Update app.py

app.py

@@ -1,119 +1,314 @@
 import streamlit as st
-import pandas as pd
+import feedparser
 from transformers import pipeline
-from datetime import datetime
+import requests
+import datetime
 
-# ================================
-# Streamlit Page Configuration
-# ================================
-st.set_page_config(
-    page_title="🌐 Multi-Language Translator",
-    layout="centered",
-    initial_sidebar_state="auto",
+# Streamlitの設定
+st.set_page_config(page_title="今日のテクノロジーニュース", layout="wide")
+st.title("📡 今日のテクノロジーニュース")
+
+# RSSフィードのURL
+rss_url = "https://rss.nytimes.com/services/xml/rss/nyt/Technology.xml"
+
+# 利用可能な翻訳モデルのリスト
+models = [
+    {
+        "name": "facebook/nllb-200-distilled-600M",
+        "description": "Translation • Updated Feb 15 • 322k • 504",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "facebook/mbart-large-50-many-to-many-mmt",
+        "description": "Translation • Updated Sep 29, 2023 • 646k • 278",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "facebook/nllb-200-3.3B",
+        "description": "Translation • Updated Feb 12, 2023 • 28.9k • 249",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "google/madlad400-10b-mt",
+        "description": "Translation • Updated Apr 12 • 1.76k • 84",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "ken11/mbart-ja-en",
+        "description": "Translation • Updated Oct 13, 2021 • 63 • 3",
+        "src_lang": "jpn_Jpan",
+        "tgt_lang": "eng_Latn"
+    },
+    {
+        "name": "facebook/nllb-200-1.3B",
+        "description": "Translation • Updated Feb 12, 2023 • 14.6k • 44",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "facebook/nllb-200-distilled-1.3B",
+        "description": "Translation • Updated Feb 12, 2023 • 101k • 98",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "alirezamsh/small100",
+        "description": "Translation • Updated Jul 23 • 1.85k • 60",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "Unbabel/wmt22-cometkiwi-da",
+        "description": "Translation • Updated Oct 13, 2023 • 1 • 24",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "ychenNLP/nllb-200-3.3B-easyproject",
+        "description": "Translation • Updated Aug 30, 2023 • 73 • 2",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/sugoi-v3.3-ja-en-ct2-float16",
+        "description": "Translation • Updated May 10, 2023 • 2",
+        "src_lang": "jpn_Jpan",
+        "tgt_lang": "eng_Latn"
+    },
+    {
+        "name": "JustFrederik/sugoi-v3.3-ja-en-ct2-int8",
+        "description": "Translation • Updated May 10, 2023 • 22 • 1",
+        "src_lang": "jpn_Jpan",
+        "tgt_lang": "eng_Latn"
+    },
+    {
+        "name": "JustFrederik/sugoi-v4-ja-en-ct2-float16",
+        "description": "Translation • Updated May 10, 2023 • 13 • 1",
+        "src_lang": "jpn_Jpan",
+        "tgt_lang": "eng_Latn"
+    },
+    {
+        "name": "JustFrederik/sugoi-v4-ja-en-ct2-int8",
+        "description": "Translation • Updated May 10, 2023",
+        "src_lang": "jpn_Jpan",
+        "tgt_lang": "eng_Latn"
+    },
+    {
+        "name": "JustFrederik/sugoi-v4-ja-en-ct2",
+        "description": "Translation • Updated May 10, 2023 • 20 • 1",
+        "src_lang": "jpn_Jpan",
+        "tgt_lang": "eng_Latn"
+    },
+    {
+        "name": "JustFrederik/sugoi-v3.3-ja-en-ct2",
+        "description": "Translation • Updated May 10, 2023",
+        "src_lang": "jpn_Jpan",
+        "tgt_lang": "eng_Latn"
+    },
+    {
+        "name": "JustFrederik/nllb-200-distilled-600M-ct2-int8",
+        "description": "Translation • Updated May 15, 2023 • 225",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-distilled-1.3B-ct2-int8",
+        "description": "Translation • Updated May 15, 2023 • 74 • 1",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-1.3B-ct2-int8",
+        "description": "Translation • Updated May 15, 2023 • 12",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-1.3B-ct2-float16",
+        "description": "Translation • Updated May 15, 2023 • 6",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-1.3B-ct2",
+        "description": "Translation • Updated May 15, 2023 • 14",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-distilled-1.3B-ct2",
+        "description": "Translation • Updated May 15, 2023 • 3",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-distilled-1.3B-ct2-float16",
+        "description": "Translation • Updated May 15, 2023 • 7 • 1",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-distilled-600M-ct2",
+        "description": "Translation • Updated May 15, 2023 • 4",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-distilled-600M-ct2-float16",
+        "description": "Translation • Updated May 15, 2023 • 8",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "JustFrederik/nllb-200-3.3B-ct2-float16",
+        "description": "Translation • Updated May 15, 2023 • 26 • 3",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "Babelscape/mrebel-large",
+        "description": "Translation • Updated Jun 21, 2023 • 67.5k • 66",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "Babelscape/mrebel-large-32",
+        "description": "Translation • Updated Jun 23, 2023 • 97 • 6",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "Babelscape/mrebel-base",
+        "description": "Translation • Updated Jun 23, 2023 • 66 • 5",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "winstxnhdw/nllb-200-distilled-1.3B-ct2-int8",
+        "description": "Translation • Updated Aug 3, 2023 • 2.42k • 4",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "michaelfeil/ct2fast-nllb-200-distilled-1.3B",
+        "description": "Translation • Updated Dec 10, 2023 • 10 • 1",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "michaelfeil/ct2fast-nllb-200-3.3B",
+        "description": "Translation • Updated Jul 21, 2023 • 36 • 11",
+        "src_lang": "eng_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    {
+        "name": "qiyuw/WSPAlign-xlm-base",
+        "description": "Translation • Updated Mar 18 • 4",
+        "src_lang": "xlm_Latn",
+        "tgt_lang": "jpn_Jpan"
+    },
+    # 既存のモデルを以下に追加できます
+]
+
+# プルダウンメニューでモデルを選択
+st.sidebar.header("翻訳モデルの選択")
+selected_model = st.sidebar.selectbox(
+    "使用する翻訳モデルを選択してください:",
+    options=models,
+    format_func=lambda x: f"{x['name']} - {x['description']}"
 )
 
-# ================================
-# Cache the Translation Pipelines
-# ================================
 @st.cache_resource
-def load_translation_pipelines():
+def load_translation_model(model_name, src_lang, tgt_lang):
     """
-    Load and cache translation pipelines to avoid reloading on every interaction.
+    選択された翻訳モデルをロードし、キャッシュします。
     """
-    enja = pipeline("translation", model="staka/fugumt-en-ja")
-    jaen = pipeline("translation", model="staka/fugumt-ja-en")
-    zhja = pipeline("translation", model="Helsinki-NLP/opus-mt-tc-big-zh-ja")
-    return {'enja': enja, 'jaen': jaen, 'zhja': zhja}
-
-# Load the translation models
-try:
-    session_models = load_translation_pipelines()
-except Exception as e:
-    st.error(f"Error loading translation models: {e}")
-    session_models = {}
-
-# ================================
-# Streamlit Application Layout
-# ================================
-st.title("🌐 Multi-Language Translator")
-
-# Initialize session state for CSV creation flag
-if 'csv_created' not in st.session_state:
-    st.session_state.csv_created = False
+    try:
+        translator = pipeline(
+            "translation",
+            model=model_name,
+            src_lang=src_lang,
+            tgt_lang=tgt_lang
+        )
+        return translator
+    except Exception as e:
+        st.error(f"翻訳モデルのロード中にエラーが発生しました: {e}")
+        return None
 
-# ================================
-# User Input Section
-# ================================
-st.header("🔤 Enter Text to Translate")
+@st.cache_data
+def translate_text(_translator, text):
+    """
+    テキストを日本語に翻訳します。
+    翻訳結果をキャッシュします。
+    """
+    if not _translator:
+        return "翻訳エラー"
+    try:
+        translation = _translator(text, max_length=500)[0]['translation_text']
+        return translation
+    except Exception as e:
+        st.error(f"翻訳中にエラーが発生しました: {e}")
+        return "翻訳エラー"
 
-# Model selection
-model_options = {
-    'English to Japanese': 'enja',
-    'Japanese to English': 'jaen',
-    'Chinese to Japanese': 'zhja'
-}
-model_display = list(model_options.keys())
-model_keys = list(model_options.values())
+@st.cache_data(ttl=3600)
+def fetch_rss_feed(url):
+    """
+    RSSフィードを取得し、XMLを保存してパースします。
+    キャッシュの有効期限は1時間です。
+    """
+    try:
+        response = requests.get(url)
+        if response.status_code != 200:
+            st.error(f"RSSフィードの取得に失敗しました。ステータスコード: {response.status_code}")
+            return None
+        # フィードのXMLを保存（データセットとして蓄積）
+        now = datetime.datetime.now()
+        filename = now.strftime("feed_%Y%m%d_%H%M%S.xml")
+        with open(filename, 'wb') as f:
+            f.write(response.content)
+        # フィードをパース
+        feed = feedparser.parse(response.content)
+        return feed
+    except Exception as e:
+        st.error(f"RSSフィードの取得中にエラーが発生しました: {e}")
+        return None
 
-selected_model_display = st.selectbox("Select Translation Model", model_display, index=0)
-selected_model = model_options[selected_model_display]
+# フィードを取得
+feed = fetch_rss_feed(rss_url)
 
-# Text input
-text = st.text_area("Input Text", height=150)
+if feed is None:
+    st.stop()  # フィードの取得に失敗した場合、アプリを停止します
 
-# ================================
-# Translation and Output
-# ================================
-if st.button("🚀 Translate"):
-    if not text.strip():
-        st.warning("Please enter text to translate.")
-    elif selected_model not in session_models:
-        st.error("Selected translation model is not available.")
-    else:
-        with st.spinner("Translating..."):
-            try:
-                translator = session_models[selected_model]
-                translation = translator(text)[0]['translation_text']
-                st.success("Translation Successful!")
-                st.subheader("📝 Translation Result")
-                st.write(translation)
+# 翻訳モデルをロード
+translator = load_translation_model(selected_model['name'], selected_model['src_lang'], selected_model['tgt_lang'])
 
-                # Prepare data for CSV
-                timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
-                data = {
-                    'Timestamp': [timestamp],
-                    'Model': [selected_model_display],
-                    'Original Text': [text],
-                    'Translated Text': [translation]
-                }
-                df = pd.DataFrame(data)
+# フィード内の記事をパースしてタイトルと説明を翻訳
+for entry in feed.entries:
+    # タイトルと説明を取得
+    title = entry.title
+    description = entry.description
 
-                # Save to CSV
-                csv_file = 'translation_data.csv'
-                if not st.session_state.csv_created:
-                    df.to_csv(csv_file, mode='w', header=True, index=False)
-                    st.session_state.csv_created = True
-                else:
-                    df.to_csv(csv_file, mode='a', header=False, index=False)
+    # タイトルと説明を日本語に翻訳（翻訳結果をキャッシュ）
+    translated_title = translate_text(translator, title)
+    translated_description = translate_text(translator, description)
 
-                st.info(f"Translation saved to `{csv_file}`.")
-            except Exception as e:
-                st.error(f"An error occurred during translation: {e}")
+    # Markdown形式で表示
+    st.markdown(f"### **タイトル（英語）:** {title}")
+    st.markdown(f"**タイトル（日本語）:** {translated_title}")
+    st.markdown(f"**概要（英語）:**")
+    st.write(description)
+    st.markdown(f"**概要（日本語）:**")
+    st.write(translated_description)
+    st.markdown(f"[🌐 続きを読む]({entry.link})")
+    st.markdown("---")
 
-# ================================
-# Optional: Download Translation Data
-# ================================
-if st.button("📥 Download Translation Data"):
-    try:
-        df = pd.read_csv('translation_data.csv')
-        csv = df.to_csv(index=False).encode('utf-8')
-        st.download_button(
-            label="Download CSV",
-            data=csv,
-            file_name='translation_data.csv',
-            mime='text/csv',
-        )
-    except FileNotFoundError:
-        st.warning("No translation data available to download.")
-    except Exception as e:
-        st.error(f"An error occurred while preparing the download: {e}")
+# キャッシュをクリアするボタン
+if st.button("キャッシュをクリア"):
+    load_translation_model.clear(selected_model['name'], selected_model['src_lang'], selected_model['tgt_lang'])
+    translate_text.clear()
+    fetch_rss_feed.clear()
+    st.success("キャッシュをクリアしました。")