구조적인 부분 수정

backend/app/services/analysis_db_service.py

@@ -0,0 +1,196 @@
+"""StockAnalysis DB 서비스 — 단일 진실 공급원 (Single Source of Truth)
+
+StockAnalysis ORM 저장 로직을 중앙화하여 3곳의 중복 제거:
+  - batch_daily_analysis.py   (bulk)
+  - batch_intraday_analysis.py (per-stock, rescore)
+
+PostgreSQL UPSERT(ON CONFLICT DO UPDATE)를 사용하여
+DELETE → INSERT 사이의 race condition을 원자적으로 해결.
+"""
+from __future__ import annotations
+
+import logging
+from datetime import date, datetime, timezone
+from typing import TYPE_CHECKING
+
+from sqlalchemy import select
+from sqlalchemy.dialects.postgresql import insert as pg_insert
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from app.core.database import AsyncSessionLocal
+from app.models.models import StockAnalysis
+
+logger = logging.getLogger(__name__)
+
+if TYPE_CHECKING:
+    from app.api.endpoints.recommendations import StockRecommendation
+
+
+def passes_price_filter(result: "StockRecommendation") -> bool:
+    """5,000원 이하 저가주 필터 (CLAUDE.md 5절).
+
+    True → 통과 (거래 가능), False → 필터됨 (5,000원 미만).
+    """
+    return not (result.current_price and result.current_price < 5000)
+
+
+# UPSERT 시 갱신할 컬럼 집합 (PK·unique key·최초등록 시각 제외)
+_UPSERT_UPDATE_COLS = frozenset({
+    "stock_name", "current_price", "change_rate", "volume",
+    "rsi", "ma5", "ma20", "ma60",
+    "stochastic_k", "stochastic_d",
+    "bollinger_upper", "bollinger_lower", "bollinger_middle",
+    "signals", "score",
+    "ml_score", "ml_confidence", "ml_signal",
+    "atr_14", "adx_14",
+    "updated_at",
+})
+
+
+def to_analysis_dict(
+    stock_code: str,
+    stock_name: str,
+    result: "StockRecommendation",
+    analysis_date: date,
+) -> dict:
+    """StockRecommendation → StockAnalysis INSERT 딕셔너리 변환.
+
+    ORM 객체 대신 dict를 반환하여 단건 UPSERT와 bulk INSERT에서 모두 재사용.
+    """
+    return {
+        "stock_code": stock_code,
+        "stock_name": stock_name,
+        "analysis_date": analysis_date,
+        "current_price": result.current_price,
+        "change_rate": result.change_rate,
+        "volume": result.volume,
+        "rsi": result.rsi,
+        "ma5": result.ma5,
+        "ma20": result.ma20,
+        "ma60": result.ma60,
+        "stochastic_k": result.stochastic_k,
+        "stochastic_d": result.stochastic_d,
+        "bollinger_upper": result.bollinger_upper,
+        "bollinger_lower": result.bollinger_lower,
+        "bollinger_middle": result.bollinger_middle,
+        "signals": result.signals,
+        "score": result.score,
+        "ml_score": result.ml_score,
+        "ml_confidence": result.ml_confidence,
+        "ml_signal": result.ml_signal,
+        "atr_14": result.atr_14,
+        "adx_14": result.adx_14,
+        "updated_at": datetime.now(timezone.utc).replace(tzinfo=None),
+    }
+
+
+async def upsert_analysis(
+    session: AsyncSession,
+    stock_code: str,
+    stock_name: str,
+    result: "StockRecommendation",
+    analysis_date: date,
+) -> None:
+    """단건 UPSERT — 동일 (stock_code, analysis_date)가 있으면 UPDATE, 없으면 INSERT.
+
+    DELETE+INSERT 대비:
+      - 원자적 연산 (race condition 없음)
+      - 단일 DB 왕복 (네트워크 절반)
+    """
+    row = to_analysis_dict(stock_code, stock_name, result, analysis_date)
+    stmt = pg_insert(StockAnalysis).values(**row)
+    stmt = stmt.on_conflict_do_update(
+        constraint="uq_stock_analysis_code_date",
+        set_={col: stmt.excluded[col] for col in _UPSERT_UPDATE_COLS},
+    )
+    await session.execute(stmt)
+
+
+async def bulk_upsert_analysis(
+    session: AsyncSession,
+    items: list[tuple[str, str, "StockRecommendation"]],
+    analysis_date: date,
+) -> int:
+    """다건 UPSERT — 단일 INSERT ... ON CONFLICT 배치 실행.
+
+    items: [(stock_code, stock_name, StockRecommendation), ...]
+    Returns: 처리된 건수
+    """
+    if not items:
+        return 0
+
+    rows = [to_analysis_dict(code, name, r, analysis_date) for code, name, r in items]
+    stmt = pg_insert(StockAnalysis).values(rows)
+    stmt = stmt.on_conflict_do_update(
+        constraint="uq_stock_analysis_code_date",
+        set_={col: stmt.excluded[col] for col in _UPSERT_UPDATE_COLS},
+    )
+    await session.execute(stmt)
+    return len(rows)
+
+
+async def run_db_scanner(analysis_date: date) -> set[str]:
+    """배치 DB 기반 스캐너 — KosdaqScanner와 동일한 기준을 단일일 데이터로 적용.
+
+    저장된 RSI / 볼린저 / 거래량 데이터만 사용하므로 추가 API 호출 없음.
+
+    조건 (성능 중심 필터):
+      - RSI(14) ≤ 30  (과매도)
+      - 종가 ≤ BB 하단 × 1.015  (BB 하단 근처)
+      - 시세 ≥ 5,000원
+      - ADX ≤ 30  (지나치게 강한 추세는 평균회귀 부적합)
+      - MA5 ≥ MA20 × 0.95  (ADX 방향 proxy — 강한 하락 추세 종목 제외)
+
+    Returns: 신호 종목 코드 set
+    """
+    try:
+        async with AsyncSessionLocal() as session:
+            result = await session.execute(
+                select(
+                    StockAnalysis.stock_code,
+                    StockAnalysis.current_price,
+                    StockAnalysis.rsi,
+                    StockAnalysis.bollinger_lower,
+                    StockAnalysis.adx_14,
+                    StockAnalysis.ma5,
+                    StockAnalysis.ma20,
+                ).where(
+                    StockAnalysis.analysis_date == analysis_date,
+                    StockAnalysis.current_price >= 5000,
+                    StockAnalysis.rsi.isnot(None),
+                    StockAnalysis.rsi <= 30,
+                    StockAnalysis.bollinger_lower.isnot(None),
+                )
+            )
+            rows = result.all()
+
+        signal_tickers: set[str] = set()
+        for row in rows:
+            price, bb_lower = row.current_price, row.bollinger_lower
+            if not bb_lower or bb_lower == 0:
+                continue
+
+            # BB 하단선 근처 확인 (-inf ~ +1.5%)
+            pct = (price - bb_lower) / bb_lower * 100
+            if pct > 1.5:
+                continue
+
+            # 구조적 추세장(ADX > 30)는 평균회귀에 부적합
+            adx = row.adx_14
+            if adx is not None and adx > 30:
+                continue
+
+            # ADX 방향 proxy: MA5 < MA20 × 0.95 → 강한 하락 추세 → 평균회귀 부적합
+            ma5, ma20 = row.ma5, row.ma20
+            if ma5 is not None and ma20 is not None and ma20 > 0:
+                if float(ma5) < float(ma20) * 0.95:
+                    continue
+
+            signal_tickers.add(row.stock_code)
+
+        logger.info("🔍 DB 스캐너: %d개 신호 종목 (RSI≤30 + BB하단 + ADX≤30 + MA방향)", len(signal_tickers))
+        return signal_tickers
+
+    except Exception as exc:
+        logger.error("❌ DB 스캐너 실패: %s", exc, exc_info=True)
+        return set()