unit-syntax 0.1.2

Literal physical units for Jupyter and IPython

unit-syntax extends the Python language in Jupyter/IPython to support expressions with physical units:

>>> speed = 5 meters/second
>>> 2 seconds * speed
10 meter

Behind the scenes this is translated into standard Python that uses the excellent Pint units library.

Getting Started

Install the package:

$ pip install unit-syntax

To enable unit-syntax in a Jupyter/IPython session run:

import unit_syntax
unit_syntax.enable_ipython()

Note: in Jupyter this must be run in its own cell before any units expressions are evaluated.

Usage

An interactive notebook to play around with units

Units apply to the preceding value (a literal, variable, function call or indexing), and have higher precedence than other operators:

x * 1.21 gigawatts

This is equivalent to x * (1.21 gigawatts), and desugars to something like x * Quantity(1.21, "gigawatts"). The high precedence means units apply to the literal not the whole expression.

Values can be converted to another measurement system:

(88 miles / hour) furlongs / fortnight

Pint transparently supports numpy when available:

velocity = [5, 7] meters/second**2
location = velocity * 2 seconds
distance_traveled = numpy.linalg.norm(location)

The Grammar

The units term follows this grammar:

units:
    | NAME '/' units_group
    | NAME '*' units_group
    | NAME units_group
    | NAME '**' NUMBER
    | NAME

units_group:
    | '(' units ')'
    | units

Why? How?

I like using Python+Jupyter Notebook as a calculator for physical problems and often wish it had the clarity and type checking of explicit units. Pint is great, but (IMO) its necessary verbosity makes it hard to see the underlying calculation that's going.

unit-syntax is an IPython/Jupyter custom input transformer that rewrites expressions with units into calls to pint.Quantity. The parser is a lightly modified version of the Python grammar using the same parser generator (pegen) as Python itself.

unit-syntax cannot (currently) be used for standalone python scripts outside of IPython/Jupyter, but that's in principle possible through meta_path import hooks.

The syntax takes advantage of the fact that that in python its illegal for a NAME to follow a "primary" (literal, function call etc), so there's no ambiguity.

Prior Art

The immediate inspriration of unit-syntax is a language called Fortress from Sun Microsystems. Fortress was intended as a modern Fortran, and had first-class support for units in both the syntax and type system.

F# (an OCaml derivative from Microsoft) also has first class support for units.

The Julia package Unitful.jl

Open questions and future work

• Fortress uses an in operator to apply units to a non-literal value, e.g x in meters. This has the advantage of being unambiguous regardless of parenthesization. In python this would conflcit with value in [a, b, c], but as is

• Move to tree-sitter, which will be simpler and has a chance of providing syntax highlighting

• Test against various ipython and python versions

• Support standalone scripts through sys.meta_path

• Check units at parse time

• Unit type hints, maybe checked with @runtime_checkable. More Pint typechecking discussion

• Pint does not do the right thing when applied to generator expressions, e.g (a for a in range(0, 4)) meters

• Demo colab notebook: https://colab.research.google.com/drive/1PInyLGZHnUzEuUVgMsLrUUNdCurXK7v1#scrollTo=JszzXmATY0TV

• Describe parsing ambuguity like 1 meters * sin(45 degrees)

• Figure out story around parenthesization

Development

To regenerate the parser:

python -m pegen grammar.txt -o unit_syntax/parser.py

Running tests:

 $ poetry install --with dev
 $ poetry run pytest

Future work and open questions

• Parenthisized units expressions
• Demo colab notebook
• Move to tree-sitter so there's a chance of getting syntax highlighting
• Jupyter syntax checks
• Typography of output
• Test against various ipython and python versions
• Support standalone scripts through sys.meta_path
• Check units at parse time
• Unit type hints, maybe checked with @runtime_checkable. More Pint typechecking discussion
• Does not do the right thing when applied to generator expressions, e.g (a for a in range(0, 4)) meters