tests/test_xpos_normalization.py

from scripts import build_dataset as BUILD


def test_valid_nine_position_tag_is_preserved():
    assert BUILD.canonical_xpos(
        "VERB", "v3siia---", "Aspect=Imp|Mood=Ind|Number=Sing|Person=3"
    ) == "v3siia---"


def test_proiel_xpos_is_rebuilt_from_upos_without_false_person():
    assert BUILD.canonical_xpos("SCONJ", "G-", "_") == "c--------"


def test_ud_features_fill_positional_morphology():
    assert BUILD.canonical_xpos(
        "VERB",
        "V-",
        "Aspect=Imp|Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin|Voice=Act",
    ) == "v3piia---"


def test_every_ordinary_token_row_is_normalized():
    source = (
        "# sent_id = test\n"
        "1\tκαί\tκαί\tSCONJ\tG-\t_\t2\tmark\t_\t_\n"
        "2\tλέγει\tλέγω\tVERB\tV-\tMood=Ind|Number=Sing|Person=3|Tense=Pres|Voice=Act\t0\troot\t_\t_\n"
    )
    normalized = BUILD.canonicalize_conllu_xpos(source)
    rows = [line.split("\t") for line in normalized.splitlines() if line and not line.startswith("#")]
    assert [row[4] for row in rows] == ["c--------", "v3spia---"]


def test_native_relations_are_mapped_to_ud_v2():
    relation = BUILD.canonical_native_deprel
    assert relation("SBJ", "NOUN", 2) == "nsubj"
    assert relation("OBJ", "NOUN", 2) == "obj"
    assert relation("ATR", "ADJ", 2) == "amod"
    assert relation("ATR", "NOUN", 2) == "nmod"
    assert relation("ADV", "VERB", 2) == "advcl"
    assert relation("AuxP", "ADP", 2) == "case"
    assert relation("AuxC", "SCONJ", 2) == "mark"
    assert relation("OBJ_CO", "NOUN", 2) == "conj"
    assert relation("unknown-native-label", "X", 2) == "dep"
    assert relation("SBJ", "NOUN", 0) == "root"