fix: support bilingual Booking Excel headers

booking_ingestion/excel.py

@@ -14,7 +14,7 @@ from typing import Iterable, Mapping, Optional
 
 
 PROCESSOR_NAME = "booking-excel-importer"
-PROCESSOR_VERSION = "2.0.0"
+PROCESSOR_VERSION = "2.1.0"
 RESULT_SCHEMA_VERSION = "1.0"
 XLSX_MIME = "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
 MAX_UPLOAD_BYTES = 25 * 1024 * 1024
@@ -183,21 +183,28 @@ def _normalized_header(value: object) -> str:
     return re.sub(r"[\s._\-/\\:：()（）]+", "", normalized)
 
 
+HEADER_LABELS = {
+    "tour_code": (
+        "Tour Code",
+        "Group Code",
+        "Gourp Code",
+        "团号",
+        "团队代码",
+        "Tour Code / 主团号",
+    ),
+    "hotel": (
+        "โรงแรม",
+        "วางข้อมูลที่นี่ / 酒店明细",
+        "Raw Hotel Detail / วางข้อมูลที่นี่ / 酒店明细",
+    ),
+}
+
+# Keep the approved source labels readable while applying one exact normalized
+# comparison to every workbook header.  This is deliberately not substring or
+# fuzzy matching: an unrecognized supplier change must fail safely.
 HEADER_ALIASES = {
-    "tour_code": frozenset(
-        {
-            "TOURCODE",
-            "GROUPCODE",
-            "GOURPCODE",
-            "团号",
-            "团队代码",
-        }
-    ),
-    "hotel": frozenset(
-        {
-            "โรงแรม",
-        }
-    ),
+    key: frozenset(_normalized_header(label) for label in labels)
+    for key, labels in HEADER_LABELS.items()
 }
 
 ROOM_BRACKET_PATTERN = re.compile(
@@ -305,7 +312,7 @@ def _header_columns(cells: Iterable[object]) -> dict[str, int] | None:
     if any(len(matches[key]) != 1 for key in matches):
         raise BookingExcelError(
             "BOOKING_EXCEL_HEADERS_AMBIGUOUS",
-            "Excel 表头重复，无法确定 Tour Code 或โรงแรม列",
+            "Excel 表头重复，无法唯一确定 Tour Code/主团号或酒店明细列",
         )
     return {key: values[0] for key, values in matches.items()}
 
@@ -643,7 +650,7 @@ def parse_excel_bytes(payload: bytes) -> ExcelDocument:
     if not data_sheets:
         raise BookingExcelError(
             "BOOKING_EXCEL_HEADERS_MISSING",
-            "Excel 中未找到同时包含 Tour Code 和โรงแรม的工作表",
+            "Excel 中未找到同时包含 Tour Code/主团号和酒店明细的工作表",
         )
     if not rows:
         raise BookingExcelError(