stringunitconverter

Goal

Get the multiplication factor required to convert from one unit to another, with the units being given as strings.

Main concepts

• Use strings to specify units (e.g. 'kN').
• Allow quasi all mathematical combinations of units (e.g. 'N/m^2'). Math execution order rules should be respected.
• Keep code short (~100 LOC) and unit-independent and put all unit-related data in a data file (JSON). It should be easy to add units.
• Focus lies on engineering applications. Currency conversion etc. is not supported.
• Working concept:
  1. Keep a data file with a multiplier for each individual unit to its SI brother without prefix. For example: kg (to g): 1e-3, mmHg (to Pa): 0.00750.
  2. Use this datafile to get the multiplier for each given side to base SI. This is done by substituting each prefix and unit in the string with the appropriate numerical value. The equation, containing only mathematical operations and numbers, is then evaluated, returning the multiplier.
  3. Multiplier for the conversion follows from dividing the two derived multipliers.

Limits and notes

• Dimensional correctness is not checked.
• Conversions are always multiplicative. Therefore, conversions between coordinate systems with a different origin are not (yet) supported. For example, between °C and K. (For this reason, °C and °F are deliberately unsupported; use K.) Conversions between Hz and s are therefore also not possible, as this would require inverting the value.
• A space is interpreted as a multiplication. Example: 'kN m' equals 'kN*m'.
• The substitution parser is designed to split two or more units (and accompanying prefixes) not separated by spaces correctly. However, in some cases this can become rather difficult. For example, 'm' is used both as a prefix (milli) and as a unit (metre). The parser may fail in such cases (please report these - see contact info). The parser works from the end to the front and assumes that a unit is preceded by a prefix. For example, the parser makes the following interpretations:
  • Nmm => N mm
  • mmN => m mN
  • Nmmm => N m mm
  • mmmN => mm mN

Supported prefixes and units

Prefixes: y to Y.
Units: g, N, s, mmHg, Pa, bar, kgf, A, K, mol, Hz, J, W, C, V, Ohm, rad, °, deg, degree, S, F, oz, psi, ft, in, L.
Also supported: %.

Examples

>>> import stringunitconverter as suc
>>>
>>> suc.multiplier('N/m^2', 'kN/cm^2')
10.0
>>> suc.multiplier('1/kPa', '1/(N/m^2)')
0.001

Contribution and development

Testing

Tests are in the directory tests in the root directory. To run the tests, navigate to the root directory and run python tests/test.py.

This requires a local copy of the repo. Note that the tests will be run on the stringunitconverter Python package. This package may or may not point to the code in the local copy of the repo, depending on how the package was installed.

Perhaps a testing framework such as pytest will be used in the future, however, currently the tests are very simple and the codebase to be tested is also very small. So, it appears that in this particular case those testing frameworks mostly create overhead.

To get this repo locally

1. Clone the repo. The directory with the project will be located in the current working directory of the terminal.

   $ git clone git@gitlab.com:abaeyens/stringunitconverter.git
   

2. Navigate to the created directory.

   $ cd stringunitconverter
   

Create a local install

A local install allows to try out the library locally. This can be useful during development. First, ensure that the project root directory is named stringunitconverter. Second, run in the project root directory:

$ python -m pip install -e .

This installs the package such that it is accessible like all other Python packages, e.g. using import stringunitconverter. The -e option denotes that it uses a symbolic link: code changes in the project directory (including branch switching) take effect at the first following import. No re-installation is required.

Note: python should refer to Python 3. One may have to write python3 to avoid using Python 2.

Note: there appear to be problems with this method on some Windows machines.

Note: if it is desired to install several versions of the same package on a single system, for example a stable version from PyPi and a development version from a local install, it may be useful to use Python virtual environments.

Contact

Have a question, suggestion, ... ? Send an email to 2arne.baeyens@gmail.com.

It is also possible to use the GitLab issues webpage.