Function parse_rule

script/compressor.py:202–240

Convert raw text to AdblockRule objects.

(rule_text: str)

# --- Compression Logic (Optimized) ---

def parse_rule(rule_text: str) -> List[AdblockRule]:
    """Convert raw text to AdblockRule objects."""
    rule_text = rule_text.strip()
    res = []
    
    # 1. Domain Lists (DOMAIN, example.com)
    m_domain = RE_DOMAIN_SUFFIX_LIST.match(rule_text)
    if m_domain:
        host = m_domain.group(1)
        if not RE_IP_LIKE.match(host):
            res.append(AdblockRule(f"||{host}^", True, host, rule_text))
        return res

    # 2. /etc/hosts
    if RE_ETC_HOSTS.match(rule_text):
        parts = rule_text.split()
        hosts = [p for p in parts if p and not p.startswith(('#', '!'))][1:]
        for h in hosts:
            if not RE_IP_LIKE.match(h):
                res.append(AdblockRule(f"||{h}^", True, h, rule_text))
        return res

    # 3. Pure Domain (example.com) - strict check
    if RE_VALID_DOMAIN_CHARS.match(rule_text) and '.' in rule_text and not rule_text.startswith('/') and not rule_text.startswith('||'):
         if not RE_IP_LIKE.match(rule_text):
             res.append(AdblockRule(f"||{rule_text}^", True, rule_text, rule_text))
         return res

    # 4. Standard Adblock matches ||...^
    if rule_text.startswith("||") and rule_text.endswith("^"):
        inner = rule_text[2:-1]
        # Only allow plain domains to be "Compressible", wildcards fall to fallback
        if not RE_IP_LIKE.match(inner) and '*' not in inner:
            res.append(AdblockRule(rule_text, True, inner, rule_text))
            return res

    # Fallback: Uncompressible (includes Wildcards, RegEx, or Invalid Formats)
    res.append(AdblockRule(rule_text, False, None, rule_text))
    return res

def compress_rules(rules: List[str], include_wildcards: bool) -> Tuple[List[str], List[str]]:
    """Compress validated rules."""