app.py

import gradio as gr
from openai import OpenAI
import os
import openai


# client = OpenAI(api_key="sk-9hWFK-_5UQgzJMqT44ng72sZbKQXVvVXeVd11yoER3T3BlbkFJmd2HDc-N7CaH6PDRCYZfqe8hkT7DnxZNwcVkf3eeUA")

openai.api_key ="sk-9hWFK-_5UQgzJMqT44ng72sZbKQXVvVXeVd11yoER3T3BlbkFJmd2HDc-N7CaH6PDRCYZfqe8hkT7DnxZNwcVkf3eeUA"

import gradio as gr

# from langchain.chat_models import ChatOpenAI
# from langchain.memory import ConversationBufferMemory
# from langchain.chains import ConversationChain
# from langchain.schema import AIMessage, HumanMessage, SystemMessage

# from langchain.document_loaders import TextLoader
# from langchain.document_loaders import PyPDFLoader


#from langchain_community.chat_models import ChatOpenAI
from langchain_openai import ChatOpenAI
# from langchain_community.memory import ConversationBufferMemory

from langchain.memory import ConversationBufferMemory



from langchain.chains import ConversationChain
from langchain.schema import AIMessage, HumanMessage, SystemMessage

from langchain_community.document_loaders import TextLoader
from langchain_community.document_loaders import PyPDFLoader
from langchain_core.runnables import RunnableWithMessageHistory


# LLM과 메모리 초기화
llm = ChatOpenAI(temperature=0.0, model='gpt-3.5-turbo', openai_api_key="sk-9hWFK-_5UQgzJMqT44ng72sZbKQXVvVXeVd11yoER3T3BlbkFJmd2HDc-N7CaH6PDRCYZfqe8hkT7DnxZNwcVkf3eeUA")

memory = ConversationBufferMemory()

#memory = RunnableWithMessageHistory()

conversation = ConversationChain(
    llm=llm,
    memory=memory)




# 세션 히스토리 정의
def get_session_history():
    return []  # 세션 히스토리를 관리하는 로직을 구현할 수 있습니다.

# 메모리 및 대화 초기화
# memory = RunnableWithMessageHistory(runnable=llm, get_session_history=get_session_history)

# conversation = RunnableWithMessageHistory(llm=llm, memory=memory)




# 상담봇 - 채팅 및 답변
def counseling_bot_chat(message, chat_history):
    if message == "":
        return "", chat_history
    else:
        result_message = ""
        if len(chat_history) <= 1:
            messages = [
                SystemMessage(content="당신은 헤이마트의 상담원입니다. 마트 상품과 관련되지 않은 질문에는 정중히 거절하세요."),
                AIMessage(content="안녕하세요, 헤이마트입니다. 상담을 도와드리겠습니다."),
                HumanMessage(content=message)
            ]
            result_message = conversation.predict(input=messages)
        else:
            result_message = conversation.predict(input=message)

        chat_history.append([message, result_message])
        return "", chat_history

# 상담봇 - 되돌리기
def counseling_bot_undo(chat_history):
    if len(chat_history) > 1:
        chat_history.pop()
    return chat_history

# 상담봇 - 초기화
def counseling_bot_reset(chat_history):
    chat_history = [[None, "안녕하세요, 헤이마트입니다. 상담을 도와드리겠습니다."]]
    return chat_history


# 번역봇
def translate_bot(output_conditions, output_language, input_text):
    if input_text == "":
        return ""
    else:
        if output_conditions == "":
            output_conditions = ""
        else:
            output_conditions = "번역할 때의 조건은 다음과 같습니다. " + output_conditions

        completion = client.chat.completions.create(
            model="gpt-3.5-turbo",
            messages=[
                {"role": "system", "content": "당신은 번역가입니다. 입력한 언어를 다른 설명없이 곧바로 {0}로 번역해서 알려주세요. 번역이 불가능한 언어라면 번역이 불가능하다고 말한 후 그 이유를 설명해주세요. {1}".format(output_language, output_conditions)},
                {"role": "user", "content": input_text}
            ])

        return completion.choices[0].message.content

# 번역봇 - Text업로드
def translate_bot_Text_upload(files):
    loader = TextLoader(files)
    document = loader.load()
    return document[0].page_content

# 번역봇 - PDF업로드
def translate_bot_PDF_upload(files):
    loader = PyPDFLoader(files)
    pages = loader.load_and_split()
    return pages[0].page_content


# 소설봇
def novel_bot(model, temperature, detail):
    completion = client.chat.completions.create(
        model=model,
        temperature=temperature,
        messages=[
            {"role": "system", "content": "당신은 소설가입니다. 요청하는 조건에 맞춰 소설을 작성해주세요."},
            {"role": "user", "content": detail}
        ])
    return completion.choices[0].message.content


# 레이아웃
with gr.Blocks(theme=gr.themes.Default()) as app:
    with gr.Tab("상담봇"):
        #1
        gr.Markdown(
            value="""
            # <center>상담봇</center>
            <center>헤이마트 상담 봇입니다. 마트에서 판매하는 상품과 관련된 질문에 답변드립니다.</center>
            """)
        #2
        cb_chatbot = gr.Chatbot(
            value=[[None, "안녕하세요, 헤이마트입니다. 상담을 도와드리겠습니다."]],
            show_label=False
        )
        with gr.Row():
            #3
            cb_user_input = gr.Text(
                lines=1,
                placeholder="입력 창",
                container=False,
                scale=9
            )
            #4
            cb_send_btn = gr.Button(
                value="보내기",
                scale=1,
                variant="primary",
                icon="https://cdn-icons-png.flaticon.com/128/12439/12439334.png"
            )
        with gr.Row():
            #5
            gr.Button(value="↩️ 되돌리기").click(fn=counseling_bot_undo, inputs=cb_chatbot, outputs=cb_chatbot)
            #6
            gr.Button(value="🔄️ 초기화").click(fn=counseling_bot_reset, inputs=cb_chatbot, outputs=cb_chatbot)
        # 보내기1
        cb_send_btn.click(fn=counseling_bot_chat, inputs=[cb_user_input, cb_chatbot], outputs=[cb_user_input, cb_chatbot])
        # 보내기2
        cb_user_input.submit(fn=counseling_bot_chat, inputs=[cb_user_input, cb_chatbot], outputs=[cb_user_input, cb_chatbot])

    with gr.Tab("번역봇"):
        #1
        gr.Markdown(
            value="""
            # <center>번역봇</center>
            <center>다국어 번역 봇입니다.</center>
            """)
        with gr.Row():
            #2
            tb_output_conditions = gr.Text(
                label="번역 조건",
                placeholder="예시: 자연스럽게",
                lines=1,
                max_lines=3
            )
            #3
            tb_output_language = gr.Dropdown(
                label="출력 언어",
                choices=["한국어", "영어", "일본어", "중국어"],
                value="한국어",
                allow_custom_value=True,
                interactive=True
            )
        with gr.Row():
            #7
            tb_TXTupload = gr.UploadButton(label="📄 Txt 업로드")
            #8
            tb_PDFupload = gr.UploadButton(label="📤 PDF 업로드")
        #4
        tb_submit = gr.Button(
            value="번역하기",
            variant="primary"
        )
        with gr.Row():
            #5
            tb_input_text = gr.Text(
                placeholder="번역할 내용을 적어주세요.",
                lines=10,
                max_lines=20,
                show_copy_button=True,
                label=""
            )
            #6
            tb_output_text = gr.Text(
                lines=10,
                max_lines=20,
                show_copy_button=True,
                label="",
                interactive=False
            )
        # 보내기
        tb_submit.click(
            fn=translate_bot,
            inputs=[tb_output_conditions,
                    tb_output_language,
                    tb_input_text],
            outputs=tb_output_text
        )
        # Text파일 업로드
        tb_TXTupload.upload(
            fn=translate_bot_Text_upload,
            inputs=tb_TXTupload,
            outputs=tb_input_text
        )
        # PDF파일 업로드
        tb_PDFupload.upload(
            fn=translate_bot_PDF_upload,
            inputs=tb_PDFupload,
            outputs=tb_input_text
        )

    with gr.Tab("소설봇"):
        #1
        gr.Markdown(
            value="""
            # <center>소설봇</center>
            <center>소설을 생성해주는 봇입니다.</center>
            """)
        with gr.Accordion(label="사용자 설정"):
            with gr.Row():
                with gr.Column(scale=1):
                    #2
                    nb_model = gr.Dropdown(
                        label="모델 선택",
                        choices=["gpt-3.5-turbo", "gpt-3.5-turbo-16k", "gpt-4", "gpt-4-32k", "gpt-4-1106-preview"],
                        value="gpt-4-1106-preview",
                        interactive=True
                    )
                    #3
                    nb_temperature = gr.Slider(
                        label="창의성",
                        info="숫자가 높을 수록 창의적",
                        minimum=0,
                        maximum=2,
                        step=0.1,
                        value=1,
                        interactive=True
                    )
                #4
                nb_detail = gr.Text(
                    container=False,
                    placeholder="소설의 세부적인 설정을 작성합니다.",
                    lines=8,
                    scale=4
                )
        #5
        nb_submit = gr.Button(
            value="생성하기",
            variant="primary"
        )
        #6
        nb_output = gr.Text(
            label="",
            placeholder="이곳에 소설의 내용이 출력됩니다.",
            lines=10,
            max_lines=200,
            show_copy_button=True
        )
        # 보내기
        nb_submit.click(
            fn=novel_bot,
            inputs=[nb_model, nb_temperature, nb_detail],
            outputs=nb_output
        )

app.launch()