code/lib_skill.py

"""Shared helpers: SKILL.md frontmatter parsing and CJK ratio."""
import re, yaml
_Loader = getattr(yaml, "CSafeLoader", yaml.SafeLoader)

FM = re.compile(r"\A?---\s*\r?\n(.*?)\r?\n---\s*(?:\r?\n|\Z)", re.S)
CJK = re.compile(r"[぀-ヿ㐀-䶿一-鿿가-힯]")


def parse_skill_md(text):
    """Returns (frontmatter_valid, name, description, frontmatter_raw, body)."""
    if not text: return False, None, None, None, text or ""
    m = FM.match(text)
    if not m: return False, None, None, None, text
    raw, body = m.group(1), text[m.end():]
    try:
        d = yaml.load(raw, Loader=_Loader)  # libyaml C loader: ~10x faster than pure-Python safe_load
        if not isinstance(d, dict): raise ValueError
        name, desc = d.get("name"), d.get("description")
        return bool(name and desc), None if name is None else str(name), None if desc is None else str(desc), raw, body
    except Exception:  # lenient fallback for invalid YAML (common: unquoted colons)
        g = lambda k: (re.search(rf"^{k}:\s*(.+)$", raw, re.M) or [None, None])[1]
        return False, g("name"), g("description"), raw, body


def cjk_ratio(text):
    return len(CJK.findall(text)) / max(len(text), 1) if text else 0.0