#!/usr/bin/env python3
"""
Analyze column widths from YAML specs.
Usage: python3 analyze_column_widths.py [stand]
  stand: test (default), dev, prod
"""
import os, sys, yaml

STAND = sys.argv[1] if len(sys.argv) > 1 else 'test'
REPO = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
YAML_DIR = f'{REPO}/generated/{STAND}/resources_yaml'

if not os.path.isdir(YAML_DIR):
    print(f'Error: {YAML_DIR} not found')
    sys.exit(1)

all_params = []
nested_count = 0

def extract_params(svc, op, params_list, depth=0):
    global nested_count
    for p in params_list:
        if not isinstance(p, dict): continue
        all_params.append((svc, op, p, depth))
        sub = p.get('sub_params', [])
        if sub:
            nested_count += len(sub)
            extract_params(svc, op, sub, depth + 1)

for fname in sorted(os.listdir(YAML_DIR)):
    if not fname.endswith('.yaml'): continue
    with open(os.path.join(YAML_DIR, fname)) as f:
        data = yaml.safe_load(f)
    svc = data.get('name', fname)
    for op in data.get('operations', []):
        if op.get('action') not in ('create', 'modify'): continue
        extract_params(svc, op.get('name', '?'), op.get('params', []))

print(f'Stand: {STAND} | Files: {len(os.listdir(YAML_DIR))}')
print(f'Total params: {len(all_params)} | Nested subparams: {nested_count}')

def stats(items, label):
    if not items: return
    lens = [len(str(t[2])) for t in items]
    n = len(lens)
    mx, avg = max(lens), sum(lens)/n
    print(f'\n--- {label} (n={n}) ---')
    print(f'  MAX: {mx} | AVG: {avg:.0f}')
    print(f'  >30: {sum(l in (l>30 for l in lens))/n*100:.0f}%  >50: {sum(1 for l in lens if l>50)/n*100:.0f}%  >100: {sum(1 for l in lens if l>100)/n*100:.0f}%  >200: {sum(1 for l in lens if l>200)/n*100:.0f}%  >500: {sum(1 for l in lens if l>500)/n*100:.0f}%')

codes   = [(s,o,p.get('code','')) for s,o,p,d in all_params]
dtypes  = [(s,o,p.get('data_type', p.get('type',''))) for s,o,p,d in all_params]
defs    = [(s,o,str(p['default'])) for s,o,p,d in all_params if 'default' in p]
descrs  = [(s,o,p.get('descr','') or p.get('man','')) for s,o,p,d in all_params if p.get('descr') or p.get('man')]

constraint_keys = ['min_value','max_value','regex','value_list','max_length','min_length','func','ref_svc_id','unique']
constraints = []
for s,o,p,d in all_params:
    parts = []
    for k in constraint_keys:
        v = p.get(k)
        if v is not None: parts.append(f'{k}={v}')
    if parts: constraints.append((s,o,' | '.join(parts)))

stats(codes, 'Code'); stats(dtypes, 'Type'); stats(defs, 'Default')
stats(descrs, 'Description'); stats(constraints, 'Constraints')

for name, arr in [('Code', codes), ('Type', dtypes), ('Default', defs),
                   ('Description', descrs), ('Constraints', constraints)]:
    top = sorted(arr, key=lambda x: len(str(x[2])), reverse=True)[:3]
    print(f'\nTop-3 longest {name}:')
    for s, o, v in top:
        print(f'  [{s}/{o}] len={len(str(v))}: {str(v)[:120]}')