import json
from datetime import datetime
import collections

data = json.load(open('/Users/dvapp/Documents/HRM/scratch/legacy_roster.json'))

# employee_code -> day_of_week (0=Mon...6=Sun) -> shift_name
patterns = collections.defaultdict(dict)

for entry in data:
    if not entry['date'].startswith('2026-01'): continue
    
    dt = datetime.strptime(entry['date'], '%Y-%m-%d')
    dow = dt.weekday() # 0 is Monday
    
    code = entry['code']
    shift = entry['shift_name']
    is_rd = entry['is_rest_day']
    
    # Store the most frequent shift for this day of week for this employee
    if code not in patterns: patterns[code] = {}
    if dow not in patterns[code]: patterns[code][dow] = []
    
    patterns[code][dow].append({'shift': shift, 'is_rd': is_rd})

# Distill patterns (take the most common assignment for each DOW)
final_patterns = {}
for code, dows in patterns.items():
    final_patterns[code] = {}
    for dow, assignments in dows.items():
        # Just take the first one for now, usually it's consistent
        final_patterns[code][dow] = assignments[0]

with open('/Users/dvapp/Documents/HRM/scratch/employee_patterns.json', 'w') as f:
    json.dump(final_patterns, f, indent=2)

print(f"Generated patterns for {len(final_patterns)} employees")