app.py

import traceback
import streamlit as st
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_openai import ChatOpenAI
from langchain_anthropic import ChatAnthropic
from langchain_google_genai import ChatGoogleGenerativeAI
import requests
from bs4 import BeautifulSoup
from urllib.parse import urlparse


SUMMARIZE_PROMPT = """以下のコンテンツについて、内容を300文字程度でわかりやすく箇条書きで要約してください。

========

{content}

========

日本語で書いてね！
"""


def init_page():
    st.set_page_config(
        page_title="Webpage Summarizer",
        page_icon="🌐"
    )
    st.header("Webpage Summarizer 🌐📝")


def select_model(temperature=0):
    models = ("GPT-4o","GPT-4o-mini", "Claude 3.5 Sonnet", "Gemini 1.5 Pro")
    model_choice = st.radio("Choose a model:", models)
    if model_choice == "GPT-4o":
        return ChatOpenAI(temperature=temperature, model_name="gpt-4o")
    elif model_choice == "GPT-4o-mini":
        return ChatOpenAI(temperature=temperature, model_name="gpt-4o-mini")
    elif model_choice == "Claude 3.5 Sonnet":
        return ChatAnthropic(temperature=temperature, model_name="claude-3-5-sonnet-20240620")
    elif model_choice == "Gemini 1.5 Pro":
        return ChatGoogleGenerativeAI(temperature=temperature, model="gemini-1.5-pro-latest")


def init_chain():
    llm = select_model()
    prompt = ChatPromptTemplate.from_messages([
        ("user", SUMMARIZE_PROMPT),
    ])
    output_parser = StrOutputParser()
    chain = prompt | llm | output_parser
    return chain


def validate_url(url):
    """ URLが有効かどうかを判定する関数 """
    try:
        result = urlparse(url)
        return all([result.scheme, result.netloc])
    except ValueError:
        return False


def get_content(url):
    try:
        with st.spinner("Fetching Website ..."):
            response = requests.get(url)
            soup = BeautifulSoup(response.text, 'html.parser')
            # なるべく本文の可能性が高い要素を取得する
            if soup.main:
                return soup.main.get_text()
            elif soup.article:
                return soup.article.get_text()
            else:
                return soup.body.get_text()
    except:
        st.write(traceback.format_exc())  # エラーが発生した場合はエラー内容を表示
        return None

def main():
    init_page()
    chain = init_chain()
    # Style adjustments (optional, remove if not needed)
    st.markdown("""<style>.st-emotion-cache-15ecox0 { display: none !important; }
    @media (max-width: 50.5rem) {.st-emotion-cache-13ln4jf {max-width: calc(0rem + 100vw);}}
    </style>""",unsafe_allow_html=True,)
    # ユーザーの入力を監視
    if url := st.text_input("URL: ", key="input"):
        is_valid_url = validate_url(url)
        if not is_valid_url:
            st.write('Please input valid url')
        else:
            if content := get_content(url):
                st.markdown("## Summary")
                st.write_stream(chain.stream({"content": content}))

if __name__ == '__main__':
    main()