第一步

This commit is contained in:
amb
2026-09-01 11:53:59 +08:00
commit 47bf6cc5ca
66 changed files with 6501 additions and 0 deletions
View File
+28
View File
@@ -0,0 +1,28 @@
"""存储服务抽象(扩展接口 2:本地文件 → MinIO/S3)。"""
from typing import Protocol, runtime_checkable
@runtime_checkable
class StorageService(Protocol):
"""文件存储统一接口。所有业务代码通过此接口读写文件,禁止直接 open()/Path()。"""
def save(self, key: str, data: bytes) -> str:
"""保存数据,返回实际存储路径。"""
...
def read(self, key: str) -> bytes:
"""读取数据。"""
...
def delete(self, key: str) -> None:
"""删除文件。"""
...
def exists(self, key: str) -> bool:
"""文件是否存在。"""
...
def get_size(self, key: str) -> int:
"""获取文件大小(字节)。"""
...
+67
View File
@@ -0,0 +1,67 @@
"""本地文件系统存储实现(MVP)。"""
from pathlib import Path
from app.core.config import get_settings
from app.core.logging import get_logger
logger = get_logger(__name__)
class LocalStorageService:
"""文件存储到服务器本地 data/ 目录。
key 格式示例:
users/{uid}/kbs/{kb_id}/original/{rand}_{safe_name}
users/{uid}/kbs/{kb_id}/markdown/{rand}.md
实际路径 = storage_root / key
"""
def __init__(self) -> None:
settings = get_settings()
self._root = settings.storage_root_path
self._root.mkdir(parents=True, exist_ok=True)
def _resolve(self, key: str) -> Path:
"""解析 key 为绝对路径,同时防路径穿越。"""
path = (self._root / key).resolve()
# 安全:确保解析后路径仍在 root 内
if not str(path).startswith(str(self._root)):
raise ValueError(f"路径穿越检测:key={key}")
return path
def save(self, key: str, data: bytes) -> str:
path = self._resolve(key)
path.parent.mkdir(parents=True, exist_ok=True)
path.write_bytes(data)
logger.debug("Saved file: %s (%d bytes)", key, len(data))
return key
def read(self, key: str) -> bytes:
path = self._resolve(key)
return path.read_bytes()
def delete(self, key: str) -> None:
path = self._resolve(key)
if path.exists():
path.unlink()
logger.debug("Deleted file: %s", key)
def exists(self, key: str) -> bool:
return self._resolve(key).exists()
def get_size(self, key: str) -> int:
path = self._resolve(key)
return path.stat().st_size
# 单例
_instance: LocalStorageService | None = None
def get_storage() -> LocalStorageService:
global _instance
if _instance is None:
_instance = LocalStorageService()
return _instance
+50
View File
@@ -0,0 +1,50 @@
"""对象存储 key 生成。
布局:users/{user_id}/kbs/{kb_id}/original/{rand8}_{safe_filename}
users/{user_id}/kbs/{kb_id}/markdown/{rand8}.md
规则(需求 §27 / §23):
- 不信任用户文件名:取 basename,剔除路径穿越与控制字符,截断长度。
- 主体是内部 UUID,公共 URL 使用 token,绝不回显此 key。
"""
import re
import secrets
import uuid
_SAFE_NAME_RE = re.compile(r"[^A-Za-z0-9._\-]+")
_MAX_FILENAME_LEN = 120
def sanitize_filename(original: str) -> str:
name = original.replace("\\", "/").split("/")[-1]
name = _SAFE_NAME_RE.sub("_", name).strip("._")
if not name:
name = "file"
return name[:_MAX_FILENAME_LEN]
def original_object_key(
*,
user_id: str,
knowledge_base_id: str,
document_id: str,
original_filename: str,
) -> str:
rand = secrets.token_urlsafe(6) # ~8 字符
safe = sanitize_filename(original_filename)
return (
f"users/{user_id}/kbs/{knowledge_base_id}/docs/{document_id}/original/{rand}_{safe}"
)
def markdown_object_key(
*,
user_id: str,
knowledge_base_id: str,
document_id: str,
) -> str:
rand = secrets.token_urlsafe(6)
return (
f"users/{user_id}/kbs/{knowledge_base_id}/docs/{document_id}/markdown/{rand}.md"
)