랭그래프 활용한 주식 정보 도출 프로젝트

- 투자 의견 조회
ㄴ 전반적인 주식 판단 기준(ex. 채무, 리스크 등등)을 기준으로 판단하여 도출
- 투자 추천
ex)
{
  "tickers": [
    "NVDA", "GOOGL", "AAPL"
  ],
  "risk_type": "aggressive"
}

project/STOCK_APP/.idea/STOCK_APP.iml

@@ -2,13 +2,10 @@
 <module type="PYTHON_MODULE" version="4">
   <component name="NewModuleRootManager">
     <content url="file://$MODULE_DIR$">
+      <sourceFolder url="file://$MODULE_DIR$" isTestSource="false" />
       <excludeFolder url="file://$MODULE_DIR$/.venv" />
     </content>
     <orderEntry type="jdk" jdkName="~/Source/project/STOCK_APP/.venv" jdkType="Python SDK" />
     <orderEntry type="sourceFolder" forTests="false" />
   </component>
-  <component name="PackageRequirementsSettings" />
-  <component name="PyDocumentationSettings" />
-  <component name="ReSTService" />
-  <component name="TestRunnerService" />
 </module>

project/STOCK_APP/backend/config/settings.py

@@ -8,5 +8,6 @@ class Settings(BaseSettings):
     watsonx_project_id: str = Field(alias="WATSONX_PROJECT_ID")
     watsonx_url: str = Field(alias="WATSONX_URL")
     hf_token: str = Field(alias="HF_TOKEN")
+    serper_api_key: str = Field(alias="SERPER_API_KEY")
 
 settings = Settings()

project/STOCK_APP/backend/graph/nodes.py

@@ -1,6 +1,13 @@
 from backend.services.news_service import get_news
 from backend.services.financial_service import get_financial_info
 from backend.services.technical_service import get_technical_info
+from backend.services.competitor_service import get_competitor_info
+from langchain_core.prompts import ChatPromptTemplate
+from backend.ai.llm import hugging_llm
+from backend.prompts.all_prompt import REPORT_PROMPT
+
+prompt = ChatPromptTemplate.from_template(REPORT_PROMPT)
+report_chain = prompt | hugging_llm
 
 async def news_node(state):
     news_list = await get_news(state["company_name"])
@@ -15,7 +22,58 @@ async def technical_node(state):
     return {"technicals": technicals}
 
 async def competitor_node(state):
-    pass
+    competitors = await get_competitor_info(state["ticker"])
+    return {"competitors": competitors}
 
 async def report_node(state):
-    pass
+    """
+    수집된 정보를 LLM 에게 넘기고 분석 요청
+    1. 뉴스기사
+    2. 재무정보(시가총액, 매출, 순이익...)
+    3. 기술적 지표(현재주가, 20일선, 60일선, rsi...)
+    4. 경쟁회사 재무 정보
+    """
+
+    news_text = "\n".join([f"제목 : {n.title}\n내용 : {n.snippet}" for n in state["news"]])
+
+    financials = state['financials']
+    financial_text = f"""
+    시가총액 : {financials['market_cap']}
+    매출 : {financials['revenue']}
+    총자산 : {financials['total_assets']}
+    순이익 : {financials['net_income']}
+    PER : {financials['pe_ratio']}
+    PBR : {financials['pb_ratio']}
+"""
+
+    technicals = state['technicals']
+    technical_text = f"""
+    현재주가 : {technicals['current_price']}
+    20일선 : {technicals['sma20']}
+    60일선 : {technicals['sma60']}
+    RSI : {technicals['rsi']}
+    MACD : {technicals['macd']}
+    MACD_SIGNAL : {technicals['macd_signal']}
+    추세 : {technicals['trend']}
+"""
+
+    competitor_text = "\n".join([f"""
+    티커 : {c['ticker']}
+    회사 : {c['company_name']}
+    시가총액 : {c['market_cap']}
+    PER : {c['pe_ratio']}
+    매출성장률 : {c['revenue_growth']}
+    순이익률 : {c['profit_margin']}
+""" for c in state['competitors']])
+
+    report = await report_chain.ainvoke(
+        {
+            "company_name" : state["company_name"],
+            "news": news_text,
+            "financials": financial_text,
+            "technicals": technical_text,
+            "competitors": competitor_text,
+        }
+    )
+
+    return {"report": report.content}

project/STOCK_APP/backend/graph/state.py

@@ -1,10 +1,10 @@
 from typing import TypedDict
 
-class StockAnalysis(TypedDict):
+class StockAnalysisState(TypedDict):
     query:str
     ticker:str
     company_name:str
-    new:list
+    news:list
     financials:dict
     technicals:dict
     competitors:list

project/STOCK_APP/backend/graph/stock_graph.py

@@ -0,0 +1,30 @@
+from langgraph.graph import StateGraph, START, END
+from backend.graph.state import StockAnalysisState
+from backend.graph.nodes import news_node, financial_node, competitor_node, report_node, technical_node
+import logging
+
+logger = logging.getLogger(__name__)
+
+def build_stock_graph():
+    # 그래프 생성
+    graph = StateGraph(StockAnalysisState)
+
+    # 노드 등록
+    graph.add_node("news", news_node)
+    graph.add_node("financial", financial_node)
+    graph.add_node("technical", technical_node)
+    graph.add_node("competitors", competitor_node)
+    graph.add_node("report", report_node)
+
+    # 연결
+    graph.add_edge(START, "news")
+    graph.add_edge("news", "financial")
+    graph.add_edge("financial", "technical")
+    graph.add_edge("technical", "competitors")
+    graph.add_edge("competitors", "report")
+    graph.add_edge("report", END)
+
+    # 실행
+    compiled = graph.compile()
+    logger.info("LangGraph 주식 분석 그래프 빌드 완료")
+    return compiled

project/STOCK_APP/backend/main.py

@@ -4,40 +4,34 @@ from contextlib import asynccontextmanager
 
 from backend.routers.stock_router import router as stock_router
 
-# from backend.repository.db_init import Base, SessionLocal, engine
+from backend.repository.db_init import Base, SessionLocal, engine
-#
+
 # from backend.routers.evalu_router import router as evalu_router
 # from backend.routers.assistant_router import router as assistant_router
 # from backend.routers.call_router import router as call_router
 # from backend.repository.seed import seed_customers
 
-# @asynccontextmanager
+@asynccontextmanager
-# async def lifespan(app: FastAPI):
+async def lifespan(app: FastAPI):
-#     print("서버 시작")
+    print("서버 시작")
-#
+
-#     # Base 에 등록된 모든 모델에 테이블 자동 생성
+    # Base 에 등록된 모든 모델에 테이블 자동 생성
-#     Base.metadata.create_all(bind=engine)
+    Base.metadata.create_all(bind=engine)
-#     print("[DB] 테이블 생성 완료 (또는 이미 존재)")
+    print("[DB] 테이블 생성 완료 (또는 이미 존재)")
-#
+
-#     db = SessionLocal()
+    yield
-#     try:
+    print("서버 종료")
-#         # 기본 user 데이터 삽입
-#         seed_customers(db)
-#     finally:
-#         db.close()
-#
-#     yield
-#     print("서버 종료")
 
 # app = FastAPI()
 # app = FastAPI(title="상담 LLM", version="1.0", lifespan=lifespan)
-app = FastAPI(title="$ Stock AI", version="1.0", description="""
+app = FastAPI(title="$ Stock AI", version="1.0", lifespan=lifespan, description="""
     #### $주식 AI 분석 FastAPI + LangGraph + yfinance + ChromaDB
     
     ### 주요 기능
     | 번호 | 기능 | 엔드포인트 |
     | --- | --- | --- |
     |1 | 기업 종합 분석 | 'POST /api/stock/analysis'
+    |2 | 투자 의견 조회 | 'GET /api/stock/opinion/{analysis_id}'
     """,
 )
 

project/STOCK_APP/backend/prompts/all_prompt.py

@@ -0,0 +1,43 @@
+REPORT_PROMPT = """
+당신은 월가 애널리스트입니다.
+
+회사명 : 
+{company_name}
+
+최근 뉴스 :
+{news}
+
+재무정보 :
+{financials}
+
+기술적 지표 :
+{technicals}
+
+경쟁사 :
+{competitors}
+
+다음 형식으로 분석하세요.
+
+1. 기업 개요
+2. 최근 뉴스 핵심 요약
+3. 재무 상태 분석
+4. 기술적 분석
+5. 경쟁사 비교
+6. 주요 리스크
+7. 종합의견
+
+반드시 구체적인 수치를 인용하세요.
+"""
+
+INVESTOR_SENTIMENT_PROMPT = """
+다음 뉴스들의 전체적인 투자 심리를 분석하시오.
+
+{news_text}
+
+반드시 JSON 형식으로 응답하세요.
+{{
+    "positive":0,
+    "negative":0,
+    "neutral":0,
+}}
+"""

project/STOCK_APP/backend/repository/db_init.py

@@ -0,0 +1,23 @@
+from sqlalchemy import create_engine
+from sqlalchemy.orm import DeclarativeBase, sessionmaker
+from pathlib import Path
+
+Path("db").mkdir(parents=True, exist_ok=True)
+
+# echo sql 구문 출력시키는
+engine = create_engine('sqlite:///db/stock.db', echo=True)
+
+# 세션 팩토리 생성
+SessionLocal = sessionmaker(bind=engine, autoflush=False, autocommit=False)
+
+# Base = declarative_Base()
+class Base(DeclarativeBase):
+    pass
+
+# session 을 다른 모듈에서 사용할 수 있도록 제공
+def get_db():
+    db = SessionLocal()
+    try:
+        yield db
+    finally:
+        db.close()

project/STOCK_APP/backend/repository/models.py

@@ -0,0 +1,28 @@
+from backend.repository.db_init import Base
+from sqlalchemy.orm import Mapped, mapped_column, relationship
+from datetime import datetime
+from sqlalchemy import JSON, ForeignKey, String
+
+class StockAnalysis(Base):
+    __tablename__ = "stock_analysis"
+
+    analysis_id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    ticker:Mapped[str]
+    company_name:Mapped[str]
+    analysis_json: Mapped[dict] = mapped_column(JSON)
+    report:Mapped[str]
+    created_at:Mapped[datetime] = mapped_column(default=datetime.now, nullable=False)
+    opinion = relationship("InvestmentOpinion", back_populates="analysis", uselist=False)
+
+class InvestmentOpinion(Base):
+    __tablename__ = "investment_opinion"
+
+    opinion_id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    analysis_id: Mapped[int] = mapped_column(ForeignKey("stock_analysis.analysis_id"), unique=True)
+    opinion: Mapped[str]
+    # 투자의견 : Buy, sell.....
+    rating:Mapped[str] = mapped_column(String(20), nullable=True)
+    score:Mapped[int] = mapped_column(default=0)
+    created_at:Mapped[datetime] = mapped_column(default=datetime.now, nullable=False)
+    analysis = relationship("StockAnalysis", back_populates="opinion")
+

project/STOCK_APP/backend/routers/stock_router.py

@@ -1,12 +1,30 @@
 from fastapi import APIRouter, Depends
-from sqlalchemy.orm import Session
-
 from backend.schemas.stock_schemas import StockAnalyzeReq
 from backend.services.stock_service import StockService
+from sqlalchemy.orm import Session
+from backend.repository.db_init import get_db
+from backend.services.stock_service import get_stock_service
+from backend.schemas.stock_schemas import RecommendStock
+from backend.services.recommend_service import generate_portfolio_report, portfolio_service
 
 router = APIRouter(prefix="/api/stock", tags=["기헙종합분석"])
 
 @router.post("/analyze", summary="기업 종합 분석", description="자연어로 해당기업을 요청하면 뉴스, 재무, 기술적, 경쟁사 분석을 수행합니다.", )
-async def stock_analyze(req:StockAnalyzeReq):
+async def stock_analyze(req:StockAnalyzeReq, db:Session=Depends(get_db)):
-    service = StockService()
+    service = get_stock_service()
-    return await service._extract(req.query)
+    return await service.analyze(req.query, db)
+
+@router.get("/opinion/{analysis_id}", summary="투자 의견 조회", description="", )
+async def stock_analyze_opinion(analysis_id:int, db:Session=Depends(get_db)):
+    service = get_stock_service()
+    return await service.opinion_service(analysis_id, db)
+
+@router.post("/recommend", summary="투자 추천", description="", )
+async def stock_recommend(req: RecommendStock, db: Session=Depends(get_db)):
+    result = portfolio_service(req, db)
+    report = await generate_portfolio_report(result['portfolio'], req.risk_type)
+    return {
+        "risk_type":req.risk_type,
+        "portfolio":result['portfolio'],
+        "report":report,
+    }

project/STOCK_APP/backend/schemas/stock_schemas.py

@@ -1,5 +1,5 @@
 from pydantic import BaseModel
-
+from enum import Enum
 
 class StockAnalyzeReq(BaseModel):
     query: str
@@ -7,3 +7,15 @@ class StockAnalyzeReq(BaseModel):
 class TickerInfo(BaseModel):
     ticker: str
     company_name: str
+
+
+# enum
+# risk_type : aggressive, balanced, conservative
+class RiskType(str, Enum):
+    AGGRESSIVE = "aggressive"
+    BALANCED = "balanced"
+    CONSERVATIVE = "conservative"
+
+class RecommendStock(BaseModel):
+    tickers: list[str]
+    risk_type : RiskType

project/STOCK_APP/backend/services/competitor_service.py

@@ -0,0 +1,41 @@
+import yfinance as yf
+
+COMPETITOR_MAP = {
+    "NVDA": ["AMD", "INTC", "TSM", "AVGO"],
+    "AMD": ["NVDA", "INTC", "TSM"],
+    "TSLA": ["RIVN", "GM", "F"],
+    "AAPL": ["MSFT", "GOOGL", "AMZN"],
+    "MSFT": ["AAPL", "GOOGL", "AMZN"],
+    "GOOGL": ["MSFT", "META", "AMZN"],
+}
+
+async def get_competitor_info(ticker: str):
+    """
+    경쟁 관계 회사에 대한 핵심 정보 추출
+    """
+    competitors_tickers = COMPETITOR_MAP.get(ticker, [])
+
+    results =[]
+    for comp in competitors_tickers:
+        company = yf.Ticker(comp)
+        info = company.info
+
+        results.append(
+            {
+                #
+                "ticker": comp,
+                # 회사 이름
+                "company_name": info.get("longName"),
+                # 시가총액
+                "market_cap": info.get("marketCap"),
+                # per
+                "pe_ratio": info.get("pe_ratio"),
+                # 매출 성장률
+                "revenue_growth": info.get("revenue_growth"),
+                # 순 이익률
+                "profit_margin": info.get("profitMargins"),
+            }
+        )
+
+    return results
+

project/STOCK_APP/backend/services/financial_service.py

@@ -1,4 +1,5 @@
-import yahooquery as yf
+import yfinance as yf
+# from yahooquery import Ticker
 
 async def get_financial_info(ticker: str):
     """
@@ -7,7 +8,7 @@ async def get_financial_info(ticker: str):
     company = yf.Ticker(ticker)
 
     info = company.info
-    income = company.income_stmtf
+    income = company.income_stmt
     balance = company.balance_sheet
     cashflow = company.cash_flow
 
@@ -31,6 +32,5 @@ async def get_financial_info(ticker: str):
         # per
         "pe_ratio": info.get("trailingPE"),
         # pbr
-        "pb_ratio":info.get("priceToBook")
+        "pb_ratio": info.get("priceToBook"),
     }
-

project/STOCK_APP/backend/services/news_service.py

@@ -1,6 +1,6 @@
 from langchain_community.utilities import GoogleSerperAPIWrapper
-
 from backend.schemas.news_schemas import NewsItem
+from backend.config.settings import settings
 
 
 async def get_news(company_name: str):
@@ -8,8 +8,9 @@ async def get_news(company_name: str):
     구글 뉴스 검색 후
     title, snippet, url, source, date 추출
     """
-    search = GoogleSerperAPIWrapper(type="news")
+    search = GoogleSerperAPIWrapper(type="news", serper_api_key=f"{settings.serper_api_key}")
-    results = await search.results(f"{company_name} stock news")
+    # results = search.aresults(f"{company_name} stock news")
+    results = await search.aresults(f"{company_name} stock news")
     news_list = []
 
     for item in results['news'][:10]:

project/STOCK_APP/backend/services/recommend_service.py

@@ -0,0 +1,71 @@
+from sqlalchemy.orm import Session
+from backend.schemas.stock_schemas import RecommendStock, RiskType
+from backend.repository.models import StockAnalysis
+from fastapi import HTTPException
+from backend.ai.llm import hugging_llm
+
+def portfolio_service(req, db):
+    """
+    사용자가 입력한 종목에 대해서 분석정보 찾기
+    risk_type: 비중 조절
+    """
+
+    analyses = db.query(StockAnalysis).filter(StockAnalysis.ticker.in_(req.tickers)).all()
+
+    portfolio_items = []
+    for analysis in analyses:
+        if not analysis.opinion:
+            continue
+
+        portfolio_items.append(
+            {
+            "ticker": analysis.ticker,
+            "score": analysis.opinion.score,
+            "rating": analysis.opinion.rating,
+            }
+        )
+
+    if not portfolio_items:
+        raise HTTPException(status_code=5000, detail="분석된 종목을 찾을 수 없습니다.")
+
+    # 투자성향
+    risk_type = req.risk_type
+    import math
+    for item in portfolio_items:
+        score = item["score"]
+
+        if risk_type == RiskType.AGGRESSIVE:
+            item['adjusted_score'] = score ** 1.5
+        elif risk_type == RiskType.CONSERVATIVE:
+            item['adjusted_score'] = math.sqrt(score)
+        else:
+            item['adjusted_score'] = score
+    total_adjusted = sum(item['adjusted_score'] for item in portfolio_items)
+
+    for item in portfolio_items:
+        item['weight'] = round(item['adjusted_score'] / total_adjusted * 100,2)
+
+    return {"risk_type": req.risk_type, "portfolio": portfolio_items}
+
+
+async def generate_portfolio_report(portfolio_items, risk_type):
+    portfolio_text = "\n".join([f"{item['ticker']}: {item['weight']}%" for item in portfolio_items])
+    prompt = f"""
+당신은 월가의 포트폴리오 전략가입니다.
+
+투자성향
+{risk_type.value}
+
+포트폴리오:
+{portfolio_text}
+
+다음을 작성하시오.
+1. 포트폴리오 요약
+2. 강점
+3. 약점
+4. 주요 리스크
+5. 추천 투자기간
+6. 리밸런싱 전략
+"""
+    result = await hugging_llm.ainvoke(prompt)
+    return result.content

project/STOCK_APP/backend/services/score_service.py

@@ -0,0 +1,137 @@
+from langchain_core.output_parsers import JsonOutputParser
+from langchain_core.prompts import ChatPromptTemplate
+from backend.ai.llm import hugging_llm
+from backend.prompts.all_prompt import INVESTOR_SENTIMENT_PROMPT
+
+def get_rating(score):
+    # 최종판단
+    if score >= 85:
+        return "Strong Buy"
+    elif score >= 70:
+        return "Buy"
+    elif score >= 50:
+        return "Hold"
+    elif score >= 30:
+        return "Sell"
+    return "Strong Sell"
+
+# 재무 정보 1. PER(20점)
+def get_per_score(per):
+    if per is None:
+        return 0
+    if per < 20:
+        return 20
+    elif per < 35:
+        return 15
+    return 10
+
+# 2. 순이익률
+def get_profit_margin(financials):
+    net_income = financials.get("net_income")
+    revenue = financials.get("revenue")
+
+    if not revenue:
+        return 0, 0
+    profit_margin = net_income / revenue
+
+    profit_score = 0
+    if profit_margin > 0.3:
+        profit_score = 20
+    elif profit_margin > 0.15:
+        profit_score = 15
+    else:
+        profit_score = 10
+
+    return profit_score, profit_margin
+
+# 3. 기술분석 (20점)
+def get_tech_score(technicals):
+    technicals_score = 0
+    trend = technicals['trend']
+    if trend == 'bullish':
+        technicals_score += 10
+
+    macd = technicals['macd']
+    macd_signal = technicals['macd_signal']
+
+    if macd > macd_signal:
+        technicals_score += 10
+    return technicals_score
+
+# 4. 경쟁사 분석(20점)
+def get_com_score(competitor, per, profit_margin):
+    # 경쟁사 per 평균 비교
+    competitors_score = 0
+    pe_list = [c["pe_ratio"] for c in competitor if c["pe_ratio"] is not None]
+
+    if not pe_list:
+        avg_pe = None
+    else:
+        avg_pe = sum(pe_list) / len(pe_list)
+
+    if avg_pe is not None and per < avg_pe:
+        competitors_score += 10
+
+    profit_margin_list = [c["profit_margin"] for c in competitor if c["profit_margin"] is not None]
+
+    avg_profit_margin = sum(profit_margin_list) / len(profit_margin_list)
+    if profit_margin > avg_profit_margin:
+        competitors_score += 10
+
+    return competitors_score
+
+# 5. 뉴스 분석(20점)
+async def get_news_score(news_list):
+    # 뉴스 투자 심리
+    news_text = "\n".join(
+        [f"제목 : {n['title']}\n내용 : {n['snippet']}" for n in news_list]
+    )
+
+    prompt = ChatPromptTemplate.from_template(INVESTOR_SENTIMENT_PROMPT)
+
+    chain = prompt | hugging_llm | JsonOutputParser()
+
+    sentiment = await chain.ainvoke({"news_text":news_text})
+
+    total = sentiment["positive"] + sentiment["negative"] + sentiment["neutral"]
+
+    if total == 0:
+        return 0
+
+    positive_ratio = sentiment["positive"] / total
+    return round(positive_ratio * 20)
+
+async def evaluation(analysis):
+    score = 0
+    data = analysis.analysis_json
+
+    # 1. per 점수
+    pe_ratio = data['financials'].get('pe_ratio')
+    per_score = get_per_score(pe_ratio)
+
+    score += per_score
+
+    # 2. 순이익률 점수
+    profit_score, profit_margin = get_profit_margin(data['financials'])
+    score += profit_score
+
+    # 3. 기술적 점수
+    technicals_score = get_tech_score(data["technicals"])
+    score += technicals_score
+
+    # 4. 경쟁사 점수
+    competitor_score = get_com_score(data['competitors'], pe_ratio, profit_margin)
+    score += competitor_score
+
+    # 5. 뉴스 점수
+    news_score = await get_news_score(data['news'])
+    score += news_score
+
+    return {
+        "total_score": score,
+        "per_score": per_score,
+        "profit_score": profit_score,
+        "technical_score": technicals_score,
+        "competitor_score": competitor_score,
+        "news_score": news_score
+    }

project/STOCK_APP/backend/services/stock_service.py

@@ -1,6 +1,12 @@
 from backend.schemas.stock_schemas import TickerInfo
 import yahooquery as yq
 from fastapi import HTTPException
+from backend.graph.stock_graph import build_stock_graph
+from backend.repository.models import StockAnalysis, InvestmentOpinion
+from fastapi.encoders import jsonable_encoder
+from typing import Optional
+from backend.services.score_service import evaluation, get_rating
+from backend.ai.llm import hugging_llm
 
 COMPANY_MAP = {
     "NVDA": {
@@ -115,6 +121,8 @@ class StockService:
                 if alias.lower() in query:
                     return TickerInfo(ticker = ticker, company_name = info["company_name"])
 
+        return None
+
     def _extract_company_keyword(self, query:str):
         """
         query : 엔비디아 분석해줘
@@ -146,3 +154,88 @@ class StockService:
 
         except Exception:
             return None
+
+    async def analyze(self, query: str, db):
+        # 티커, 회사명 추출
+        ticker_info = await self._extract(query)
+
+        # 그래프 실행
+        stock_graph = build_stock_graph()
+        result = await stock_graph.ainvoke(
+            {
+            "query": query,
+            "ticker": ticker_info.ticker,
+            "company_name": ticker_info.company_name
+            }
+        )
+
+        # 데이터베이스 저장
+        # 테이블과 관련있는 모델 객체 생성
+        entity = StockAnalysis(
+            ticker = result['ticker'],
+            company_name = result['company_name'],
+            analysis_json = jsonable_encoder(result),
+            report = result['report']
+        )
+        db.add(entity)
+        db.commit()
+        db.refresh(entity)
+
+        return {"analysis_id":entity.analysis_id, **result}
+
+    async def opinion_service(self, analysis_id, db):
+        # analysis_id 디비 조회
+        analysis = db.get(StockAnalysis, analysis_id)
+
+        score_result = await evaluation(analysis)
+
+        #
+        total_score = score_result['total_score']
+        rating = get_rating(total_score)
+
+        # AI 투자 의견서
+        opinion = await self.generate_opinion(analysis.report, total_score, rating)
+
+        entity = InvestmentOpinion(
+            analysis_id = analysis_id,
+            opinion = opinion,
+            rating = rating,
+            score = total_score,
+        )
+        db.add(entity)
+        db.commit()
+        db.refresh(entity)
+
+        return {"opinion_id":entity.opinion_id, "analysis_id":analysis_id, "opinion":opinion, "rating":rating, **score_result,}
+
+    async def generate_opinion(self, report, total_score, rating):
+        prompt = f"""
+        당신은 월가의 수석 애널리스트입니다.
+        
+        종합점수 : 
+        {total_score} / 100
+        투자등급 :
+        {rating}
+        
+        다음 기업 분석 보고서를 기반으로 투자 의견서를 작성하세요.
+        {report}
+        
+        반드시 아래 형식으로 작성하세요.
+        1. 투자등급
+        2. 투자근거
+        3. 핵심리스크
+        4. 단기전망
+        5. 장기전망
+        6. 최종의견
+        """
+        result = await hugging_llm.ainvoke(prompt)
+        return result.content
+
+# 싱글톤(객체 하나만 생성) 서비스 인스턴스
+_service:Optional[StockService] = None
+
+def get_stock_service():
+    global _service
+    if _service is None:
+        _service = StockService()
+    return _service

project/STOCK_APP/backend/services/technical_service.py

@@ -1,4 +1,4 @@
-import yahooquery as yf
+import yfinance as yf
 from ta.trend import MACD, SMAIndicator
 from ta.momentum import RSIIndicator
 
@@ -10,7 +10,7 @@ async def get_technical_info(ticker: str):
     # 주가 데이터 다운로드
     df = yf.download(ticker, period="1y", interval="1d")
     # 종가 가져오기
-    close = df['Close'].squeeze()
+    close = df["Close"].squeeze()
 
     macd = MACD(close)
     macd_value = float(macd.macd().iloc[-1])
@@ -25,12 +25,11 @@ async def get_technical_info(ticker: str):
         # 60일선
         "sma60": float(SMAIndicator(close, window=60).sma_indicator().iloc[-1]),
         # rsi
-        "rsi":float(SMAIndicator(close, window=14).rsi().iloc[-1]),
+        "rsi": float(RSIIndicator(close, window=14).rsi().iloc[-1]),
         # macd
-        "macd":macd.macd(),
+        "macd": macd_value,
         # macd_signal
         "macd_signal": signal_value,
         # trend
         "trend": trend,
     }
-