iterwrite

Have you ever run some code which printed output like this?

It's hard to read, and makes it difficult to compare results between iterations. Enter iterwrite! With no additional lines, have your model output looking clean instead.

See how iterwrite does magically cleans up your output:

• Alignment grows when necessary, but does not shrink back down;
• Floats shown as floats;
• Separation between arguments.

Usage

To use iterwrite, simply create a writer object with the messages you wish to print during training. Then, simply provide the new values to the writer whenever you want a tidy string for printing.

import iterwrite

writer = iterwrite.Iterwriter(bing="Bing: {:.2f}", bang="Bang: {:.2f}", boom="(total {:d})")
print(writer.format(bing=0.0, bang=0.0, boom=0))
print(writer.format(bing=12.34, bang=567.890, boom=672349))
print(writer.format(bing=0.0, bang=0.0, boom=0))

You'll get something like the below.

Bing: 0.00 ‖ Bang: 0.00 ‖ (total 0)
Bing: 12.34 ‖ Bang: 567.89 ‖ (total 672349)
Bing:  0.00 ‖ Bang:   0.00 ‖ (total      0)

That's it! For another example, consider checking resources/model.py for the code used to create the example blocks in the introduction.

Installation

As yet, iterwriter is a GitHub only package. You can still install it directly with pip:

pip install git+https://github.com/jamesross2/iterwrite.git

Contributing

This package is a piece of code I wrote for myself, loosely wrapped up into a very basic package. I welcome all contributions; there are several raise NotImplementedError statements that would make a sensible starting point.

Environment

We recommend using pyenv to set up an install of Python. Then, create a virtual environment with

python3.8 -m pip install --upgrade pip virtualenv
python3.8 -m virtualenv --prompt="(iterwrite-venv3.8) " .venv

# install packages into source environment
source .venv/bin/activate && pip install -r requirements.txt && deactivate

When working on the package, be sure to work in the virtual environment.

source .venv/bin/activate

Better yet, use direnv to activate the environment automatically whenever you set the repository as your working directory.

cat "source .venv/bin/activate" > .envrc
direnv allow

Testing

It is recommended to use the Makefile to test any code additions.

make help  # to explain possible commands
make format  # apply various formatting requirements
make test  # check that model passes linting and testing using current virtual environment  

Tests are run using tox and Travis CI. Run the tox tests using either make tox or just tox.