228 lines
6.8 KiB
Python
228 lines
6.8 KiB
Python
import base64
|
||
from flask import Flask, request, render_template, redirect, url_for, jsonify
|
||
import os
|
||
import uuid
|
||
from PIL import Image
|
||
import re
|
||
import json
|
||
from ESConnect import *
|
||
from openai import OpenAI
|
||
# import config
|
||
|
||
# 创建Flask应用实例
|
||
app = Flask(__name__)
|
||
# app.config.from_object(config.Config)
|
||
|
||
# OCR和信息提取函数，使用大模型API处理图片并提取结构化信息
|
||
def ocr_and_extract_info(image_path):
|
||
"""
|
||
使用大模型API进行OCR识别并提取图片中的结构化信息
|
||
|
||
参数:
|
||
image_path (str): 图片文件路径
|
||
|
||
返回:
|
||
dict: 包含提取信息的字典，格式为 {'id': '', 'name': '', 'students': '', 'teacher': ''}
|
||
"""
|
||
def encode_image(image_path):
|
||
"""
|
||
将图片编码为base64格式
|
||
|
||
参数:
|
||
image_path (str): 图片文件路径
|
||
|
||
返回:
|
||
str: base64编码的图片字符串
|
||
"""
|
||
with open(image_path, "rb") as image_file:
|
||
return base64.b64encode(image_file.read()).decode('utf-8')
|
||
|
||
# 将图片转换为base64编码
|
||
base64_image = encode_image(image_path)
|
||
|
||
# 初始化OpenAI客户端，使用百度AI Studio的API
|
||
client = OpenAI(
|
||
api_key="188f57db3766e02ed2c7e18373996d84f4112272",
|
||
# 含有 AI Studio 访问令牌的环境变量，https://aistudio.baidu.com/account/accessToken,
|
||
base_url="https://aistudio.baidu.com/llm/lmapi/v3", # aistudio 大模型 api 服务域名
|
||
)
|
||
|
||
# 调用大模型API进行图片识别和信息提取
|
||
chat_completion = client.chat.completions.create(
|
||
messages=[
|
||
{'role': 'system', 'content': '你是一个能理解图片和文本的助手，请根据用户提供的信息进行回答。'},
|
||
{'role': 'user', "content": [
|
||
{"type": "text", "text": "请识别这张图片中的信息，只显示json不显示其它信息便于解析"
|
||
"以JSON格式返回（id对应比赛名称或论文名称，name对应项目名称，students对应参赛学生，teacher对应指导老师,出现多个名字用列表存储）"
|
||
"：{'id':'', 'name':'','students':'','teacher':''}"},
|
||
{
|
||
"type": "image_url",
|
||
"image_url": {
|
||
"url": f"data:image/png;base64,{base64_image}"
|
||
}
|
||
}
|
||
]}
|
||
],
|
||
model="ernie-4.5-turbo-vl-32k", # 使用百度文心大模型
|
||
)
|
||
|
||
# 获取API返回的文本内容
|
||
response_text = chat_completion.choices[0].message.content
|
||
|
||
def parse_respound(text):
|
||
"""
|
||
解析API返回的文本，提取JSON数据
|
||
|
||
参数:
|
||
text (str): API返回的文本
|
||
|
||
返回:
|
||
dict or None: 解析成功返回字典，失败返回None
|
||
"""
|
||
# 尝试直接解析标准JSON
|
||
try:
|
||
result=json.loads(text)
|
||
if result:
|
||
print("success")
|
||
return result
|
||
except json.JSONDecodeError:
|
||
print("无法解析标准json")
|
||
pass
|
||
|
||
# 提取markdown代码块中的内容
|
||
code_block = re.search(r'```json\n(.*?)```', text, re.DOTALL)
|
||
if code_block:
|
||
try:
|
||
result=json.loads(code_block.group(1))
|
||
if result:
|
||
print("success")
|
||
return result
|
||
except json.JSONDecodeError:
|
||
print("无法解析markdown")
|
||
pass
|
||
|
||
# 尝试替换单引号并解析
|
||
try:
|
||
fixed_json = text.replace("'", "\"")
|
||
result=json.loads(fixed_json)
|
||
if(result):
|
||
print("success")
|
||
return result
|
||
except json.JSONDecodeError:
|
||
print("无法替换单引号")
|
||
pass
|
||
|
||
# 解析API返回的文本
|
||
result_data = parse_respound(response_text)
|
||
return result_data
|
||
|
||
"""
|
||
模拟大模型识别图像并返回结构化JSON。
|
||
实际应调用Qwen-VL或其他OCR+解析服务。
|
||
"""
|
||
|
||
|
||
# 首页路由
|
||
@app.route('/')
|
||
def index():
|
||
"""
|
||
渲染首页模板
|
||
|
||
返回:
|
||
str: 渲染后的HTML页面
|
||
"""
|
||
return render_template('index.html')
|
||
|
||
# 图片上传路由
|
||
@app.route('/upload', methods=['POST'])
|
||
def upload_image():
|
||
"""
|
||
处理图片上传请求，调用OCR识别并存储结果
|
||
|
||
返回:
|
||
JSON: 上传成功或失败的响应
|
||
"""
|
||
# 获取上传的文件
|
||
file = request.files.get('file')
|
||
if not file:
|
||
return jsonify({"error": "No file uploaded"}), 400
|
||
|
||
# 保存上传的图片
|
||
filename = f"{uuid.uuid4()}_{file.filename}"
|
||
image_path = os.path.join("image", filename)
|
||
file.save(image_path)
|
||
|
||
# 调用大模型进行识别
|
||
try:
|
||
data = ocr_and_extract_info(image_path) # 替换为真实AI接口调用
|
||
insert_data(data) # 存入ES
|
||
return jsonify({"message": "成功录入", "data": data})
|
||
except Exception as e:
|
||
return jsonify({"error": str(e)}), 500
|
||
|
||
# 搜索路由
|
||
@app.route('/search')
|
||
def search():
|
||
"""
|
||
处理搜索请求，从Elasticsearch中检索匹配的数据
|
||
|
||
返回:
|
||
JSON: 搜索结果列表
|
||
"""
|
||
keyword = request.args.get('q')
|
||
if not keyword:
|
||
return jsonify([])
|
||
results = search_by_any_field(keyword)
|
||
print(results)
|
||
return jsonify(results)
|
||
|
||
# 结果页面路由
|
||
@app.route('/results')
|
||
def results_page():
|
||
"""
|
||
渲染搜索结果页面
|
||
|
||
返回:
|
||
str: 渲染后的HTML页面
|
||
"""
|
||
return render_template('results.html')
|
||
|
||
# 显示所有数据路由
|
||
@app.route('/all')
|
||
def show_all():
|
||
"""
|
||
获取所有数据并渲染到页面
|
||
|
||
返回:
|
||
str: 渲染后的HTML页面，包含所有数据
|
||
"""
|
||
all_data = search_all()
|
||
return render_template('all.html', data=all_data)
|
||
|
||
# 删除数据路由
|
||
@app.route('/delete/<doc_id>', methods=['POST'])
|
||
def delete_entry(doc_id):
|
||
"""
|
||
根据文档ID删除数据
|
||
|
||
参数:
|
||
doc_id (str): 要删除的文档ID
|
||
|
||
返回:
|
||
重定向到所有数据页面或错误信息
|
||
"""
|
||
if delete_by_id(doc_id):
|
||
return redirect(url_for('show_all'))
|
||
else:
|
||
return "删除失败", 500
|
||
|
||
|
||
|
||
# 主程序入口
|
||
if __name__ == '__main__':
|
||
# 创建Elasticsearch索引
|
||
create_index_with_mapping()
|
||
# 创建图片存储目录
|
||
os.makedirs("image", exist_ok=True)
|
||
# 启动Flask应用
|
||
app.run(use_reloader=False) |