siwei_ai/app.py

from flask import Flask, render_template, request, jsonify
import psycopg2
from psycopg2.extras import DictCursor
import logging
import ollama
import json
import datetime
import uuid
import os
from vocal import voice_text
from voice_translation_test import vocal_text
from flask_cors import CORS
from dotenv import load_dotenv
from embed import embed
from query import query
from get_vector_db import get_vector_db
import time
from funasr import AutoModel
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
from langchain_community.chat_models import ChatOllama
from langchain.prompts import ChatPromptTemplate, PromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough
from langchain.retrievers.multi_query import MultiQueryRetriever
from get_vector_db import get_vector_db
from pypinyin import lazy_pinyin
import re


LLM_MODEL = os.getenv('LLM_MODEL', 'qwen2:7b')


load_dotenv()
TEMP_FOLDER = os.getenv('TEMP_FOLDER', './_temp')
os.makedirs(TEMP_FOLDER, exist_ok=True)


app = Flask(__name__)
CORS(app)


# 配置日志
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# 连接数据库
conn = psycopg2.connect(
    dbname="real3d",
    user="postgres",
    password="postgis",
    # host="192.168.100.30",
    host="192.168.60.2",
    port="5432"
)


# Function to get the prompt templates for generating alternative questions and answering based on context
def get_prompt():
    QUERY_PROMPT = PromptTemplate(
        input_variables=["question"],
        template="""你是一名AI语言模型助理。你的任务是生成五个
        从中检索相关文档的给定用户问题的不同版本
        矢量数据库。通过对用户问题生成多个视角
        目标是帮助用户克服基于距离的一些局限性
        相似性搜索。请提供这些用换行符分隔的备选问题。
        Original question: {question}""",
    )

    template = """仅根据以下上下文用中文回答问题：
    {context},请严格以markdown格式输出并保障寄送格式正确无误，
    Question: {question}
    """
    # Question: {question}


    prompt = ChatPromptTemplate.from_template(template)
    return QUERY_PROMPT, prompt

# 文件保存路径
UPLOAD_FOLDER = 'data/audio'
os.makedirs(UPLOAD_FOLDER, exist_ok=True)

#预加载模型权重到内存加快模型转文本速度

#模型1
model = AutoModel(model="E:\\yuyin_model\\Voice_translation", model_revision="v2.0.4",
                    vad_model="E:\\yuyin_model\\Endpoint_detection", vad_model_revision="v2.0.4",
                    punc_model="E:\\yuyin_model\\Ct_punc", punc_model_revision="v2.0.4",
                    use_cuda=True,use_fast = True,
                    )

#模型2
# inference_pipeline = pipeline(
#     task=Tasks.auto_speech_recognition,
#     # model='iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch',
#     model='C:\\Users\\siwei\\.cache\\modelscope\\hub\\iic\\speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch',
#     # model="model\punc_ct-transformer_cn-en-common-vocab471067-large",
#     model_revision="v2.0.4",
#     device='gpu')



# 后台接口

@app.route('/embed', methods=['POST'])
def route_embed():
    start_time = time.time()
    if 'file' not in request.files:
        return jsonify({"error": "No file part"}), 400

    file = request.files['file']

    if file.filename == '':
        return jsonify({"error": "No selected file"}), 400

    embedded = embed(file)
    end_time = time.time()
    print("Time taken for embedding: ", end_time - start_time)

    if embedded:
        return jsonify({"message": "File embedded successfully"}), 200

    return jsonify({"error": "File embedded unsuccessfully"}), 400



def route_query(msg):
    response = query(msg)
    # print(response)
    # if response:
    #     resObj = {}
    #     resObj["data"] = response
    #     resObj["code"] = 200
    #     resObj["type"] = "answer"
    #     return resObj
    # return {"error": "Something went wrong"}, 400
    return response

@app.route('/delete', methods=['DELETE'])
def route_delete():
    db = get_vector_db()
    db.delete_collection()

    return jsonify({"message": "Collection deleted successfully"}), 200

@app.route("/")
def home():
    return render_template('index.html')

# 后台接口





#定义需要替换的词
target_word = "抱坡"
target_word_pinyin = lazy_pinyin(target_word)

#判断拼音是否相同
def is_same_pinyin(word1,word2):
    return lazy_pinyin(word1) == lazy_pinyin(word2)

#替换同音字
def replace_word(text,target_word):
    words = re.findall(r'\b\w+\b', text)
    for word in words:
        if is_same_pinyin(word,target_word):
            text = text.replace(word,target_word)
    return text

# 文件上传
@app.route('/upload', methods=['POST'])
def upload_file():
    if 'file' not in request.files:
        return jsonify({"error": "No file part in the request"}), 400

    file = request.files['file']

    if file.filename == '':
        return jsonify({"error": "No file selected for uploading"}), 400

    # 生成UUID文件名
    file_ext = os.path.splitext(file.filename)[1]
    filename = f"{uuid.uuid4()}{file_ext}"

    # 保存文件
    file_path = os.path.join(UPLOAD_FOLDER, filename)
    file.save(file_path)

    #语音转文字模型1
    res = model.generate(file_path,
                batch_size_s=30,
                hotword='test')
    texts = [item['text'] for item in res]
    msg = ' '.join(texts)
    # print(msg)
    #语音转文字模型2
    # res = inference_pipeline(file_path)
    # # print(res)
    # texts = [item['text'] for item in res]
    # # print(texts)
    # msg = ' '.join(texts)


    # msg = vocal_text(file_path)
    os.remove(file_path)
    msg = replace_word(msg,target_word)
    words_to_replace = ["爆破", "爆坡","高坡"]
    for word in words_to_replace:
        msg = msg.replace(word, "抱坡")
    print(msg)


    return jsonify({"msg": "上传成功",
                    "code": 200,
                    "filename": filename,
                    "voiceMsg": msg
                    }), 200


# 接收消息，大模型解析


chat_history = "用户：你好，我是智能助手，请问有什么可以帮助您？\\n智能助手：好的，请问您有什么需求？"
sys_xuanzhi = """请扮演文本提取工具,根据输入和聊天上下文信息,基于以下因子选择、选址范围和用地类型提取这句话中的关键信息，提取到的结果请严格以json格式字符串输出并保障寄送格式正确无误，
选址范围 = ['抱坡区','天涯区','崖州区','海棠区','吉阳区' ],
因子选择 = ["高程","坡度","永久基本农田","城镇开发边界内","生态保护红线","文化保护区","自然保护地","风景名胜区","国有使用权","河道管理线","水库","公益林","火葬场","垃圾处理场","污水处理场","高压线","变电站","古树","城市道路","主要出入口","文化活动设施","体育运动场所","排水","供水","燃气","电力","电信","十五分钟社区生活圈邻里中心","社区服务设施","零售商业场所","医疗卫生设施","幼儿园服务半径","小学服务半径","为老服务设施"],
用地类型 = ['园地','耕地','林地','草地','湿地','公共卫生用地','老年人社会福利用地','儿童社会福利用地','残疾人社会福利用地','其他社会福利用地','零售商业用地','批发市场用地','餐饮用地','旅馆用地','公用设施营业网点用地','娱乐用地','康体用地','一类工业用地','二类工业用地','广播电视设施用地','环卫用地','消防用地','干渠','水工设施用地','其他公用设施用地','公园绿地','防护绿地','广场用地','军事设施用地','使领馆用地','宗教用地','文物古迹用地','监教场所用地','殡葬用地','其他特殊用地','河流水面','湖泊水面','水库水面','坑塘水面','沟渠','冰川及常年积雪','渔业基础设施用海','增养殖用海','捕捞海域','工业用海','盐田用海','固体矿产用海','油气用海','可再生能源用海','海底电缆管道用海','港口用海','农业设施建设用地','工矿用地','畜禽养殖设施建设用地','水产养殖设施建设用地','城镇住宅用地','特殊用地','居住用地','绿地与开敞空间用地','水田','水浇地','旱地','果园','茶园','橡胶园','其他园地','乔木林地','竹林地','城镇社区服务设施用地','农村宅基地','农村社区服务设施用地','机关团体用地','科研用地','文化用地','教育用地','体育用地','医疗卫生用地','社会福利用地','商业用地','商务金融用地','二类农村宅基地','图书与展览用地','文化活动用地','高等教育用地','中等职业教育用地','体育训练用地','其他交通设施用地','供水用地','排水用地','供电用地','供燃气用地','供热用地','通信用地','邮政用地','医院用地','基层医疗卫生设施用地','田间道','盐碱地','沙地','裸土地','裸岩石砾地','村道用地','村庄内部道路用地','公共管理与公共服务用地','仓储用地','交通运输用地','公用设施用地','交通运输用海','航运用海','路桥隧道用海','风景旅游用海','文体休闲娱乐用海','军事用海','其他特殊用海','空闲地','田坎','港口码头用地','管道运输用地','城市轨道交通用地','城镇道路用地','交通场站用地','一类城镇住宅用地','二类城镇住宅用地','三类城镇住宅用地','一类农村宅基地','商业服务业用地','三类工业用地','一类物流仓储用地','二类物流仓储用地','三类物流仓储用地','盐田','对外交通场站用地','公共交通场站用地','社会停车场用地','中小学用地','幼儿园用地','其他教育用地','体育场馆用地','灌木林地','其他林地','天然牧草地','人工牧草地','其他草地','森林沼泽','灌丛沼泽','沼泽草地','其他沼泽地','沿海滩涂','内陆滩涂','红树林地','乡村道路用地','种植设施建设用地','娱乐康体用地','其他商业服务业用地','工业用地','采矿用地','物流仓储用地','储备库用地','铁路用地','公路用地','机场用地'],
landType是用地类型
districtName是选址范围
area是用地大小，单位统一转换为亩
factors是因子选择
其他公里、千米的单位转换为米
输出json格式数据如下：
{
    "districtName": "抱坡区",
    "landType": "居住用地",
    "area": {
        "min": 30,
        "max": 50
    },
    "factors": [
        {
            "type": "医疗卫生设施",
            "condition": "lt",
            "value": "500"
        },
        {
            "type": "永久基本农田",
            "condition": "not_intersect"
        },
        {
            "type": "火葬场",
            "condition": "gt",
            "value": "1000"
        }，
        {
            "type": "幼儿园服务半径",
            "condition": "lt",
            "value": "1000"
        }，
        {
            "type": "小学服务半径",
            "condition": "lt",
            "value": "1000"
        }，
    ]
}
json中"condition"的值为"gt"、"lt"、"get"、"let"、"between","not_intersect"、"intersect"、"not_contain"、"contain"、"between"
"""
sys_question = """请扮演问答工具,对用户输入信息进行回答，请严格以markdown格式输出并保障寄送格式正确无误"""
# 智能选址
def update_chat_history(user_message):
    global chat_history  # 使用全局变量以便更新
    prompt = chat_history + "\\n用户：" + user_message

    # 生成回复，并加入聊天上下文
    res = ollama.generate(
        model="qwen2.5:7b",
        stream=False,
        system=sys_xuanzhi,
        prompt=prompt,
        options={"temperature": 0, "num_ctx": 32000, },
        keep_alive=-1
    )

    # 获取机器人回复
    bot_message = res["response"]

    # 更新聊天历史
    chat_history += "\\n智能助手：" + bot_message

    # 返回机器人的回复
    return bot_message

#简单知识问答，未关联本地知识库
def update_chat_history_simple(user_message):
    global chat_history  # 使用全局变量以便更新
    prompt = chat_history + "\\n用户：" + user_message

    # 生成回复，并加入聊天上下文
    res = ollama.generate(
        model="qwen2.5:7b",
        stream=False,
        system=sys_question,
        prompt=prompt,
        options={"temperature": 0, "num_ctx": 32000, },
        keep_alive=-1
    )

    # 获取机器人回复
    bot_message = res["response"]

    # 更新聊天历史
    chat_history += "\\n智能助手：" + bot_message

    # 返回机器人的回复
    return bot_message

@app.route('/closeMsg', methods=['DELETE'])
def delMsg():
    global chat_history
    chat_history = ""
    return jsonify({"msg": "清除成功",
                    "code": 200,
                    "chat_history": chat_history,
                    })


@app.route('/msg', methods=['POST'])
def inputMsg():
    # 从请求中获取JSON数据
    data = request.get_json()

    # 检查是否接收到数据
    if not data:
        return jsonify({"error": "No data received"}), 400

    # 打印接收到的消息
    print(data['msg'])

    msg = data['msg']
    msg = replace_word(msg,target_word)
    words_to_replace1 = ["爆破", "爆坡"]
    for word in words_to_replace1:
        msg = msg.replace(word, "抱坡")
    print(msg)

    type = data['type']
    if type == 'selectLand':
    # 调用大模型解析
    # 这里调用大模型，并返回解析结果

    # 示例：用户输入一条消息
    # msg= "我计划在抱坡区选取适宜地块作为工业用地，要求其在城市开发边界内，离小学大于1000m,坡度小于25度，用地面积在80-100亩之间。"
        res = update_chat_history(msg)
        print(res)  # 打印生成的回复

        addtress = ['抱坡区', '天涯区', '崖州区', '海棠区', '吉阳区']
        land = ['园地','耕地','林地','草地','湿地','公共卫生用地','老年人社会福利用地','儿童社会福利用地','残疾人社会福利用地','其他社会福利用地','零售商业用地','批发市场用地','餐饮用地','旅馆用地','公用设施营业网点用地','娱乐用地','康体用地','一类工业用地','二类工业用地','广播电视设施用地','环卫用地','消防用地','干渠','水工设施用地','其他公用设施用地','公园绿地','防护绿地','广场用地','军事设施用地','使领馆用地','宗教用地','文物古迹用地','监教场所用地','殡葬用地','其他特殊用地','河流水面','湖泊水面','水库水面','坑塘水面','沟渠','冰川及常年积雪','渔业基础设施用海','增养殖用海','捕捞海域','工业用海','盐田用海','固体矿产用海','油气用海','可再生能源用海','海底电缆管道用海','港口用海','农业设施建设用地','工矿用地','畜禽养殖设施建设用地','水产养殖设施建设用地','城镇住宅用地','特殊用地','居住用地','绿地与开敞空间用地','水田','水浇地','旱地','果园','茶园','橡胶园','其他园地','乔木林地','竹林地','城镇社区服务设施用地','农村宅基地','农村社区服务设施用地','机关团体用地','科研用地','文化用地','教育用地','体育用地','医疗卫生用地','社会福利用地','商业用地','商务金融用地','二类农村宅基地','图书与展览用地','文化活动用地','高等教育用地','中等职业教育用地','体育训练用地','其他交通设施用地','供水用地','排水用地','供电用地','供燃气用地','供热用地','通信用地','邮政用地','医院用地','基层医疗卫生设施用地','田间道','盐碱地','沙地','裸土地','裸岩石砾地','村道用地','村庄内部道路用地','公共管理与公共服务用地','仓储用地','交通运输用地','公用设施用地','交通运输用海','航运用海','路桥隧道用海','风景旅游用海','文体休闲娱乐用海','军事用海','其他特殊用海','空闲地','田坎','港口码头用地','管道运输用地','城市轨道交通用地','城镇道路用地','交通场站用地','一类城镇住宅用地','二类城镇住宅用地','三类城镇住宅用地','一类农村宅基地','商业服务业用地','三类工业用地','一类物流仓储用地','二类物流仓储用地','三类物流仓储用地','盐田','对外交通场站用地','公共交通场站用地','社会停车场用地','中小学用地','幼儿园用地','其他教育用地','体育场馆用地','灌木林地','其他林地','天然牧草地','人工牧草地','其他草地','森林沼泽','灌丛沼泽','沼泽草地','其他沼泽地','沿海滩涂','内陆滩涂','红树林地','乡村道路用地','种植设施建设用地','娱乐康体用地','其他商业服务业用地','工业用地','采矿用地','物流仓储用地','储备库用地','铁路用地','公路用地','机场用地']
        json_res = res.replace("json","")
        json_res = json_res.replace("```","")
        if json_res != "未找到相关数据":
            try:
                json_res = json.loads(json_res)
                districtName = json_res["districtName"]
                landType = json_res["landType"]
                # if landType != "未找到相关数据" and landType != "" and districtName  != "未找到相关数据"and districtName != "":
                if landType in land and districtName in addtress:
                    json_res = jsonResToDict(json_res)
                    # print(json_res)
                else:
                    json_res = "未找到相关数据"
                    json_res = jsonResToDict_wrong(json_res)
            except:
                json_res = "未找到相关数据"
                json_res = jsonResToDict_wrong(json_res)
        else:
            json_res = "未找到相关数据"
            json_res = jsonResToDict_wrong(json_res)
    elif type == 'answer':
        # json_res = route_query(msg)
        # json_res = jsonResToDict_questions(json_res)
        # print(json_res)  # 打印生成的回复

        json_res = update_chat_history_simple(msg)
        json_res = jsonResToDict_questions(json_res)
        print(json_res)  # 打印生成的回复

    # 返回响应
    return jsonify(json_res)


# 将大模型解析的结果转换为选址需要的数据格式

def jsonResToDict(json_res):

    # 1.查询选址范围信息
    districtName = json_res["districtName"]
    ewkt = getAiDistrict(districtName)
    # 2.保存选址范围信息
    geomId = saveGeom(ewkt)
    # 3.获取用地类型信息
    landType = json_res["landType"]
    landType = getLandType(landType, "YDYHFLDM")
    # 4.获取模板信息
    factorTemplates = getTemplateByCode(landType)

    # TODO 以哪个因子列表为准，模版和因子个数怎么匹配

    now = datetime.datetime.now()
    formatted_time = now.strftime("%Y%m%d%H%M%S")
    res = {
        "xzmj": 1500,
        "xmmc": "规划选址项目_"+formatted_time,
        "jsdw": "建设单位",
        "ydxz_bsm": landType,
        "ydmjbegin": json_res["area"]["min"],
        "ydmjend":  json_res["area"]["max"],
        "geomId": geomId,
        "yxyz": [],
        # TODO: 循环遍历
        # "yxyz": [
        #     {
        #         "id": "259e5bbaab434dbfb9c679bd44d4bfa4",
        #         "name": "幼儿园服务半径",
        #         "bsm": "TB_YEY",
        #         "conditionInfo": {
        #             "spatial_type": "distance",
        #             "default": "lt",
        #             "hasValue": true,
        #             "defaultValue": "300",
        #             "unit": "米",
        #             "clip": false
        #         }
        #     }
        # ],
        # "useMultiple": json_res["useMultiple"],
        "useLandType": True,
        # "multipleDistance": json_res["multipleDistance"]
    }
    # 循环遍历输入因子
    factors = json_res["factors"]
    input_factors = {}
    for factor in factors:
        factorInfo = getFactorByName(factor["type"])
        if factorInfo == None:
            continue
        factorId = factorInfo["id"]
        factorBsm = factorInfo["bsm"]
        conditionInfo = factorInfo["condition_info"]
        conditionObj = json.loads(conditionInfo)

        defaultValue = '0'
        default = 'lt'
        if "value" in factor:
            defaultValue = str(factor["value"])
        if "condition" in factor:
            default = factor["condition"]
        # if defaultValue == '':
        #     defaultValue = '0'

        factor_info = {
            "id": factorId,
            "name": factor["type"],
            "bsm": factorBsm,
            "conditionInfo": {
                "spatial_type": conditionObj["spatial_type"],
                "default": default,
                "hasValue": conditionObj["hasValue"],
                "defaultValue": defaultValue,
                "unit": conditionObj["unit"],
                "clip": conditionObj["clip"]
            }
        }
        input_factors[factor_info["id"]] = factor_info
    # 循环遍历模板
    # 记录已经添加的因子 ID
    added_factor_ids = set()

    # 首先处理模板
    for factorTemplate in factorTemplates:
        factorId = factorTemplate["id"]
        factorTemplate["conditionInfo"] = json.loads(factorTemplate["conditionInfo"])
        res["yxyz"].append(factorTemplate)
        added_factor_ids.add(factorId)  # 记录已添加的因子 ID

    # 然后检查 input_factors 并添加未在模板中的因子
    for factor_id, factor_info in input_factors.items():
        if factor_id not in added_factor_ids:
            res["yxyz"].append(factor_info)

    resObj = {}
    resObj["data"] = res
    resObj["code"] = 200
    resObj["type"] = "selectLand"

    return resObj

#返回问答信息
def jsonResToDict_questions(json_res):
    resObj = {}
    resObj["data"] = json_res
    resObj["code"] = 200
    resObj["type"] = "answer"
    return resObj

# 返回错误信息
def jsonResToDict_wrong(json_res):
    resObj = {}
    resObj["data"] = json_res
    resObj["code"] = 500
    resObj["type"] = "selectLand"
    return resObj

# 获取因子信息


def getFactorByName(name):
    with conn.cursor(cursor_factory=DictCursor) as cur:
        sql = "SELECT * FROM base.t_fzss_fzxz_factor WHERE name = %s"
        complete_sql = cur.mogrify(sql, (name,)).decode('utf-8')
        logger.info(f"Executing SQL: {complete_sql}")
        cur.execute(sql, (name,))
        res = cur.fetchone()
    return res

# 获取内置模板信息


def getTemplateByCode(code):
    with conn.cursor(cursor_factory=DictCursor) as cur:
        sql = 'SELECT factor_id as id,factor_name as name,factor_bsm as bsm,condition_info as "conditionInfo" FROM base.t_fzss_fzxz_factor_temp WHERE land_type_code = %s'
        complete_sql = cur.mogrify(sql, (code,)).decode('utf-8')
        logger.info(f"Executing SQL: {complete_sql}")
        cur.execute(sql, (code,))
        res = cur.fetchall()
        # 将查询结果转换为字典列表
        result_list = [dict(row) for row in res]
    return result_list

# 获取选址范围信息


def getAiDistrict(name):
    with conn.cursor(cursor_factory=DictCursor) as cur:
        sql = "SELECT public.st_asewkt(geom) as geom FROM base.t_fzss_fzxz_ai_district WHERE name = %s"
        complete_sql = cur.mogrify(sql, (name,)).decode('utf-8')
        logger.info(f"Executing SQL: {complete_sql}")
        cur.execute(sql, (name,))
        res = cur.fetchone()
    return res["geom"]


# 保存选址范围信息
def saveGeom(ewkt):
    new_uuid = str(uuid.uuid4())  # 生成一个新的 UUID
    from_type = 3
    with conn.cursor() as cur:
        sql = "INSERT INTO base.t_fzss_zhxz_file(id,geom,from_type,create_time,area) VALUES (%s,public.st_geomfromewkt(%s),%s,now(),public.st_area(public.st_geomfromewkt(%s)::public.geography))"
        complete_sql = cur.mogrify(
            sql, (new_uuid, ewkt, from_type, ewkt)).decode('utf-8')
        logger.info(f"Executing SQL: {complete_sql}")
        cur.execute(sql, (new_uuid, ewkt, from_type, ewkt))
        conn.commit()
    return new_uuid

# 获取用地类型信息


def getLandType(landName, fzbs):
    with conn.cursor(cursor_factory=DictCursor) as cur:
        sql = "SELECT dm,mc,fzbs FROM base.t_fzss_fzxz_dict WHERE mc = %s and fzbs=%s"
        complete_sql = cur.mogrify(sql, (landName, fzbs)).decode('utf-8')
        logger.info(f"Executing SQL: {complete_sql}")
        cur.execute(sql, (landName, fzbs))
        res = cur.fetchone()
    return res["dm"]


# getTemplateByCode("08")

# getAiDistrict("抱坡区")

# ewkt="SRID=4326;POLYGON ((109.568515723151 18.2729002407864, 109.564270326708 18.2607742953866, 109.580087492139 18.2571512198688, 109.588461804591 18.2570597503377, 109.58884305979 18.2645363088176, 109.582107142538 18.2732736518031, 109.568515723151 18.2729002407864))"
# saveGeom(ewkt)

# getFactorByName("幼儿园服务半径")

# msg=voice_text('data/audio/1364627f-5a9b-42d7-b7f6-b99c094606cd.mp3')
# msg=vocal_text('data/audio/1364627f-5a9b-42d7-b7f6-b99c094606cd.mp3')
# print(msg)
if __name__ == '__main__':
    # app.run()
    app.run(
        host='0.0.0.0',
        port=4000
    )