tests/test_convert_goldenset.py

import importlib.util
import sys
from pathlib import Path

from openpyxl import Workbook

# The converter is a standalone script; it lives at submission/ in the working
# repo and at the bundle root in the published code repo.
_ROOT = Path(__file__).resolve().parents[1]
_SCRIPT = next(
    p for p in (_ROOT / "submission" / "convert_goldenset_to_jsonl.py",
                _ROOT / "convert_goldenset_to_jsonl.py")
    if p.exists()
)
_spec = importlib.util.spec_from_file_location("convert_goldenset_to_jsonl", _SCRIPT)
converter = importlib.util.module_from_spec(_spec)
sys.modules["convert_goldenset_to_jsonl"] = converter
_spec.loader.exec_module(converter)


def _workbook(rows: list[dict]) -> Path:
    header = ["case_id", "Link", "full_text", *converter.SCHEMA_FIELDS,
              "comment", "original_input"]
    wb = Workbook()
    ws = wb.active
    ws.title = "GOLDENSET"
    ws.append(header)
    for row in rows:
        ws.append([row.get(h) for h in header])
    return wb


def _convert(tmp_path: Path, rows: list[dict]) -> list[dict]:
    path = tmp_path / "Goldenset_Test_final.xlsx"
    _workbook(rows).save(path)
    return converter.convert_workbook(path, fallback={})


def test_scrape_only_rows_are_dropped(tmp_path: Path) -> None:
    records = _convert(tmp_path, [
        {"case_id": "A1", "Link": "http://x", "full_text": "t"},
        {"case_id": "A2", "full_text": "t", "legal_subject_judgement": "Contract"},
    ])
    assert [r["case_id"] for r in records] == ["A2"]


def test_reviewed_rows_without_legal_subject_are_kept(tmp_path: Path) -> None:
    records = _convert(tmp_path, [
        # reviewed: real values but no legal subject (the marker convention fails
        # here) — kept, because its cells enter the scoring denominators
        {"case_id": "B1", "full_text": "t", "trial_end_date": "2024-01-31",
         "plaintiffs_all_count": 2, "legal_subject_judgement": "None"},
        # only 'None' markers: contributes no scoreable cell -> dropped, exactly
        # like the scoring pipeline's row-inclusion rule
        {"case_id": "B2", "full_text": "t", "legal_subject_judgement": "None"},
    ])
    assert [r["case_id"] for r in records] == ["B1"]
    # the 'None' marker publishes as null
    assert records[0]["legal_subject_judgement"] is None
    assert records[0]["trial_end_date"] == "2024-01-31"


def test_zero_values_keep_a_row(tmp_path: Path) -> None:
    records = _convert(tmp_path, [
        {"case_id": "C1", "full_text": "t", "plaintiff_loosing_share": 0},
    ])
    assert [r["case_id"] for r in records] == ["C1"]
    assert records[0]["plaintiff_loosing_share"] == 0


def test_currency_only_rows_are_dropped_like_scoring(tmp_path: Path) -> None:
    # scoring ignores Currency_* for row inclusion; the converter must match
    records = _convert(tmp_path, [
        {"case_id": "D1", "full_text": "t", "Currency_dispute_value_nominal": "CHF"},
    ])
    assert records == []


def test_nonpecuniary_is_normalised_to_the_schema_literal(tmp_path: Path) -> None:
    records = _convert(tmp_path, [
        {"case_id": "E1", "full_text": "t", "legal_subject_judgement": "Claim",
         "dispute_value_nominal": "Nonpecuniary"},
    ])
    assert records[0]["dispute_value_nominal"] == "nonpecuniary"