Seera-Unified-SFDA/sfda_parser/sfda_scraper/test_detail_pdf_expand.py

"""Tests for splitting comma-separated device identifiers into separate rows."""

from __future__ import annotations

import os
import unittest

from sfda_parser.sfda_scraper.detail_pdf import (
    DeviceRow,
    _expand_device_row,
    _expand_device_rows,
    _split_multi_values,
    parse_affected_device_list,
)


class TestSplitMultiValues(unittest.TestCase):
    def test_splits_comma_separated_values(self):
        self.assertEqual(
            _split_multi_values("24B0729, 25B0770, 25H0171, 26B0542"),
            ["24B0729", "25B0770", "25H0171", "26B0542"],
        )

    def test_single_value_unchanged(self):
        self.assertEqual(_split_multi_values("24B0729"), ["24B0729"])

    def test_splits_semicolon_separated_values(self):
        self.assertEqual(_split_multi_values("A1; A2; A3"), ["A1", "A2", "A3"])

    def test_splits_space_separated_numeric_ids(self):
        value = (
            "0231547054 0231591827 0231758880 0231758898 0231758932 "
            "0231792414 0231792421 0231912903 0232670901 0232709219 "
            "0232709254 0232744928 0232784925 0232784989 0232805458 0232837110"
        )
        parts = _split_multi_values(value)
        self.assertEqual(len(parts), 16)
        self.assertEqual(parts[0], "0231547054")
        self.assertEqual(parts[-1], "0232837110")

    def test_does_not_split_prose_with_spaces(self):
        self.assertEqual(
            _split_multi_values("Surgical Instrument Kit"),
            ["Surgical Instrument Kit"],
        )


class TestExpandDeviceRow(unittest.TestCase):
    def _row(self, **kwargs) -> DeviceRow:
        base = {
            "material": "Test Device",
            "material_description": "Description",
            "catalog_number": "",
            "udi": "",
            "serial_no": "",
            "gstn": "",
            "batch": "",
        }
        base.update(kwargs)
        return DeviceRow(**base)

    def test_expands_serial_numbers_into_separate_rows(self):
        rows = _expand_device_row(
            self._row(serial_no="24B0729, 25B0770, 25H0171, 26B0542")
        )
        self.assertEqual(len(rows), 4)
        self.assertEqual([row.serial_no for row in rows], [
            "24B0729",
            "25B0770",
            "25H0171",
            "26B0542",
        ])
        for row in rows:
            self.assertEqual(row.material, "Test Device")
            self.assertEqual(row.material_description, "Description")

    def test_expands_catalog_numbers_into_separate_rows(self):
        rows = _expand_device_row(self._row(catalog_number="CAT-1, CAT-2, CAT-3"))
        self.assertEqual(len(rows), 3)
        self.assertEqual([row.catalog_number for row in rows], ["CAT-1", "CAT-2", "CAT-3"])

    def test_expands_batch_values_into_separate_rows(self):
        rows = _expand_device_row(self._row(batch="LOT1, LOT2"))
        self.assertEqual(len(rows), 2)
        self.assertEqual([row.batch for row in rows], ["LOT1", "LOT2"])

    def test_zips_serial_and_catalog_when_counts_match(self):
        rows = _expand_device_row(
            self._row(serial_no="S1, S2", catalog_number="C1, C2")
        )
        self.assertEqual(len(rows), 2)
        self.assertEqual(rows[0].serial_no, "S1")
        self.assertEqual(rows[0].catalog_number, "C1")
        self.assertEqual(rows[1].serial_no, "S2")
        self.assertEqual(rows[1].catalog_number, "C2")

    def test_single_value_row_unchanged(self):
        row = self._row(serial_no="ONLY-ONE", catalog_number="CAT-9")
        self.assertEqual(_expand_device_row(row), [row])

    def test_expand_device_rows_flattens_list(self):
        devices = [
            self._row(serial_no="A, B"),
            self._row(catalog_number="X, Y, Z"),
        ]
        expanded = _expand_device_rows(devices)
        self.assertEqual(len(expanded), 5)

    def test_expands_space_separated_udi_values(self):
        udi_value = "0231547054 0231591827 0231758880"
        rows = _expand_device_row(self._row(udi=udi_value, material="Monitor"))
        self.assertEqual(len(rows), 3)
        self.assertEqual(rows[1].udi, "0231591827")
        self.assertEqual(rows[2].material, "Monitor")


class TestParseAffectedDeviceListExpansion(unittest.TestCase):
    def test_parse_expands_comma_separated_serials_from_table_like_input(self):
        # Minimal synthetic check via direct expansion path used by parser output.
        devices = _expand_device_rows([
            DeviceRow(
                material="Pump",
                material_description="",
                catalog_number="REF-100",
                udi="",
                serial_no="24B0729, 25B0770, 25H0171, 26B0542",
                gstn="",
                batch="",
            )
        ])
        self.assertEqual(len(devices), 4)
        self.assertEqual(devices[0].catalog_number, "REF-100")
        self.assertEqual(devices[-1].serial_no, "26B0542")


class TestSparseTablePdfExpansion(unittest.TestCase):
    def test_medtronic_pdf_expands_sparse_table_lot_lists(self):
        pdf_path = os.path.abspath(
            os.path.join(
                os.path.dirname(__file__),
                "../../../../sites/erp.seeraunified.com/public/files/"
                "SA-28-05-26-1408-ade-detailad786ead786e.pdf",
            )
        )
        if not os.path.isfile(pdf_path):
            self.skipTest(f"fixture PDF not found: {pdf_path}")

        with open(pdf_path, "rb") as pdf_file:
            devices = parse_affected_device_list(pdf_file.read())

        self.assertGreaterEqual(len(devices), 40)
        serial_numbers = [device.serial_no for device in devices if device.serial_no]
        self.assertIn("0012508285", serial_numbers)
        self.assertIn("0013272796", serial_numbers)


if __name__ == "__main__":
    unittest.main()