Update app.py

app.py

@@ -1,9 +1,7 @@
-import gradio as gr
 from youtubesearchpython import VideosSearch
-import whisper
+import gradio as gr
 import openai
-import os
-from pytube import YouTube
+from langchain_community.document_loaders import YoutubeLoader
 
 # OpenAI API 키 설정
 openai.api_key = os.getenv('O_API_KEY')
@@ -14,41 +12,37 @@ def search_youtube_videos(keyword, limit=5, order='date'):
     video_urls = [video['link'] for video in results['result']]
     return video_urls
 
-def download_audio_from_youtube(url, output_path="downloaded_audio.mp4"):
-    yt = YouTube(url)
-    audio_stream = yt.streams.get_audio_only()
-    audio_stream.download(output_filename=output_path)
-    return output_path
-
-def convert_audio_to_text(audio_path):
-    model = whisper.load_model("base")
-    result = model.transcribe(audio_path)
-    return result["text"]
+def get_transcript(url):
+    loader = YoutubeLoader()
+    transcript = loader.load(url)
+    text = " ".join([segment['text'] for segment in transcript['segments']])
+    return text
 
 def summarize_text(text):
     response = openai.Completion.create(
         engine="text-davinci-003",
-        prompt=f"요약해줘: {text}",
+        prompt=f"요약: {text}",
         max_tokens=150
     )
     return response.choices[0].text.strip()
 
 def process_keyword(keyword):
-    urls = search_youtube_videos(keyword)
+    video_urls = search_youtube_videos(keyword)
     summaries = []
-    for url in urls[:1]:  # 데모를 위해 첫 번째 URL만 처리
-        audio_path = download_audio_from_youtube(url)
-        text = convert_audio_to_text(audio_path)
-        summary = summarize_text(text)
-        summaries.append(summary)
-        os.remove(audio_path)  # 임시 오디오 파일 삭제
-    return summaries
+    for url in video_urls:
+        try:
+            text = get_transcript(url)
+            summary = summarize_text(text)
+            summaries.append(f"URL: {url}\nSummary: {summary}\n")
+        except Exception as e:
+            summaries.append(f"URL: {url}\nError: {str(e)}\n")
+    return "\n".join(summaries)
 
 # Gradio 인터페이스
 interface = gr.Interface(
     fn=process_keyword,
     inputs=gr.Textbox(label="검색 키워드"),
-    outputs=gr.Textbox(label="요약"),
+    outputs=gr.Textbox(label="결과"),
 )
 
 interface.launch()