52 lines
1.9 KiB
Python
52 lines
1.9 KiB
Python
import os
|
|
import uuid
|
|
import shutil
|
|
import subprocess
|
|
from fastapi import FastAPI, UploadFile, File
|
|
from fastapi.responses import FileResponse
|
|
import uvicorn
|
|
|
|
app = FastAPI(title="MinerU Extraction API")
|
|
|
|
@app.post("/extract")
|
|
def extract_document(file: UploadFile = File(...)):
|
|
# 1. 创建独立的任务文件夹，防止多个请求同时发生时文件冲突
|
|
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. 接收并保存客户端上传的文件
|
|
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. 调用你刚刚跑通的 MinerU 命令行
|
|
# 相当于在终端执行 mineru -p xxx -o xxx
|
|
subprocess.run(["mineru", "-p", input_file_path, "-o", out_dir], check=True)
|
|
|
|
# 4. 将 MinerU 的输出结果打包成 ZIP 文件
|
|
print(f"[Task {task_id}] 解析完成，正在打包结果...")
|
|
zip_base_path = os.path.join(work_dir, "result")
|
|
shutil.make_archive(zip_base_path, 'zip', out_dir)
|
|
zip_file_path = f"{zip_base_path}.zip"
|
|
|
|
# 5. 将 ZIP 文件作为响应返回给客户端
|
|
return FileResponse(
|
|
zip_file_path,
|
|
media_type="application/zip",
|
|
filename=f"parsed_{file.filename}.zip"
|
|
)
|
|
|
|
except Exception as e:
|
|
return {"status": "error", "message": str(e)}
|
|
|
|
if __name__ == "__main__":
|
|
# 绑定到 0.0.0.0 代表监听本机所有 IP (包括 192.168.4.6)
|
|
# 端口设置为你要求的 5000
|
|
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)
|