test_wiktor_ner.py

"""Tests for WiktorNERInputParser."""
from src.annotations import MorphosyntacticAnnotation, NerAnnotation
from src.input_parsers.wiktor_ner import WiktorNERInputParser
import clarin_json


def test_wiktor_ner_input_parser():
    """Test WiktorNERInputParser."""
    parser = WiktorNERInputParser()
    with clarin_json.open('example_inputs/ner_test.jsonl', 'r') as f:
        for example_json in f:
            text, annotations = parser.parse(example_json)

    assert text == "Marek Kowalski pojechał do Wrocławia."

    assert len(annotations) == 8

    assert (0, 14, NerAnnotation("nam_liv_person")) in annotations
    assert (27, 36, NerAnnotation("nam_loc_gpe_city")) in annotations

    assert (0, 5, MorphosyntacticAnnotation("subst:sg:nom:m1", "Marek")) in annotations
    assert (
        6,
        14,
        MorphosyntacticAnnotation("subst:sg:nom:m1", "Kowalski"),
    ) in annotations
    assert (
        15,
        23,
        MorphosyntacticAnnotation("praet:sg:m1:perf", "pojechać"),
    ) in annotations
    assert (24, 26, MorphosyntacticAnnotation("prep:gen", "do")) in annotations
    assert (
        27,
        36,
        MorphosyntacticAnnotation("subst:sg:gen:m3", "Wrocław"),
    ) in annotations
    assert (36, 37, MorphosyntacticAnnotation("interp", ".")) in annotations