I have a lot of first party python packages in a monorepo and in their pyproject.toml files all have their versions set to 0.0.0 and the local deps are referenced using paths, i.e.

[tool.poetry]
name = "package-name"
version = "0.0.0"
description = ""
authors = [""]
packages = [
    { include = "package_name" },
]

[tool.poetry.dependencies]
python = ">=3.9,<3.10"
dep-a = { path = "../../packages/dep-a", develop = true }
dep-b = { path = "../../packages/dep-b", develop = true }

How can I do something like this in pants before packaging the dist? I think the answer is a custom plugin as described in the python_distribution docs but I want to be sure just in case that won’t work or is the wrong path.

import tomlkit

# Path to TOML file
file_path = "pyproject.toml"

# The version number to update the package and it's dependencies to
new_version = "1.2.3"

# Read the existing TOML file
with open(file_path, "r") as file:
    data = tomlkit.parse(file.read())

# Update the version in the [tool.poetry] section
data["tool"]["poetry"]["version"] = new_version

# Loop through each dependency in [tool.poetry.dependencies]
for dep, details in data["tool"]["poetry"]["dependencies"].items():
    # Check if this dependency has a 'path' property
    if isinstance(details, dict) and "path" in details:
        # Update this dependency to the new version, removing the 'path' and 'develop' keys
        data["tool"]["poetry"]["dependencies"][dep] = new_version

# Write the updated TOML data back to the file
with open(file_path, "w") as file:
    file.write(tomlkit.dumps(data))

I don't know the answer to the actual question you're asking... but, just checking: are these distributions published as libraries, or are they only published as executable artifacts (e.g. pex, docker images)? If the latter, I think the typical approach is to skip the `python_distribution`s and just let pants dependency inference pull individual files together, rather than wrap them up in distributions.

These are published as libraries and then slurped up and used in Databricks notebooks. Eventually we aim to use only docker container with PEX but I need to support both models during the transition.