OurCodesAreAllRight/FilesReadSystem
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

添加任务状态双写机制和历史记录功能

Browse Source 
- 实现任务状态同时写入Redis和MongoDB的双写机制
- 添加MongoDB任务集合及CRUD操作接口
- 新增任务历史记录查询、列表展示和删除功能
- 重构任务状态更新逻辑，统一使用update_task_status函数
- 添加模板填服务中AI审核字段值的功能
- 优化前端任务历史页面显示和交互体验
This commit is contained in:
KiriAky 107
2026-04-10 01:15:53 +08:00
parent ed0f51f2a4
commit 858b594171
7 changed files with 638 additions and 138 deletions
Download Patch File Download Diff File Expand all files Collapse all files

121
backend/app/api/endpoints/documents.py
View File

@@ -23,6 +23,52 @@ logger = logging.getLogger(__name__)
router = APIRouter(prefix="/upload", tags=["文档上传"])
# ==================== 辅助函数 ====================
async def update_task_status(
task_id: str,
status: str,
progress: int = 0,
message: str = "",
result: dict = None,
error: str = None
):
"""
更新任务状态，同时写入 Redis 和 MongoDB
Args:
task_id: 任务ID
status: 状态
progress: 进度
message: 消息
result: 结果
error: 错误信息
"""
meta = {"progress": progress, "message": message}
if result:
meta["result"] = result
if error:
meta["error"] = error
# 尝试写入 Redis
try:
await redis_db.set_task_status(task_id, status, meta)
except Exception as e:
logger.warning(f"Redis 任务状态更新失败: {e}")
# 尝试写入 MongoDB（作为备用）
try:
await mongodb.update_task(
task_id=task_id,
status=status,
message=message,
result=result,
error=error
)
except Exception as e:
logger.warning(f"MongoDB 任务状态更新失败: {e}")
# ==================== 请求/响应模型 ====================
class UploadResponse(BaseModel):
@@ -77,6 +123,17 @@ async def upload_document(
task_id = str(uuid.uuid4())
try:
# 保存任务记录到 MongoDB（如果 Redis 不可用时仍能查询）
try:
await mongodb.insert_task(
task_id=task_id,
task_type="document_parse",
status="pending",
message=f"文档 {file.filename} 已提交处理"
)
except Exception as mongo_err:
logger.warning(f"MongoDB 保存任务记录失败: {mongo_err}")
content = await file.read()
saved_path = file_service.save_uploaded_file(
content,
@@ -122,6 +179,17 @@ async def upload_documents(
saved_paths = []
try:
# 保存任务记录到 MongoDB
try:
await mongodb.insert_task(
task_id=task_id,
task_type="batch_parse",
status="pending",
message=f"已提交 {len(files)} 个文档处理"
)
except Exception as mongo_err:
logger.warning(f"MongoDB 保存批量任务记录失败: {mongo_err}")
for file in files:
if not file.filename:
continue
@@ -159,9 +227,9 @@ async def process_document(
"""处理单个文档"""
try:
# 状态: 解析中
await redis_db.set_task_status(
await update_task_status(
task_id, status="processing",
meta={"progress": 10, "message": "正在解析文档"}
progress=10, message="正在解析文档"
)
# 解析文档
@@ -172,9 +240,9 @@ async def process_document(
raise Exception(result.error or "解析失败")
# 状态: 存储中
await redis_db.set_task_status(
await update_task_status(
task_id, status="processing",
meta={"progress": 30, "message": "正在存储数据"}
progress=30, message="正在存储数据"
)
# 存储到 MongoDB
@@ -191,9 +259,9 @@ async def process_document(
# 如果是 Excel，存储到 MySQL + AI生成描述 + RAG索引
if doc_type in ["xlsx", "xls"]:
await redis_db.set_task_status(
await update_task_status(
task_id, status="processing",
meta={"progress": 50, "message": "正在存储到MySQL并生成字段描述"}
progress=50, message="正在存储到MySQL并生成字段描述"
)
try:
@@ -215,9 +283,9 @@ async def process_document(
else:
# 非结构化文档
await redis_db.set_task_status(
await update_task_status(
task_id, status="processing",
meta={"progress": 60, "message": "正在建立索引"}
progress=60, message="正在建立索引"
)
# 如果文档中有表格数据，提取并存储到 MySQL + RAG
@@ -238,17 +306,13 @@ async def process_document(
await index_document_to_rag(doc_id, original_filename, result, doc_type)
# 完成
await redis_db.set_task_status(
await update_task_status(
task_id, status="success",
meta={
"progress": 100,
"message": "处理完成",
progress=100, message="处理完成",
result={
"doc_id": doc_id,
"result": {
"doc_id": doc_id,
"doc_type": doc_type,
"filename": original_filename
}
"doc_type": doc_type,
"filename": original_filename
}
)
@@ -256,18 +320,19 @@ async def process_document(
except Exception as e:
logger.error(f"文档处理失败: {str(e)}")
await redis_db.set_task_status(
await update_task_status(
task_id, status="failure",
meta={"error": str(e)}
progress=0, message="处理失败",
error=str(e)
)
async def process_documents_batch(task_id: str, files: List[dict]):
"""批量处理文档"""
try:
await redis_db.set_task_status(
await update_task_status(
task_id, status="processing",
meta={"progress": 0, "message": "开始批量处理"}
progress=0, message="开始批量处理"
)
results = []
@@ -318,21 +383,23 @@ async def process_documents_batch(task_id: str, files: List[dict]):
results.append({"filename": file_info["filename"], "success": False, "error": str(e)})
progress = int((i + 1) / len(files) * 100)
await redis_db.set_task_status(
await update_task_status(
task_id, status="processing",
meta={"progress": progress, "message": f"已处理 {i+1}/{len(files)}"}
progress=progress, message=f"已处理 {i+1}/{len(files)}"
)
await redis_db.set_task_status(
await update_task_status(
task_id, status="success",
meta={"progress": 100, "message": "批量处理完成", "results": results}
progress=100, message="批量处理完成",
result={"results": results}
)
except Exception as e:
logger.error(f"批量处理失败: {str(e)}")
await redis_db.set_task_status(
await update_task_status(
task_id, status="failure",
meta={"error": str(e)}
progress=0, message="批量处理失败",
error=str(e)
)

99
backend/app/api/endpoints/tasks.py
View File

@@ -1,13 +1,13 @@
"""
任务管理 API 接口
提供异步任务状态查询
提供异步任务状态查询和历史记录
"""
from typing import Optional
from fastapi import APIRouter, HTTPException
from app.core.database import redis_db
from app.core.database import redis_db, mongodb
router = APIRouter(prefix="/tasks", tags=["任务管理"])
@@ -23,25 +23,94 @@ async def get_task_status(task_id: str):
Returns:
任务状态信息
"""
# 优先从 Redis 获取
status = await redis_db.get_task_status(task_id)
if not status:
# Redis不可用时，假设任务已完成（文档已成功处理）
# 前端轮询时会得到这个响应
if status:
return {
"task_id": task_id,
"status": "success",
"progress": 100,
"message": "任务处理完成",
"result": None,
"error": None
"status": status.get("status", "unknown"),
"progress": status.get("meta", {}).get("progress", 0),
"message": status.get("meta", {}).get("message"),
"result": status.get("meta", {}).get("result"),
"error": status.get("meta", {}).get("error")
}
# Redis 不可用时，尝试从 MongoDB 获取
mongo_task = await mongodb.get_task(task_id)
if mongo_task:
return {
"task_id": mongo_task.get("task_id"),
"status": mongo_task.get("status", "unknown"),
"progress": 100 if mongo_task.get("status") == "success" else 0,
"message": mongo_task.get("message"),
"result": mongo_task.get("result"),
"error": mongo_task.get("error")
}
# 任务不存在或状态未知
return {
"task_id": task_id,
"status": status.get("status", "unknown"),
"progress": status.get("meta", {}).get("progress", 0),
"message": status.get("meta", {}).get("message"),
"result": status.get("meta", {}).get("result"),
"error": status.get("meta", {}).get("error")
"status": "unknown",
"progress": 0,
"message": "无法获取任务状态（Redis和MongoDB均不可用）",
"result": None,
"error": None
}
@router.get("/")
async def list_tasks(limit: int = 50, skip: int = 0):
"""
获取任务历史列表
Args:
limit: 返回数量限制
skip: 跳过数量
Returns:
任务列表
"""
try:
tasks = await mongodb.list_tasks(limit=limit, skip=skip)
return {
"success": True,
"tasks": tasks,
"count": len(tasks)
}
except Exception as e:
# MongoDB 不可用时返回空列表
return {
"success": False,
"tasks": [],
"count": 0,
"error": str(e)
}
@router.delete("/{task_id}")
async def delete_task(task_id: str):
"""
删除任务
Args:
task_id: 任务ID
Returns:
是否删除成功
"""
try:
# 从 Redis 删除
if redis_db._connected and redis_db.client:
key = f"task:{task_id}"
await redis_db.client.delete(key)
# 从 MongoDB 删除
deleted = await mongodb.delete_task(task_id)
return {
"success": True,
"deleted": deleted
}
except Exception as e:
raise HTTPException(status_code=500, detail=f"删除任务失败: {str(e)}")

69
backend/app/api/endpoints/templates.py
View File

@@ -23,6 +23,44 @@ logger = logging.getLogger(__name__)
router = APIRouter(prefix="/templates", tags=["表格模板"])
# ==================== 辅助函数 ====================
async def update_task_status(
task_id: str,
status: str,
progress: int = 0,
message: str = "",
result: dict = None,
error: str = None
):
"""
更新任务状态，同时写入 Redis 和 MongoDB
"""
from app.core.database import redis_db
meta = {"progress": progress, "message": message}
if result:
meta["result"] = result
if error:
meta["error"] = error
try:
await redis_db.set_task_status(task_id, status, meta)
except Exception as e:
logger.warning(f"Redis 任务状态更新失败: {e}")
try:
await mongodb.update_task(
task_id=task_id,
status=status,
message=message,
result=result,
error=error
)
except Exception as e:
logger.warning(f"MongoDB 任务状态更新失败: {e}")
# ==================== 请求/响应模型 ====================
class TemplateFieldRequest(BaseModel):
@@ -244,6 +282,17 @@ async def upload_joint_template(
# 3. 异步处理源文档到MongoDB
task_id = str(uuid.uuid4())
if source_file_info:
# 保存任务记录到 MongoDB
try:
await mongodb.insert_task(
task_id=task_id,
task_type="source_process",
status="pending",
message=f"开始处理 {len(source_file_info)} 个源文档"
)
except Exception as mongo_err:
logger.warning(f"MongoDB 保存任务记录失败: {mongo_err}")
background_tasks.add_task(
process_source_documents,
task_id=task_id,
@@ -282,12 +331,10 @@ async def upload_joint_template(
async def process_source_documents(task_id: str, files: List[dict]):
"""异步处理源文档，存入MongoDB"""
from app.core.database import redis_db
try:
await redis_db.set_task_status(
await update_task_status(
task_id, status="processing",
meta={"progress": 0, "message": "开始处理源文档"}
progress=0, message="开始处理源文档"
)
doc_ids = []
@@ -316,22 +363,24 @@ async def process_source_documents(task_id: str, files: List[dict]):
logger.error(f"源文档处理异常: {file_info['filename']}, error: {str(e)}")
progress = int((i + 1) / len(files) * 100)
await redis_db.set_task_status(
await update_task_status(
task_id, status="processing",
meta={"progress": progress, "message": f"已处理 {i+1}/{len(files)}"}
progress=progress, message=f"已处理 {i+1}/{len(files)}"
)
await redis_db.set_task_status(
await update_task_status(
task_id, status="success",
meta={"progress": 100, "message": "源文档处理完成", "doc_ids": doc_ids}
progress=100, message="源文档处理完成",
result={"doc_ids": doc_ids}
)
logger.info(f"所有源文档处理完成: {len(doc_ids)}个")
except Exception as e:
logger.error(f"源文档批量处理失败: {str(e)}")
await redis_db.set_task_status(
await update_task_status(
task_id, status="failure",
meta={"error": str(e)}
progress=0, message="源文档处理失败",
error=str(e)
)