from pathlib import Path

from review_agent.regulatory_review.services.text_extract import extract_text


def test_extract_text_reads_plain_text(tmp_path):
    path = tmp_path / "说明书.txt"
    path.write_text("产品名称：甲胎蛋白检测试剂盒\n储存条件：2-8℃", encoding="utf-8")

    result = extract_text(path)

    assert "甲胎蛋白" in result.text
    assert result.status == "success"
    assert result.content_hash


def test_extract_text_reports_unsupported_file(tmp_path):
    path = tmp_path / "image.png"
    path.write_bytes(b"png")

    result = extract_text(path)

    assert result.status == "unsupported"
    assert result.text == ""