scripts/check.py

#!/usr/bin/env python3
"""Scan human-facing text for a small set of exposure-aware hard leaks."""

from __future__ import annotations

import argparse
import json
import re
import sys
from pathlib import Path
from typing import Any


EXPOSURES = ("private", "internal", "external", "public")
ALLOWED_PROFILE_KEYS = {
    "profile_version",
    "author_label",
    "languages",
    "mode_preferences",
    "global_prefer",
    "global_avoid",
}
PATTERNS = {
    "local_path": re.compile(r"(?<![A-Za-z])[A-Za-z]:[\\/]|(?<![\w.])/(?:home|Users|tmp)/"),
    "unc_path": re.compile(
        r'(?<!\\)\\\\[A-Za-z0-9][A-Za-z0-9._-]*(?:\\[^\\/:*?"<>|\r\n]+)+'
    ),
    "internal_marker": re.compile(
        r"【(?:待校正|TODO|信息缺失)|(?:internal|内部)\s+(?:SOP|workflow|marker)",
        re.IGNORECASE,
    ),
    "credential_assignment": re.compile(
        r"(?<![\w-])"
        r"(?P<label>(?:[A-Za-z][A-Za-z0-9]*[_-]){0,2}api[_ -]?(?:key|token)|"
        r"access[_ -]?token|auth[_ -]?token|refresh[_ -]?token|"
        r"client[_ -]?secret|secret|token|password|passwd)"
        r"(?![\w-])\s*[:=]\s*"
        r"(?P<quote>[\"']?)"
        r"(?P<value>[A-Za-z0-9][A-Za-z0-9_./+=:@!#$%&*?~-]{7,})"
        r"(?P=quote)",
        re.IGNORECASE,
    ),
    "credential_bearer": re.compile(
        r"(?<![\w-])(?P<label>authorization)\s*:\s*bearer\s+"
        r"(?P<value>[A-Za-z0-9][A-Za-z0-9._~+/\-=]{7,})",
        re.IGNORECASE,
    ),
}
EXTERNAL_ONLY_FINDINGS = {"local_path", "unc_path", "internal_marker"}
CREDENTIAL_FINDINGS = {"credential_assignment", "credential_bearer"}
NON_SECRET_VALUES = {
    "authentication",
    "authorization",
    "changeme",
    "example",
    "hidden",
    "masked",
    "omitted",
    "placeholder",
    "redacted",
    "your_token",
    "your_token_here",
}


class CheckError(RuntimeError):
    pass


def validate_profile(path: Path) -> dict[str, Any]:
    profile = json.loads(path.read_text(encoding="utf-8"))
    if not isinstance(profile, dict):
        raise CheckError("voice profile must be a JSON object")
    extra = sorted(set(profile) - ALLOWED_PROFILE_KEYS)
    if extra:
        raise CheckError(f"voice profile has unknown fields: {extra}")
    if not isinstance(profile.get("profile_version"), str):
        raise CheckError("voice profile requires string profile_version")
    languages = profile.get("languages")
    if (
        not isinstance(languages, list)
        or not languages
        or not set(languages) <= {"zh-CN", "en"}
    ):
        raise CheckError("voice profile languages must be a non-empty zh-CN/en list")
    if len(languages) != len(set(languages)):
        raise CheckError("voice profile languages must not contain duplicates")
    if "author_label" in profile and not isinstance(profile["author_label"], str):
        raise CheckError("voice profile author_label must be a string")
    for field in ("global_prefer", "global_avoid"):
        if field in profile and (
            not isinstance(profile[field], list)
            or not all(isinstance(item, str) for item in profile[field])
        ):
            raise CheckError(f"voice profile {field} must be a string list")
    preferences = profile.get("mode_preferences", {})
    if not isinstance(preferences, dict):
        raise CheckError("voice profile mode_preferences must be an object")
    for mode, values in preferences.items():
        if not isinstance(values, dict) or set(values) - {"prefer", "avoid"}:
            raise CheckError(f"voice profile mode {mode} must contain prefer/avoid only")
        for field, items in values.items():
            if not isinstance(items, list) or not all(isinstance(item, str) for item in items):
                raise CheckError(f"voice profile mode {mode}.{field} must be a string list")
    return profile


def scan_text(text: str, exposure: str) -> list[dict[str, str]]:
    findings: list[dict[str, str]] = []
    for kind, pattern in PATTERNS.items():
        if kind in EXTERNAL_ONLY_FINDINGS and exposure not in {"external", "public"}:
            continue
        match = next(
            (
                candidate
                for candidate in pattern.finditer(text)
                if kind not in CREDENTIAL_FINDINGS
                or candidate.group("value").casefold() not in NON_SECRET_VALUES
            ),
            None,
        )
        if match:
            if kind == "credential_bearer":
                preview = f"{match.group('label')}: Bearer [REDACTED]"
            elif kind == "credential_assignment":
                preview = f"{match.group('label')}=[REDACTED]"
            else:
                preview = match.group(0)[:80]
            findings.append(
                {
                    "kind": kind,
                    "severity": "P0",
                    "preview": preview,
                }
            )
    return findings


def build_parser() -> argparse.ArgumentParser:
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("--exposure", required=True, choices=EXPOSURES)
    source = parser.add_mutually_exclusive_group(required=True)
    source.add_argument("--text")
    source.add_argument("--file")
    parser.add_argument("--voice-profile")
    return parser


def main() -> int:
    args = build_parser().parse_args()
    try:
        if args.file:
            text = Path(args.file).read_text(encoding="utf-8")
        else:
            text = args.text
        profile = validate_profile(Path(args.voice_profile)) if args.voice_profile else None
        result = {
            "exposure": args.exposure,
            "voice_profile_loaded": profile is not None,
            "findings": scan_text(text, args.exposure),
        }
    except (OSError, CheckError, ValueError, json.JSONDecodeError) as exc:
        print(f"check: {exc}", file=sys.stderr)
        return 1
    print(json.dumps(result, ensure_ascii=False, indent=2))
    return 2 if result["findings"] else 0


if __name__ == "__main__":
    raise SystemExit(main())