add workers check

gen_cons.py

@@ -31,6 +31,44 @@ ROTA_START_DATE = "2026-08-03"
 sites = ("rota a", "rota b", "rota c")
 
 
+def _normalize_worker_key(value):
+    text = str(value or "").strip()
+    if not text:
+        return ""
+    return re.sub(r"\s+", " ", text).lower()
+
+
+def _validate_worker_alignment(base_workers, other_workers, base_label, other_label):
+    """Validate that two worker collections contain the same members.
+
+    Comparison is case-insensitive and whitespace-normalized.
+    Raises ValueError with a clear diff when the sets do not match.
+    """
+    base_clean = [str(w).strip() for w in base_workers if str(w).strip()]
+    other_clean = [str(w).strip() for w in other_workers if str(w).strip()]
+
+    base_norm_to_raw = {_normalize_worker_key(w): w for w in base_clean}
+    other_norm_to_raw = {_normalize_worker_key(w): w for w in other_clean}
+
+    base_norm = set(base_norm_to_raw.keys())
+    other_norm = set(other_norm_to_raw.keys())
+
+    only_in_base_norm = sorted(base_norm - other_norm)
+    only_in_other_norm = sorted(other_norm - base_norm)
+
+    if not only_in_base_norm and not only_in_other_norm:
+        return
+
+    only_in_base = [base_norm_to_raw[k] for k in only_in_base_norm]
+    only_in_other = [other_norm_to_raw[k] for k in only_in_other_norm]
+
+    raise ValueError(
+        "Worker mismatch detected between sources. "
+        f"{base_label} only: {only_in_base}. "
+        f"{other_label} only: {only_in_other}."
+    )
+
+
 def _parse_sheet_date(value):
     if pd.isna(value):
         return None
@@ -185,7 +223,7 @@ def extract_leave_and_rota_from_calender(calender_df):
 
 
     #print(leave_requests)
-    return leave_requests, work_requests#, rota_data
+    return leave_requests, work_requests, worker_names
 
 def extract_shift_assignments_from_rota(rota_df):
     """
@@ -319,11 +357,43 @@ def load_workers():
         workers_days.append({"initial": initial, "name": name, "availability": availability, "requests": requests})
 
 
-    leave_requests, work_requests = extract_leave_and_rota_from_calender(calender_df)
+    leave_requests, work_requests, calendar_workers = extract_leave_and_rota_from_calender(calender_df)
 
     assignments, assignments_by_worker = extract_shift_assignments_from_rota(rota_df)
     print(assignments)
 
+    # Validate cross-source worker matching before constructing Worker objects.
+    days_workers = [w["name"] for w in workers_days]
+    days_initials = [w["initial"] for w in workers_days]
+
+    _validate_worker_alignment(
+        base_workers=days_workers,
+        other_workers=calendar_workers,
+        base_label="Days tab worker names",
+        other_label="Calendar tab worker names",
+    )
+
+    _validate_worker_alignment(
+        base_workers=days_initials,
+        other_workers=list(assignments_by_worker.keys()),
+        base_label="Days tab worker initials",
+        other_label="Rota tab assigned initials",
+    )
+
+    _validate_worker_alignment(
+        base_workers=days_initials,
+        other_workers=list(rota_b_assignments.keys()),
+        base_label="Days tab worker initials",
+        other_label="Rota B assignments initials",
+    )
+
+    _validate_worker_alignment(
+        base_workers=days_initials,
+        other_workers=list(priors_map.keys()),
+        base_label="Days tab worker initials",
+        other_label="Running totals worker initials",
+    )
+
     logger.debug("Using Running totals priors for oncall/twilight/weekend A only")
 
     workers = []
@@ -455,7 +525,7 @@ def main(
         weeks_to_rota=weeks,
         balance_offset_modifier=bom,
         use_previous_shifts=True,
-        name="cons rota 2026 july",
+        name="cons rota test",
         allow_force_assignment_with_leave_conflict=True,
         constraint_options=RotaConstraintOptions(
             balance_weekends=True,