mitai-jinkendo/backend/scripts/inspect_activity_eav.py

"""
Diagnose: Was liegt in activity_session_metrics (EAV) vs. activity_log?

Ausführung (mit gesetzten DB_*-Variablen wie die App, z. B. aus .env):

  cd backend
  python scripts/inspect_activity_eav.py

Lokal ohne Docker-Hostname: z. B. ``set DB_HOST=127.0.0.1`` (Windows) / ``export DB_HOST=127.0.0.1``,
Port/User/Pass wie in der laufenden Postgres-Instanz.

Im Backend-Container (Compose-Service meist ``backend``, Arbeitsverzeichnis ``/app``):

  docker compose exec backend python /app/scripts/inspect_activity_eav.py

Optional:
  python scripts/inspect_activity_eav.py --limit 30
  python scripts/inspect_activity_eav.py --profile <uuid>
  python scripts/inspect_activity_eav.py --activity <activity_log uuid>

Keine Schreibzugriffe — nur SELECT.
"""
from __future__ import annotations

import argparse
import os
import sys

# backend/ als Import-Root
_BACKEND_ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
if _BACKEND_ROOT not in sys.path:
    sys.path.insert(0, _BACKEND_ROOT)


def _val_row(r: dict) -> str | None:
    dt = r.get("data_type")
    if dt == "integer":
        v = r.get("value_int")
        return str(v) if v is not None else None
    if dt == "float":
        v = r.get("value_num")
        return str(v) if v is not None else None
    if dt == "string":
        v = r.get("value_text")
        return repr(v) if v is not None else None
    if dt == "boolean":
        v = r.get("value_bool")
        return str(v) if v is not None else None
    return None


def main() -> None:
    parser = argparse.ArgumentParser(description="EAV activity_session_metrics inspizieren")
    parser.add_argument("--limit", type=int, default=40, help="Zeilen Report A/B")
    parser.add_argument("--profile", type=str, default=None, help="profile_id filtern")
    parser.add_argument("--activity", type=str, default=None, help="activity_log.id (einzelne Session)")
    args = parser.parse_args()

    from db import get_db, get_cursor

    if args.activity:
        with get_db() as conn:
            with get_cursor(conn) as cur:
                cur.execute(
                    """
                    SELECT al.id, al.profile_id, al.date, al.start_time, al.source,
                           al.training_category, al.training_type_id, al.activity_type,
                           al.duration_min, al.kcal_active, al.hr_avg, al.hr_max, al.distance_km
                    FROM activity_log al
                    WHERE al.id = %s::uuid
                    """,
                    (args.activity,),
                )
                h = cur.fetchone()
                if not h:
                    print("activity_log: keine Zeile für diese id")
                    return
                print("=== activity_log (Kopfzeile) ===")
                for k, v in dict(h).items():
                    print(f"  {k}: {v}")

                cur.execute(
                    """
                    SELECT m.id AS metric_id, tp.key, tp.data_type, tp.source_field,
                           m.value_num, m.value_int, m.value_text, m.value_bool, m.updated_at
                    FROM activity_session_metrics m
                    JOIN training_parameters tp ON tp.id = m.training_parameter_id
                    WHERE m.activity_log_id = %s::uuid
                    ORDER BY tp.key
                    """,
                    (args.activity,),
                )
                rows = cur.fetchall()
                print(f"\n=== activity_session_metrics ({len(rows)} Zeilen) ===")
                for r in rows:
                    d = dict(r)
                    print(
                        f"  {d['key']} ({d['data_type']}) "
                        f"value={_val_row(d)!r} source_field={d.get('source_field')!r} "
                        f"updated_at={d.get('updated_at')}"
                    )
                if not rows:
                    print("  (keine EAV-Zeilen)")
        return

    prof_filter = ""
    if args.profile:
        prof_filter = " AND al.profile_id = %s::uuid "
    params_a: tuple = (args.profile, args.limit) if args.profile else (args.limit,)
    params_b: tuple = (args.profile, args.limit) if args.profile else (args.limit,)

    q_recent_eav = f"""
        SELECT
            al.id AS activity_id,
            al.profile_id,
            al.date,
            al.start_time,
            al.source,
            al.training_type_id,
            al.training_category,
            tp.key AS parameter_key,
            tp.data_type,
            tp.source_field AS tp_source_field,
            m.value_num,
            m.value_int,
            m.value_text,
            m.value_bool,
            m.updated_at
        FROM activity_session_metrics m
        JOIN activity_log al ON al.id = m.activity_log_id
        JOIN training_parameters tp ON tp.id = m.training_parameter_id
        WHERE 1=1 {prof_filter}
        ORDER BY m.updated_at DESC NULLS LAST, al.date DESC, al.start_time DESC
        LIMIT %s
    """

    q_csv_no_eav = f"""
        SELECT
            al.id AS activity_id,
            al.profile_id,
            al.date,
            al.start_time,
            al.source,
            al.training_type_id,
            al.training_category,
            (SELECT COUNT(*) FROM activity_session_metrics m WHERE m.activity_log_id = al.id) AS eav_count
        FROM activity_log al
        WHERE al.source = 'csv' {prof_filter}
        ORDER BY al.date DESC, al.start_time DESC
        LIMIT %s
    """

    with get_db() as conn:
        with get_cursor(conn) as cur:
            print("=== A) Neueste EAV-Zeilen (join activity_log + training_parameters) ===\n")
            cur.execute(q_recent_eav, params_a)
            for r in cur.fetchall():
                d = dict(r)
                v = _val_row(d)
                print(
                    f"{d['date']} {d['start_time']} | {d['activity_id']} | src={d['source']!r} | "
                    f"type={d['training_type_id']} cat={d['training_category']!r} | "
                    f"{d['parameter_key']}={v!r} (tp.source_field={d.get('tp_source_field')!r})"
                )

            print("\n=== B) Neueste CSV-importierte Sessions: EAV-Anzahl pro Zeile ===\n")
            cur.execute(q_csv_no_eav, params_b)
            for r in cur.fetchall():
                d = dict(r)
                print(
                    f"{d['date']} {d['start_time']} | {d['activity_id']} | "
                    f"type={d['training_type_id']} | eav_count={d['eav_count']}"
                )

    print("\nFertig. Für eine Session im Detail: --activity <uuid>")


if __name__ == "__main__":
    main()