test_ru.py

from src.annotations import DateAnnotation
from src.detectors.date.date import DateDetector


def test_detect_dates_pl():
    detector = DateDetector("ru")
    
    text = "1.01.2022 я продал кошку. 5 апреля 2021 я купил собаку."
    found_dates = detector.detect(text, dict())
    
    format_date1 = [
        (DateAnnotation.AnnotationPart.TWO_DIGITS_DAY, "01"), # Only supports two digits for now
        (DateAnnotation.AnnotationPart.OTHER, "."),
        (DateAnnotation.AnnotationPart.TWO_DIGIT_MONTH, "01"),
        (DateAnnotation.AnnotationPart.OTHER, "."),
        (DateAnnotation.AnnotationPart.FOUR_DIGIT_YEAR, "2022")
    ]
    
    format_date2 = [
        (DateAnnotation.AnnotationPart.TWO_DIGITS_DAY, "05"),
        (DateAnnotation.AnnotationPart.OTHER, " "),
        (DateAnnotation.AnnotationPart.TEXT_MONTH, "апреля"), # Only supports two digits for now
        (DateAnnotation.AnnotationPart.OTHER, " "),
        (DateAnnotation.AnnotationPart.FOUR_DIGIT_YEAR, "2021"),
    ]
    
    assert found_dates == [(0, 9, DateAnnotation(format_date1)), (26, 39, DateAnnotation(format_date2))]