suhyun1's picture
Upload app.py
6169786 verified
import os
import gradio as gr
import bs4
from langchain_community.document_loaders import WebBaseLoader
from langchain.text_splitter import CharacterTextSplitter
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import FAISS
from langchain.chains import RetrievalQA
from langchain_groq import ChatGroq
# ํ™˜๊ฒฝ ๋ณ€์ˆ˜๋กœ๋ถ€ํ„ฐ Groq API Key ๋ถˆ๋Ÿฌ์˜ค๊ธฐ
groq_api_key = os.environ.get("GROQ_API_KEY", "")
# ๊ตญ๊ฐ€๊ธฐ๋ก์› ์›น ๋ฌธ์„œ ๋ชฉ๋ก
urls = [
"https://archives.go.kr/next/newsearch/listSubjectContent.do?subjectFieldId=000011",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003140&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003288&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003290&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003292&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=008757&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003293&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003294&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003295&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003289&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=010816&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=010817&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=009154&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003260&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003278&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003281&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003283&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003284&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003280&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003282&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003287&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003286&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003285&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003279&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003141&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003143&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003144&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003142&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=008653&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=010827&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=008582&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=008663&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=008581&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=010828&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=010830&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=010831&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003145&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=009425&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003146&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=010821&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003151&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003149&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003148&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=008655&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=008654&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newsearch/listSubjectDescription.do?id=003150&pageFlag=A&sitePage=1-2-1",
"https://archives.go.kr/next/newmanager/recodeRegister.do",
"https://archives.go.kr/next/newtour/tourCourse.do",
"https://archives.go.kr/next/newrecordsMngPro/recordsDonateInfo.do",
"https://archives.go.kr/next/newdata/pepoleRecodPresentIntro.do",
"https://archives.go.kr/next/newsearch/searchGuideList.do",
"https://archives.go.kr/next/newsearch/searchGuideList.do?page=2",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=441",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=381",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=341",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=261",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=227",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=59",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=30",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=64",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=321",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=124",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=267",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=141",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=149",
"https://archives.go.kr/next/newsearch/searchGuideDetail.do?guideSeq=22"
]
# ์›น๋ฌธ์„œ ๋กœ๋”ฉ
loader = WebBaseLoader(web_paths=urls, bs_kwargs=dict(parse_only=bs4.SoupStrainer()))
docs = loader.load()
# ๋ฌธ์„œ ๋ถ„ํ• 
splitter = CharacterTextSplitter(separator="\n", chunk_size=500, chunk_overlap=50)
split_docs = splitter.split_documents(docs)
# ์ž„๋ฒ ๋”ฉ ๋ฐ ๋ฒกํ„ฐ ์ €์žฅ ๋ฐ ๋ฆฌํŠธ๋ฆฌ๋ฒ„ ์„ค์ •
embedding_model = HuggingFaceEmbeddings(model_name="snunlp/KR-SBERT-V40K-klueNLI-augSTS")
vectorstore = FAISS.from_documents(split_docs, embedding_model)
retriever = vectorstore.as_retriever()
# LLM + QA ์ฒด์ธ
llm = ChatGroq(groq_api_key=groq_api_key, model_name="llama3-70b-8192")
qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=retriever, chain_type="stuff")
# Gradio ์ฑ„ํŒ… ํ•จ์ˆ˜
def chat_with_history(user_input, history):
if history is None:
history = []
query = user_input.strip() + " ํ•œ๊ตญ์–ด๋กœ ๋‹ตํ•ด์ฃผ์„ธ์š”."
result = qa_chain({"query": query})
answer = result.get("result", "๋‹ต๋ณ€์„ ์ฐพ์„ ์ˆ˜ ์—†์Šต๋‹ˆ๋‹ค.")
history.append((user_input, answer))
return "", history, history
# Gradio ์ธํ„ฐํŽ˜์ด์Šค ๊ตฌ์„ฑ
with gr.Blocks() as demo:
gr.Markdown("## ๐Ÿ“š ๊ตญ๊ฐ€๊ธฐ๋ก์› ์ •๋ณด ์ฑ—๋ด‡")
chatbot = gr.Chatbot(label="๊ธฐ๋ก์› ์ฑ—๋ด‡")
msg = gr.Textbox(placeholder="์งˆ๋ฌธ์„ ์ž…๋ ฅํ•˜์„ธ์š”", label="๐Ÿ’ฌ ์งˆ๋ฌธ ์ž…๋ ฅ")
state = gr.State([])
msg.submit(chat_with_history, inputs=[msg, state], outputs=[msg, chatbot, state])
demo.launch()