Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Crane04/esem/llms.txt

Use this file to discover all available pages before exploring further.

The python() helper accepts either a file path to a local .py file or the name of a pip-installed package. When you pass a package name, the worker calls importlib.import_module() internally — the same mechanism Python itself uses for all standard imports. This means any package that works in a regular Python script will work through the bridge.

​
Importing an installed package

Pass the package name as a string instead of a file path: 
import { python } from "esem-bridge";

const np = await python("numpy");
The package must already be installed in the Python interpreter that esem-bridge is using. If it is not installed, Python will raise a ModuleNotFoundError which crosses the bridge as a PythonError.

​
Using a virtualenv

By default the bridge spawns python3 from your system PATH. Set the ESEM_PYTHON environment variable to point to any Python binary — including one inside a virtual environment — and the bridge will use that interpreter instead: 
python -m venv .venv
.venv/bin/pip install numpy scikit-learn
ESEM_PYTHON=.venv/bin/python node index.js

​
Using a conda environment

The same ESEM_PYTHON variable works for conda environments: 
ESEM_PYTHON=/opt/conda/envs/myenv/bin/python node index.js

​
How module loading works

The Python worker decides how to load a module spec based on simple path detection:
  • If the spec starts with ./ or /, or ends with .py → loaded as a file path using importlib.util.spec_from_file_location
  • Otherwise → loaded as an installed package using importlib.import_module()
This means "./tools.py", "/abs/path/tools.py", and "tools.py" are all treated as file paths, while "numpy", "sklearn", and "pandas" are treated as package names.
Set ESEM_PYTHON in a .env file or in your deployment environment’s configuration (e.g. Vercel environment variables, a Docker ENV instruction, or a systemd service file). This way you never need to prefix individual node commands, and the correct interpreter is always used regardless of how the process is started.
Packages with C extensions or native compiled code — such as numpy, pandas, scikit-learn, or cryptography — work fine through the bridge. However, the package must be installed for the exact Python interpreter pointed to by ESEM_PYTHON. A package installed in one virtualenv is not visible to another interpreter, and a package installed for python3.11 may not be visible to python3.12.

Build docs developers (and LLMs) love

Get started for freeTalk to us