notebook-terminal 0.4.2

A cross-platform PTY/ConPTY terminal for Jupyter with bundled xterm.js

notebook-terminal

Cross-platform xterm.js terminal for Jupyter Notebook and JupyterLab, with local PTY/ConPTY and remote JupyterHub/Jupyter Server backends.

Repository: https://github.com/ZawadzkiR/notebook-terminal

Installation

pip install notebook\_terminal-0.4.1-py3-none-any.whl

Restart the kernel and refresh JupyterLab after replacing an older build.

Local terminal

from notebook\_terminal import terminal

term = terminal(height=450)
term.run("python main.py", cwd="/home/user/project")

On Windows:

term.run("python main.py", cwd=r"C:\\Users\\User\\project")

Python code with a working directory

term.run\_python(
    """
from utils import helper
helper()
""",
    cwd="/home/user/project",
)

Python file, arguments and working directory

term.run\_python\_file(
    "main.py",
    cwd="/home/user/project",
    args=\["input.csv", "two words"],
    kwargs={"limit": 100, "verbose": True},
)

For a local backend, a relative file path is resolved against cwd. The child Python process is run from that directory, so imports such as from utils import helper work when utils.py or the package is located there.

Remote JupyterHub terminal

import os
from notebook\_terminal import remote\_terminal

remote = remote\_terminal(
    server\_url="https://jupyter.example.com/user/<user>/",
    token=os.environ\["JUPYTERHUB\_TOKEN"],
)

remote.run("python main.py", cwd="/home/user/project")
remote.run\_python("from utils import helper; helper()", cwd="/home/user/project")
remote.run\_python\_file(
    "main.py",
    cwd="/home/user/project",
    args=\["input.csv"],
)

A default remote directory can also be set once:

remote = remote\_terminal(
    server\_url="https://jupyter.example.com/user/user/",
    token=os.environ\["JUPYTERHUB\_TOKEN"],
    cwd="/home/user/project",
)

remote.run("git status")
remote.run\_python\_file("main.py")

The remote backend uses the Jupyter Server terminal REST API and terminal WebSocket. It does not require SSH. Remote cwd currently targets POSIX shells, which are standard on Linux JupyterHub installations.

cwd behavior

• cwd on a single method call applies only to that command.
• The current interactive shell directory is restored after the command.
• terminal(cwd=...) sets the initial directory for a local terminal session.
• remote\_terminal(cwd=...) sets the default directory used for remote commands.
• A per-call cwd overrides the remote default.

Main API

term.run(command, cwd=None)
term.run\_python(code, cwd=None, rich\_output=False)
term.run\_python\_file(path, cwd=None, args=None, kwargs=None, rich\_output=False)
term.send\_text(text)
term.interrupt()
term.history()
term.success("Done")

Security

Treat a JupyterHub token as a password. Do not commit it to Git or place it directly in a notebook that will be shared.

Author

Robert Zawadzki — GitHub: ZawadzkiR

Function-level working directory and Python interpreter

cwd is supplied to each operation. Remote connection creation does not set a working directory.

term.run("git status", cwd="/home/user/project")
term.run\_python("from helpers import load\_data", cwd="/home/user/project", interpreter="python3.12")
term.run\_python\_file(
    "main.py",
    cwd="/home/user/project",
    interpreter="/home/user/.venvs/project/bin/python",
    args=\["input.csv"],
)

The same calls work with remote\_terminal(...). executable= remains available as a backward-compatible alias for interpreter=; do not pass different values to both.