Class RuleValidator

script/compressor.py:61–198

Port of the JavaScript Validator class. Performs strict validation and cleans up invalid rules + preceding comments.

# --- Logic Ported from JavaScript (Validator) ---

class RuleValidator:
    """
    Port of the JavaScript Validator class.
    Performs strict validation and cleans up invalid rules + preceding comments.
    """
    def __init__(self, allow_ip: bool = False):
        self.allow_ip = allow_ip
        self._prev_removed = False

    @staticmethod
    def is_comment(line: str) -> bool:
        return line.lstrip().startswith(('!', '#'))

    def _valid_hostname(self, hostname: str, rule_text: str, has_limit_mod: bool) -> bool:
        """Checks if hostname is valid. Imitates tldts logic with standard lib."""
        hostname = hostname.strip().lower()
        
        if not hostname:
            return False

        # IP Check
        is_ip = bool(re.match(r"^\d{1,3}(\.\d{1,3}){3}$", hostname))
        if is_ip and not self.allow_ip:
            logger.debug(f"Invalid hostname (IP not allowed): {hostname} in {rule_text}")
            return False

        # Public Suffix Check (Heuristic approximation for stdlib)
        if '.' not in hostname and not is_ip and not has_limit_mod and hostname != "localhost":
            logger.debug(f"Matching whole TLD/Suffix not allowed: {hostname} in {rule_text}")
            return False

        return True

    def _valid_etc_hosts(self, rule: str) -> bool:
        parts = rule.split()
        # Remove comments/IP
        hosts = [p for p in parts if not p.startswith(('#', '!'))]
        if len(hosts) < 2: 
            return False
        
        hostnames = hosts[1:] # 0 is IP
        return all(self._valid_hostname(h, rule, False) for h in hostnames)

    def _valid_adblock(self, rule: str) -> bool:
        # 1. Parse Modifiers (Simplified parser)
        has_limit_mod = False
        if '$' in rule:
            try:
                base, opts = rule.split('$', 1)
                options = opts.split(',')
                for opt in options:
                    name = opt.split('=', 1)[0].strip()
                    if name not in SUPPORTED_MODIFIERS:
                        logger.debug(f"Unsupported modifier {name}: {rule}")
                        return False
                    if name in ANY_PATTERN_MODIFIER:
                        has_limit_mod = True
            except ValueError: