test_pl.py

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

def test_detect_dates_pl():
    detector = DateDetector("pl")

    text = "W dniu 1.01.2022 sprzedałem kota. 5 kwietnia 2021 roku kupiłem psa."
    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"), # Only supports two digits for now
        (DateAnnotation.AnnotationPart.OTHER, " "),
        (DateAnnotation.AnnotationPart.TEXT_MONTH, "kwietnia"),
        (DateAnnotation.AnnotationPart.OTHER, " "),
        (DateAnnotation.AnnotationPart.FOUR_DIGIT_YEAR, "2021"),
    ]
    
    assert found_dates == [(7, 16, DateAnnotation(format_date1)), (34, 49, DateAnnotation(format_date2))]