Python Development¶

Qip provide a few features to ease the development process.

Editable mode¶

Qip provides an editable mode to allow for rapid development when developing a Python package:

>>> qip install -e .

See also

Pip “Editable” installs

Let’s clone the popular Pystache repository to demonstrate this feature:

>>> git clone https://github.com/defunkt/pystache.git
>>> cd ./pystache
>>> qip install -e .

The pystache definition will ensure that the install-location value will target the cloned repository:

{
    "identifier": "pystache",
    "version": "0.5.4",
    "namespace": "library",
    "description": "Mustache for Python",
    "command": {
        "pystache": "python -m pystache.commands.render",
        "pystache-test": "python -m pystache.commands.test"
    },
    "environ": {
        "PYTHONPATH": "${INSTALL_LOCATION}:${PYTHONPATH}"
    },
    "variants": [
        {
            "identifier": "3.8",
            "install-location": "/path/to/pystache/",
            "requirements": [
                "python >= 3.8, < 3.9"
            ]
        }
    ]
}

It is then possible to execute the following command:

>>> wiz -add /tmp/qip/definitions run pystache -- 'Hello {{world}}' '{"world": "everybody"}'
info: Start command: python -m pystache.commands.render 'Hello {{world}}' '{"world": "everybody"}'
Hello everybody

You can modify the pystache/commands/render.py file to print a different output and ensure that it works as expected:

diff --git a/pystache/commands/render.py b/pystache/commands/render.py
index 1a9c309..9738c5e 100644
--- a/pystache/commands/render.py
+++ b/pystache/commands/render.py
@@ -88,7 +88,7 @@ def main(sys_argv=sys.argv):
         context = json.loads(context)

     rendered = renderer.render(template, context)
-    print rendered
+    print "test " + rendered

The command will then return:

>>> wiz -add /tmp/qip/definitions run pystache -- 'Hello {{world}}' '{"world": "everybody"}'
info: Start command: python -m pystache.commands.render 'Hello {{world}}' '{"world": "everybody"}'
test Hello everybody

Optional dependencies¶

Qip supports extra keywords provided in the request. Many Python packages use this installation feature, like Gunicorn.

>>> qip install gunicorn[gevent]

When using extra keywords, the nature of the package installed is fundamentally modified. Not only does it change the dependencies, but it could also change the Entry-Points created.

Custom Wiz definition¶

A default Wiz definition will be created for each Python packages to install. It will contain:

identifier based of the project name.
version based on the package version.
description based on the package description.
commands based on entry points defined in setup.py.
requirements based on package dependencies.
install-location based on relative library path.

It is possible to add a custom definition within the repository to extend this Wiz definition. The custom definition should be included in the source code under package_data/wiz.json.

Let’s use again the Pystache repository to demonstrate this feature:

>>> git clone https://github.com/defunkt/pystache.git
>>> cd ./pystache

Add the following definition in pystache/package_data/wiz.json

{
    "command": {
        "say_hello": "python -m pystache.commands.render 'Hello {{world}}' '{\"world\": \"everybody\"}'"
    }
}

Now install the definition as follows:

>>> qip install .

It is then possible to execute the following command:

>>> wiz -add /tmp/qip/definitions run say_hello
info: Start command: python -m pystache.commands.render 'Hello {{world}}' '{"world": "everybody"}'
Hello everybody

Using a custom definition could be particularly helpful when a Python package depends on a non-Python library.

Note

It is also possible to add optional custom definition matching extra keywords provided in the request. For instance, if a non-Python library must be set as a dependency when the “gevent” keyword is passed when installing the Gunicorn library, a targeted definition can be set in gunicorn/package_data/wiz-gevent.json.

Working with DCCs¶

When writing a Python plugin for a Digital content creation tool, a custom Wiz definition should be used to ease the development and deployment process.

Here are a few usage examples:

Maya (Autodesk)

{
    "identifier": "foo",
    "namespace": "maya",
    "environ": {
        "MAYA_PLUG_IN_PATH": "${INSTALL_LOCATION}/foo/package_data/maya/plugin:${MAYA_PLUG_IN_PATH}",
        "MAYA_SCRIPT_PATH": "${INSTALL_LOCATION}/foo/package_data/maya/script/mel:${MAYA_SCRIPT_PATH}",
        "PYTHONPATH": "${INSTALL_LOCATION}/foo/package_data/maya/script/python:${PYTHONPATH}"
    },
    "requirements": [
        "maya"
    ]
}

Flame (Autodesk)

{
    "identifier": "foo",
    "namespace": "flame",
    "environ": {
        "DL_PYTHON_HOOK_PATH": "${INSTALL_LOCATION}/foo/package_data/python:${DL_PYTHON_HOOK_PATH}"
    },
    "requirements": [
        "flame"
    ]
}

Nuke (Foundry)

{
    "identifier": "foo",
    "namespace": "nuke",
    "environ": {
        "NUKE_PATH": "${INSTALL_LOCATION}/foo/package_data/nuke:${NUKE_PATH}"
    },
    "requirements": [
        "nuke"
    ]
}

Houdini (SideFX)

{
    "identifier": "foo",
    "namespace": "houdini",
    "environ": {
        "HOUDINI_PATH": "${INSTALL_LOCATION}/foo/package_data/houdini:${HOUDINI_PATH}"
    },
    "requirements": [
        "houdini"
    ]
}

RV (Shotgun)

{
    "identifier": "foo",
    "namespace": "rv",
    "environ": {
        "RV_SUPPORT_PATH": "${INSTALL_LOCATION}/foo/package_data/rv:${RV_SUPPORT_PATH}"
    },
    "requirements": [
        "rv"
    ]
}