still want to keep a single source of truth.

It won't be helpful in other use cases like managing local app environments with Poetry.

Alternatives

If you are building a package library and want this functionality but you don't really need anything else from Poetry you are probably better off using Hatch, PDM, or another alternative that comes with this functionality built in without requiring plugins.

How to use

Make sure you have Poetry version 1.2.0 or above. Read below for instructions to install it if you haven't.

Install Poetry Version Plugin

Install this plugin to your Poetry:

$ poetry self add poetry-version-plugin

--> 100%

Set version in init file

Set your package version in your file __init__.py, for example:

from .main import do_awesome_stuff, AwesomeClass

__version__ = "0.2.3"

And then edit your pyproject.toml with a section containing:

[tool.poetry-version-plugin]
source = "init"

Next, build your project. It will show an output like:

$ poetry build
Using __init__.py file at my_awesome_package/__init__.py for dynamic version
Setting package dynamic version to __version__ variable from __init__.py: 0.1.9
Building my-awesome-package (0.1.9)
  - Building sdist
  - Built my-awesome-package-0.1.9.tar.gz
  - Building wheel
  - Built my-awesome-package-0.1.9-py3-none-any.whl

Set the version in a Git tag

Alternatively, to extract the version to use from a Git tag, add a section:

[tool.poetry-version-plugin]
source = "git-tag"

Then create a git tag, for example:

$ git tag 0.1.3

In this case, when building your project, it will show an output like:

$ poetry build
Git tag found, setting dynamic version to: 0.1.3
Building my-awesome-package (0.1.3)
  - Building sdist
  - Built my-awesome-package-0.1.3.tar.gz
  - Building wheel
  - Built my-awesome-package-0.1.3-py3-none-any.whl

Version in pyproject.toml

Currently (2021-05-24) Poetry requires a version configuration in the pyproject.toml, even if you use this plugin.

When using this plugin, that version config won't be used, but Poetry still requires it to be present in the pyproject.toml.

To make it more evident that you are not using that version you can set it to 0.

[tool.poetry]
name = "my-awesome-package"
version = "0"

That way, you will more easily notice if the plugin is not installed, as it will show that you are building a package with version 0 instead of the dynamic version set.

An example pyproject.toml

A short, minimal example pyproject.toml could look like:

[tool.poetry]
name = "my-awesome-package"
version = "0"
description = ""
authors = ["Rick Sanchez <[email protected]>"]
readme = "README.md"

[tool.poetry.dependencies]
python = "^3.7"

[build-system]
requires = ["poetry-core"]
build-backend = "poetry.core.masonry.api"

[tool.poetry-version-plugin]
source = "init"

Why

By default, Poetry expects you to set your package version in pyproject.toml. And that would work in most cases.

But imagine you want to expose the version of your package in a __version__ variable so that your users can do things like:

import my_awesome_package
print(my_awesome_package.__version__)

You could manually write the __version__ variable and handle the synchronization between it and the pyproject.toml yourself, which is very error-prone.

The current official way of doing it without duplicating the value is with importlib.metadata.

But that module is only available in Python 3.8 and above. So, for Python 3.7 you have to install a backport as a dependency of your package:

[tool.poetry.dependencies]
importlib-metadata = {version = "^1.0", python = "<3.8"}

But then, when they release each new version of the backport (currently 4.0.1), you have to update it (or not). And your users would have to manually handle conflicts with any other packages that also depend on importlib-metadata which could be multiple, as many packages could be doing the same trick (I've dealt with that).

The other option is not to pin any version range of your importlib-metadata in your pyproject.toml and hope for the best.

And then your __init__.py would have to include code using it, like:

# I don't want this extra complexity 😔
# And it doesn't work in Docker 🐋
try:
    import importlib.metadata as importlib_metadata
except ModuleNotFoundError:
    import importlib_metadata

__version__ = importlib_metadata.version(__name__)

But that code is extra complexity and logic needed in your code, in each of your packages.

🚨 Additionally, this only works when your package is installed in a Python environment. It won't work if, for example, you simply put your code in a container, which is common for web apps and distributed systems.

How this plugin solves it

With this plugin, your package doesn't depend on importlib-metadata, so your users won't need to handle conflicts or extra dependencies.

Instead, your build system (Poetry) is what needs to have this plugin installed.

That avoids the extra code complexity on your side, dependency conflicts for your users, and support for other use cases like code copied directly inside a container.

Version from Git tag

Alternatively, this plugin can also extract the version from a Git tag.

So, you could only create each version in a Git tag (for example, a GitHub release) instead of writing it in code.

And then build the package on Continuous Integration (e.g. GitHub Actions). And this plugin would get the version of the package from that Git tag.

Install Poetry 1.2.0

For this plugin to work, you need Poetry version 1.2.0 or above.

Poetry 1.2.0 was released recently.

There's a high chance you already have installed Poetry 1.1.x.

The first step is to uninstall it:

$ curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py -O
--> 100%

$ python get-poetry.py --uninstall
--> 100%

And then install the new Poetry with the new installer:

$ curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/install-poetry.py -O
--> 100%

$ python install-poetry.py --preview
--> 100%

🔍 Notice that the new installer file is named install-poetry.py instead of get-poetry.py. Also, notice that, currently, you need to set --preview for it to install the alpha version 1.2.0.

You can check that it worked with:

$ poetry --version
Poetry (version 1.2.0)

Support for version in init file

When using a __version__ variable in your __init__.py you can have more logic in that file, import modules, and do more things above and below the declaration of that variable.

But the value has to be a literal string, like:

___version___ = "0.2.0"

...instead of calling a function or something similar.

And the variable has to be at the top-level, so it can't be inside an if statement or similar.

This is all fine and supported in your __init__.py:

# __init__.py

# This is all valid 👍✅

from .main import do_awesome_stuff, AwesomeClass

awesome = AwesomeClass()

# Some comment explaining why this is commented out
# __version__ = "1.0.0"

__version__ = "0.2.3"

if __name__ == "__main__":
    awesome.run()

This example is all valid and supported, and it includes:

  • Imports
  • Other objects and variables
  • Comments
  • The same string __version__ inside a comment
  • If blocks around

But this is not supported:

# 🚨 Not supported

if 2 == 2:
    __version__ = "0.1.0

And this is not supported:

# 🚨 Not supported

def get_version():
    return "0.2.0"

__version__ = get_version()

How the plugin works

Poetry runs the plugin when building a package, and it sets the version right before creating the "package distributable" (e.g., the wheel).

How the version variable works

If you have a package (a single package) declared in the packages config in your pyproject.toml, the plugin will use that package's __init__.py to find the __version__ variable.

If you don't have any packages config, the plugin will assume that you have a single package named as your project, but in the module version (changing - for _). So, if your package is my-awesome-project, the plugin will use the file at my_awesome_project/__init__.py to find the __version__ variable.

This file structure is the default if you create a new project with the command poetry new, so it should work as expected. ✨

The way the plugin works internally is by parsing the __init__.py file. Reading the Python's "Abstract Syntax Tree" using the ast standard module and extracting the literal value of the string. So, it doesn't execute the code in __init__.py, it only reads it as Python code.

The plugin doesn't try to import and execute that __init__.py file because that could require extra computation, external dependencies, etc. And it doesn't try to extract the __version__ with regular expressions, as that would be prone to errors if, for example, there was some other __version__ somewhere in the code, in a comment or inside a string.

Warning

🚨 Consider this in the alpha stage. Poetry 1.2.0a1 with support for plugins was released on 2021-05-21. I started writing this plugin 3 days later, on 2021-05-24.

Things might break in Poetry or in this plugin. So, please try it and test it very carefully before fully adopting it for delicate systems.

The way it works might change, and the specific configuration might change.

Also, if you don't find the following sections intuitive:

[tool.poetry-version-plugin]
source = "init"

and

[tool.poetry-version-plugin]
source = "git-tag"

let me know what alternative configuration would make more sense and be more intuitive to you.

👍 The good news is, assuming you are building packages to then upload them to PyPI for your users to download and use them, the worst that could happen if something broke is that you wouldn't be able to build a new version until something is fixed or changed. But your users shouldn't be affected in any way.

Release Notes

Latest Changes

  • 📝 Update README, update install command, recommend Hatch and PDM. PR #31 by @tiangolo.

0.2.0

  • ✨ Add support for Poetry 1.2.0 and above (including 1.5.1), deprecate support for Python 3.6. PR #28 by @mbeacom.
  • ✏️ Fix typos and rewording in README.md. PR #8 by @Gl0deanR.

0.1.3

License

This project is licensed under the terms of the MIT license.

More Repositories

1

fastapi

FastAPI framework, high performance, easy to learn, fast to code, ready for production
Python
69,216
star
2

full-stack-fastapi-postgresql

Full stack, modern web application generator. Using FastAPI, PostgreSQL as database, Docker, automatic HTTPS and more.
TypeScript
14,514
star
3

typer

Typer, build great CLIs. Easy to code. Based on Python type hints.
Python
13,112
star
4

sqlmodel

SQL databases in Python, designed for simplicity, compatibility, and robustness.
Python
12,169
star
5

uwsgi-nginx-flask-docker

Docker image with uWSGI and Nginx for Flask applications in Python running in a single container. Optionally with Alpine Linux.
Python
2,926
star
6

uvicorn-gunicorn-fastapi-docker

Docker image with Uvicorn managed by Gunicorn for high-performance FastAPI web applications in Python with performance auto-tuning. Optionally with Alpine Linux.
Python
2,468
star
7

asyncer

Asyncer, async and await, focused on developer experience.
Python
1,329
star
8

pydantic-sqlalchemy

Tools to convert SQLAlchemy models to Pydantic models
Python
1,075
star
9

dockerswarm.rocks

Docker Swarm mode rocks! Ideas, tools and recipes. Get a production-ready, distributed, HTTPS served, cluster in minutes, not weeks.
Shell
1,054
star
10

nginx-rtmp-docker

Docker image with Nginx using the nginx-rtmp-module module for live multimedia (video) streaming.
Dockerfile
952
star
11

uwsgi-nginx-docker

Docker image with uWSGI and Nginx for applications in Python (as Flask) in a single container. Optionally with Alpine Linux.
Python
627
star
12

uvicorn-gunicorn-docker

Docker image with Uvicorn managed by Gunicorn for high-performance web applications in Python with performance auto-tuning. Optionally with Alpine Linux.
Python
580
star
13

full-stack

Full stack, modern web application generator. Using Flask, PostgreSQL DB, Docker, Swagger, automatic HTTPS and more.
Python
520
star
14

meinheld-gunicorn-flask-docker

Docker image with Meinheld and Gunicorn for Flask applications in Python. Optionally with Alpine Linux.
Python
479
star
15

full-stack-fastapi-couchbase

Full stack, modern web application generator. Using FastAPI, Couchbase as database, Docker, automatic HTTPS and more.
Python
424
star
16

typer-cli

Run Typer scripts with completion, without having to create a package, using Typer CLI.
Python
349
star
17

blog-posts

Blog posts and related code by Sebastián Ramírez (@tiangolo)
Python
261
star
18

babun-docker

Use Docker Toolbox with Babun (Cygwin) in Windows
Shell
171
star
19

docker-with-compose

Docker image with Docker Compose installed for CI.
Shell
159
star
20

meinheld-gunicorn-docker

Docker image with Meinheld managed by Gunicorn for high-performance WSGI (Flask, Django, etc) web applications in Python with performance auto-tuning. Optionally with Alpine Linux.
Python
157
star
21

uvicorn-gunicorn-starlette-docker

Docker image with Uvicorn managed by Gunicorn for high-performance Starlette web applications in Python with performance auto-tuning. Optionally with Alpine Linux.
Python
155
star
22

node-frontend

Node.js Docker image with all Puppeteer dependencies installed for frontend Chrome Headless testing and default Nginx config, for multi-stage Docker building
Dockerfile
134
star
23

latest-changes

A GitHub Action to add latest changes after each PR merged automatically
Python
132
star
24

flask-frontend-docker

Minimal project generator with a Flask backend, a modern frontend (Vue, React or Angular), a Traefik load balancer with HTTPS, all based on Docker.
Vue
131
star
25

python-machine-learning-docker

Docker image with Python 3.6 and 3.7 using Conda, with CUDA variants. To serve as base image for Machine Learning projects.
Dockerfile
80
star
26

uvicorn-gunicorn-machine-learning-docker

Docker image for high-performance Machine Learning web applications. With Uvicorn managed by Gunicorn in Python 3.7 and 3.6, using Conda, with CUDA and TensorFlow variants.
Python
64
star
27

full-stack-flask-couchbase

Full stack, modern web application generator. Using Flask, Couchbase as database, Docker, Swagger, automatic HTTPS and more.
Python
57
star
28

issue-manager

Automatically close issues that have a label, after a custom delay, if no one replies back.
Python
54
star
29

full-stack-flask-couchdb

Full stack, modern web application generator. Using Flask, CouchDB as database, Docker, Swagger, automatic HTTPS and more.
Python
30
star
30

label-approved

Label a Pull Request after a number of approvals
Python
21
star
31

angular-docker-multi-stage-example

Angular in Docker with Nginx, supporting environments, built with multi-stage Docker builds
16
star
32

github-actions-sandbox

Not useful for you. It's just a sandbox GitHub repo for me to try out stuff and develop GitHub Actions.
Python
15
star
33

docker-auto-labels

Generate each Docker constraint label in random nodes in the cluster.
Python
14
star
34

tiangolo

12
star
35

compose-to-rancher

Convert Docker Compose V2 to Rancher compatible Docker Compose V1
Python
10
star
36

ngx-http-client

Angular (4.3+) HttpClientModule with parameter encodings compatible with back ends (Node.js, Python, PHP, etc)
TypeScript
10
star
37

tiangolo.com

TypeScript
7
star
38

wunderlist2csv

Convert from Wunderlist backup json file to a CSV file importable by TaskCoach
Python
6
star
39

anaconda_cluster_install

Automatically Install Anaconda Python in a cluster of machines, for a specified user.
Shell
5
star
40

bitbucket_issues_to_redmine_csv

Python
3
star