api_utils.py

"""
API 유틸리티 모듈
실무에서 바로 사용 가능한 조위 예측 API 함수들
"""

from datetime import datetime, timedelta
from typing import Dict, List, Optional, Union
import pytz
from supabase_utils import get_supabase_client
from config import STATION_NAMES

# API 응답 표준 포맷
def create_api_response(success: bool, data: any = None, error: str = None, meta: Dict = None) -> Dict:
    """표준 API 응답 포맷 생성"""
    response = {
        "success": success,
        "timestamp": datetime.now(pytz.timezone('Asia/Seoul')).isoformat(),
    }
    
    if meta:
        response["meta"] = meta
    
    if success:
        response["data"] = data
    else:
        response["error"] = error or "Unknown error"
    
    return response

def get_station_meta(station_id: str) -> Dict:
    """관측소 메타 정보 반환"""
    # 관측소 좌표 정보 (실제 좌표)
    STATION_COORDS = {
        "DT_0001": {"lat": 37.452, "lon": 126.592},
        "DT_0002": {"lat": 36.9669, "lon": 126.823},
        "DT_0003": {"lat": 35.4262, "lon": 126.421},
        "DT_0008": {"lat": 37.1922, "lon": 126.647},
        "DT_0017": {"lat": 37.0075, "lon": 126.353},
        "DT_0018": {"lat": 35.9755, "lon": 126.563},
        "DT_0024": {"lat": 36.0069, "lon": 126.688},
        "DT_0025": {"lat": 36.4064, "lon": 126.486},
        "DT_0037": {"lat": 36.1173, "lon": 125.985},
        "DT_0043": {"lat": 37.2394, "lon": 126.429},
        "DT_0050": {"lat": 36.9131, "lon": 126.239},
        "DT_0051": {"lat": 36.1289, "lon": 126.495},
        "DT_0052": {"lat": 37.3382, "lon": 126.586},
        "DT_0065": {"lat": 37.2394, "lon": 126.155},
        "DT_0066": {"lat": 35.6858, "lon": 126.334},
        "DT_0067": {"lat": 36.6737, "lon": 126.132},
        "DT_0068": {"lat": 35.6181, "lon": 126.302},
    }
    
    coords = STATION_COORDS.get(station_id, {"lat": 0, "lon": 0})
    
    return {
        "obs_post_id": station_id,
        "obs_post_name": STATION_NAMES.get(station_id, "Unknown"),
        "obs_lat": str(coords["lat"]),
        "obs_lon": str(coords["lon"]),
        "data_type": "prediction"  # 예측 데이터임을 명시
    }

# 1. 현재/미래 조위 조회 (조화 예측 폴백 포함)
def api_get_tide_level(
    station_id: str, 
    target_time: Optional[str] = None,
    use_harmonic_fallback: bool = True
) -> Dict:
    """
    특정 시간의 조위 정보 조회
    
    Args:
        station_id: 관측소 ID
        target_time: 조회 시간 (ISO format, None이면 현재 시간)
        use_harmonic_fallback: 최종 예측이 없을 때 조화 예측 사용 여부
    
    Returns:
        API 응답 (최종 예측 우선, 없으면 조화 예측)
    """
    supabase = get_supabase_client()
    if not supabase:
        return create_api_response(False, error="Database connection failed")
    
    try:
        # 대상 시간 파싱
        kst = pytz.timezone('Asia/Seoul')
        if target_time:
            query_time = datetime.fromisoformat(target_time.replace('Z', '+00:00'))
            if query_time.tzinfo is None:
                query_time = kst.localize(query_time)
        else:
            query_time = datetime.now(kst)
        
        # UTC로 변환하여 쿼리 (중요!)
        query_time_utc = query_time.astimezone(pytz.UTC)
        query_str = query_time_utc.strftime('%Y-%m-%dT%H:%M:%S')
        
        # 가장 가까운 5분 단위로 반올림
        minutes = query_time.minute
        rounded_minutes = round(minutes / 5) * 5
        if rounded_minutes == 60:
            query_time_rounded = query_time.replace(minute=0, second=0, microsecond=0) + timedelta(hours=1)
        else:
            query_time_rounded = query_time.replace(minute=rounded_minutes, second=0, microsecond=0)
        
        query_time_rounded_utc = query_time_rounded.astimezone(pytz.UTC)
        
        # 1차: 정확한 시간 매칭 시도
        result = supabase.table('tide_predictions')\
            .select('*')\
            .eq('station_id', station_id)\
            .eq('predicted_at', query_time_rounded_utc.strftime('%Y-%m-%dT%H:%M:%S'))\
            .execute()
        
        # 2차: 정확한 매칭이 없으면 전후 5분 범위에서 가장 가까운 것
        if not result.data:
            start_time = (query_time_rounded_utc - timedelta(minutes=5)).strftime('%Y-%m-%dT%H:%M:%S')
            end_time = (query_time_rounded_utc + timedelta(minutes=5)).strftime('%Y-%m-%dT%H:%M:%S')
            
            result = supabase.table('tide_predictions')\
                .select('*')\
                .eq('station_id', station_id)\
                .gte('predicted_at', start_time)\
                .lte('predicted_at', end_time)\
                .order('predicted_at')\
                .limit(1)\
                .execute()
        
        if result.data:
            # 최종 예측 데이터가 있는 경우
            data = result.data[0]
            # UTC를 KST로 변환
            time_utc = datetime.fromisoformat(data['predicted_at'].replace('Z', '+00:00'))
            time_kst = time_utc.astimezone(pytz.timezone('Asia/Seoul'))
            
            return create_api_response(
                success=True,
                data={
                    "record_time": time_kst.isoformat(),  # KST로 변환
                    "record_time_kst": time_kst.strftime('%Y-%m-%d %H:%M:%S KST'),
                    "final_value": round(data.get('final_tide_level', 0), 1),
                    "residual_value": round(data.get('predicted_residual', 0), 1),
                    "harmonic_value": round(data.get('harmonic_level', 0), 1),
                    "data_source": "final_prediction",
                    "confidence": "high"
                },
                meta=get_station_meta(station_id)
            )
        
        # 2차: 조화 예측 (harmonic_predictions) 폴백
        if use_harmonic_fallback and not result.data:
            # 1차: 정확한 시간 매칭 시도
            result = supabase.table('harmonic_predictions')\
                .select('*')\
                .eq('station_id', station_id)\
                .eq('predicted_at', query_time_rounded_utc.strftime('%Y-%m-%dT%H:%M:%S'))\
                .execute()
            
            # 2차: 정확한 매칭이 없으면 전후 5분 범위
            if not result.data:
                start_time = (query_time_rounded_utc - timedelta(minutes=5)).strftime('%Y-%m-%dT%H:%M:%S')
                end_time = (query_time_rounded_utc + timedelta(minutes=5)).strftime('%Y-%m-%dT%H:%M:%S')
                
                result = supabase.table('harmonic_predictions')\
                    .select('*')\
                    .eq('station_id', station_id)\
                    .gte('predicted_at', start_time)\
                    .lte('predicted_at', end_time)\
                    .order('predicted_at')\
                    .limit(1)\
                    .execute()
            
            if result.data:
                data = result.data[0]
                # UTC를 KST로 변환
                time_utc = datetime.fromisoformat(data['predicted_at'].replace('Z', '+00:00'))
                time_kst = time_utc.astimezone(pytz.timezone('Asia/Seoul'))
                
                return create_api_response(
                    success=True,
                    data={
                        "record_time": time_kst.isoformat(),  # KST로 변환
                        "record_time_kst": time_kst.strftime('%Y-%m-%d %H:%M:%S KST'),
                        "final_value": round(data.get('harmonic_level', 0), 1),
                        "residual_value": None,  # 잔차 예측 없음
                        "harmonic_value": round(data.get('harmonic_level', 0), 1),
                        "data_source": "harmonic_only",
                        "confidence": "medium",
                        "note": "잔차 예측이 없어 조화 예측만 제공됩니다",
                        "query_time": query_time.strftime('%Y-%m-%d %H:%M:%S KST'),
                        "matched_time_diff_seconds": abs((time_utc - query_time_utc).total_seconds())
                    },
                    meta=get_station_meta(station_id)
                )
        
        return create_api_response(
            success=False,
            error=f"No data available for {query_str}",
            meta=get_station_meta(station_id)
        )
        
    except Exception as e:
        return create_api_response(False, error=str(e))

# 2. 시간대별 조위 조회 (공공 API 형식)
def api_get_tide_series(
    station_id: str,
    start_time: Optional[str] = None,
    end_time: Optional[str] = None,
    interval_minutes: int = 60
) -> Dict:
    """
    시간대별 조위 정보 조회 (공공 API 형식과 유사)
    
    Args:
        station_id: 관측소 ID
        start_time: 시작 시간 (None이면 현재)
        end_time: 종료 시간 (None이면 24시간 후)
        interval_minutes: 데이터 간격 (기본 60분)
    
    Returns:
        시계열 데이터
    """
    supabase = get_supabase_client()
    if not supabase:
        return create_api_response(False, error="Database connection failed")
    
    try:
        # 시간 범위 설정
        kst = pytz.timezone('Asia/Seoul')
        if start_time:
            start_dt = datetime.fromisoformat(start_time.replace('Z', '+00:00'))
        else:
            start_dt = datetime.now(kst)
        
        if end_time:
            end_dt = datetime.fromisoformat(end_time.replace('Z', '+00:00'))
        else:
            end_dt = start_dt + timedelta(hours=24)
        
        # 최종 예측 조회
        result = supabase.table('tide_predictions')\
            .select('predicted_at, final_tide_level, predicted_residual, harmonic_level')\
            .eq('station_id', station_id)\
            .gte('predicted_at', start_dt.strftime('%Y-%m-%dT%H:%M:%S'))\
            .lte('predicted_at', end_dt.strftime('%Y-%m-%dT%H:%M:%S'))\
            .order('predicted_at')\
            .execute()
        
        data_points = []
        data_source = "final_prediction"
        
        if result.data:
            # 간격에 맞춰 데이터 필터링
            for i, item in enumerate(result.data):
                if i % (interval_minutes // 5) == 0:  # 5분 간격 데이터 기준
                    # UTC를 KST로 변환
                    time_utc = datetime.fromisoformat(item['predicted_at'].replace('Z', '+00:00'))
                    time_kst = time_utc.astimezone(pytz.timezone('Asia/Seoul'))
                    
                    data_points.append({
                        "record_time": time_kst.strftime('%Y-%m-%d %H:%M:%S'),  # KST
                        "real_value": str(round(item['final_tide_level'], 0)),  # 정수로 표시
                        "pre_value": str(round(item['harmonic_level'], 0)),
                        "residual": str(round(item['predicted_residual'], 0))
                    })
        else:
            # 조화 예측 폴백
            result = supabase.table('harmonic_predictions')\
                .select('predicted_at, harmonic_level')\
                .eq('station_id', station_id)\
                .gte('predicted_at', start_dt.strftime('%Y-%m-%dT%H:%M:%S'))\
                .lte('predicted_at', end_dt.strftime('%Y-%m-%dT%H:%M:%S'))\
                .order('predicted_at')\
                .execute()
            
            if result.data:
                data_source = "harmonic_only"
                for i, item in enumerate(result.data):
                    if i % (interval_minutes // 5) == 0:
                        # UTC를 KST로 변환
                        time_utc = datetime.fromisoformat(item['predicted_at'].replace('Z', '+00:00'))
                        time_kst = time_utc.astimezone(pytz.timezone('Asia/Seoul'))
                        
                        data_points.append({
                            "record_time": time_kst.strftime('%Y-%m-%d %H:%M:%S'),  # KST
                            "real_value": str(round(item['harmonic_level'], 0)),
                            "pre_value": str(round(item['harmonic_level'], 0)),
                            "residual": "0"
                        })
        
        meta = get_station_meta(station_id)
        meta["data_source"] = data_source
        meta["data_count"] = len(data_points)
        meta["interval_minutes"] = interval_minutes
        
        return {
            "result": {
                "meta": meta,
                "data": data_points
            }
        }
        
    except Exception as e:
        return create_api_response(False, error=str(e))

# 3. 만조/간조 정보
def api_get_extremes_info(
    station_id: str,
    date: Optional[str] = None,
    include_secondary: bool = False
) -> Dict:
    """
    특정 날짜의 만조/간조 정보
    
    Args:
        station_id: 관측소 ID
        date: 날짜 (YYYY-MM-DD, None이면 오늘)
        include_secondary: 부차 만조/간조 포함 여부
    
    Returns:
        만조/간조 시간과 수위
    """
    supabase = get_supabase_client()
    if not supabase:
        return create_api_response(False, error="Database connection failed")
    
    try:
        # 날짜 범위 설정
        if date:
            target_date = datetime.strptime(date, '%Y-%m-%d')
        else:
            target_date = datetime.now(pytz.timezone('Asia/Seoul'))
        
        start_dt = target_date.replace(hour=0, minute=0, second=0)
        end_dt = target_date.replace(hour=23, minute=59, second=59)
        
        # 데이터 조회 (최종 예측 우선)
        result = supabase.table('tide_predictions')\
            .select('predicted_at, final_tide_level')\
            .eq('station_id', station_id)\
            .gte('predicted_at', start_dt.strftime('%Y-%m-%dT%H:%M:%S'))\
            .lte('predicted_at', end_dt.strftime('%Y-%m-%dT%H:%M:%S'))\
            .order('predicted_at')\
            .execute()
        
        data_source = "final_prediction"
        
        # 데이터가 없으면 조화 예측 사용
        if not result.data:
            result = supabase.table('harmonic_predictions')\
                .select('predicted_at, harmonic_level')\
                .eq('station_id', station_id)\
                .gte('predicted_at', start_dt.strftime('%Y-%m-%dT%H:%M:%S'))\
                .lte('predicted_at', end_dt.strftime('%Y-%m-%dT%H:%M:%S'))\
                .order('predicted_at')\
                .execute()
            
            if result.data:
                # 컬럼명 통일
                for item in result.data:
                    item['final_tide_level'] = item.pop('harmonic_level')
                data_source = "harmonic_only"
        
        if not result.data or len(result.data) < 3:
            return create_api_response(False, error="Insufficient data for extremes")
        
        # 극값 찾기
        extremes = []
        data = result.data
        
        for i in range(1, len(data) - 1):
            prev_level = data[i-1]['final_tide_level']
            curr_level = data[i]['final_tide_level']
            next_level = data[i+1]['final_tide_level']
            
            # 만조 (극대값)
            if curr_level > prev_level and curr_level > next_level:
                # UTC를 KST로 변환
                time_utc = datetime.fromisoformat(data[i]['predicted_at'].replace('Z', '+00:00'))
                time_kst = time_utc.astimezone(pytz.timezone('Asia/Seoul'))
                
                extremes.append({
                    'type': 'high_tide',
                    'time': time_kst.isoformat(),  # KST ISO format
                    'time_kst': time_kst.strftime('%Y-%m-%d %H:%M:%S KST'),
                    'level': round(curr_level, 2)
                })
            # 간조 (극소값)
            elif curr_level < prev_level and curr_level < next_level:
                # UTC를 KST로 변환
                time_utc = datetime.fromisoformat(data[i]['predicted_at'].replace('Z', '+00:00'))
                time_kst = time_utc.astimezone(pytz.timezone('Asia/Seoul'))
                
                extremes.append({
                    'type': 'low_tide',
                    'time': time_kst.isoformat(),  # KST ISO format
                    'time_kst': time_kst.strftime('%Y-%m-%d %H:%M:%S KST'),
                    'level': round(curr_level, 2)
                })
        
        # 주요 만조/간조만 필터링 (부차 제외)
        if not include_secondary and len(extremes) > 4:
            # 수위 차이가 큰 것들만 선택
            high_tides = sorted([e for e in extremes if e['type'] == 'high_tide'], 
                              key=lambda x: x['level'], reverse=True)[:2]
            low_tides = sorted([e for e in extremes if e['type'] == 'low_tide'], 
                             key=lambda x: x['level'])[:2]
            extremes = sorted(high_tides + low_tides, key=lambda x: x['time'])
        
        meta = get_station_meta(station_id)
        meta["date"] = target_date.strftime('%Y-%m-%d')
        meta["data_source"] = data_source
        
        return create_api_response(
            success=True,
            data={
                "extremes": extremes,
                "summary": {
                    "high_tide_count": len([e for e in extremes if e['type'] == 'high_tide']),
                    "low_tide_count": len([e for e in extremes if e['type'] == 'low_tide']),
                    "max_level": max([e['level'] for e in extremes]) if extremes else None,
                    "min_level": min([e['level'] for e in extremes]) if extremes else None
                }
            },
            meta=meta
        )
        
    except Exception as e:
        return create_api_response(False, error=str(e))

# 4. 위험 수위 알림
def api_check_tide_alert(
    station_id: str,
    hours_ahead: int = 24,
    warning_level: float = 700.0,
    danger_level: float = 750.0
) -> Dict:
    """
    위험 수위 체크 및 알림
    
    Args:
        station_id: 관측소 ID
        hours_ahead: 확인할 시간 (기본 24시간)
        warning_level: 주의 수위 (cm)
        danger_level: 경고 수위 (cm)
    
    Returns:
        위험 수위 정보
    """
    supabase = get_supabase_client()
    if not supabase:
        return create_api_response(False, error="Database connection failed")
    
    try:
        now = datetime.now(pytz.timezone('Asia/Seoul'))
        end_time = now + timedelta(hours=hours_ahead)
        
        # 위험 수위 데이터 조회
        result = supabase.table('tide_predictions')\
            .select('predicted_at, final_tide_level')\
            .eq('station_id', station_id)\
            .gte('predicted_at', now.strftime('%Y-%m-%dT%H:%M:%S'))\
            .lte('predicted_at', end_time.strftime('%Y-%m-%dT%H:%M:%S'))\
            .gte('final_tide_level', warning_level)\
            .order('predicted_at')\
            .execute()
        
        alerts = []
        alert_level = "safe"
        
        if result.data:
            for item in result.data:
                level = item['final_tide_level']
                
                if level >= danger_level:
                    severity = "danger"
                    alert_level = "danger"
                elif level >= warning_level:
                    severity = "warning"
                    if alert_level != "danger":
                        alert_level = "warning"
                else:
                    continue
                
                # UTC를 KST로 변환
                time_utc = datetime.fromisoformat(item['predicted_at'].replace('Z', '+00:00'))
                time_kst = time_utc.astimezone(pytz.timezone('Asia/Seoul'))
                
                alerts.append({
                    "time": time_kst.isoformat(),  # KST ISO format
                    "time_kst": time_kst.strftime('%Y-%m-%d %H:%M:%S KST'),
                    "level": round(level, 2),
                    "severity": severity
                })
        
        # 첫 위험 시간 계산
        first_alert_time = None
        first_alert_time_kst = None
        if alerts:
            first_alert_time = alerts[0]['time']  # 이미 KST
            first_alert_time_kst = alerts[0]['time_kst']
            time_until = (datetime.fromisoformat(first_alert_time) - now).total_seconds() / 3600
        else:
            time_until = None
        
        meta = get_station_meta(station_id)
        meta["check_time"] = now.isoformat()
        meta["hours_ahead"] = hours_ahead
        
        return create_api_response(
            success=True,
            data={
                "alert_level": alert_level,
                "alert_count": len(alerts),
                "first_alert_time": first_alert_time,
                "hours_until_first": round(time_until, 1) if time_until else None,
                "alerts": alerts[:10],  # 최대 10개만
                "thresholds": {
                    "warning": warning_level,
                    "danger": danger_level
                }
            },
            meta=meta
        )
        
    except Exception as e:
        return create_api_response(False, error=str(e))

# 5. 다중 관측소 비교
def api_compare_stations(
    station_ids: List[str],
    target_time: Optional[str] = None
) -> Dict:
    """
    여러 관측소 동시 비교
    
    Args:
        station_ids: 관측소 ID 리스트
        target_time: 비교 시간 (None이면 현재)
    
    Returns:
        관측소별 조위 비교 정보
    """
    if not station_ids:
        return create_api_response(False, error="No station IDs provided")
    
    try:
        comparison_data = []
        
        for station_id in station_ids[:10]:  # 최대 10개 관측소
            result = api_get_tide_level(station_id, target_time)
            
            if result.get("success") and result.get("data"):
                data = result["data"]
                comparison_data.append({
                    "station_id": station_id,
                    "station_name": STATION_NAMES.get(station_id, "Unknown"),
                    "tide_level": data.get("final_value"),
                    "data_source": data.get("data_source"),
                    "time": data.get("record_time")
                })
            else:
                comparison_data.append({
                    "station_id": station_id,
                    "station_name": STATION_NAMES.get(station_id, "Unknown"),
                    "tide_level": None,
                    "data_source": "no_data",
                    "time": None
                })
        
        # 수위 기준 정렬
        comparison_data.sort(key=lambda x: x['tide_level'] if x['tide_level'] else 0, reverse=True)
        
        # 통계 계산
        valid_levels = [d['tide_level'] for d in comparison_data if d['tide_level']]
        
        stats = {
            "max_level": max(valid_levels) if valid_levels else None,
            "min_level": min(valid_levels) if valid_levels else None,
            "avg_level": round(sum(valid_levels) / len(valid_levels), 1) if valid_levels else None,
            "station_count": len(comparison_data),
            "valid_count": len(valid_levels)
        }
        
        return create_api_response(
            success=True,
            data={
                "comparison": comparison_data,
                "statistics": stats
            },
            meta={
                "query_time": target_time or datetime.now(pytz.timezone('Asia/Seoul')).isoformat(),
                "station_count": len(station_ids)
            }
        )
        
    except Exception as e:
        return create_api_response(False, error=str(e))

# 6. 건강 체크 / 상태 확인
def api_health_check() -> Dict:
    """
    API 및 데이터베이스 상태 확인
    
    Returns:
        시스템 상태 정보
    """
    try:
        supabase = get_supabase_client()
        db_status = "connected" if supabase else "disconnected"
        
        # 데이터 가용성 체크
        data_availability = {}
        
        if supabase:
            # 최종 예측 데이터 확인
            result = supabase.table('tide_predictions')\
                .select('station_id', count='exact')\
                .limit(1)\
                .execute()
            
            tide_count = result.count if hasattr(result, 'count') else 0
            
            # 조화 예측 데이터 확인
            result = supabase.table('harmonic_predictions')\
                .select('station_id', count='exact')\
                .limit(1)\
                .execute()
            
            harmonic_count = result.count if hasattr(result, 'count') else 0
            
            data_availability = {
                "tide_predictions": tide_count,
                "harmonic_predictions": harmonic_count
            }
        
        return create_api_response(
            success=True,
            data={
                "status": "healthy" if db_status == "connected" else "degraded",
                "database": db_status,
                "data_availability": data_availability,
                "api_version": "1.0.0",
                "endpoints": [
                    "/api/tide_level",
                    "/api/tide_series",
                    "/api/extremes",
                    "/api/alert",
                    "/api/compare",
                    "/api/health"
                ]
            }
        )
        
    except Exception as e:
        return create_api_response(
            success=False,
            error=str(e),
            data={"status": "error"}
        )