scrypt/sandbox/mirror.py

"""Opt-in mirror: copy a shallow, filtered snapshot of the player's real
home directory into the VFS, so the Warden's threats land personally.

Strictly read-only against the host. Hard caps on depth, entry count, and
file size; secret-looking paths and non-text files are never ingested
(names only). This module is the ONLY sandbox code allowed to import os.
"""

from __future__ import annotations

import os
from pathlib import Path

from .vfs import VFS

MAX_DEPTH = 3
MAX_ENTRIES = 250
MAX_TEXT_BYTES = 8_192
TEXT_SUFFIXES = {".txt", ".md", ".rst", ".csv", ".log"}

# Never descend into these; never ingest even the names of their contents.
FORBIDDEN_DIRS = {
    ".ssh", ".aws", ".gnupg", ".password-store", ".config", ".local", ".cache",
    ".mozilla", ".thunderbird", "node_modules", ".git", ".venv", "venv",
}
SECRET_MARKERS = ("key", "token", "secret", "passwd", "password", "credential", ".env")


def _suspicious(name: str) -> bool:
    lower = name.lower()
    return any(marker in lower for marker in SECRET_MARKERS)


def mirror_home(vfs: VFS, source: Path | None = None) -> int:
    """Ingest a snapshot under /home/drifter. Returns entries mirrored.

    Call only after explicit player consent.
    """
    source = source or Path.home()
    count = 0

    def walk(real_dir: Path, virtual_dir: str, depth: int) -> None:
        nonlocal count
        if depth > MAX_DEPTH or count >= MAX_ENTRIES:
            return
        try:
            entries = sorted(real_dir.iterdir(), key=lambda p: p.name.lower())
        except (PermissionError, OSError):
            return
        for entry in entries:
            if count >= MAX_ENTRIES:
                return
            name = entry.name
            if name in FORBIDDEN_DIRS or _suspicious(name):
                continue
            if entry.is_symlink():
                continue
            virtual_path = f"{virtual_dir}/{name}"
            if entry.is_dir():
                vfs.mkdir(virtual_path)
                count += 1
                walk(entry, virtual_path, depth + 1)
            elif entry.is_file():
                content = "<mirrored file>"
                try:
                    if (
                        entry.suffix.lower() in TEXT_SUFFIXES
                        and entry.stat().st_size <= MAX_TEXT_BYTES
                    ):
                        content = entry.read_text(encoding="utf-8", errors="replace")
                except (PermissionError, OSError):
                    pass
                vfs.write(virtual_path, content)
                count += 1

    walk(source, "/home/drifter", 1)
    vfs.chdir("/home/drifter")
    return count