paperfree、知网论文查重API接口

{
  "error_no": 0,
  "message": "",
  "data": "task_123456789"
}

{
  "error_no": -1,
  "message": "文件格式错误，请上传Doc或docx文件",
  "data": ""
}

{
  "error_no": 0,
  "message": "",
  "data": [
    {
      "task_id": "task_123456789",
      "check_status": 3,
      "similar": "18.5%",
      "check_result": "http://xxx/report/123.zip",
      "check_type": "pf",
      "word_count": 8500
    }
  ]
}

import requests
import time

# 配置信息（替换为实际值）
API_KEY = "your_api_key"
API_SECRET = "your_api_secret"
SUBMIT_URL = "http://example.com/submit_check"  # 替换为实际提交地址
QUERY_URL = "http://example.com/query_status"   # 替换为实际查询地址

def submit_paper(file_path, title, author):
    """提交论文查重任务"""
    # 构造表单数据
    files = {"doc": open(file_path, "rb")}
    data = {
        "title": title,
        "author": author,
        "checkType": "pf",
        "apiKey": API_KEY,
        "apiSecret": API_SECRET
    }
    # 发送请求
    response = requests.post(SUBMIT_URL, files=files, data=data)
    result = response.json()
    if result["error_no"] == 0:
        return result["data"]  # 返回任务ID
    else:
        raise Exception(f"提交失败：{result['message']}")

def query_status(task_ids):
    """查询任务状态"""
    data = {
        "task_id_list": task_ids,
        "apiKey": API_KEY,
        "apiSecret": API_SECRET
    }
    response = requests.post(QUERY_URL, data=data)
    result = response.json()
    if result["error_no"] == 0:
        return result["data"]
    else:
        raise Exception(f"查询失败：{result['message']}")

if __name__ == "__main__":
    # 1. 提交查重任务
    task_id = submit_paper(
        file_path="test_paper.docx",
        title="基于机器学习的图像识别研究",
        author="张明"
    )
    print(f"提交成功，任务ID：{task_id}")

    # 2. 轮询查询结果（间隔5秒）
    while True:
        status_list = query_status(task_id)
        task_status = status_list[0]
        print(f"当前状态：{task_status['check_status']}（{get_status_desc(task_status['check_status'])}）")
        
        if task_status["check_status"] == 3:
            # 检测完成，输出结果
            print(f"重复率：{task_status['similar']}")
            print(f"报告下载地址：{task_status['check_result']}")
            break
        elif task_status["check_status"] in [-1, -2, 5]:
            # 任务失败或已删除
            print(f"任务异常：{get_status_desc(task_status['check_status'])}")
            break
        
        time.sleep(5)  # 间隔5秒查询一次

def get_status_desc(status_code):
    """状态码描述映射"""
    desc_map = {
        0: "待检测",
        1: "检测中",
        2: "生成报告中",
        3: "检测完成",
        5: "报告已删除",
        -1: "检测失败",
        -2: "生成报告失败"
    }
    return desc_map.get(status_code, "未知状态")