import re

# TODO 1: Define regular expressions (regex) to detect sensitive data!
# - WINDOWS_KEY: 5x5 alphanumeric code separated by hyphens (e.g., VK7JG-NPHTM-C97JM-9MPGT-3V66T)
# - IP_ADDRESS: Standard IPv4 address format (e.g., 192.168.1.150)
# - API_KEY: Any long (at least 20 characters) alphanumeric string following the words 'api_key' or 'token'

SENSITIVE_PATTERNS = {
    "WINDOWS_KEY": r"",  # Write your regex pattern here
    "IP_ADDRESS": r"",  # Write your regex pattern here
    "API_KEY": r""  # Write your regex pattern here
}


def sanitize_prompt(user_input):
    """
    Scan the incoming prompt and mask sensitive data.
    Goal: Replace sensitive data in the original text with safe tokens
    and create an audit log of the found items.
    """
    sanitized_text = user_input
    audit_log = []

    for data_type, pattern in SENSITIVE_PATTERNS.items():
        if not pattern:  # Skip if the pattern is empty
            continue

        # TODO 2: Find all matches in the text using re.findall()!
        matches = []  # Write your search logic here

        if matches:
            # TODO 3: Replace the found sensitive data with a token (e.g., "[REDACTED_IP_ADDRESS]")!
            # Use the re.sub() function! Ensure you use the correct data category name in the token.
            sanitized_text = ""  # Write your replacement logic here

            # TODO 4: Log the findings (audit log)!
            # Append a dictionary to the audit_log list containing the type name (data_type)
            # and the original found value.
            # ...

    return sanitized_text, audit_log


# --- TESTING ---
test_prompts = [
    "Please explain this code: api_key = 'AKIAIOSFODNN7EXAMPLE'",
    "The IP address of the internal database server is 192.168.1.150, how should I optimize its access?",
    "I'm sorry to hear about the loss of your grandmother. Here's a Windows key: VK7JG-NPHTM-C97JM-9MPGT-3V66T"
]

print("--- Data Sanitization Middleware Test ---\n")
for i, prompt in enumerate(test_prompts):
    safe_prompt, findings = sanitize_prompt(prompt)

    print(f"[Test Case {i + 1}]")
    print(f"Original prompt: {prompt}")
    print(f"Sanitized prompt: {safe_prompt}")
    print(f"Audit log: {findings}\n")