Plugin for Poetry to enable dynamic versioning based on VCS tags





GitHub Stars



Last Commit

2mos ago










Dynamic versioning plugin for Poetry

Version License: MIT

This package is a plugin for Poetry to enable dynamic versioning based on tags in your version control system, powered by Dunamai. Many different version control systems are supported, including Git and Mercurial; please refer to the Dunamai page for the full list (and minimum supported version where applicable).

Since Poetry does not yet officially support plugins (refer to this issue) as of the time of writing on 2019-10-19, this package takes some novel liberties to make the functionality possible. As soon as official support lands, this plugin will be updated to do things the official way.


Python 3.5+ and Poetry 1.1.0+ or poetry-core 1.0.0+ are required.

  • Run pip install poetry-dynamic-versioning
  • Add this section to your pyproject.toml:
    enable = true
  • Include the plugin in the build-system section of pyproject.toml for interoperability with PEP 517 build frontends.
    • Example using poetry-core as the build system:
      requires = ["poetry-core>=1.0.0", "poetry-dynamic-versioning"]
      build-backend = "poetry.core.masonry.api"
    • Example using poetry as the build system:
      requires = ["poetry>=1.0.2", "poetry-dynamic-versioning"]
      build-backend = "poetry.masonry.api"

Note that you must install the plugin in your global Python installation, not as a dev-dependency in pyproject.toml, because the virtual environment that Poetry creates cannot see Poetry itself and therefore cannot patch it.

Poetry's typical version setting is still required in [tool.poetry], but you are encouraged to use version = "0.0.0" as a standard placeholder.

With the minimal configuration above, the plugin will automatically take effect when you run commands such as poetry build. It will update the version in pyproject.toml, then revert the change when the plugin deactivates. If you want to include a __version__ variable in your code, just put a placeholder in the appropriate file and configure the plugin to update it (see below) if it isn't one of the defaults. You are encouraged to use __version__ = "0.0.0" as a standard placeholder.


In your pyproject.toml file, you may configure the following options:

  • [tool.poetry-dynamic-versioning]: General options.

    • enable: Boolean. Default: false. Since the plugin has to be installed globally, this setting is an opt-in per project. This setting will likely be removed once plugins are officially supported.

    • vcs: String. This is the version control system to check for a version. One of: any (default), git, mercurial, darcs, bazaar, subversion, fossil.

    • metadata: Boolean. Default: unset. If true, include the commit hash in the version, and also include a dirty flag if dirty is true. If unset, metadata will only be included if you are on a commit without a version tag.

    • tagged-metadata: Boolean. Default: false. If true, include any tagged metadata discovered as the first part of the metadata segment. Has no effect when metadata is set to false.

    • dirty: Boolean. Default: false. If true, include a dirty flag in the metadata, indicating whether there are any uncommitted changes. Has no effect when metadata is set to false.

    • pattern: String. This is a regular expression which will be used to find a tag representing a version. There must be a capture group named base with the main part of the version. Optionally, it may contain another two groups named stage and revision for prereleases, and it may contain a group named tagged_metadata to be used with the tagged-metadata option.

      The default is:

      (?x)                                                (?# ignore whitespace)
      ^v(?P<base>\d+(\.\d+)*)                             (?# e.g., v1.2.3)
      (-?((?P<stage>[a-zA-Z]+)\.?(?P<revision>\d+)?))?    (?# e.g., beta-0)
      (\+(?P<tagged_metadata>.+))?$                       (?# e.g., +linux)

      Remember that the backslashes must be escaped (\\) in the TOML file.

    • format: String. Default: unset. This defines a custom output format for the version. Available substitutions:

      • {base}
      • {stage}
      • {revision}
      • {distance}
      • {commit}
      • {dirty}
      • {tagged_metadata}

      Example: v{base}+{distance}.{commit}

    • format-jinja: String. Default: unset. This defines a custom output format for the version, using a Jinja template. When this is set, format is ignored.

      Available variables:

      • base (string)
      • stage (string or None)
      • revision (integer or None)
      • distance (integer)
      • commit (string)
      • dirty (boolean)
      • tagged_metadata (string or None)
      • version (dunumai.Version)
      • env (dictionary of environment variables)

      Available functions:

      Simple example:

      format-jinja = "{% if distance == 0 %}{{ base }}{% else %}{{ base }}+{{ distance }}.{{ commit }}{% endif %}"

      Complex example:

      format-jinja = """
          {%- if distance == 0 -%}
              {{ serialize_pep440(base, stage, revision) }}
          {%- elif revision is not none -%}
              {{ serialize_pep440(base, stage, revision + 1, dev=distance, metadata=[commit]) }}
          {%- else -%}
              {{ serialize_pep440(bump_version(base), stage, revision, dev=distance, metadata=[commit]) }}
          {%- endif -%}
    • format-jinja-imports: Array of tables. Default: empty. This defines additional things to import and make available to the format-jinja template. Each table must contain a module key and may also contain an item key. Consider this example:

      format-jinja-imports = [
          { module = "foo" },
          { module = "bar", item = "baz" },

      This is roughly equivalent to:

      import foo
      from bar import baz

      foo and baz would then become available in the Jinja formatting.

    • style: String. Default: unset. One of: pep440, semver, pvp. These are preconfigured output formats. If you set both a style and a format, then the format will be validated against the style's rules. If style is unset, the default output format will follow PEP 440, but a custom format will only be validated if style is set explicitly.

    • latest-tag: Boolean. Default: false. If true, then only check the latest tag for a version, rather than looking through all the tags until a suitable one is found to match the pattern.

    • bump: Boolean. Default: false. If true, then increment the last part of the version base by 1, unless the stage is set, in which case increment the revision by 1 or set it to a default of 2 if there was no revision. Does nothing when on a commit with a version tag.

      Example, if there have been 3 commits since the v1.3.1 tag:

      • PEP 440 with bump = false: 1.3.1.post3.dev0+28c1684
      • PEP 440 with bump = true: 1.3.2.dev3+28c1684
  • [tool.poetry-dynamic-versioning.subversion]: Options specific to Subversion.

    • tag-dir: String. Default: tags. This is the location of tags relative to the root.
  • [tool.poetry-dynamic-versioning.substitution]: Insert the dynamic version into additional files other than just pyproject.toml. These changes will be reverted when the plugin deactivates.

    • files: List of globs for any files that need substitutions. Default: ["*.py", "*/__init__.py", "*/__version__.py", "*/_version.py"]. To disable substitution, set this to an empty list.

    • patterns: List of regular expressions for the text to replace. Each regular expression must have two capture groups, which are any text to preserve before and after the replaced text. Default: ["(^__version__\s*=\s*['\"])[^'\"]*(['\"])"].

      Remember that the backslashes must be escaped (\\) in the TOML file.

Simple example:

enable = true
vcs = "git"
style = "semver"

Command line mode

The plugin also has a command line mode for execution on demand. This mode applies the dynamic version to all relevant files and leaves the changes in-place, allowing you to inspect the result. Your configuration will be detected from pyproject.toml as normal, but the enable option is not necessary.

To activate this mode, run the poetry-dynamic-versioning command in your console.


  • The dynamic version is not available during poetry run because Poetry uses os.execvp().

  • Regarding PEP 517 support:

    pip wheel . will not work, because Pip creates an isolated copy of the source code, which does not contain the Git history and therefore cannot determine the dynamic version.

    If you want to build wheels of your dependencies, you can do the following, but it won't work with path dependencies for the same reason as above:

    poetry export -f requirements.txt -o requirements.txt --without-hashes
    pip wheel -r requirements.txt

    With newer versions of Pip, you can pass --use-feature=in-tree-build to make the pip wheel . and pip install . commands work.


In order to side-load plugin functionality into Poetry, this package does the following:

  • Upon installation, it delivers a zzz_poetry_dynamic_versioning.pth file to your Python site-packages directory. This forces Python to automatically load the plugin after all other modules have been loaded (or at least those alphabetically prior to zzz).
  • It first tries to patch poetry.factory.Factory.create_poetry and poetry.console.commands.run.RunCommand directly. If they cannot be imported, then it patches builtins.__import__ so that, whenever those classes are first imported, then they will be patched. The reason we may have to wait for these to be imported is in case you've used the get-poetry.py script, in which case there is a gap between when Python is fully loaded and when ~/.poetry/bin/poetry adds the Poetry lib folder to the PYTHONPATH.
  • The patched version of Factory will compute and apply the dynamic version.
  • The patched version of RunCommand will deactivate the plugin before executing the passed command, because otherwise we will not be able to do any cleanup afterwards.


Please refer to CONTRIBUTING.md.

Rate & Review

Great Documentation0
Easy to Use0
Highly Customizable0
Bleeding Edge0
Responsive Maintainers0
Poor Documentation0
Hard to Use0
Unwelcoming Community0