inference_example.py

"""Runnable demo for the Old English lemmatiser (Nerthus resource-enhanced release).

Usage:
    pip install -r requirements.txt
    python inference_example.py
"""

from lemmatizer import OldEnglishLemmatizer


def main() -> None:
    lem = OldEnglishLemmatizer.from_pretrained(".")

    words = ["cyninges", "cwæð", "godan", "þurh", "cyning", "folce", "halgan"]
    print("word -> (lemma, POS)")
    print("-" * 34)
    for w, pred in zip(words, lem.lemmatize_batch(words)):
        print(f"{w:12} -> {pred}")

    print("\nText:")
    for item in lem.lemmatize_text("se cyning cwæð to his folce"):
        print(f"  {item['word']:10} {item['lemma']:12} {item['pos']}")

    # The prior assumes the gold lemma is an Old English headword. Switch it off
    # for embedded Latin and dense onomastic material, where the corpus leaves
    # the inflected form standing as its own lemma.
    print("\nWithout the lexicon prior (lexicon_bonus=0):")
    for item in lem.lemmatize_text("israeles daniele", lexicon_bonus=0):
        print(f"  {item['word']:10} {item['lemma']:12} {item['pos']}")


if __name__ == "__main__":
    main()