tests/scrapers/test_at.py

from datetime import date

from legex.models.base import Case
from legex.scrapers.at import (
    ATScraper,
    _extract_text_from_xml,
    _first_geschaeftszahl,
    _norm_gz,
    _parse_doc_id_date,
)


def _make(case_id: str, rechtsgebiete: str = "Zivilrecht") -> Case:
    return Case(
        case_id=case_id,
        jurisdiction="at",
        metadata={"rechtsgebiete": rechtsgebiete},
    )


def test_parse_doc_id_date_jjr():
    assert _parse_doc_id_date("JJR_20241219_OGH0002_0010OB00108_24Z0000_000") == date(2024, 12, 19)


def test_parse_doc_id_date_jjt():
    assert _parse_doc_id_date("JJT_20230515_OGH0002_0040OB00042_23X0000_000") == date(2023, 5, 15)


def test_parse_doc_id_date_invalid():
    assert _parse_doc_id_date("not-a-real-id") is None
    assert _parse_doc_id_date("") is None


def test_first_geschaeftszahl_single():
    assert _first_geschaeftszahl("4Ob402/85") == "4Ob402/85"


def test_first_geschaeftszahl_list_string():
    # API often joins related cases with semicolons; we take the first.
    raw = {"item": "4Ob402/85; 4Ob402/87; 4Ob128/89"}
    assert _first_geschaeftszahl(raw) == "4Ob402/85"


def test_civil_filter_keeps_zivilrecht():
    cases = [_make("1Ob1/24z", "Zivilrecht"), _make("2Os1/24x", "Strafrecht")]
    kept = ATScraper.civil_filter(cases)
    assert [c.case_id for c in kept] == ["1Ob1/24z"]


def test_civil_filter_accepts_multi_value():
    case = _make("1Ob1/24z", "Zivilrecht | Arbeitsrecht")
    assert ATScraper.civil_filter([case])


def test_norm_gz_strips_whitespace():
    # RIS prints "1 Ob 108/24z" in some payloads and "1Ob108/24z" in others.
    assert _norm_gz("1 Ob 108/24z") == _norm_gz("1Ob108/24z") == "1ob108/24z"


def test_extract_text_from_xml_concatenates_text_nodes():
    xml = b"""<?xml version="1.0"?><root><a>Hello</a><b>  world </b><c></c><d>foo bar</d></root>"""
    out = _extract_text_from_xml(xml)
    assert out == "Hello world foo bar"


def test_extract_text_from_xml_invalid():
    assert _extract_text_from_xml(b"<not xml>") is None


def test_enrich_skips_cases_with_existing_text():
    # If full_text is already set, enrich shouldn't make a network call.
    c = Case(case_id="1Ob1/24z", jurisdiction="at", full_text="cached", metadata={"doc_id": ""})
    out = ATScraper.enrich([c])
    assert out[0].full_text == "cached"