fix: support bilingual Booking Excel headers

This commit is contained in:
Wyndham ARR
2026-08-11 15:29:42 +08:00
parent ca3e8e18fa
commit c947230471
15 changed files with 366 additions and 30 deletions

View File

@@ -9,6 +9,7 @@ from openpyxl.styles import PatternFill
from booking_ingestion.excel import (
BookingExcelError,
PROCESSOR_VERSION,
REVIEW_CONFIRMED,
REVIEW_PENDING,
build_excel_parse_result,
@@ -113,6 +114,128 @@ class BookingExcelParserTests(unittest.TestCase):
self.assertEqual(parsed["status"], "needs_review")
self.assertEqual(len(parsed["room_items"]), 2)
self.assertEqual(parsed["warnings"][0]["code"], "BOOKING_ROOM_TYPE_REVIEW_REQUIRED")
self.assertEqual(parsed["processor_version"], "2.1.0")
self.assertEqual(PROCESSOR_VERSION, "2.1.0")
def test_bilingual_headers_accept_lian_tai_and_qbd_variants(self) -> None:
payload = workbook_bytes(
[
(
"AUTO_CALC",
[["BLOCK_CODE", "RATE_CODE"], ["LT-NEW", "RACK"]],
),
(
"LianTai",
[
["Booking export"],
[
"Tour Code \n主团号",
"Tour Days",
"Notes",
"วางข้อมูลที่นี่ / 酒店明细",
],
["LT-NEW", "", "", "酒店 (【U-TWN8.5】 2)"],
],
),
(
"QBD",
[
["QBD export"],
[None],
[
"Month",
"NO.",
"Tour Code \n主团号",
"Tour Days",
"Raw Hotel Detail / วางข้อมูลที่นี่ / 酒店明细",
],
["", 1, "QBD-NEW", "", "酒店 (【U-DBL12】 3)"],
],
),
]
)
document = parse_excel_bytes(payload)
self.assertEqual(document.worksheets, ("LianTai", "QBD"))
self.assertEqual(
[row.group_code_raw for row in document.rows], ["LT-NEW", "QBD-NEW"]
)
self.assertEqual(document.extracted_item_count, 2)
self.assertEqual(document.pending_item_count, 0)
def test_legacy_tour_code_aliases_remain_supported(self) -> None:
for tour_header in (
"Tour Code",
"Group Code",
"Gourp Code",
"团号",
"团队代码",
):
with self.subTest(tour_header=tour_header):
payload = workbook_bytes(
[
(
"Legacy",
[
[tour_header, "โรงแรม"],
["LEGACY", "酒店 (【U-TWN8.5】 1)"],
],
)
]
)
document = parse_excel_bytes(payload)
self.assertEqual(document.worksheets, ("Legacy",))
self.assertEqual(document.rows[0].group_code_raw, "LEGACY")
def test_lookalike_bilingual_header_is_rejected(self) -> None:
payload = workbook_bytes(
[
(
"Data",
[
[
"Tour Code / 主团号(备用)",
"Raw Hotel Detail / วางข้อมูลที่นี่ / 酒店明细",
],
["SYN", "酒店 (【U-TWN8.5】 1)"],
],
)
]
)
with self.assertRaises(BookingExcelError) as raised:
parse_excel_bytes(payload)
self.assertEqual(raised.exception.code, "BOOKING_EXCEL_HEADERS_MISSING")
self.assertIn("Tour Code/主团号", raised.exception.safe_message)
self.assertIn("酒店明细", raised.exception.safe_message)
def test_duplicate_recognized_headers_are_ambiguous(self) -> None:
payload = workbook_bytes(
[
(
"Data",
[
[
"Tour Code",
"Tour Code / 主团号",
"วางข้อมูลที่นี่ / 酒店明细",
],
["SYN", "SYN", "酒店 (【U-TWN8.5】 1)"],
],
)
]
)
with self.assertRaises(BookingExcelError) as raised:
parse_excel_bytes(payload)
self.assertEqual(raised.exception.code, "BOOKING_EXCEL_HEADERS_AMBIGUOUS")
self.assertIn("Tour Code/主团号", raised.exception.safe_message)
self.assertIn("酒店明细", raised.exception.safe_message)
def test_physically_last_tour_row_replaces_earlier_and_cancel_removes_it(self) -> None:
payload = booking_rows(