Skip to content

v0.3.1

Choose a tag to compare

View all tags
@er2es er2es released this 25 Sep 18:10
· 44 commits to main since this release
v0.3.1
d564522

A patch release: a stop-word fix for accent-folded languages, 100% test coverage, and a bigger, multilingual demo.

After upgrading, run fuzzphony:schema --apply, then a full fuzzphony:reindex. See UPGRADE.md.

Fixed

  • Accented stop words are ignored again. With accent folding (the default), unaccent ran before the stemmer's stop-word check. German für, Hungarian és, French à and their counterparts in every language with a stop-word list were indexed and had to match like ordinary words, so Tasche für Laptop did not find Laptop Tasche. The fuzzphony_<language> configuration now drops stop words first, using a new fuzzphony_<language>_stop dictionary. schema --apply repairs existing configurations, and fuzzphony:doctor reports one that still keeps them.

Changed

  • The unit and integration suites cover 100% of the lines in src/: 539 tests, up from 400 tests and 91.5%. CI fails below 90%, and Codecov reports the coverage of every pull request's changes.
  • The README now opens with the demo, says who Fuzzphony is for and when it is not the right tool, and has an extended 1.0 roadmap: search-as-you-type, synonyms, facets, "did you mean", zero-downtime reindex and search analytics.

Demo

  • A new Languages page shows the same search in English, German, French, Spanish and Hungarian, with one-click examples and the lexeme PostgreSQL produced for every word.
  • The default catalogue is now 500 000 products, measured at that size in demo/README.md ("Proven on 500 000 rows"). The Benchmark and compare pages say how many statements they run, and on how many rows.

Full list: CHANGELOG

Assets 2
Loading