databases.py

import os
import sqlite3
from contextlib import contextmanager
from datetime import datetime
from pathlib import Path
from statistics import median
from typing import Optional, Dict, Any, List, Tuple

# Reutilizamos la misma lógica que antes, pero centralizada en este módulo

DEFAULT_DB_PATH = None  # set by set_db_path at runtime

# Flag global per decidir si es fa servir blockchain (AWS QLDB) per als esdeveniments
USE_BLOCKCHAIN_FOR_EVENTS = False

# Ruta a la base de dades de feedback agregat (separa de users.db)
FEEDBACK_DB_PATH = Path(__file__).resolve().parent / "temp" / "feedback.db"

# Ruta a la base de dades de captions per als scores
CAPTIONS_DB_PATH = Path(__file__).resolve().parent / "temp" / "captions.db"

# Ruta a la base de dades d'esdeveniments (events.db) a demo/temp
EVENTS_DB_PATH = Path(__file__).resolve().parent / "temp" / "events.db"

# Ruta a la base de dades de vídeos (videos.db) a demo/temp
VIDEOS_DB_PATH = Path(__file__).resolve().parent / "temp" / "videos.db"


def set_db_path(db_path: str):
    global DEFAULT_DB_PATH
    DEFAULT_DB_PATH = db_path
    os.makedirs(os.path.dirname(db_path), exist_ok=True)


def set_blockchain_enabled(enabled: bool) -> None:
    """Activa o desactiva l'ús de blockchain per registrar esdeveniments.

    Quan està desactivat (per defecte), els esdeveniments es registren a
    demo/temp/events.db. Quan està activat, s'envien a aws_qldb.
    """

    global USE_BLOCKCHAIN_FOR_EVENTS
    USE_BLOCKCHAIN_FOR_EVENTS = bool(enabled)


def get_connection():
    if not DEFAULT_DB_PATH:
        raise ValueError("Database path not set. Call set_db_path(path) first.")
    return sqlite3.connect(DEFAULT_DB_PATH)


@contextmanager
def get_conn(db_path: Optional[str] = None):
    path = db_path or DEFAULT_DB_PATH
    conn = sqlite3.connect(path, check_same_thread=False)
    conn.row_factory = sqlite3.Row
    try:
        yield conn
        conn.commit()
    finally:
        conn.close()


def init_schema():
    with get_conn() as conn:
        c = conn.cursor()
        # (tus tablas existentes)
        c.execute(
            """
        CREATE TABLE IF NOT EXISTS users (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
            username TEXT UNIQUE NOT NULL,
            password_hash TEXT,
            role TEXT NOT NULL,
            created_at TEXT NOT NULL
        );
        """
        )
        # Migraciones: asegurar columnas esperadas
        try:
            c.execute("PRAGMA table_info(users)")
            cols = {row[1] for row in c.fetchall()}  # set de nombres de columnas
            if "password_hash" not in cols:
                c.execute("ALTER TABLE users ADD COLUMN password_hash TEXT")
            if "role" not in cols:
                c.execute("ALTER TABLE users ADD COLUMN role TEXT NOT NULL DEFAULT 'verd'")
            if "created_at" not in cols:
                c.execute("ALTER TABLE users ADD COLUMN created_at TEXT NOT NULL DEFAULT ''")
        except sqlite3.OperationalError:
            pass
        # Intento de limpieza de columna antigua si existiera (SQLite no permite DROP COLUMN en versiones antiguas)
        try:
            c.execute("ALTER TABLE users DROP COLUMN pw_hash;")
        except sqlite3.OperationalError:
            pass
        # (opcional: tus otras tablas)

        # Esquema per a demo/temp/events.db (registre d'esdeveniments)
        EVENTS_DB_PATH.parent.mkdir(parents=True, exist_ok=True)
        with sqlite3.connect(str(EVENTS_DB_PATH)) as econn:
            ec = econn.cursor()
            ec.execute(
                """
            CREATE TABLE IF NOT EXISTS events (
                id INTEGER PRIMARY KEY AUTOINCREMENT,
                timestamp TEXT NOT NULL,
                session TEXT,
                ip TEXT,
                user TEXT,
                password TEXT,
                phone TEXT,
                action TEXT NOT NULL,
                sha1sum TEXT,
                visibility TEXT
            );
            """
            )
            # Afegir columna visibility si la taula ja existia sense aquest camp
            try:
                ec.execute("ALTER TABLE events ADD COLUMN visibility TEXT")
            except sqlite3.OperationalError:
                # La columna ja existeix
                pass
            econn.commit()

        # >>> TABLA PARA FEEDBACK DE AD (no depende de videos)
        c.execute(
            """
        CREATE TABLE IF NOT EXISTS feedback_ad (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
            video_name TEXT NOT NULL,          -- nombre de carpeta dentro de videos/completed
            user_id INTEGER NOT NULL REFERENCES users(id) ON DELETE CASCADE,
            transcripcio INTEGER NOT NULL,     -- 1..10
            identificacio INTEGER NOT NULL,    -- 1..10
            localitzacions INTEGER NOT NULL,   -- 1..10
            activitats INTEGER NOT NULL,       -- 1..10
            narracions INTEGER NOT NULL,       -- 1..10
            expressivitat INTEGER NOT NULL,    -- 1..10
            comments TEXT,
            created_at TEXT NOT NULL
        );
        """
        )
        # Add column if it doesn't exist, for backwards compatibility
        try:
            c.execute(
                "ALTER TABLE feedback_ad ADD COLUMN expressivitat INTEGER NOT NULL DEFAULT 7;"
            )
        except sqlite3.OperationalError:
            pass  # column already exists


def add_feedback_ad(
    video_name: str,
    user_id: int,
    transcripcio: int,
    identificacio: int,
    localitzacions: int,
    activitats: int,
    narracions: int,
    expressivitat: int,
    comments: str | None,
):
    with get_conn() as conn:
        conn.execute(
            """INSERT INTO feedback_ad
               (video_name, user_id, transcripcio, identificacio, localitzacions, activitats, narracions, expressivitat, comments, created_at)
               VALUES (?,?,?,?,?,?,?,?,?,?)""",
            (
                video_name,
                user_id,
                transcripcio,
                identificacio,
                localitzacions,
                activitats,
                narracions,
                expressivitat,
                comments,
                now_str(),
            ),
        )


def get_feedback_ad_for_video(video_name: str):
    with get_conn() as conn:
        cur = conn.execute(
            """SELECT * FROM feedback_ad WHERE video_name=? ORDER BY created_at DESC""",
            (video_name,),
        )
        return cur.fetchall()


def get_accessible_videos_for_session(session_id: str | None) -> List[str]:
    """Retorna els noms de vídeo accessibles per a una sessió.

    Regles:
      - Sempre inclou vídeos amb visibility='public' a videos.db.
      - Afegeix vídeos per als quals el camp owner coincideix amb algun phone
        registrat a events.db per a la mateixa session.

    Args:
        session_id: Identificador de sessió (st.session_state.session_id).
    """

    # 1) Vídeos públics
    public_videos: set[str] = set()
    with _connect_videos_db() as vconn:
        try:
            for row in vconn.execute(
                "SELECT DISTINCT video_name FROM videos WHERE visibility = 'public'"
            ):
                public_videos.add(row["video_name"])
        except sqlite3.OperationalError:
            # Si la taula no existeix encara, no hi ha vídeos
            return []

    if not session_id:
        return sorted(public_videos)

    # 2) Telèfons associats a la sessió actual
    phones: set[str] = set()
    with _connect_events_db() as econn:
        for row in econn.execute(
            "SELECT DISTINCT phone FROM events WHERE session = ? AND phone IS NOT NULL AND phone != ''",
            (session_id,),
        ):
            phones.add(row["phone"])

    if not phones:
        return sorted(public_videos)

    # 3) Vídeos amb owner associat a algun d'aquests telèfons
    owner_videos: set[str] = set()
    with _connect_videos_db() as vconn:
        q_marks = ",".join("?" for _ in phones)
        params: Tuple[Any, ...] = tuple(phones)
        query = (
            f"SELECT DISTINCT video_name FROM videos WHERE owner IN ({q_marks})"
        )
        for row in vconn.execute(query, params):
            owner_videos.add(row["video_name"])

    all_videos = public_videos | owner_videos
    return sorted(all_videos)


def _connect_feedback_db() -> sqlite3.Connection:
    """Connexió directa a demo/data/feedback.db.

    És independent de DEFAULT_DB_PATH perquè aquesta BD és específica de feedback
    agregat importat des de engine.
    """

    FEEDBACK_DB_PATH.parent.mkdir(parents=True, exist_ok=True)
    conn = sqlite3.connect(str(FEEDBACK_DB_PATH))
    conn.row_factory = sqlite3.Row
    return conn


def _connect_captions_db() -> sqlite3.Connection:
    """Connexió a demo/data/captions.db i creació de la taula si cal.

    Estructura:
      - variable TEXT PRIMARY KEY (p.ex. "score_1")
      - caption  TEXT (etiqueta humana)
    """

    CAPTIONS_DB_PATH.parent.mkdir(parents=True, exist_ok=True)
    conn = sqlite3.connect(str(CAPTIONS_DB_PATH))
    cur = conn.cursor()
    cur.execute(
        """
        CREATE TABLE IF NOT EXISTS captions (
            variable TEXT PRIMARY KEY,
            caption  TEXT NOT NULL
        );
        """
    )
    conn.commit()
    return conn


def insert_demo_feedback_row(
    *,
    user: str,
    session: str,
    video_name: str,
    version: str,
    une_ad: str,
    free_ad: str,
    comments: str | None,
    transcripcio: int,
    identificacio: int,
    localitzacions: int,
    activitats: int,
    narracions: int,
    expressivitat: int,
) -> None:
    """Insereix una valoració detallada a demo/data/feedback.db.

    Escala els sliders de 0-7 a 0-100 i desa els textos d'UNE i narració lliure.
    Les columnes de sliders tenen per nom el caption del slider a la UI.
    """

    # Escalat 0-7 -> 0-100
    def scale(v: int) -> int:
        v = max(0, min(7, int(v)))
        return int(round(v * 100.0 / 7.0))

    slider_values = {
        "Precisió Descriptiva": scale(transcripcio),
        "Sincronització Temporal": scale(identificacio),
        "Claredat i Concisió": scale(localitzacions),
        "Inclusió de Diàleg": scale(activitats),
        "Contextualització": scale(narracions),
        "Flux i Ritme de la Narració": scale(expressivitat),
    }

    ts = datetime.utcnow().strftime("%Y-%m-%d %H:%M:%S")

    with _connect_feedback_db() as conn:
        conn.execute(
            """
            INSERT INTO feedback (
                timestamp, user, session, video_name, version, une_ad, free_ad, comments,
                score_1, score_2, score_3, score_4, score_5, score_6,
                "Precisió Descriptiva",
                "Sincronització Temporal",
                "Claredat i Concisió",
                "Inclusió de Diàleg",
                "Contextualització",
                "Flux i Ritme de la Narració"
            ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?);
            """,
            (
                ts,
                user,
                session,
                video_name,
                version,
                une_ad,
                free_ad,
                comments or "",
                slider_values["Precisió Descriptiva"],
                slider_values["Sincronització Temporal"],
                slider_values["Claredat i Concisió"],
                slider_values["Inclusió de Diàleg"],
                slider_values["Contextualització"],
                slider_values["Flux i Ritme de la Narració"],
                slider_values["Precisió Descriptiva"],
                slider_values["Sincronització Temporal"],
                slider_values["Claredat i Concisió"],
                slider_values["Inclusió de Diàleg"],
                slider_values["Contextualització"],
                slider_values["Flux i Ritme de la Narració"],
            ),
        )


def _connect_events_db() -> sqlite3.Connection:
    """Connexió directa a demo/temp/events.db.

    Es fa independent de DEFAULT_DB_PATH per mantenir aquesta BD separada
    de users.db, igual que feedback.db.
    """

    EVENTS_DB_PATH.parent.mkdir(parents=True, exist_ok=True)
    conn = sqlite3.connect(str(EVENTS_DB_PATH))
    conn.row_factory = sqlite3.Row
    return conn


def _connect_videos_db() -> sqlite3.Connection:
    """Connexió directa a demo/temp/videos.db.

    Aquesta BD conté metadades dels vídeos (video_name, owner, visibility, sha1sum...).
    """

    VIDEOS_DB_PATH.parent.mkdir(parents=True, exist_ok=True)
    conn = sqlite3.connect(str(VIDEOS_DB_PATH))
    conn.row_factory = sqlite3.Row
    return conn


def log_event(
    *,
    session: str,
    ip: str,
    user: str,
    password: str,
    phone: str,
    action: str,
    sha1sum: str,
    visibility: str | None = None,
    timestamp: Optional[str] = None,
) -> None:
    """Insereix un registre a demo/temp/events.db.

    - timestamp: si no s'especifica, es fa servir UTC "YYYY-MM-DD HH:MM:SS".
    - session, ip, user, password, phone, sha1sum es guarden com a TEXT.
    """

    ts = timestamp or datetime.utcnow().strftime("%Y-%m-%d %H:%M:%S")

    if not USE_BLOCKCHAIN_FOR_EVENTS:
        # Mode per defecte: registrar en demo/data/events.db
        with _connect_events_db() as conn:
            conn.execute(
                """INSERT INTO events
                   (timestamp, session, ip, user, password, phone, action, sha1sum, visibility)
                   VALUES (?,?,?,?,?,?,?,?,?)""",
                (
                    ts,
                    session or "",
                    ip or "",
                    user or "",
                    password or "",
                    phone or "",
                    action,
                    sha1sum or "",
                    visibility or "",
                ),
            )
    else:
        # Mode blockchain: delegar a aws_qldb (simulat fins activació real)
        try:
            from aws_qldb import qldb_manager

            payload = {
                "timestamp": ts,
                "session": session or "",
                "ip": ip or "",
                "user": user or "",
                "password": password or "",
                "phone": phone or "",
                "action": action,
                "sha1sum": sha1sum or "",
                "visibility": visibility or "",
            }

            # Mètode específic per a esdeveniments generics (simulat)
            if hasattr(qldb_manager, "record_event"):
                qldb_manager.record_event(payload)
            else:
                # Fallback: registrar com a log simulat
                print(f"[QLDB EVENTS - SIMULATED] {payload}")
        except Exception as e:
            # No interrompre el flux de l'aplicació per errors de blockchain
            print(f"[QLDB EVENTS ERROR] No s'ha pogut registrar l'esdeveniment: {e}")


def get_feedback_video_stats(agg: str = "mitjana") -> List[Dict[str, Any]]:
    """Retorna estadístiques agregades per vídeo de demo/data/feedback.db.

    Es basa exclusivament en les columnes numèriques score_1..score_6 (0-100).

    agg pot ser:
    - "mitjana": mitjana dels scores per vídeo.
    - "mediana": mediana dels scores per vídeo.
    - "inicial": primer registre (per timestamp) per vídeo.
    - "actual": darrer registre (per timestamp) per vídeo.
    """

    agg = (agg or "mitjana").lower()
    with _connect_feedback_db() as conn:
        cur = conn.execute(
            """
            SELECT
              video_name,
              timestamp,
              score_1,
              score_2,
              score_3,
              score_4,
              score_5,
              score_6
            FROM feedback
            """
        )
        rows = cur.fetchall()

    by_video: Dict[str, List[Dict[str, Any]]] = {}
    for row in rows:
        vn = row["video_name"]
        parsed_scores = [
            row["score_1"],
            row["score_2"],
            row["score_3"],
            row["score_4"],
            row["score_5"],
            row["score_6"],
        ]
        enriched = {
            "video_name": vn,
            "timestamp": row["timestamp"],
            "scores": parsed_scores,
        }
        by_video.setdefault(vn, []).append(enriched)

    def parse_ts(ts: str) -> datetime:
        # Format des d'init_feedback.py: "YYYY-MM-DD HH:MM:SS"
        try:
            return datetime.strptime(ts, "%Y-%m-%d %H:%M:%S")
        except Exception:
            return datetime.min

    result: List[Dict[str, Any]] = []
    for video_name, vrows in by_video.items():
        if not vrows:
            continue

        # Ordenem per timestamp per als modes "inicial" i "actual"
        vrows_sorted = sorted(vrows, key=lambda r: parse_ts(r["timestamp"]))

        def agg_index(idx: int) -> Optional[float]:
            vals = [r["scores"][idx] for r in vrows if r["scores"][idx] is not None]
            if not vals:
                return None
            if agg == "mitjana":
                return float(sum(vals) / len(vals))
            if agg == "mediana":
                return float(median(vals))
            if agg == "inicial":
                first = vrows_sorted[0]["scores"][idx]
                return float(first) if first is not None else None
            if agg == "actual":
                last = vrows_sorted[-1]["scores"][idx]
                return float(last) if last is not None else None
            # fallback a mitjana si el mode no és reconegut
            return float(sum(vals) / len(vals))

        row_out: Dict[str, Any] = {
            "video_name": video_name,
            "n": len(vrows),
        }
        for i in range(6):
            row_out[f"score_{i+1}"] = agg_index(i)

        result.append(row_out)

    # Ordenació per defecte alfabètica pel nom; l'ordre final es decidirà a la UI
    result.sort(key=lambda r: r["video_name"])
    return result


def _init_captions_from_eval() -> None:
    """Inicialitza captions.db agafant etiquetes des d'un eval.csv.

    Per simplicitat, intentem llegir `demo/data/media/parella/MoE/eval.csv`.
    Si no existeix o falla, es deixen etiquetes per defecte.
    """

    base_demo = Path(__file__).resolve().parent
    eval_path = base_demo / "data" / "media" / "parella" / "MoE" / "eval.csv"

    default_labels = [f"score_{i}" for i in range(1, 7)]
    labels = default_labels[:]

    if eval_path.exists():
        try:
            import csv

            with eval_path.open("r", encoding="utf-8") as f:
                reader = csv.DictReader(f)
                tmp: List[str] = []
                for row in reader:
                    if len(tmp) >= 6:
                        break
                    name = (row.get("Caracteristica") or "").strip().strip('"')
                    if name:
                        tmp.append(name)
                if tmp:
                    labels = tmp
                    while len(labels) < 6:
                        labels.append(default_labels[len(labels)])
                    labels = labels[:6]
        except Exception:
            pass

    with _connect_captions_db() as conn:
        cur = conn.cursor()
        cur.execute("DELETE FROM captions")
        for i in range(6):
            cur.execute(
                "INSERT OR REPLACE INTO captions (variable, caption) VALUES (?, ?)",
                (f"score_{i+1}", labels[i]),
            )


def get_feedback_score_labels() -> List[str]:
    """Retorna les etiquetes humanes per a score_1..score_6 des de captions.db.

    Si captions.db és buit, s'intenta inicialitzar-lo a partir d'un eval.csv.
    """

    default_labels = [f"score_{i}" for i in range(1, 7)]

    with _connect_captions_db() as conn:
        cur = conn.cursor()
        cur.execute("SELECT variable, caption FROM captions ORDER BY variable")
        rows = cur.fetchall()

        if not rows:
            # Inicialitzar des d'un eval.csv i tornar-ho a intentar
            _init_captions_from_eval()
            cur.execute("SELECT variable, caption FROM captions ORDER BY variable")
            rows = cur.fetchall()

    if not rows:
        return default_labels

    labels: List[str] = []
    for _, caption in rows:
        labels.append(caption)

    while len(labels) < 6:
        labels.append(default_labels[len(labels)])

    return labels[:6]


def get_feedback_ad_stats():
    # medias por vídeo y ranking
    with get_conn() as conn:
        cur = conn.execute(
            """
            SELECT
              video_name,
              COUNT(*) AS n,
              AVG(transcripcio)   AS avg_transcripcio,
              AVG(identificacio)  AS avg_identificacio,
              AVG(localitzacions) AS avg_localitzacions,
              AVG(activitats)     AS avg_activitats,
              AVG(narracions)     AS avg_narracions,
              AVG(expressivitat)  AS avg_expressivitat,
              (AVG(transcripcio)+AVG(identificacio)+AVG(localitzacions)+AVG(activitats)+AVG(narracions)+AVG(expressivitat))/6.0 AS avg_global
            FROM feedback_ad
            GROUP BY video_name
            ORDER BY avg_global DESC, n DESC;
        """
        )
        return cur.fetchall()


def now_str():
    return datetime.utcnow().isoformat(timespec="seconds") + "Z"


# Users

def create_user(username: str, password_hash: str, role: str):
    with get_conn() as conn:
        conn.execute(
            "INSERT INTO users(username, password_hash, role, created_at) VALUES (?,?,?,?)",
            (username, password_hash, role, now_str()),
        )


def get_user(username: str):
    with get_conn() as conn:
        cur = conn.execute("SELECT * FROM users WHERE username=?", (username,))
        return cur.fetchone()


def get_all_users() -> List[Dict[str, Any]]:
    with get_conn() as conn:
        cur = conn.execute("SELECT id, username, role FROM users ORDER BY username")
        return cur.fetchall()


def update_user_password(username: str, password_hash: str):
    with get_conn() as conn:
        conn.execute(
            "UPDATE users SET password_hash = ? WHERE username = ?",
            (password_hash, username),
        )