Version 1.1.1

Bug Fix

Fix UnicodeEncodeError for non-Latin-1 characters in DAWG lookups

Words containing characters outside the Latin-1 encoding (such as Polish words with ł or ę) would cause a UnicodeEncodeError when passed to the DAWG lookup functions (__contains__ and find_combinations).

This release gracefully handles such words by:

• Returning False from __contains__ for words with non-Latin-1 characters
• Returning an empty list from find_combinations for such words

This allows the library to safely process text that may contain non-Icelandic words without crashing.

Version 1.1.0

• Performance improvements: C++ implementations of lookup_id(), lemma_forms(), and DawgDictionary with significant speedups
• Hash-based deduplication in lookup_id() for faster duplicate detection
• Python Stable ABI (abi3) support for portable CPython wheels across Python 3.9-3.14+
• Updated with latest BÍN data
• CFFI upgraded to >=1.17.1
• PyPy 3.11 support (dropped PyPy 3.9/3.10 - end of life)
• Optimized wheel builds with shared data preparation in CI

Full Changelog: 1.0.4-7...1.1.0

Version 1.0.4

• Updated with latest BÍN data
• Explicit support for Python 3.14 and PyPy 3.11
• Project metadata migrated to pyproject.toml
• Minor refactoring, including improvements to BÍN data sanitation

Full Changelog: 1.0.3-4...1.0.4-7

Version 1.0.3

• Explicit support for Python 3.13
• Now requires Python 3.9+
• Minor vocabulary tweaks
• Project metadata now in pyproject.toml

Full Changelog: 1.0.2...1.0.3-4

Version 1.0.2

• Minor tweak

Version 1.0.1

• Various minor tweaks and fixes
• Updated with latest BÍN data

Full Changelog: 1.0.0...1.0.1

Version 1.0.0

• Version bumped from 0.4.8 to 1.0.0 to reflect that at this point this is mature, battle-tested software
• Binary wheels now provided for CPython 3.12
• Added get_compound function which looks up a word in the database and returns its meaning(s), prioritizing returning its compound structure
• Updated with latest BÍN data

Full Changelog: 0.4.8...1.0.0

Version 0.4.8

• Updated with latest BÍN data
• Now requires Python 3.8 or later

Full Changelog: 0.4.7...0.4.8

Version 0.4.7

• Words added and removed from resources/ord.add.csv, resources/ord.suffixes.csv and config/BinErrata.conf to support custom tone-of-voice corrections in GreynirCorrect

Version 0.4.6

• Upgraded BÍN source data to the latest version from Stofnun Árna Magnússonar í íslenskum fræðum