461 lines
16 KiB
Python
461 lines
16 KiB
Python
import hashlib
|
||
import os
|
||
import time
|
||
import traceback
|
||
from pathlib import Path
|
||
|
||
import aiofiles
|
||
from langchain_text_splitters import MarkdownTextSplitter
|
||
|
||
from yuxi import config
|
||
from yuxi.config.static.models import EmbedModelInfo
|
||
from yuxi.utils import hashstr, logger
|
||
from yuxi.utils.datetime_utils import utc_isoformat
|
||
|
||
|
||
def validate_file_path(file_path: str, db_id: str = None) -> str:
|
||
"""
|
||
验证文件路径安全性，防止路径遍历攻击 - 支持本地文件和MinIO URL
|
||
|
||
Args:
|
||
file_path: 要验证的文件路径或MinIO URL
|
||
db_id: 数据库ID，用于获取知识库特定的上传目录
|
||
|
||
Returns:
|
||
str: 规范化后的安全路径
|
||
|
||
Raises:
|
||
ValueError: 如果路径不安全
|
||
"""
|
||
try:
|
||
# 检测是否是MinIO URL，如果是则直接返回（不进行路径遍历检查）
|
||
if is_minio_url(file_path):
|
||
logger.debug(f"MinIO URL detected, skipping path validation: {file_path}")
|
||
return file_path
|
||
|
||
# 规范化路径（仅对本地文件）
|
||
normalized_path = os.path.abspath(os.path.realpath(file_path))
|
||
|
||
# 获取允许的根目录
|
||
from yuxi.knowledge import knowledge_base
|
||
|
||
allowed_dirs = [
|
||
os.path.abspath(os.path.realpath(config.save_dir)),
|
||
]
|
||
|
||
# 如果指定了db_id，添加知识库特定的上传目录
|
||
if db_id:
|
||
try:
|
||
allowed_dirs.append(os.path.abspath(os.path.realpath(knowledge_base.get_db_upload_path(db_id))))
|
||
except Exception:
|
||
# 如果无法获取db路径，使用通用上传目录
|
||
allowed_dirs.append(
|
||
os.path.abspath(os.path.realpath(os.path.join(config.save_dir, "database", "uploads")))
|
||
)
|
||
|
||
# 检查路径是否在允许的目录内
|
||
is_safe = False
|
||
for allowed_dir in allowed_dirs:
|
||
try:
|
||
if normalized_path.startswith(allowed_dir):
|
||
is_safe = True
|
||
break
|
||
except Exception:
|
||
continue
|
||
|
||
if not is_safe:
|
||
logger.warning(f"Path traversal attempt detected: {file_path} (normalized: {normalized_path})")
|
||
raise ValueError(f"Access denied: Invalid file path: {file_path}")
|
||
|
||
return normalized_path
|
||
|
||
except Exception as e:
|
||
logger.error(f"Path validation failed for {file_path}: {e}")
|
||
raise ValueError(f"Invalid file path: {file_path}")
|
||
|
||
|
||
def _unescape_separator(separator: str | None) -> str | None:
|
||
"""将前端传入的字面量转义字符转换为实际字符
|
||
|
||
例如: "\\n\\n\\n" -> "\n\n\n"
|
||
"""
|
||
if not separator:
|
||
return None
|
||
|
||
# 处理常见的转义序列
|
||
separator = separator.replace("\\n", "\n")
|
||
separator = separator.replace("\\r", "\r")
|
||
separator = separator.replace("\\t", "\t")
|
||
separator = separator.replace("\\\\", "\\")
|
||
|
||
return separator
|
||
|
||
|
||
def sanitize_processing_params(params: dict | None) -> dict | None:
|
||
"""移除批次级临时字段，避免写入单文件元数据。"""
|
||
if not params:
|
||
return None
|
||
|
||
sanitized = params.copy()
|
||
sanitized.pop("_preprocessed_map", None)
|
||
sanitized.pop("content_hashes", None)
|
||
return sanitized
|
||
|
||
|
||
def split_text_into_chunks(text: str, file_id: str, filename: str, params: dict = {}) -> list[dict]:
|
||
"""
|
||
将文本分割成块，使用 LangChain 的 MarkdownTextSplitter 进行智能分割
|
||
"""
|
||
chunks = []
|
||
chunk_size = params.get("chunk_size", 1000)
|
||
chunk_overlap = params.get("chunk_overlap", 200)
|
||
|
||
# 获取分隔符并转换为实际字符
|
||
separator = params.get("qa_separator")
|
||
separator = _unescape_separator(separator)
|
||
|
||
# 向后兼容：如果旧配置设置了 use_qa_split=True 但未指定 separator，使用默认分隔符
|
||
use_qa_split = params.get("use_qa_split", False)
|
||
if use_qa_split and not separator:
|
||
separator = "\n\n\n"
|
||
logger.debug("启用了向后兼容模式：use_qa_split=True，使用默认分隔符 \\n\\n\\n")
|
||
|
||
# 使用 MarkdownTextSplitter 进行智能分割
|
||
# MarkdownTextSplitter 会尝试沿着 Markdown 格式的标题进行分割
|
||
text_splitter = MarkdownTextSplitter(
|
||
chunk_size=chunk_size,
|
||
chunk_overlap=chunk_overlap,
|
||
)
|
||
|
||
# 如果设置了分隔符，先分割后以当前的分割逻辑处理
|
||
if separator:
|
||
# 转换分隔符为可视格式（换行符显示为 \n）
|
||
separator_display = separator.replace("\n", "\\n").replace("\r", "\\r").replace("\t", "\\t")
|
||
logger.debug(f"启用预分割模式，使用分隔符: '{separator_display}'")
|
||
pre_chunks = text.split(separator)
|
||
text_chunks = []
|
||
for pre_chunk in pre_chunks:
|
||
if pre_chunk.strip():
|
||
text_chunks.extend(text_splitter.split_text(pre_chunk))
|
||
else:
|
||
text_chunks = text_splitter.split_text(text)
|
||
|
||
# 转换为标准格式
|
||
for chunk_index, chunk_content in enumerate(text_chunks):
|
||
if chunk_content.strip(): # 跳过空块
|
||
chunks.append(
|
||
{
|
||
"id": f"{file_id}_chunk_{chunk_index}",
|
||
"content": chunk_content, # .strip(),
|
||
"file_id": file_id,
|
||
"filename": filename,
|
||
"chunk_index": chunk_index,
|
||
"source": filename,
|
||
"chunk_id": f"{file_id}_chunk_{chunk_index}",
|
||
}
|
||
)
|
||
|
||
logger.debug(f"Successfully split text into {len(chunks)} chunks using MarkdownTextSplitter")
|
||
return chunks
|
||
|
||
|
||
async def calculate_content_hash(data: bytes | bytearray | str | os.PathLike[str] | Path) -> str:
|
||
"""
|
||
计算文件内容的 SHA-256 哈希值。
|
||
|
||
Args:
|
||
data: 文件内容的二进制数据或文件路径
|
||
|
||
Returns:
|
||
str: 十六进制哈希值
|
||
"""
|
||
sha256 = hashlib.sha256()
|
||
|
||
if isinstance(data, (bytes, bytearray)):
|
||
sha256.update(data)
|
||
return sha256.hexdigest()
|
||
|
||
if isinstance(data, (str, os.PathLike, Path)):
|
||
path = Path(data)
|
||
async with aiofiles.open(path, "rb") as file_handle:
|
||
chunk = await file_handle.read(8192)
|
||
while chunk:
|
||
sha256.update(chunk)
|
||
chunk = await file_handle.read(8192)
|
||
|
||
return sha256.hexdigest()
|
||
|
||
# 理论上不会执行到这里，但保留作为防御性编程
|
||
raise TypeError(f"Unsupported data type for hashing: {type(data)!r}") # type: ignore[unreachable]
|
||
|
||
|
||
async def prepare_item_metadata(item: str, content_type: str, db_id: str, params: dict | None = None) -> dict:
|
||
"""
|
||
准备文件或URL的元数据 - 支持本地文件和MinIO文件
|
||
|
||
Args:
|
||
item: 文件路径或MinIO URL
|
||
content_type: 内容类型 ("file" 或 "url")
|
||
db_id: 数据库ID
|
||
params: 处理参数，可选
|
||
"""
|
||
# 检查是否有预处理信息 (针对 URL 转 HTML 文件的情况)
|
||
if params and "_preprocessed_map" in params and item in params["_preprocessed_map"]:
|
||
pre_info = params["_preprocessed_map"][item]
|
||
|
||
# 使用预处理信息
|
||
filename = pre_info.get("filename", item) # 通常是原始 URL
|
||
|
||
# 截断文件名以适应数据库限制 (512 chars)，保留部分后缀信息如果可能
|
||
if len(filename) > 500:
|
||
filename_display = filename[:400] + "..." + filename[-90:]
|
||
else:
|
||
filename_display = filename
|
||
|
||
file_type = "html" # 强制转换为 html 类型，以便后续作为文件处理
|
||
item_path = pre_info["path"] # MinIO path
|
||
content_hash = pre_info["content_hash"]
|
||
|
||
# 使用 item(url) 生成 ID，保证同一 URL 即使多次添加 ID 也不同（配合 time）
|
||
# 或者我们应该基于 hash？不，基于 time 更符合上传逻辑
|
||
file_id = f"file_{hashstr(item + str(time.time()), 6)}"
|
||
|
||
metadata = {
|
||
"database_id": db_id,
|
||
"filename": filename_display,
|
||
"path": item_path,
|
||
"file_type": file_type,
|
||
"status": "processing",
|
||
"created_at": utc_isoformat(),
|
||
"file_id": file_id,
|
||
"content_hash": content_hash,
|
||
"parent_id": params.get("parent_id"),
|
||
}
|
||
|
||
if params:
|
||
safe_params = sanitize_processing_params(params) or {}
|
||
# 覆盖 content_type 为 file，确保后续解析走文件流程（MinIO 下载 -> HTML 解析）
|
||
# 而不是再次尝试作为 URL 抓取
|
||
safe_params["content_type"] = "file"
|
||
safe_params["original_source"] = item # 保存完整 URL 到 JSON 字段，避免数据库字段长度限制
|
||
metadata["processing_params"] = safe_params
|
||
|
||
return metadata
|
||
|
||
if content_type == "file":
|
||
# 检测是否是MinIO URL还是本地文件路径
|
||
if is_minio_url(item):
|
||
# MinIO文件处理
|
||
logger.debug(f"Processing MinIO file: {item}")
|
||
# 从MinIO URL中提取文件名
|
||
if "?" in item:
|
||
# URL可能包含查询参数，去掉它们
|
||
item_clean = item.split("?")[0]
|
||
else:
|
||
item_clean = item
|
||
|
||
# 获取文件名（从路径的最后部分）
|
||
filename = item_clean.split("/")[-1]
|
||
|
||
# 如果文件名包含时间戳，提取原始文件名
|
||
import re
|
||
|
||
timestamp_pattern = r"^(.+)_(\d{13})(\.[^.]+)$"
|
||
match = re.match(timestamp_pattern, filename)
|
||
if match:
|
||
original_filename = match.group(1) + match.group(3)
|
||
# 存储原始文件名用于显示
|
||
filename_display = original_filename
|
||
else:
|
||
filename_display = filename
|
||
|
||
file_type = filename.split(".")[-1].lower() if "." in filename else ""
|
||
item_path = item # 保持MinIO URL作为路径
|
||
|
||
# 从 content_hashes 映射中获取 content_hash
|
||
content_hash = None
|
||
if params and "content_hashes" in params and isinstance(params["content_hashes"], dict):
|
||
content_hash = params["content_hashes"].get(item)
|
||
|
||
if not content_hash:
|
||
raise ValueError(f"Missing content_hash for file: {item}")
|
||
|
||
else:
|
||
# 本地文件处理
|
||
file_path = Path(item)
|
||
file_type = file_path.suffix.lower().replace(".", "")
|
||
filename = file_path.name
|
||
filename_display = filename
|
||
item_path = os.path.relpath(file_path, Path.cwd())
|
||
content_hash = None
|
||
try:
|
||
if file_path.exists():
|
||
content_hash = await calculate_content_hash(file_path)
|
||
except Exception as exc: # noqa: BLE001
|
||
logger.warning(f"Failed to calculate content hash for {file_path}: {exc}")
|
||
|
||
# 生成文件ID
|
||
file_id = f"file_{hashstr(str(item_path) + str(time.time()), 6)}"
|
||
|
||
elif content_type == "url":
|
||
# URL 处理
|
||
filename = item # 使用完整 URL 作为文件名
|
||
filename_display = item
|
||
file_type = "url"
|
||
item_path = item
|
||
content_hash = None # URL 没有 content_hash
|
||
file_id = f"url_{hashstr(item + str(time.time()), 6)}"
|
||
|
||
else:
|
||
raise ValueError(f"Unsupported content_type: {content_type}")
|
||
|
||
metadata = {
|
||
"database_id": db_id,
|
||
"filename": filename_display, # 使用显示用的文件名
|
||
"path": item_path,
|
||
"file_type": file_type,
|
||
"status": "processing",
|
||
"created_at": utc_isoformat(),
|
||
"file_id": file_id,
|
||
"content_hash": content_hash,
|
||
"parent_id": params.get("parent_id") if params else None,
|
||
}
|
||
|
||
# 保存处理参数到元数据
|
||
if params:
|
||
metadata["processing_params"] = sanitize_processing_params(params)
|
||
|
||
return metadata
|
||
|
||
|
||
def merge_processing_params(metadata_params: dict | None, request_params: dict | None) -> dict:
|
||
"""
|
||
合并处理参数：优先使用请求参数，缺失时使用元数据中的参数
|
||
|
||
Args:
|
||
metadata_params: 元数据中保存的参数
|
||
request_params: 请求中提供的参数
|
||
|
||
Returns:
|
||
dict: 合并后的参数
|
||
"""
|
||
merged_params = {}
|
||
|
||
# 首先使用元数据中的参数作为默认值
|
||
if metadata_params:
|
||
merged_params.update(metadata_params)
|
||
|
||
# 然后使用请求参数覆盖（如果提供）
|
||
if request_params:
|
||
merged_params.update(request_params)
|
||
|
||
logger.debug(f"Merged processing params: {metadata_params=}, {request_params=}, {merged_params=}")
|
||
return merged_params
|
||
|
||
|
||
def get_embedding_config(embed_info: dict) -> dict:
|
||
"""
|
||
获取嵌入模型配置（兼容性入口，新代码请直接使用 get_embedding_model_info_by_id）。
|
||
|
||
Args:
|
||
embed_info: 嵌入信息字典
|
||
|
||
Returns:
|
||
dict: 标准化的嵌入配置
|
||
"""
|
||
try:
|
||
assert isinstance(embed_info, dict), f"embed_info must be a dict, got {type(embed_info)}"
|
||
assert "model_id" in embed_info, f"embed_info must contain 'model_id', got {embed_info}"
|
||
logger.warning(f"Using model_id: {embed_info['model_id']}")
|
||
from yuxi.models.embed import get_embedding_model_info_by_id
|
||
|
||
return get_embedding_model_info_by_id(embed_info["model_id"])
|
||
|
||
except AssertionError as e:
|
||
logger.error(f"AssertionError in get_embedding_config: {e}, embed_info={embed_info}")
|
||
|
||
# 兼容性检查：旧版配置字段
|
||
try:
|
||
# 1. 检查 embed_info 是否有效
|
||
if not embed_info or ("model" not in embed_info and "name" not in embed_info):
|
||
logger.error(f"Invalid embed_info: {embed_info}, using default embedding model config")
|
||
raise ValueError("Invalid embed_info: must be a non-empty dictionary")
|
||
|
||
# 2. 检查是否是 EmbedModelInfo 对象（在某些情况下可能直接传入对象）
|
||
if hasattr(embed_info, "name") and isinstance(embed_info, EmbedModelInfo):
|
||
logger.debug(f"Using EmbedModelInfo object: {embed_info.name}, {traceback.format_exc()}")
|
||
config_dict = embed_info.model_dump()
|
||
config_dict["api_key"] = os.getenv(config_dict["api_key"]) or config_dict["api_key"]
|
||
return config_dict
|
||
|
||
raise ValueError(f"Unsupported embed_info format: {embed_info}")
|
||
|
||
except Exception as e:
|
||
logger.error(f"Error in get_embedding_config: {e}, embed_info={embed_info}")
|
||
# 返回默认配置作为fallback
|
||
logger.warning("Falling back to default embedding model config")
|
||
try:
|
||
config_dict = config.embed_model_names[config.embed_model].model_dump()
|
||
config_dict["api_key"] = os.getenv(config_dict["api_key"]) or config_dict["api_key"]
|
||
return config_dict
|
||
except Exception as fallback_error:
|
||
logger.error(f"Failed to get default embedding config: {fallback_error}")
|
||
raise ValueError(f"Failed to get embedding config and fallback failed: {e}")
|
||
|
||
|
||
def is_minio_url(file_path: str) -> bool:
|
||
"""
|
||
检测是否是MinIO URL
|
||
|
||
Args:
|
||
file_path: 文件路径或URL
|
||
|
||
Returns:
|
||
bool: 是否是MinIO URL
|
||
"""
|
||
return file_path.startswith(("http://", "https://", "s3://")) or "minio" in file_path.lower()
|
||
|
||
|
||
def parse_minio_url(file_path: str) -> tuple[str, str]:
|
||
"""
|
||
解析MinIO URL，提取bucket名称和对象名称
|
||
|
||
支持标准 HTTP/HTTPS URL 格式：
|
||
- http(s)://host/bucket-name/path/to/object
|
||
|
||
Args:
|
||
file_path: MinIO文件URL (http:// 或 https://)
|
||
|
||
Returns:
|
||
tuple[str, str]: (bucket_name, object_name)
|
||
|
||
Raises:
|
||
ValueError: 如果无法解析URL
|
||
"""
|
||
try:
|
||
from urllib.parse import urlparse
|
||
|
||
# 解析URL
|
||
parsed_url = urlparse(file_path)
|
||
|
||
# 对于 minio:// 协议，bucket名称在netloc中
|
||
if parsed_url.scheme == "minio":
|
||
bucket_name = parsed_url.netloc
|
||
object_name = parsed_url.path.lstrip("/")
|
||
else:
|
||
# 对于 http/https 协议，bucket名称在path的第一部分
|
||
object_name = parsed_url.path.lstrip("/")
|
||
path_parts = object_name.split("/", 1)
|
||
if len(path_parts) > 1:
|
||
bucket_name = path_parts[0]
|
||
object_name = path_parts[1]
|
||
else:
|
||
raise ValueError(f"无法解析MinIO URL中的bucket名称: {file_path}")
|
||
|
||
logger.debug(f"Parsed MinIO URL: bucket_name={bucket_name}, object_name={object_name}")
|
||
return bucket_name, object_name
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to parse MinIO URL {file_path}: {e}")
|
||
raise ValueError(f"无法解析MinIO URL: {file_path}")
|