"""Checks for names and file contents before anything goes into a draft (SPEC C-10, C-12, C-16). No code runs here."""

import ast
import difflib
import re
from dataclasses import dataclass

MAX_FILE_BYTES = 100 * 1024
NAME = re.compile(r"^[A-Za-z0-9._-]{1,100}$")
# Bot users can't touch the CI of the uploads repo or its helper scripts (SPEC C-16, ADR-014)
PROTECTED_PREFIXES = (".sourcecraft/", ".gitea/", ".tgit/", ".git/")
PROTECTED_FILES = {"ruff.toml"}
_FENCE = re.compile(r"^\s*```[\w+-]*[ \t]*\n(.*?)\n?```\s*$", re.DOTALL)


class ValidationError(ValueError):
    pass


@dataclass(frozen=True)
class SyntaxIssue:
    line: int
    message: str


def check_name(name: str) -> str:
    name = name.strip()
    if not NAME.match(name) or name in {".", ".."}:
        raise ValidationError("name")
    return name


def is_protected(path: str) -> bool:
    return path in PROTECTED_FILES or any(path.startswith(p) or path + "/" == p for p in PROTECTED_PREFIXES)


def join(directory: str, name: str) -> str:
    return f"{directory}/{name}" if directory else name


def strip_fence(text: str) -> str:
    """Code pasted as a Telegram message often comes wrapped in ``` — keep only what's inside."""
    match = _FENCE.match(text)
    return match.group(1) if match else text


def check_content(data: bytes) -> str:
    """Size, encoding, NUL bytes. Returns the decoded text."""
    if len(data) > MAX_FILE_BYTES:
        raise ValidationError("too_big")
    if b"\0" in data:
        raise ValidationError("binary")
    try:
        return data.decode("utf-8")
    except UnicodeDecodeError:
        raise ValidationError("encoding") from None


def python_syntax(path: str, text: str) -> SyntaxIssue | None:
    if not path.endswith(".py"):
        return None
    try:
        ast.parse(text, filename=path)
    except SyntaxError as e:
        return SyntaxIssue(line=e.lineno or 0, message=e.msg)
    return None


def diff_stat(old: str, new: str) -> tuple[int, int]:
    added = removed = 0
    for line in difflib.ndiff(old.splitlines(), new.splitlines()):
        if line.startswith("+ "):
            added += 1
        elif line.startswith("- "):
            removed += 1
    return added, removed


def check_message(message: str) -> str:
    message = message.strip()
    if not 1 <= len(message) <= 200 or len(message.splitlines()[0]) > 72:
        raise ValidationError("message")
    return message