triple-extraxtion/server.py

import requests
from nltk.tokenize import sent_tokenize

from fastapi import FastAPI

from return_result import get_ner_result
from return_result import get_re_result
from pdfReader import extract_text_from_pdf
from pdfReader import split_sentences

from threading import Lock
app = FastAPI()
lock = Lock()



@app.post("/extractKnowledge")
async def process_doc(data: dict):
    # 尝试获取线程锁，如果已经被其他线程获取，则等待
    locked = lock.acquire(blocking=False)
    if not locked:
        return {"code": 500,
                "msg": "Task is already running. Please try again later."}

    try:
        # 在锁的保护下执行任务
        doc_address = data.get("docAddress")
        if not doc_address.endswith(".pdf"):
            return {"code": 500,
                    "msg": "Invalid file format"}
        
        text = extract_text_from_pdf(doc_address)
        sentences = split_sentences(text)
        results = []
        for raw_text in sentences:
            entities = get_ner_result(raw_text)
            result = get_re_result(entities, raw_text)
            for i in result:
                results.append(i)
        jsonRes =  {
                    "docInfo": {
                        "docId": data.get("docId"), 
                        "taskId": data.get("taskId"),
                        "subTaskId": data.get("subTaskId"),
                        "docName": data.get("docName"),
                        "docAddress":data.get("docAddress") 
                    },
                    "knowledgeList": results,
                    "code": 200, 
                    "msg": "success"
                }
        
        url = "http://example.com/kg/saveKnowledge"
        headers = {"Content-Type": "application/json"}
        data = {"key1": "value1", "key2": "value2"} # 替换为您的json数据

        try:
            response = requests.post(url, headers=headers, data=jsonRes)
            response.raise_for_status()  # 如果响应状态码不是 200，则会抛出异常
            if response.json()["code"] != 200:
                result = {"code": 500, "msg": response.json()["msg"]}
            else:
                result = {"code": 200, "msg": "成功"}
        except requests.exceptions.RequestException as e:
            result = {"code": 500, "msg": str(e)}

        return result
    
    except requests.exceptions.RequestException as e:
            return {"code": 500, "msg": str(e)}
    finally:
        # 释放线程锁
        lock.release()
    
    

if __name__ == '__main__':
    import uvicorn
    uvicorn.run(app=app,
                host="0.0.0.0",
                port=9999,
                workers=1)