vault backup: 2026-04-14 14:01:52

2026-04-14 14:01:53 +08:00
parent 3ea5708f6c
commit b388511dd7
130 changed files with 57616 additions and 12 deletions
--- a/Dehaze/Mineru_api_server-V2.py
+++ b/Dehaze/Mineru_api_server-V2.py
@@ -0,0 +1,66 @@
+import os
+import uuid
+import shutil
+import subprocess
+from fastapi import FastAPI, UploadFile, File, BackgroundTasks # 1. 导入 BackgroundTasks
+from fastapi.responses import FileResponse
+import uvicorn
+
+app = FastAPI(title="MinerU Extraction API")
+
+# 定义一个清理函数
+def cleanup_task_dir(dir_path: str):
+    """删除指定的文件夹及其所有内容"""
+    if os.path.exists(dir_path):
+        shutil.rmtree(dir_path)
+        print(f"🧹 已自动清理临时目录: {dir_path}")
+
+@app.post("/extract")
+async def extract_document(background_tasks: BackgroundTasks, file: UploadFile = File(...)): # 2. 注入后台任务参数
+    # 1. 创建独立的任务文件夹
+    task_id = str(uuid.uuid4())[:8]
+    work_dir = os.path.abspath(f"./api_workspace/{task_id}")
+    out_dir = os.path.join(work_dir, "OUT")
+    os.makedirs(work_dir, exist_ok=True)
+    
+    input_file_path = os.path.join(work_dir, file.filename)
+    
+    try:
+        # 2. 接收并保存文件
+        print(f"[Task {task_id}] 收到文件: {file.filename}，开始处理...")
+        with open(input_file_path, "wb") as buffer:
+            shutil.copyfileobj(file.file, buffer)
+            
+        # 3. 执行 MinerU 命令
+        subprocess.run(["mineru", "-p", input_file_path, "-o", out_dir], check=True)
+        
+        # 4. 穿透外层冗余文件夹，寻找真实的输出目录
+        target_zip_dir = out_dir
+        for root, dirs, files in os.walk(out_dir):
+            if any(f.endswith('.md') for f in files):
+                target_zip_dir = root
+                break
+                
+        # 5. 打包结果
+        zip_base_path = os.path.join(work_dir, "result")
+        shutil.make_archive(zip_base_path, 'zip', target_zip_dir) 
+        zip_file_path = f"{zip_base_path}.zip"
+        
+        # 6. 【核心改进】：添加后台任务，在响应发送后删除 work_dir
+        background_tasks.add_task(cleanup_task_dir, work_dir)
+        
+        # 7. 返回给客户端
+        return FileResponse(
+            zip_file_path,
+            media_type="application/zip",
+            filename=f"parsed_{file.filename}.zip"
+        )
+        
+    except Exception as e:
+        # 如果出错了，也尝试清理一下该任务的文件夹，防止空间占用
+        background_tasks.add_task(cleanup_task_dir, work_dir)
+        return {"status": "error", "message": str(e)}
+
+if __name__ == "__main__":
+    print("🚀 MinerU API 启动成功！监听地址: http://192.168.4.6:5000")
+    uvicorn.run(app, host="0.0.0.0", port=5000)