GarminHealthLab/backend/services/health.py

"""
Health data service: read endpoints + upsert helpers used by the Garmin sync.

Mirrors the original Node HealthService, including the camelCase JSON mapping.
Upserts use backend-specific SQL because SQLite does not support
`ON DUPLICATE KEY UPDATE` (it uses `ON CONFLICT ... DO UPDATE`).
"""
import datetime
import uuid

from db import execute, query_one, query_all
from config import DB_TYPE


def _range_sql(user_id, start=None, end=None):
    params = [user_id]
    sql = "WHERE user_id = ?"
    if start:
        sql += " AND date >= ?"
        params.append(start)
    if end:
        sql += " AND date <= ?"
        params.append(end)
    return sql, params


def get_summary(user_id, start=None, end=None):
    """Every stored metric for each day, in the camelCase the UI and the AI
    prompt consume."""
    sql, params = _range_sql(user_id, start, end)
    columns = ", ".join(HEALTH_COLUMNS)
    rows = query_all(
        f"SELECT date, {columns} FROM health_data {sql} ORDER BY date ASC", params
    )

    out = []
    for r in rows:
        day = {"date": r["date"]}
        for column, key in HEALTH_COLUMNS.items():
            value = r.get(column)
            # Null metrics are omitted rather than sent as null. A year of 40
            # metrics is 394 KB over the tunnel, and most of it is nulls for
            # sensors this watch does not have; the client already treats a
            # missing key and a null the same way.
            if value is not None:
                day[key] = value
        # Sleep stays nested for backwards compatibility with the UI and the
        # existing recommendation rules.
        day["sleep"] = (
            {
                "duration": r.get("sleep_duration"),
                "quality": r.get("sleep_quality"),
                "deepSeconds": r.get("sleep_deep_seconds"),
                "lightSeconds": r.get("sleep_light_seconds"),
                "remSeconds": r.get("sleep_rem_seconds"),
                "awakeSeconds": r.get("sleep_awake_seconds"),
            }
            if r.get("sleep_duration") is not None
            else None
        )
        out.append(day)
    return out


def get_steps(user_id, start=None, end=None):
    sql, params = _range_sql(user_id, start, end)
    rows = query_all(
        f"SELECT date, steps FROM health_data {sql} AND steps IS NOT NULL ORDER BY date ASC",
        params,
    )
    return [{"date": r["date"], "steps": r["steps"]} for r in rows]


def get_heart_rate(user_id, start=None, end=None):
    sql, params = _range_sql(user_id, start, end)
    rows = query_all(
        f"SELECT date, heart_rate, heart_rate_variability FROM health_data {sql} "
        "AND heart_rate IS NOT NULL ORDER BY date ASC",
        params,
    )
    return [
        {
            "date": r["date"],
            "heartRate": r["heart_rate"],
            "heartRateVariability": r["heart_rate_variability"],
        }
        for r in rows
    ]


def get_sleep(user_id, start=None, end=None):
    sql, params = _range_sql(user_id, start, end)
    rows = query_all(
        f"SELECT date, sleep_duration, sleep_quality FROM health_data {sql} "
        "AND sleep_duration IS NOT NULL ORDER BY date ASC",
        params,
    )
    return [
        {"date": r["date"], "duration": r["sleep_duration"], "quality": r["sleep_quality"]}
        for r in rows
    ]


def get_activities(user_id, start=None, end=None):
    """Activities in a date range.

    Filters on start_time, not `date`: the activities table has no `date`
    column, so reusing the daily-metrics range clause raised
    "Unknown column 'date'". It went unnoticed while the only caller asked for
    every activity, which produced no date predicate at all.
    """
    params = [user_id]
    sql = "WHERE user_id = ?"
    if start:
        sql += " AND start_time >= ?"
        params.append(start)
    if end:
        # Exclusive upper bound at the next midnight rather than "end
        # 23:59:59": SQLite compares these as strings, and the stored
        # separator may be 'T' (0x54) or a space (0x20), so a same-day
        # 18:30 timestamp sorts *after* an end bound written with a space
        # and would be dropped from its own day.
        sql += " AND start_time < ?"
        params.append(
            (datetime.date.fromisoformat(end) + datetime.timedelta(days=1)).isoformat()
        )

    return query_all(
        "SELECT id, activity_type, start_time, end_time, duration, distance, "
        "calories, heart_rate_average, heart_rate_max "
        f"FROM activities {sql} ORDER BY start_time DESC",
        params,
    )


# Column name -> key in the record dict produced by the Garmin extractor.
# Keeping the mapping in one place means adding a metric touches this table
# and the extractor, and nothing else.
HEALTH_COLUMNS = {
    "steps": "steps",
    "step_goal": "stepGoal",
    "distance_meters": "distanceMeters",
    "calories_burned": "caloriesBurned",
    "active_calories": "activeCalories",
    "bmr_calories": "bmrCalories",
    "floors_ascended": "floorsAscended",
    "floors_descended": "floorsDescended",
    "intensity_minutes": "intensityMinutes",
    "sedentary_seconds": "sedentarySeconds",
    "active_seconds": "activeSeconds",
    "heart_rate": "heartRate",
    "heart_rate_max": "heartRateMax",
    "heart_rate_min": "heartRateMin",
    "heart_rate_variability": "heartRateVariability",
    "stress": "stress",
    "stress_max": "stressMax",
    "body_battery_high": "bodyBatteryHigh",
    "body_battery_low": "bodyBatteryLow",
    "body_battery_charged": "bodyBatteryCharged",
    "body_battery_drained": "bodyBatteryDrained",
    "spo2_avg": "spo2Avg",
    "spo2_min": "spo2Min",
    "respiration_avg": "respirationAvg",
    "respiration_min": "respirationMin",
    "respiration_max": "respirationMax",
    "sleep_duration": "sleepDuration",
    "sleep_quality": "sleepQuality",
    "sleep_deep_seconds": "sleepDeepSeconds",
    "sleep_light_seconds": "sleepLightSeconds",
    "sleep_rem_seconds": "sleepRemSeconds",
    "sleep_awake_seconds": "sleepAwakeSeconds",
    "sleep_spo2_avg": "sleepSpo2Avg",
    "sleep_respiration_avg": "sleepRespirationAvg",
    "sleep_stress_avg": "sleepStressAvg",
    "training_readiness": "trainingReadiness",
    "vo2max": "vo2max",
    "endurance_score": "enduranceScore",
    "hill_score": "hillScore",
    "hydration_ml": "hydrationMl",
    "hydration_goal_ml": "hydrationGoalMl",
    "sweat_loss_ml": "sweatLossMl",
    "weight_kg": "weightKg",
    "body_fat_pct": "bodyFatPct",
    "bmi": "bmi",
    "blood_pressure_systolic": "bloodPressureSystolic",
    "blood_pressure_diastolic": "bloodPressureDiastolic",
}


def _upsert(table, key_cols, cols, values):
    """INSERT ... ON CONFLICT/DUPLICATE UPDATE, written for both backends."""
    placeholders = ", ".join(["?"] * len(cols))
    updatable = [c for c in cols if c not in key_cols]
    if DB_TYPE == "mariadb":
        updates = ", ".join(f"{c}=VALUES({c})" for c in updatable)
        sql = (f"INSERT INTO {table} ({', '.join(cols)}) VALUES ({placeholders}) "
               f"ON DUPLICATE KEY UPDATE {updates}")
    else:
        conflict = ", ".join(key_cols)
        updates = ", ".join(f"{c}=excluded.{c}" for c in updatable)
        sql = (f"INSERT INTO {table} ({', '.join(cols)}) VALUES ({placeholders}) "
               f"ON CONFLICT({conflict}) DO UPDATE SET {updates}")
    execute(sql, values)


def upsert_health_daily(user_id, record):
    hid = f"{user_id}-{record['date']}"
    cols = ["id", "user_id", "date"] + list(HEALTH_COLUMNS)
    values = [hid, user_id, record.get("date")] + [
        record.get(key) for key in HEALTH_COLUMNS.values()
    ]
    _upsert("health_data", ("user_id", "date"), cols, values)
    return hid


def upsert_badge(user_id, badge):
    cols = ["id", "user_id", "badge_key", "name", "category_id",
            "difficulty_id", "earned_date", "earned_count", "points"]
    values = [
        badge["id"], user_id, badge.get("badgeKey"), badge.get("name"),
        badge.get("categoryId"), badge.get("difficultyId"),
        badge.get("earnedDate"), badge.get("earnedCount"), badge.get("points"),
    ]
    _upsert("badges", ("user_id", "id"), cols, values)
    return badge["id"]


def upsert_personal_record(user_id, record):
    cols = ["id", "user_id", "type_id", "activity_id", "activity_name",
            "activity_type", "value", "achieved_at"]
    values = [
        record["id"], user_id, record.get("typeId"), record.get("activityId"),
        record.get("activityName"), record.get("activityType"),
        record.get("value"), record.get("achievedAt"),
    ]
    _upsert("personal_records", ("user_id", "id"), cols, values)
    return record["id"]


def get_badges(user_id):
    return query_all(
        "SELECT id, badge_key, name, category_id, difficulty_id, earned_date, "
        "earned_count, points FROM badges WHERE user_id = ? "
        "ORDER BY earned_date DESC",
        [user_id],
    )


def get_personal_records(user_id):
    return query_all(
        "SELECT id, type_id, activity_id, activity_name, activity_type, value, "
        "achieved_at FROM personal_records WHERE user_id = ? "
        "ORDER BY achieved_at DESC",
        [user_id],
    )


def insert_activity(user_id, activity):
    # Prefer Garmin's own activity id when the caller has one: it is stable
    # across syncs, which is what lets a re-synced window skip what is already
    # stored instead of inserting it again.
    aid = str(activity.get("id") or uuid.uuid4())
    cols = [
        "id", "user_id", "activity_type", "start_time", "end_time",
        "duration", "distance", "calories", "heart_rate_average", "heart_rate_max",
    ]
    placeholders = ", ".join(["?"] * len(cols))
    vals = [
        aid, user_id, activity.get("activityType"), activity.get("startTime"),
        activity.get("endTime"), activity.get("duration"), activity.get("distance"),
        activity.get("calories"), activity.get("heartRateAverage"),
        activity.get("heartRateMax"),
    ]
    execute(
        f"INSERT INTO activities ({', '.join(cols)}) VALUES ({placeholders})",
        vals,
    )
    return aid