hfst-optimized-lookup¶

A pip-installable library version of hfst-optimized-lookup, originally built for itwêwina.

Install¶

pip install hfst-optimized-lookup

This requires that the machine running pip have a working C++ compiler. If enough people ask for us to upload binary ‘wheels’ so that you don’t need a compiler at install time, we could start doing so.

Usage¶

Full API Documentation

Import the library:

>>> import hfst_optimized_lookup

Then load an FST!

>>> fst = hfst_optimized_lookup.TransducerFile('../crk-relaxed-analyzer-for-dictionary.hfstol')

Hint: Download crk-relaxed-analyzer-for-dictionary.hfstol by cloning https://github.com/UAlbertaALTLab/cree-intelligent-dictionary/tree/main/src/crkeng/resources/fst to follow along! The file itself is stored in Git LFS so is tricky to link to directly.

Do an ordinary lookup, to get a list of concatenated analyses for a wordform:

>>> fst.lookup('atim')
['atim+N+A+Sg', 'atimêw+V+TA+Imp+Imm+2Sg+3SgO']

Or get each parsed analysis from the wordform

>>> analysis = fst.lookup_lemma_with_affixes('atim')[0]
>>> analysis.lemma
'atim'
>>> analysis.suffixes
('+N', '+A', '+Sg')

You can also lookup the analyses with symbols separated:

>>> fst.lookup_symbols('atim')
[['a', 't', 'i', 'm', '+N', '+A', '+Sg'], ['a', 't', 'i', 'm', 'ê', 'w', '+V', '+TA', '+Imp', '+Imm', '+2Sg', '+3SgO']]

Releasing¶

Run ./release.py --help for details of the release process.

A typical release is:

Make sure the “Unreleased” section of CHANGELOG.md is up-to-date

Run the script:

./release.py --release-timezone=America/Edmonton --push --release