Fix: Simplify dependencies and add app_gradio.py for HuggingFace Spaces

- Copy app.py to app_gradio.py as main Space file
- Remove unnecessary dependencies (FastAPI, pydantic, numpy, requests)
- Fix python-multipart version conflict with Gradio
- Update README.md to use app_gradio.py as app_file
- Keep only essential packages: gradio, anthropic, chromadb, sentence-transformers

README.md

@@ -1,11 +1,11 @@
 ---
-title: Financial RAG with Metacognitive Agent
-emoji: 💼
+title: Financial RAG Chatbot with Metacognition
+emoji: 🏦
 colorFrom: blue
 colorTo: green
 sdk: gradio
 sdk_version: 5.0.0
-app_file: app.py
+app_file: app_gradio.py
 pinned: false
 license: mit
 ---

app_gradio.py

@@ -0,0 +1,346 @@
+"""
+HuggingFace Spaces용 Gradio 앱
+Financial RAG with Metacognitive Agent
+"""
+
+import gradio as gr
+import os
+import sys
+from loguru import logger
+import asyncio
+from typing import Dict, Tuple
+
+# 로깅 설정
+logger.remove()
+logger.add(
+    sys.stdout,
+    format="<green>{time:YYYY-MM-DD HH:mm:ss}</green> | <level>{level: <8}</level> | <level>{message}</level>",
+    level="INFO"
+)
+
+# 프로젝트 루트를 Python 경로에 추가
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+
+from app.metacognitive_agent import MetaCognitiveAgent
+from app.rag_pipeline import RAGPipeline
+from services.vector_store import VectorStore
+from services.embedder import Embedder
+from utils.config import settings
+
+# 글로벌 변수
+rag_pipeline = None
+
+
+def setup_vector_db():
+    """벡터 DB 자동 설정 (없으면 다운로드 또는 생성)"""
+    db_path = settings.chroma_persist_directory
+
+    # 벡터 DB가 이미 존재하고 비어있지 않은지 확인
+    if os.path.exists(db_path):
+        if os.listdir(db_path):
+            logger.info("✅ Vector DB already exists. Skipping setup.")
+            return True
+
+    logger.info("📥 Vector DB not found. Setting up...")
+    os.makedirs(db_path, exist_ok=True)
+
+    # 옵션 1: GitHub Release에서 다운로드 시도
+    try:
+        import urllib.request
+        import tarfile
+
+        release_url = "https://github.com/csjjin2025/Hallucination_and_Deception_for_financial_RAG/releases/download/v1.0/chroma_db.tar.gz"
+        tar_path = "./data/chroma_db.tar.gz"
+
+        logger.info(f"Attempting to download from {release_url}...")
+        urllib.request.urlretrieve(release_url, tar_path)
+
+        # 파일 크기 확인
+        file_size = os.path.getsize(tar_path)
+        if file_size > 1000:
+            logger.info(f"📦 Extracting vector DB ({file_size} bytes)...")
+            with tarfile.open(tar_path, 'r:gz') as tar:
+                tar.extractall(path='./data/')
+            os.remove(tar_path)
+            logger.info("✅ Vector DB downloaded and extracted!")
+            return True
+        else:
+            logger.warning(f"Downloaded file too small ({file_size} bytes)")
+            os.remove(tar_path)
+    except Exception as e:
+        logger.warning(f"Failed to download from Release: {e}")
+
+    # 옵션 2: 테스트 DB 생성
+    try:
+        logger.info("⚠️ Creating test DB with sample data...")
+        import subprocess
+        result = subprocess.run(
+            ["python", "scripts/quick_setup_test_db.py"],
+            capture_output=True,
+            text=True,
+            timeout=300
+        )
+        if result.returncode == 0:
+            logger.info("✅ Test DB created successfully!")
+            return True
+        else:
+            logger.error(f"Test DB creation failed: {result.stderr}")
+            return False
+    except Exception as e:
+        logger.error(f"Failed to create test DB: {e}")
+        return False
+
+
+def initialize_rag_system():
+    """RAG 시스템 초기화"""
+    global rag_pipeline
+
+    try:
+        logger.info("=" * 80)
+        logger.info("🚀 Financial RAG 시스템 초기화 중...")
+        logger.info("=" * 80)
+
+        # 0. Vector DB 자동 설정
+        logger.info("0️⃣ Vector DB 설정 확인 중...")
+        if not setup_vector_db():
+            logger.error("❌ Vector DB 설정 실패")
+            return False
+
+        # 1. Vector Store 초기화
+        logger.info("1️⃣ Vector Store 로딩 중...")
+        vector_store = VectorStore(
+            persist_directory=settings.chroma_persist_directory,
+            collection_name=settings.collection_name
+        )
+        doc_count = vector_store.collection.count()
+        logger.info(f"✅ Vector Store 로딩 완료 ({doc_count}개 문서)")
+
+        # 2. Embedder 초기화
+        logger.info("2️⃣ Embedder 초기화 중...")
+        embedder = Embedder(
+            model_type=settings.embedding_model,
+            model_name=settings.embedding_model_name,
+            openai_api_key=settings.openai_api_key,
+            cohere_api_key=settings.cohere_api_key
+        )
+        logger.info(f"✅ Embedder 초기화 완료 ({embedder.get_embedding_dimension()}차원)")
+
+        # 3. Metacognitive Agent 초기화
+        logger.info("3️⃣ Metacognitive Agent 초기화 중...")
+        agent = MetaCognitiveAgent(api_key=settings.anthropic_api_key)
+        logger.info(f"✅ Agent 초기화 완료 ({agent.model})")
+
+        # 4. RAG Pipeline 생성
+        logger.info("4️⃣ RAG Pipeline 생성 중...")
+        rag_pipeline = RAGPipeline(
+            vector_store=vector_store,
+            embedder=embedder,
+            metacognitive_agent=agent
+        )
+        logger.info("✅ RAG Pipeline 생성 완료")
+
+        logger.info("=" * 80)
+        logger.info("✨ 시스템 준비 완료!")
+        logger.info(f"📚 Vector DB: {doc_count}개 문서")
+        logger.info(f"🤖 Model: {agent.model}")
+        logger.info("=" * 80)
+
+        return True
+
+    except Exception as e:
+        logger.error(f"❌ 초기화 실패: {str(e)}")
+        import traceback
+        logger.error(traceback.format_exc())
+        return False
+
+
+def format_sources(sources: list) -> str:
+    """출처 문서 포맷팅"""
+    if not sources:
+        return "출처 문서가 없습니다."
+
+    formatted = "### 📚 참고 문서\n\n"
+    for idx, source in enumerate(sources[:3], 1):  # 상위 3개만 표시
+        similarity = source.get('similarity', 0) * 100
+        filename = source.get('source_filename', 'unknown')
+        text = source.get('text', '')[:300]  # 앞 300자만
+
+        formatted += f"**{idx}. {filename}** (유사도: {similarity:.1f}%)\n"
+        formatted += f"> {text}...\n\n"
+
+    return formatted
+
+
+def format_metacognition(metacognition: Dict) -> str:
+    """메타인지 과정 포맷팅"""
+    if not metacognition:
+        return ""
+
+    history = metacognition.get('thinking_history', [])
+    iterations = metacognition.get('iterations', 0)
+
+    formatted = f"\n\n### 🧠 메타인지 과정 ({iterations}회 반복)\n\n"
+
+    for idx, step in enumerate(history, 1):
+        stage = step.get('stage', 'unknown')
+        content = step.get('content', '')
+
+        stage_emoji = {
+            'planning': '📋',
+            'monitoring': '👁️',
+            'evaluation': '✅',
+            'revision': '🔄'
+        }.get(stage, '💭')
+
+        formatted += f"**{stage_emoji} {stage.capitalize()}**\n"
+        formatted += f"{content}\n\n"
+
+    return formatted
+
+
+async def process_query_async(question: str, top_k: int, enable_metacognition: bool) -> Tuple[str, str]:
+    """비동기 쿼리 처리"""
+    if not rag_pipeline:
+        return "❌ 시스템이 초기화되지 않았습니다.", ""
+
+    if not question.strip():
+        return "❌ 질문을 입력해주세요.", ""
+
+    try:
+        logger.info(f"📝 질문: {question}")
+
+        # RAG 파이프라인으로 쿼리 처리
+        result = await rag_pipeline.query(
+            question=question,
+            top_k=top_k,
+            enable_metacognition=enable_metacognition
+        )
+
+        # 답변 포맷팅
+        answer = result.get('answer', '답변을 생성할 수 없습니다.')
+        sources = result.get('sources', [])
+        metacognition = result.get('metacognition', None)
+
+        # 출력 구성
+        main_output = f"## 💬 답변\n\n{answer}\n\n"
+        main_output += format_sources(sources)
+
+        # 메타인지 과정 (별도 탭)
+        meta_output = format_metacognition(metacognition) if metacognition else "메타인지가 비활성화되었습니다."
+
+        logger.info("✅ 답변 생성 완료")
+        return main_output, meta_output
+
+    except Exception as e:
+        error_msg = f"❌ 오류 발생: {str(e)}"
+        logger.error(error_msg)
+        import traceback
+        logger.error(traceback.format_exc())
+        return error_msg, ""
+
+
+def process_query(question: str, top_k: int, enable_metacognition: bool) -> Tuple[str, str]:
+    """Gradio용 동기 래퍼"""
+    loop = asyncio.new_event_loop()
+    asyncio.set_event_loop(loop)
+    try:
+        return loop.run_until_complete(process_query_async(question, top_k, enable_metacognition))
+    finally:
+        loop.close()
+
+
+# Gradio 인터페이스 구성
+def create_interface():
+    """Gradio 인터페이스 생성"""
+
+    with gr.Blocks(theme=gr.themes.Soft(), title="Financial RAG with Metacognitive Agent") as demo:
+        gr.Markdown("""
+        # 💼 Financial RAG System
+        ### 메타인지 에이전트 기반 금융/경제 질의응답 시스템
+
+        이 시스템은 금융/경제 논문을 기반으로 질문에 답변합니다.
+        메타인지 기능을 활성화하면 더 깊이 있는 사고 과정을 거쳐 답변을 생성합니다.
+        """)
+
+        with gr.Row():
+            with gr.Column(scale=2):
+                question_input = gr.Textbox(
+                    label="💬 질문을 입력하세요",
+                    placeholder="예: 포트폴리오 다각화는 무엇인가요?",
+                    lines=3
+                )
+
+                with gr.Row():
+                    top_k_slider = gr.Slider(
+                        minimum=1,
+                        maximum=10,
+                        value=5,
+                        step=1,
+                        label="🔍 검색할 문서 개수"
+                    )
+
+                    metacognition_check = gr.Checkbox(
+                        label="🧠 메타인지 활성화",
+                        value=True,
+                        info="더 깊이 있는 사고 과정 (처리 시간 증가)"
+                    )
+
+                submit_btn = gr.Button("🚀 질문하기", variant="primary", size="lg")
+
+                gr.Markdown("""
+                ### 💡 사용 팁
+                - **메타인지 활성화**: Planning → Monitoring �� Evaluation → Revision 과정을 거쳐 신중한 답변 생성
+                - **메타인지 비활성화**: 빠른 답변 생성
+                - **검색 문서 개수**: 많을수록 더 많은 정보를 참고하지만 처리 시간 증가
+                """)
+
+            with gr.Column(scale=3):
+                with gr.Tabs():
+                    with gr.Tab("📝 답변 및 출처"):
+                        answer_output = gr.Markdown(label="답변")
+
+                    with gr.Tab("🧠 메타인지 과정"):
+                        metacognition_output = gr.Markdown(label="사고 과정")
+
+        # 이벤트 핸들러
+        submit_btn.click(
+            fn=process_query,
+            inputs=[question_input, top_k_slider, metacognition_check],
+            outputs=[answer_output, metacognition_output]
+        )
+
+        # Enter 키로도 제출
+        question_input.submit(
+            fn=process_query,
+            inputs=[question_input, top_k_slider, metacognition_check],
+            outputs=[answer_output, metacognition_output]
+        )
+
+        gr.Markdown("""
+        ---
+        ### 📌 시스템 정보
+        - **모델**: Claude 3.5 Sonnet
+        - **임베딩**: sentence-transformers/all-MiniLM-L6-v2
+        - **벡터 DB**: ChromaDB
+        """)
+
+    return demo
+
+
+# 메인 실행
+if __name__ == "__main__":
+    # 시스템 초기화
+    logger.info("시스템 초기화 시작...")
+    success = initialize_rag_system()
+
+    if not success:
+        logger.error("시스템 초기화 실패. 종료합니다.")
+        sys.exit(1)
+
+    # Gradio 앱 실행
+    demo = create_interface()
+    demo.launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        share=False
+    )

requirements.txt

@@ -1,12 +1,6 @@
 # Gradio (HuggingFace Spaces)
 gradio>=5.0.0
 
-# FastAPI and Web Server (optional for Render.com)
-fastapi>=0.109.0
-uvicorn[standard]>=0.27.0
-pydantic>=2.5.0
-pydantic-settings>=2.1.0
-
 # Anthropic Claude
 anthropic>=0.18.0
 
@@ -17,7 +11,3 @@ sentence-transformers>=2.3.0
 # Utilities
 python-dotenv>=1.0.0
 loguru>=0.7.0
-numpy>=1.26.0
-
-# Required by ChromaDB
-requests>=2.28.0