mirror of
https://github.com/MemeMeow-Studio/MemeMeow.git
synced 2026-10-17 04:41:01 +08:00
infstellar
93 lines
4.2 KiB
Python
93 lines
4.2 KiB
Python
"""executor 结果文件协议。
|
|
|
|
该模块只处理受控结果目录、普通文件、大小限制和 JSON/schema 验证；HTTP handler
|
|
和进程 supervisor 不应直接读取任意路径。
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import json
|
|
import os
|
|
import re
|
|
import stat
|
|
from pathlib import Path
|
|
from typing import Any, Callable
|
|
|
|
|
|
class ExecutorResultStoreError(RuntimeError):
|
|
"""结果文件无法安全读取或通过 schema 校验。"""
|
|
|
|
def __init__(self, code: str, *, reason_code: str | None = None) -> None:
|
|
"""创建稳定结果错误，并可携带不含输入内容的内部原因码。"""
|
|
self.code = code
|
|
self.reason_code = reason_code if isinstance(reason_code, str) and re.fullmatch(r"[a-z0-9][a-z0-9_.-]{0,127}", reason_code) else None
|
|
super().__init__(code)
|
|
|
|
|
|
class ExecutorResultStore:
|
|
"""绑定一个根目录和结果大小上限的 executor 结果读取器。"""
|
|
|
|
def __init__(self, root: Path, *, filename: str, max_bytes: int) -> None:
|
|
"""创建结果读取器；root 不会被自动解析到符号链接外部。"""
|
|
self.root = Path(root)
|
|
self.filename = filename
|
|
self.max_bytes = max_bytes
|
|
|
|
def read(self, path: Path, *, required_fields: set[str], validator: Callable[[dict[str, Any]], None] | None = None) -> dict[str, Any]:
|
|
"""读取固定层级结果文件并执行有限 schema 验证。"""
|
|
current = self.root
|
|
try:
|
|
root_info = current.lstat()
|
|
relative = Path(path).relative_to(current)
|
|
except (OSError, ValueError) as exc:
|
|
raise ExecutorResultStoreError("agent_result_path_invalid") from exc
|
|
if stat.S_ISLNK(root_info.st_mode) or not stat.S_ISDIR(root_info.st_mode) or len(relative.parts) != 2 or relative.name != self.filename:
|
|
raise ExecutorResultStoreError("agent_result_path_invalid")
|
|
for part in relative.parts:
|
|
current = current / part
|
|
try:
|
|
info = current.lstat()
|
|
except FileNotFoundError as exc:
|
|
raise ExecutorResultStoreError("agent_result_file_missing") from exc
|
|
except OSError as exc:
|
|
raise ExecutorResultStoreError("agent_result_file_unreadable") from exc
|
|
if stat.S_ISLNK(info.st_mode):
|
|
raise ExecutorResultStoreError("agent_result_path_invalid")
|
|
try:
|
|
info = Path(path).lstat()
|
|
if not stat.S_ISREG(info.st_mode):
|
|
raise ExecutorResultStoreError("agent_result_file_unreadable")
|
|
if info.st_size > self.max_bytes:
|
|
raise ExecutorResultStoreError("agent_result_file_too_large")
|
|
descriptor = os.open(path, os.O_RDONLY | getattr(os, "O_NOFOLLOW", 0))
|
|
try:
|
|
raw = os.read(descriptor, self.max_bytes + 1)
|
|
finally:
|
|
os.close(descriptor)
|
|
except ExecutorResultStoreError:
|
|
raise
|
|
except FileNotFoundError as exc:
|
|
raise ExecutorResultStoreError("agent_result_file_missing") from exc
|
|
except OSError as exc:
|
|
raise ExecutorResultStoreError("agent_result_file_unreadable") from exc
|
|
if len(raw) > self.max_bytes:
|
|
raise ExecutorResultStoreError("agent_result_file_too_large")
|
|
try:
|
|
value = json.loads(raw.decode("utf-8"))
|
|
except (UnicodeDecodeError, json.JSONDecodeError, RecursionError) as exc:
|
|
raise ExecutorResultStoreError("agent_result_file_invalid_json") from exc
|
|
if not isinstance(value, dict):
|
|
raise ExecutorResultStoreError("agent_result_file_schema_invalid", reason_code="result_object_required")
|
|
if not required_fields.issubset(value):
|
|
raise ExecutorResultStoreError("agent_result_file_schema_invalid", reason_code="result_schema_invalid")
|
|
if validator is not None:
|
|
try:
|
|
validator(value)
|
|
except Exception as exc: # noqa: BLE001 - validator 只允许映射为稳定错误
|
|
reason_code = getattr(exc, "code", None)
|
|
raise ExecutorResultStoreError("agent_result_file_schema_invalid", reason_code=reason_code) from exc
|
|
return value
|
|
|
|
|
|
__all__ = ["ExecutorResultStore", "ExecutorResultStoreError"]
|