Attendance filter: enrich-then-filter so badge matches result set

The previous filter applied status_code conditions on raw DB rows, then
enriched the page of results. That meant a missing record with a leave
request would surface under '缺勤' (raw code='missing') but render a
'請假'/'公假' badge (enriched status_code).

Move status_text filter to AFTER enrichment so it matches the badge.
status_text sort_by='status_code' also moves to Python sort since
post-enrich codes aren't in the DB. Log a warning if the in-memory
filter scans >5000 rows so we know when to revisit.

backend/main.py

@@ -1248,7 +1248,14 @@ async def attendance_records(
     db: Session = Depends(get_db),
     current_user: User = Depends(get_current_user),
 ):
-    """Paginated records from attendance_records (SQL)."""
+    """Paginated records from attendance_records (SQL).
+
+    status_text filter runs AFTER enrichment so it matches the badge rendered
+    from post-enrich status_code. Otherwise missing-day-with-leave records
+    would surface under '缺勤' but render as '請假'/'公假' badges (mismatch).
+    Trade-off: status_text/sort_by filter runs in Python after enrich instead
+    of SQL — fine while dataset stays small (~80 records), log if it grows.
+    """
     from datetime import datetime as _dt
     from sqlalchemy import or_, and_
 
@@ -1267,39 +1274,6 @@ async def attendance_records(
         q = q.filter(AttendanceRecord.employee_name == staff)
     if status:
         q = q.filter(AttendanceRecord.status_code == status)
-    if status_text:
-        # Map status_text filter to status_code conditions so the result set
-        # stays aligned with the badge rendered from (post-enrich) status_code.
-        # Previously used late/early/OT thresholds which mismatched because
-        # the import path can leave late_minutes=0 while status_code='abnormal'
-        # (e.g. act_in=18:00 vs exp_in=11:00), and missing records also have
-        # all three == 0 despite status_code='missing'.
-        # Accept both English status_code values AND the Chinese basic-status
-        # labels returned by /api/attendance/status_texts dropdown.
-        abnormal_codes = ('abnormal', 'late', 'early', 'ot',
-                          'late_early', 'late_ot', 'early_ot')
-        if status_text in ('abnormal', '異常'):
-            q = q.filter(AttendanceRecord.status_code.in_(abnormal_codes))
-        elif status_text == 'late':
-            q = q.filter(AttendanceRecord.status_code.in_(('late', 'late_early', 'late_ot')))
-        elif status_text == 'early':
-            q = q.filter(AttendanceRecord.status_code.in_(('early', 'late_early', 'early_ot')))
-        elif status_text == 'ot':
-            q = q.filter(AttendanceRecord.status_code.in_(('ot', 'late_ot', 'early_ot')))
-        elif status_text in ('missing', '缺勤'):
-            q = q.filter(AttendanceRecord.status_code == 'missing')
-        elif status_text in ('holiday', '公假'):
-            q = q.filter(AttendanceRecord.status_code == 'holiday')
-        elif status_text in ('al', 'sl', 'cl', 'mixed_leave'):
-            q = q.filter(AttendanceRecord.status_code == status_text)
-        elif status_text in ('leave', '請假'):
-            # Any leave type — group filter
-            q = q.filter(AttendanceRecord.status_code.in_(['al', 'sl', 'cl', 'mixed_leave']))
-        elif status_text in ('normal', '正常'):
-            q = q.filter(AttendanceRecord.status_code == 'normal')
-        else:
-            # Fallback to text search
-            q = q.filter(AttendanceRecord.status_text.like(f"%{status_text}%"))
     if search:
         like = f"%{search}%"
         q = q.filter(or_(
@@ -1308,20 +1282,62 @@ async def attendance_records(
             AttendanceRecord.shift_code.like(like),
         ))
 
-    # sort
+    # Pre-sort by raw date so enrich-then-filter ordering is deterministic.
-    sortable = {
+    sortable_raw = {
         "date": AttendanceRecord.date,
         "employee_name": AttendanceRecord.employee_name,
         "weekday": AttendanceRecord.weekday,
         "shift_code": AttendanceRecord.shift_code,
-        "status_code": AttendanceRecord.status_code,
     }
-    col = sortable.get(sort_by, AttendanceRecord.date)
+    col = sortable_raw.get(sort_by, AttendanceRecord.date)
     q = q.order_by(col.desc() if sort_order == "desc" else col.asc())
 
-    total = q.count()
+    all_rows = q.all()
-    rows = q.offset((page - 1) * per_page).limit(per_page).all()
+    if len(all_rows) > 5000:
-    enrich_attendance_with_leave(rows, db)
+        logger.warning(
+            "attendance_records: in-memory filter over %d rows — consider prefiltering via date range",
+            len(all_rows),
+        )
+
+    # Enrich ALL rows so we can filter by post-enrich status_code.
+    enrich_attendance_with_leave(all_rows, db)
+
+    # Apply status_text filter on enriched status_code (post-leave/holiday override)
+    if status_text:
+        abnormal_codes = ('abnormal', 'late', 'early', 'ot',
+                          'late_early', 'late_ot', 'early_ot')
+        leave_codes = ('al', 'sl', 'cl', 'mixed_leave')
+        if status_text in ('abnormal', '異常'):
+            rows = [r for r in all_rows if (r.status_code or '').lower() in abnormal_codes]
+        elif status_text == 'late':
+            rows = [r for r in all_rows if (r.status_code or '').lower() in ('late', 'late_early', 'late_ot')]
+        elif status_text == 'early':
+            rows = [r for r in all_rows if (r.status_code or '').lower() in ('early', 'late_early', 'early_ot')]
+        elif status_text == 'ot':
+            rows = [r for r in all_rows if (r.status_code or '').lower() in ('ot', 'late_ot', 'early_ot')]
+        elif status_text in ('missing', '缺勤'):
+            rows = [r for r in all_rows if (r.status_code or '').lower() == 'missing']
+        elif status_text in ('holiday', '公假'):
+            rows = [r for r in all_rows if (r.status_code or '').lower() == 'holiday']
+        elif status_text in leave_codes:
+            rows = [r for r in all_rows if (r.status_code or '').lower() == status_text]
+        elif status_text in ('leave', '請假'):
+            rows = [r for r in all_rows if (r.status_code or '').lower() in leave_codes]
+        elif status_text in ('normal', '正常'):
+            rows = [r for r in all_rows if (r.status_code or '').lower() == 'normal']
+        else:
+            # Fallback to raw status_text substring search
+            rows = [r for r in all_rows
+                    if r.status_text and status_text in r.status_text]
+    else:
+        rows = all_rows
+
+    # Python sort by status_code when requested (raw DB only handled the basic columns)
+    if sort_by == 'status_code':
+        rows.sort(key=lambda r: (r.status_code or ''), reverse=(sort_order == 'desc'))
+
+    total = len(rows)
+    page_rows = rows[(page - 1) * per_page: (page - 1) * per_page + per_page]
 
     def to_dict(r):
         return {
@@ -1347,7 +1363,7 @@ async def attendance_records(
         }
 
     return {
-        "records": [to_dict(r) for r in rows],
+        "records": [to_dict(r) for r in page_rows],
         "total": total,
         "page": page,
         "per_page": per_page,