A collection of components to handle virtual 3nvironments and external python packages from withing touchdesigner with a hands-off approach.

All of them are designed with the same philosophy and API, so to some degree they can be used interchangeable!

TD-Pip implements pip in a complete native way without any external dependencies. It uses TouchDesigners existing python-interpreter and does not download or install any addtitional third party software.

Mount()
# A contextmanager to setup sys.env-path and if not already done, init TD_PIP and make sure this happens in time.
# Example
with op("TD_PIP").Moun():
  import installedModule

MountPackage( moduleName:str, pipPackageName:str = '',additionalSettings = [] )
# A contextmanager that prepared a packages, i.E. it checks if the passed module is already installed and will do that if the test fails.
with op("TD_PIP").MounModule( "MyModule", pipPackageName = "MyModulesPackage" ):
  import MyModule

TestModule( moduleName:str, silent:bool = False) -> bool
# Tests if a odule is available for import. If silent is False, a messagebox will pop up.

ImportModule( moduleName:str, pipPackageName:str = '', additionalSettings:List[str]=[] ) -> Module
# Returns the module as a result. Good for inline or in paameters. In regular code, use ContextManager

Implements the UV-PackageManager https://github.com/astral-sh/uv

No manual isnatllation required but it does either use the gobal installed UV or downloads and prepares a UV-Installation on demand.

NOTE: This will init your projects as a UV-Project, creating an _init_.py and pyproject.toml where it saves all installed dependencies. It also creates a fitting VENV inside of your project folder.

This allows the use of UV easily from the outside, for example when deploying the project or for creating pre-start scripts in the init.py

EACH PACkAGE INSTALLED WILL BE ADDED TO THE PYPROJECT.TOML

Mount()
# A contextmanager to setup sys.env-path and if not already done, init TD_PIP and make sure this happens in time.
# Example
with op("TD_Uv").Moun():
  import installedModule

MountPackage( moduleName:str, pipPackageName:str = '',additionalSettings = [] )
# A contextmanager that prepared a packages, i.E. it checks if the passed module is already installed and will do that if the test fails.
with op("TD_Uv").MounModule( "MyModule", pipPackageName = "MyModulesPackage" ):
  import MyModule

TestModule( moduleName:str, silent:bool = False) -> bool
# Tests if a odule is available for import. If silent is False, a messagebox will pop up.

ImportModule( moduleName:str, pipPackageName:str = '', additionalSettings:List[str]=[] ) -> Module
# Returns the module as a result. Good for inline or in paameters. In regular code, use ContextManager

op("TD_Uv").ext.extTDuv.runUvCommand("add foobar")
# Allows running of uv commands. Actually only for internal use but you know what, go wild :=

Hands off installer for conda. Can be used to install and run subprocesses that require many dependencies. In genera the heaviest solution in this repo.

If Auto Setup is set to true, the comp will Setup the first time it is initialised. The setupprcoess

• Downloads the latest miniconda-version and unpacks it in to the specified folder.
• Creates a conda environment in the specified folder und the specified name.
• Creates a refference to the ENV-Sitepackagesfolder in the .vscode/settings.json

This process can take a while, so wait a moment.

To use the componnt in TD there is a pretty streight forward API. You ca use it in two ways: Launch subprocesses or install and import python modules.

To install a module if ot already existing use op("TD_Conda").InstallPackage( packageName, installer = "conda"|"pip") to check if a module is already installed, if not nstall it, use op("TD_Conda").PrepareModule( moduleName, packageName = moduleName, installer = "conda"|"pip")

To import the module you need to mount the env. This keeps TD_Conda from poluting other potentia env or TD_Internal modules. When passing clearModules = False, this will remove the already imported modules from the importCache, resulting in a fresh new import. This is reset after the contextmanager is left. This can bue usefull when working with modules that are relying on specifric version of numpy or openCV.

with op("TD_Conda").Mount():
  import moduleName

Example:

op("TD_Conda").PrepareModule("qrcode")
with op("TD_Conda").Mount():
  import qrcode

If you have a script that is terminated in the same action, you can use another contextManager.

with op("TD_Conda").EnvShell() as envShell:
  envShell.Execute("aimg videogen --start-image rocket.png")

If you want to run the process as a damon, use the run-method.

op("TD_Conda").InstallPackage("cuda")
op("TD_Conda").InstallPackage(
       "whisper-live", 
        installer = "pip")
whisperProcess = op("TD_Conda").Run(
        "run_server.py"
    )

or, going bare-metal, the spawnMethod

process = op("TD_Conda").SpawnEnvShell()
process.Write("conda install cuda")
process.Write("pip install whisper-live")
process.Write('pip install "numpy==1.26.4 --force')
process.Write("python run_server.py")

You can now invoke other commands using the execute function. (Not the best example, but whatever).