import re, json, sys, io
from openpyxl import load_workbook
from unidecode import unidecode

sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')

ps = open(r'D:\VNPT-M\Tool\audit_tool\NEW\Update\audit_lhdsin\audit_check_script\audit_cis_windows.ps1', 'r', encoding='utf-8').read()

# 1. Static checks
static_checks = re.findall(r'Write-CheckResult\s+"([^$"].+?)"\s+"(.+?)"', ps)
criteria = [(cid, desc) for cid, desc in static_checks]
print(f'Static checks: {len(criteria)}')

# 2. Variable-based checks - expand from function calls
# Test-FirewallProfile
fw_calls = re.findall(r'Test-FirewallProfile\s+"(\w+)"\s+"([\d.]+)"\s+"(\w+)"', ps)
for profile, prefix, label in fw_calls:
    criteria.append((f'{prefix}.1', f"Thiet lap trang thai 'Windows Firewall: {label} : Firewall state'"))
    criteria.append((f'{prefix}.2', f"Thiet lap trang thai 'Windows Firewall: {label} : Inbound connections'"))
    criteria.append((f'{prefix}.3', f"Thiet lap trang thai 'Windows Firewall: {label} : Outbound connections'"))

# Test-EventLogPolicy
log_calls = re.findall(r'Test-EventLogPolicy\s+"(\w+)"\s+"([\d.]+)"\s+"(\w+)"', ps)
for logname, prefix, label in log_calls:
    criteria.append((f'{prefix}.1', f"Thiet lap chinh sach '{label} : Control Event Log behavior when the log file reaches its maximum size'"))
    criteria.append((f'{prefix}.2', f"Thiet lap chinh sach '{label} : Specify the maximum log file size (KB)'"))

print(f'Total after expanding: {len(criteria)}')

# Build lookup: normalized_full_text -> full_text  
ps_criteria = {}
for cid, desc in criteria:
    full = f"{cid}. {desc}"
    key = unidecode(full)
    ps_criteria[key] = full

# Load Excel
wb = load_workbook(r'D:\VNPT-M\Tool\audit_tool\NEW\Update\audit_lhdsin\config\Windows_Checklist.xlsx')
ws = wb.worksheets[0]

excel_rows = {}
unmatched = {}

# Pass 1: match by full text (column B has the full criteria text)
for row in range(1, ws.max_row + 1):
    val_a = str(ws.cell(row=row, column=1).value or "").strip()
    val_b = str(ws.cell(row=row, column=2).value or "").strip()
    combined = f"{val_a} {val_b}".strip()
    norm = unidecode(combined)
    if norm in ps_criteria:
        excel_rows[row] = ps_criteria[norm]
        del ps_criteria[norm]

# Pass 2: match just the CheckId in column A
still_unmatched = dict(ps_criteria)
for cid_key, full_text in still_unmatched.items():
    cid = full_text.split('. ')[0] if '. ' in full_text else full_text.split()[0]
    for row in range(1, ws.max_row + 1):
        val_a = str(ws.cell(row=row, column=1).value or "").strip()
        if val_a and cid in val_a and row not in excel_rows:
            excel_rows[row] = full_text
            del ps_criteria[cid_key]
            break

# Pass 3: try matching by normalized column B only
still_unmatched2 = dict(ps_criteria)
for cid_key, full_text in still_unmatched2.items():
    cid = full_text.split('. ')[0]
    norm_cid = unidecode(cid)
    for row in range(1, ws.max_row + 1):
        val_b = str(ws.cell(row=row, column=2).value or "").strip()
        if val_b:
            norm_b = unidecode(val_b)
            if norm_cid in norm_b.split()[0] and row not in excel_rows:
                excel_rows[row] = full_text
                del ps_criteria[cid_key]
                break

print(f'\nMatched rows: {len(excel_rows)}')
print(f'Unmatched: {len(ps_criteria)}')

if ps_criteria:
    print('\nUnmatched criteria (first 10):')
    for i, (k, v) in enumerate(list(ps_criteria.items())[:10]):
        print(f'  [{v[:70]}]')

# Build new config
new_data = []
for row in sorted(excel_rows.keys()):
    new_data.append({str(row): excel_rows[row]})

new_config = {"data": new_data}

output_path = r'D:\VNPT-M\Tool\audit_tool\NEW\Update\audit_lhdsin\config\windows_config.json'
with open(output_path, 'w', encoding='utf-8') as f:
    json.dump(new_config, f, ensure_ascii=False, indent=4)

print(f'\nNew config written to: {output_path}')
print(f'Total entries: {len(new_data)}')