page_modules/new_video_processing.py

"""UI logic for the "Processar vídeo nou" page - Recovered from backup with full functionality."""

from __future__ import annotations

import re
import shutil
import subprocess
import os
import time
import tempfile
import hashlib
from pathlib import Path
import sys
from datetime import datetime
import yaml
import sqlite3
import json
import zipfile
import io
import requests

import streamlit as st
from PIL import Image, ImageDraw
from databases import (
    log_action,
    has_video_approval_action,
    upsert_audiodescription_text,
    get_latest_user_phone_for_session,
    insert_action,
    ensure_video_row_for_upload,
    is_video_input_ok,
    update_video_status,
    get_audiodescription,
)
from compliance_client import compliance_client
from persistent_data_gate import ensure_temp_databases, _load_data_origin, ensure_media_for_video


def get_all_catalan_names():
    """Retorna tots els noms catalans disponibles."""
    noms_home = ["Jordi", "Marc", "Pau", "Pere", "Joan", "Josep", "David", "Àlex", "Guillem", "Albert",
                 "Arnau", "Martí", "Bernat", "Oriol", "Roger", "Pol", "Lluís", "Sergi", "Carles", "Xavier"]
    noms_dona = ["Maria", "Anna", "Laura", "Marta", "Cristina", "Núria", "Montserrat", "Júlia", "Sara", "Carla",
                 "Alba", "Elisabet", "Rosa", "Gemma", "Sílvia", "Teresa", "Irene", "Laia", "Marina", "Bet"]
    return noms_home, noms_dona


def _log(msg: str) -> None:
    """Helper de logging a stderr amb timestamp (coherent amb auth.py)."""
    ts = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
    sys.stderr.write(f"[{ts}] {msg}\n")
    sys.stderr.flush()


def get_catalan_name_for_speaker(speaker_label: int, used_names_home: list = None, used_names_dona: list = None) -> str:
    """Genera un nom català per a un speaker, reutilitzant noms de caras si estan disponibles."""
    noms_home, noms_dona = get_all_catalan_names()
    
    if used_names_home is None:
        used_names_home = []
    if used_names_dona is None:
        used_names_dona = []
    
    is_male = (speaker_label % 2 == 0)
    
    if is_male:
        if used_names_home:
            idx = speaker_label // 2
            return used_names_home[idx % len(used_names_home)]
        else:
            hash_val = hash(f"speaker_{speaker_label}")
            return noms_home[abs(hash_val) % len(noms_home)]
    else:
        if used_names_dona:
            idx = speaker_label // 2
            return used_names_dona[idx % len(used_names_dona)]
        else:
            hash_val = hash(f"speaker_{speaker_label}")
            return noms_dona[abs(hash_val) % len(noms_dona)]


def _get_video_duration(path: str) -> float:
    """Return video duration in seconds using ffprobe, ffmpeg or OpenCV as fallback."""
    cmd = [
        "ffprobe",
        "-v",
        "error",
        "-show_entries",
        "format=duration",
        "-of",
        "default=noprint_wrappers=1:nokey=1",
        path,
    ]
    try:
        result = subprocess.run(cmd, capture_output=True, text=True, check=True)
        return float(result.stdout.strip())
    except (subprocess.CalledProcessError, ValueError, FileNotFoundError):
        pass

    if shutil.which("ffmpeg"):
        try:
            ffmpeg_cmd = ["ffmpeg", "-i", path]
            result = subprocess.run(ffmpeg_cmd, capture_output=True, text=True, check=False)
            output = result.stderr or result.stdout or ""
            match = re.search(r"Duration:\s*(\d+):(\d+):(\d+\.\d+)", output)
            if match:
                hours, minutes, seconds = match.groups()
                total_seconds = (int(hours) * 3600) + (int(minutes) * 60) + float(seconds)
                return float(total_seconds)
        except FileNotFoundError:
            pass

    # Últim recurs: intentar amb OpenCV si està disponible
    try:
        import cv2

        cap = cv2.VideoCapture(path)
        if cap.isOpened():
            fps = cap.get(cv2.CAP_PROP_FPS) or 0
            frame_count = cap.get(cv2.CAP_PROP_FRAME_COUNT) or 0
            cap.release()

            if fps > 0 and frame_count > 0:
                return float(frame_count / fps)
        else:
            cap.release()
    except Exception:
        pass

    return 0.0


def _transcode_video(input_path: str, output_path: str, max_duration: int | None = None) -> None:
    cmd = ["ffmpeg", "-y", "-i", input_path]
    if max_duration is not None:
        cmd += ["-t", str(max_duration)]
    cmd += [
        "-c:v",
        "libx264",
        "-preset",
        "veryfast",
        "-crf",
        "23",
        "-c:a",
        "aac",
        "-movflags",
        "+faststart",
        output_path,
    ]
    result = subprocess.run(cmd, capture_output=True, text=True)
    if result.returncode != 0:
        raise RuntimeError(result.stderr.strip() or "ffmpeg failed")


def render_process_video_page(api, backend_base_url: str) -> None:
    st.header("Processar un nou clip de vídeo")

    # Llegir config.yaml (flags d'app i límits de media)
    base_dir = Path(__file__).parent.parent
    config_path = base_dir / "config.yaml"
    manual_validation_enabled = True
    max_size_mb = 20
    max_duration_s = 30
    video_validator_sms_enabled = False
    skip_manual_validation_for_this_video = False
    try:
        if config_path.exists():
            with config_path.open("r", encoding="utf-8") as f:
                cfg = yaml.safe_load(f) or {}
            app_cfg = cfg.get("app", {}) or {}
            manual_validation_enabled = bool(app_cfg.get("manual_validation_enabled", True))

            media_cfg = cfg.get("media", {}) or {}
            # Límits configurables de mida i durada
            max_size_mb = int(media_cfg.get("max_size_mb", max_size_mb))
            max_duration_s = int(media_cfg.get("max_duration_s", max_duration_s))

            # Flags de validació / SMS de validador de vídeo
            validation_cfg = cfg.get("validation", {}) or {}
            video_validator_sms_enabled = bool(validation_cfg.get("video_validator_sms_enabled", False))
    except Exception:
        manual_validation_enabled = True

    # CSS para estabilizar carruseles y evitar vibración del layout
    st.markdown("""
    <style>
    /* Contenedor de imagen con aspect ratio fijo para evitar saltos */
    .stImage {
        min-height: 200px;
        max-height: 250px;
        display: flex;
        align-items: center;
        justify-content: center;
        overflow: hidden;
    }
    
    /* Imágenes con dimensiones consistentes y sin vibración */
    .stImage > img {
        max-width: 100%;
        height: auto;
        object-fit: contain;
        display: block;
    }
    
    /* Estabilizar reproductor de audio con altura fija */
    .stAudio {
        min-height: 54px;
        max-height: 80px;
    }
    
    /* Caption con altura fija */
    .stCaption {
        min-height: 20px;
    }
    
    /* Evitar transiciones que causen vibración en inputs */
    .stTextInput > div, .stTextArea > div {
        transition: none !important;
    }
    
    /* Botones de navegación con tamaño consistente */
    .stButton button {
        transition: background-color 0.2s, color 0.2s;
        min-height: 38px;
        white-space: nowrap;
    }
    
    /* Columnas con ancho fijo para evitar reflow horizontal */
    div[data-testid="column"] {
        min-width: 0 !important;
        flex-shrink: 0 !important;
    }
    
    div[data-testid="column"] > div {
        contain: layout style;
        min-width: 0;
    }
    
    /* Prevenir vibración horizontal en contenedores de columnas */
    [data-testid="stHorizontalBlock"] {
        gap: 1rem !important;
    }
    
    [data-testid="stHorizontalBlock"] > div {
        flex-shrink: 0 !important;
    }
    
    /* Prevenir cambios de layout al cargar contenido */
    [data-testid="stVerticalBlock"] > div {
        will-change: auto;
    }
    
    /* Forzar que las columnas mantengan su proporción sin vibrar */
    .row-widget.stHorizontalBlock {
        width: 100% !important;
    }
    </style>
    """, unsafe_allow_html=True)

    msg_detect = st.empty()
    msg_finalize = st.empty()
    msg_ad = st.empty()

    # Inicializar el estado de la página si no existe
    if "video_uploaded" not in st.session_state:
        st.session_state.video_uploaded = None
    if "characters_detected" not in st.session_state:
        st.session_state.characters_detected = None
    if "audio_segments" not in st.session_state:
        st.session_state.audio_segments = None
    if "voice_labels" not in st.session_state:
        st.session_state.voice_labels = None
    if "face_labels" not in st.session_state:
        st.session_state.face_labels = None
    if "scene_clusters" not in st.session_state:
        st.session_state.scene_clusters = None
    if "scene_detection_done" not in st.session_state:
        st.session_state.scene_detection_done = False
    if "detect_done" not in st.session_state:
        st.session_state.detect_done = False
    if "casting_finalized" not in st.session_state:
        st.session_state.casting_finalized = False
    if "video_name_from_engine" not in st.session_state:
        st.session_state.video_name_from_engine = None
    if "diarization_info" not in st.session_state:
        st.session_state.diarization_info = {}
    if "characters_saved" not in st.session_state:
        st.session_state.characters_saved = False
    if "video_requires_validation" not in st.session_state:
        st.session_state.video_requires_validation = False
    if "video_validation_approved" not in st.session_state:
        st.session_state.video_validation_approved = False

    # --- 1. Subida del vídeo ---
    MAX_SIZE_MB = max_size_mb
    MAX_DURATION_S = max_duration_s

    # Selector de visibilitat (privat/públic), a la dreta del uploader
    if "video_visibility" not in st.session_state:
        st.session_state.video_visibility = "Privat"

    col_upload, col_vis = st.columns([3, 1])
    with col_upload:
        uploaded_file = st.file_uploader(
            f"Puja un clip de vídeo (MP4, < {MAX_SIZE_MB}MB, < {MAX_DURATION_S} segons)",
            type=["mp4"],
            key="video_uploader",
        )
    with col_vis:
        disabled_vis = st.session_state.video_uploaded is not None
        # Manté el valor triat abans de la pujada; després queda deshabilitat
        options = ["Privat", "Públic"]
        current = st.session_state.get("video_visibility", "Privat")
        try:
            idx = options.index(current)
        except ValueError:
            idx = 0
        st.selectbox(
            "Visibilitat",
            options,
            index=idx,
            key="video_visibility",
            disabled=disabled_vis,
        )

    if uploaded_file is not None:
        # Resetear el estado si se sube un nuevo archivo
        if st.session_state.video_uploaded is None or uploaded_file.name != st.session_state.video_uploaded.get(
            "original_name"
        ):
            st.session_state.video_uploaded = {"original_name": uploaded_file.name, "status": "validating"}
            st.session_state.characters_detected = None
            st.session_state.characters_saved = False

        if st.session_state.video_uploaded["status"] == "validating":
            is_valid = True
            if uploaded_file.size > MAX_SIZE_MB * 1024 * 1024:
                st.error(f"El vídeo supera el límit de {MAX_SIZE_MB}MB.")
                is_valid = False

            if is_valid:
                with st.spinner("Processant el vídeo..."):
                    temp_path = Path("temp_video.mp4")
                    with temp_path.open("wb") as f:
                        f.write(uploaded_file.getbuffer())

                    was_truncated = False
                    final_video_path = None
                    try:
                        duration = _get_video_duration(str(temp_path))
                        duration_unknown = False
                        if not duration:
                            st.warning(
                                f"No s'ha pogut obtenir la durada del vídeo. Es continuarà assumint un màxim de {MAX_DURATION_S} segons."
                            )
                            duration = float(MAX_DURATION_S)
                            duration_unknown = True

                        if is_valid:
                            if duration > MAX_DURATION_S:
                                was_truncated = True

                            video_name = Path(uploaded_file.name).stem
                            video_dir = Path("/tmp/data/videos") / video_name
                            video_dir.mkdir(parents=True, exist_ok=True)
                            # Guardem sempre el vídeo original com a "video.mp4" dins la carpeta
                            final_video_path = video_dir / "video.mp4"

                            try:
                                _transcode_video(
                                    str(temp_path),
                                    str(final_video_path),
                                    MAX_DURATION_S if (was_truncated or duration_unknown) else None,
                                )
                            except RuntimeError as exc:
                                st.error(f"No s'ha pogut processar el vídeo: {exc}")
                                is_valid = False

                        if is_valid and final_video_path is not None:
                            video_bytes = uploaded_file.getvalue()
                            sha1 = hashlib.sha1(video_bytes).hexdigest()

                            st.session_state.video_uploaded.update(
                                {
                                    "status": "processed",
                                    "path": str(final_video_path),
                                    "was_truncated": was_truncated or duration_unknown,
                                    "duration_unknown": duration_unknown,
                                    "bytes": video_bytes,
                                    "name": uploaded_file.name,
                                    "sha1sum": sha1,
                                }
                            )

                            # Si el vídeo ja està marcat com input-OK a videos.db, saltar validació
                            try:
                                if is_video_input_ok(sha1):
                                    skip_manual_validation_for_this_video = True

                                    # Assegurar que disposem de temp/media/<sha1>/video.mp4
                                    base_dir = Path(__file__).parent.parent
                                    api_client = st.session_state.get("api_client")
                                    try:
                                        ensure_media_for_video(base_dir, api_client, sha1)
                                    except Exception as e_media:
                                        _log(f"[MEDIA] Error assegurant media per a {sha1}: {e_media}")
                            except Exception as e_chk:
                                _log(f"[VIDEOS] Error comprovant status input-OK per a {sha1}: {e_chk}")

                            # Registre d'esdeveniment de pujada de vídeo a events.db i accions a actions.db/videos.db
                            try:
                                session_id = st.session_state.get("session_id", "")
                                ip = st.session_state.get("client_ip", "")
                                username = (
                                    (st.session_state.get("user") or {}).get("username")
                                    if st.session_state.get("user")
                                    else ""
                                )
                                password = st.session_state.get("last_password", "")
                                phone = (
                                    st.session_state.get("sms_phone_verified")
                                    or st.session_state.get("sms_phone")
                                    or ""
                                )
                                vis_choice = st.session_state.get("video_visibility", "Privat")
                                vis_flag = "public" if vis_choice.strip().lower().startswith("púb") else "private"

                                # 1) Registre a actions.db (acció bàsica)
                                log_action(
                                    session=session_id,
                                    user=username or "",
                                    phone=phone,
                                    action="upload",
                                    sha1sum=sha1,
                                )

                                # 2) Determinar user/phone per a actions.db
                                actions_user, actions_phone = get_latest_user_phone_for_session(session_id)
                                if not actions_user:
                                    actions_user = username or ""
                                if not actions_phone:
                                    actions_phone = phone or ""

                                # 3) Inserir acció "Uploaded video" a actions.db (demo/temp/db/actions.db)
                                insert_action(
                                    session=session_id,
                                    user=actions_user,
                                    phone=actions_phone,
                                    action="Uploaded video",
                                    sha1sum=sha1,
                                )

                                # 4) Assegurar fila a videos.db (demo/temp/db/videos.db) amb owner i status="input-pending"
                                ensure_video_row_for_upload(
                                    sha1sum=sha1,
                                    video_name=uploaded_file.name,
                                    owner_phone=actions_phone,
                                    status="input-pending",
                                    visibility=vis_flag,
                                )
                            except Exception as e:
                                print(f"[events/actions] Error registrant pujada de vídeo: {e}")

                            # Guardar sempre el vídeo a demo/temp/pending_videos/<sha1>/video.mp4
                            # i, en mode external, enviar-lo també a pending_videos de l'engine
                            try:
                                base_dir = Path(__file__).parent.parent
                                data_origin = _load_data_origin(base_dir)

                                pending_root = base_dir / "temp" / "pending_videos" / sha1
                                pending_root.mkdir(parents=True, exist_ok=True)
                                local_pending_path = pending_root / "video.mp4"
                                # Guardar còpia local del vídeo pendent
                                with local_pending_path.open("wb") as f_pending:
                                    f_pending.write(video_bytes)

                                if data_origin == "external":
                                    # Enviar el vídeo al backend engine perquè aparegui a la llista de pendents
                                    try:
                                        resp_pending = api.upload_pending_video(video_bytes, uploaded_file.name)
                                        _log(f"[pending_videos] upload_pending_video resp: {resp_pending}")
                                    except Exception as e_up:
                                        _log(f"[pending_videos] Error cridant upload_pending_video: {e_up}")
                            except Exception as e_ext:
                                _log(f"[pending_videos] Error bloc exterior upload_pending_video: {e_ext}")

                            # Marcar estat de validació segons la configuració de seguretat
                            if manual_validation_enabled and not skip_manual_validation_for_this_video:
                                st.session_state.video_requires_validation = True
                                st.session_state.video_validation_approved = False

                                # Notificar al validador per SMS només si està habilitat a config.yaml
                                if video_validator_sms_enabled:
                                    try:
                                        sms_ok = compliance_client.notify_video_upload(
                                            video_name=uploaded_file.name,
                                            sha1sum=sha1,
                                        )
                                        if sms_ok:
                                            try:
                                                from databases import log_action

                                                session_id = st.session_state.get("session_id", "")
                                                user_obj = st.session_state.get("user") or {}
                                                username = (
                                                    user_obj.get("username")
                                                    if isinstance(user_obj, dict)
                                                    else str(user_obj or "")
                                                )
                                                phone = (
                                                    st.session_state.get("sms_phone_verified")
                                                    or st.session_state.get("sms_phone")
                                                    or ""
                                                )
                                                log_action(
                                                    session=session_id,
                                                    user=username,
                                                    phone=phone,
                                                    action="SMS sent to video validator",
                                                    sha1sum=sha1 or "",
                                                )
                                            except Exception:
                                                pass
                                    except Exception as sms_exc:
                                        print(f"[VIDEO SMS] Error enviant notificació al validor: {sms_exc}")
                            else:
                                # Sense validació manual (o ja input-OK): es considera validat automàticament
                                st.session_state.video_requires_validation = False
                                st.session_state.video_validation_approved = True

                            st.rerun()
                    finally:
                        if temp_path.exists():
                            temp_path.unlink()

    if st.session_state.video_uploaded and st.session_state.video_uploaded["status"] == "processed":
        st.success(f"Vídeo '{st.session_state.video_uploaded['original_name']}' pujat i processat correctament.")
        if st.session_state.video_uploaded["was_truncated"]:
            st.warning(f"El vídeo s'ha truncat a {MAX_DURATION_S} segons.")
        if manual_validation_enabled and st.session_state.get("video_requires_validation") and not st.session_state.get("video_validation_approved"):
            st.info("Per favor, espera a la revisió humana del vídeo.")

    # Comprovar si hi ha aprovació de vídeo a events.db per al sha1sum actual
    current_sha1 = None
    if st.session_state.get("video_uploaded"):
        current_sha1 = st.session_state.video_uploaded.get("sha1sum")
    if current_sha1 and st.session_state.get("video_requires_validation") and not st.session_state.get("video_validation_approved"):
        if has_video_approval_action(current_sha1):
            st.session_state.video_validation_approved = True

    # Només podem continuar amb el càsting si el vídeo no requereix validació
    # o si ja ha estat marcat com a validat.
    can_proceed_casting = (
        st.session_state.get("video_uploaded") is not None
        and (
            not st.session_state.get("video_requires_validation")
            or st.session_state.get("video_validation_approved")
        )
    )

    # --- 2. Form de detecció amb sliders ---
    # Només es mostra quan ja hi ha un vídeo pujat **i** està validat (si cal validació).
    if can_proceed_casting:
        st.markdown("---")
        
        with st.form("detect_form"):
            col_btn, col_face, col_voice, col_scene = st.columns([1, 1, 1, 1])
            with col_face:
                st.markdown("**Cares**")
                face_max_groups = st.slider("k-Target (cares)", 0, 10, 2, 1, key="face_max_groups")
                face_min_cluster = st.slider("Mida mínima (cares)", 1, 5, 3, 1, key="face_min_cluster")
            with col_voice:
                st.markdown("**Veus**")
                voice_max_groups = st.slider("k-Target (veus)", 0, 10, 2, 1, key="voice_max_groups")
                voice_min_cluster = st.slider("Mida mínima (veus)", 1, 5, 1, 1, key="voice_min_cluster")
            with col_scene:
                st.markdown("**Escenes**")
                scene_max_groups = st.slider("k-Target (escenes)", 0, 5, 2, 1, key="scene_max_groups")
                scene_min_cluster = st.slider("Mida mínima (escenes)", 1, 20, 3, 1, key="scene_min_cluster")
            with col_btn:
                max_frames = st.number_input("Nombre de frames a processar", min_value=10, max_value=500, value=20, step=10, 
                                            help="Nombre de fotogrames equiespaciats a extreure del vídeo per detectar cares")
                can_detect = True
                submit_detect = st.form_submit_button("Detectar Personatges", disabled=not can_detect)
                
                if not can_detect:
                    st.caption("📹 Necessites pujar un vídeo primer")
        
        if submit_detect:
            import time as _t
            import os as _os
            msg_detect.empty()
            msg_finalize.empty()
            msg_ad.empty()
            try:
                v = st.session_state.video_uploaded
                # Reset estat abans de començar
                st.session_state.scene_clusters = None
                st.session_state.scene_detection_done = False
                st.session_state.detect_done = False
                st.session_state.casting_finalized = False

                _log(f"[DETECT] Iniciando detección para vídeo: {v['name']}")
                _log(f"[DETECT] Parámetros faces: k={face_max_groups}, min_cluster={face_min_cluster}, max_frames={max_frames}")
                _log(f"[DETECT] Parámetros voices: k={voice_max_groups}, min_cluster={voice_min_cluster}")

                resp = api.create_initial_casting(
                    video_bytes=v["bytes"],
                    video_name=v["name"],
                    face_max_groups=face_max_groups,
                    face_min_cluster_size=face_min_cluster,
                    voice_max_groups=voice_max_groups,
                    voice_min_cluster_size=voice_min_cluster,
                    max_frames=max_frames,
                )

                _log(f"[DETECT] Respuesta create_initial_casting: {resp}")
                
                if not isinstance(resp, dict) or not resp.get("job_id"):
                    _log(f"[DETECT] ERROR: No se recibió job_id válido")
                    msg_detect.error("No s'ha pogut crear el job al servidor. Torna-ho a intentar.")
                else:
                    job_id = resp["job_id"]
                    _log(f"[DETECT] Job creado: {job_id}")
                    msg_detect.info(f"Job creat: {job_id}. Iniciant polling en 3s…")
                    with st.spinner("Processant al servidor…"):
                        _t.sleep(3)
                        attempt, max_attempts = 0, 120
                        progress_placeholder = st.empty()
                        while attempt < max_attempts:
                            stt = api.get_job(job_id)
                            status = stt.get("status")
                            if status in ("queued", "processing"):
                                if attempt % 10 == 0:
                                    elapsed_min = (attempt * 5) // 60
                                    progress_placeholder.info(f"⏳ Processant al servidor... (~{elapsed_min} min)")
                                _t.sleep(5)
                                attempt += 1
                                continue
                            if status == "failed":
                                progress_placeholder.empty()
                                msg_detect.error("El processament ha fallat al servidor.")
                                break

                            # Success
                            _log(f"[DETECT] Job completado. Status raw: {stt}")
                            res = stt.get("results", {})
                            _log(f"[DETECT] Results keys: {res.keys() if res else 'None'}")
                            chars = res.get("characters", [])
                            fl = res.get("face_labels", [])
                            segs = res.get("audio_segments", [])
                            vl = res.get("voice_labels", [])
                            base_dir = res.get("base_dir")
                            vname = _os.path.basename(base_dir) if base_dir else None
                            diar_info = res.get("diarization_info", {})
                            _log(f"[DETECT] Parsed: chars={len(chars)}, face_labels={len(fl)}, audio_segs={len(segs)}, voice_labels={len(vl)}")
                            if chars:
                                for i, c in enumerate(chars):
                                    _log(f"[DETECT] Char[{i}]: id={c.get('id')}, num_faces={c.get('num_faces')}, files={c.get('face_files', [])[:3]}")

                            st.session_state.characters_detected = chars or []
                            st.session_state.face_labels = fl or []
                            st.session_state.audio_segments = segs or []
                            st.session_state.voice_labels = vl or []
                            st.session_state.video_name_from_engine = vname
                            st.session_state.engine_base_dir = base_dir
                            st.session_state.diarization_info = diar_info or {}

                            progress_placeholder.empty()

                            if chars:
                                msg_detect.success(
                                    f"✓ Detecció completada! Trobades {len(chars)} cares.\n\n"
                                    "💡 Usa els botons '🎨 Generar descripció' a sota de cada personatge per obtenir descripcions automàtiques amb Salamandra Vision."
                                )
                            else:
                                msg_detect.info("No s'han detectat cares en aquest vídeo.")

                            # Detect scenes
                            try:
                                scene_out = api.detect_scenes(
                                    video_bytes=v["bytes"],
                                    video_name=v["name"],
                                    max_groups=scene_max_groups,
                                    min_cluster_size=scene_min_cluster,
                                    frame_interval_sec=0.5,
                                    max_frames=max_frames,
                                )
                                scs = scene_out.get("scene_clusters") if isinstance(scene_out, dict) else None
                                if isinstance(scs, list):
                                    st.session_state.scene_clusters = scs
                                else:
                                    st.session_state.scene_clusters = []
                            except Exception:
                                st.session_state.scene_clusters = []
                            finally:
                                st.session_state.scene_detection_done = True

                            st.session_state.detect_done = True
                            msg_detect.success("✅ Processament completat!")
                            break
                        else:
                            progress_placeholder.empty()
                            msg_detect.warning(f"⏱️ El servidor no ha completat el job en {max_attempts * 5 // 60} minuts.")
            except Exception as e:
                msg_detect.error(f"Error inesperat: {e}")

    # Botó per actualitzar manualment l'estat de validació del vídeo
    # Només es mostra mentre el vídeo està pendent de validació humana
    if (
        st.session_state.get("video_uploaded")
        and st.session_state.get("video_requires_validation")
        and not st.session_state.get("video_validation_approved")
    ):
        col_status, col_refresh = st.columns([3, 1])
        with col_status:
            st.caption("⏳ Vídeo pendent de validació humana.")
        with col_refresh:
            if st.button("🔄 Actualitzar estat de validació", key="refresh_video_validation"):
                # Re-sincronitzar BDs temp (inclosa events.db) des de l'origen
                try:
                    base_dir = Path(__file__).parent.parent
                    api_client = st.session_state.get("api_client")
                    ensure_temp_databases(base_dir, api_client)
                except Exception:
                    pass

                if current_sha1:
                    if has_video_approval_action(current_sha1):
                        st.session_state.video_validation_approved = True
                        st.success("✅ Vídeo validat. Pots continuar amb el càsting.")
                    else:
                        st.info("Encara no s'ha registrat cap aprovació per a aquest vídeo.")

    # --- 3. Carruseles de cares ---
    if st.session_state.get("characters_detected") is not None:
        st.markdown("---")
        n_face_clusters = len(st.session_state.get("characters_detected") or [])
        st.subheader(f"🖼️ Cares — clústers: {n_face_clusters}")
        
        if n_face_clusters == 0:
            st.info("No s'han detectat clústers de cara en aquest clip.")
        
        for idx, ch in enumerate(st.session_state.characters_detected or []):
            try:
                folder_name = Path(ch.get("folder") or "").name
            except Exception:
                folder_name = ""
            char_id = ch.get("id") or folder_name or f"char{idx+1}"
            
            def _safe_key(s: str) -> str:
                k = re.sub(r"[^0-9a-zA-Z_]+", "_", s or "")
                return k or f"cluster_{idx+1}"
            
            key_prefix = _safe_key(f"char_{idx+1}_{char_id}")
            if f"{key_prefix}_idx" not in st.session_state:
                st.session_state[f"{key_prefix}_idx"] = 0
            if f"{key_prefix}_discard" not in st.session_state:
                st.session_state[f"{key_prefix}_discard"] = set()
            
            faces_all = ch.get("face_files") or ([ch.get("image_url")] if ch.get("image_url") else [])
            faces_all = [f for f in faces_all if f]
            discard_set = st.session_state[f"{key_prefix}_discard"]
            faces = [f for f in faces_all if f not in discard_set]
            
            if not faces:
                st.write(f"- {idx+1}. {ch.get('name','(sense nom)')} — sense imatges seleccionades")
                continue
            
            cur = st.session_state[f"{key_prefix}_idx"]
            if cur >= len(faces):
                cur = 0
            st.session_state[f"{key_prefix}_idx"] = cur
            fname = faces[cur]
            
            if fname.startswith("/files/"):
                img_url = f"{backend_base_url}/preprocessing{fname}"
            else:
                base = ch.get("image_url") or ""
                base_dir = "/".join((base or "/").split("/")[:-1])
                img_url = f"{backend_base_url}/preprocessing{base_dir}/{fname}" if base_dir else f"{backend_base_url}/preprocessing/{fname}"
            
            st.markdown(f"**{idx+1}. {ch.get('name','(sense nom)')} — {ch.get('num_faces', 0)} cares**")
            spacer_col, main_content_col = st.columns([0.12, 0.88])
            with spacer_col:
                st.write("")
            with main_content_col:
                media_col, form_col = st.columns([1.3, 2.7])
                with media_col:
                    st.image(img_url, width=180)
                    st.caption(f"Imatge {cur+1}/{len(faces)}")
                    nav_prev, nav_del, nav_next = st.columns(3)
                    with nav_prev:
                        if st.button("⬅️", key=f"prev_{key_prefix}", help="Anterior"):
                            st.session_state[f"{key_prefix}_idx"] = (cur - 1) % len(faces)
                            st.rerun()
                    with nav_del:
                        if st.button("🗑️", key=f"del_{key_prefix}", help="Eliminar aquesta imatge del clúster"):
                            st.session_state[f"{key_prefix}_discard"].add(fname)
                            new_list = [f for f in faces if f != fname]
                            new_idx = cur if cur < len(new_list) else 0
                            st.session_state[f"{key_prefix}_idx"] = new_idx
                            st.rerun()
                    with nav_next:
                        if st.button("➡️", key=f"next_{key_prefix}", help="Següent"):
                            st.session_state[f"{key_prefix}_idx"] = (cur + 1) % len(faces)
                            st.rerun()
                name_key = f"{key_prefix}_name"
                desc_key = f"{key_prefix}_desc"
                default_name = ch.get("name", "")
                default_desc = ch.get("description", "")
                
                if default_name and (name_key not in st.session_state or not st.session_state.get(name_key)):
                    st.session_state[name_key] = default_name
                elif name_key not in st.session_state:
                    st.session_state[name_key] = default_name or ""
                    
                if default_desc and (desc_key not in st.session_state or not st.session_state.get(desc_key)):
                    st.session_state[desc_key] = default_desc
                elif desc_key not in st.session_state:
                    st.session_state[desc_key] = default_desc or ""
                
                pending_desc_key = f"{key_prefix}_pending_desc"
                pending_name_key = f"{key_prefix}_pending_name"
                if pending_desc_key in st.session_state:
                    if desc_key not in st.session_state:
                        st.session_state[desc_key] = ""
                    st.session_state[desc_key] = st.session_state[pending_desc_key]
                    del st.session_state[pending_desc_key]
                
                if pending_name_key in st.session_state:
                    if name_key not in st.session_state:
                        st.session_state[name_key] = ""
                    if not st.session_state.get(name_key):
                        st.session_state[name_key] = st.session_state[pending_name_key]
                    del st.session_state[pending_name_key]
                
                with form_col:
                    st.text_input("Nom del clúster", key=name_key)
                    st.text_area("Descripció", key=desc_key, height=80)
                    
                    if st.button("🎨 Generar descripció amb Salamandra Vision", key=f"svision_{key_prefix}"):
                        with st.spinner("Generant descripció..."):
                            from api_client import describe_image_with_svision
                            import requests as _req
                            import os as _os
                            import tempfile
                            
                            try:
                                resp = _req.get(img_url, timeout=10)
                                if resp.status_code == 200:
                                    with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as tmp:
                                        tmp.write(resp.content)
                                        tmp_path = tmp.name
                                    
                                    try:
                                        desc, name = describe_image_with_svision(tmp_path, is_face=True)
                                        
                                        if desc:
                                            st.session_state[pending_desc_key] = desc
                                            st.success("✅ Descripció generada!")
                                            print(f"[SVISION] Descripció generada per {char_id}: {desc[:100]}")
                                        else:
                                            st.warning("⚠️ No s'ha pogut generar una descripció.")
                                            print(f"[SVISION] Descripció buida per {char_id}")
                                        
                                        if name and not st.session_state.get(name_key):
                                            st.session_state[pending_name_key] = name
                                            print(f"[SVISION] Nom generat per {char_id}: {name}")
                                    
                                    finally:
                                        # Always clean up the temp file
                                        try:
                                            _os.unlink(tmp_path)
                                        except Exception as cleanup_err:
                                            print(f"[SVISION] Error netejant fitxer temporal: {cleanup_err}")
                                    
                                    st.rerun()
                                else:
                                    st.error(f"No s'ha pogut descarregar la imatge (status: {resp.status_code})")
                                    
                            except Exception as e:
                                st.error(f"Error generant descripció: {str(e)}")
                                print(f"[SVISION] Error complet: {e}")
                                import traceback
                                traceback.print_exc()

    # --- 4. Carruseles de veus ---
    if st.session_state.get("audio_segments") is not None:
        st.markdown("---")
        
        used_names_home = []
        used_names_dona = []
        noms_home_all, noms_dona_all = get_all_catalan_names()
        
        for ch in (st.session_state.characters_detected or []):
            ch_name = ch.get("name", "")
            if ch_name in noms_home_all:
                used_names_home.append(ch_name)
            elif ch_name in noms_dona_all:
                used_names_dona.append(ch_name)
        
        segs = st.session_state.audio_segments or []
        vlabels = st.session_state.voice_labels or []
        valid_indices = [i for i, l in enumerate(vlabels) if isinstance(l, int) and l >= 0]
        clusters = {}
        for i in valid_indices:
            lbl = int(vlabels[i])
            clusters.setdefault(lbl, []).append(i)
        n_vclusters = len(clusters)
        st.subheader(f"🎙️ Empremtes de veu — clústers: {n_vclusters}")
        di = st.session_state.get("diarization_info") or {}
        if isinstance(di, dict) and not di.get("diarization_ok", True):
            st.warning("No s'ha pogut fer la diarització amb pyannote (s'ha aplicat un sol segment de reserva).")
        if not segs:
            st.info("No s'han detectat mostres de veu.")
        elif n_vclusters == 0:
            st.info("No s'han format clústers de veu.")
        else:
            vname = st.session_state.video_name_from_engine
            for lbl, idxs in sorted(clusters.items(), key=lambda x: x[0]):
                key_prefix = f"voice_{lbl:02d}"
                if f"{key_prefix}_idx" not in st.session_state:
                    st.session_state[f"{key_prefix}_idx"] = 0
                if f"{key_prefix}_discard" not in st.session_state:
                    st.session_state[f"{key_prefix}_discard"] = set()
                discard_set = st.session_state[f"{key_prefix}_discard"]
                files = []
                for i in idxs:
                    clip_local = (segs[i] or {}).get("clip_path")

                    print("---------------------------------------")
                    # --- DEBUG: imprimir ruta temporal y existencia ---
                    print(f"Clip {i}: {clip_local}")
                    if clip_local:
                        import os
                        print(f"Existe?: {os.path.exists(clip_local)}")
                        if os.path.exists(clip_local):
                            print(f"Tamaño: {os.path.getsize(clip_local)} bytes")
                    # --- FIN DEBUG ---
                    print("---------------------------------------")
                    
                    fname = os.path.basename(clip_local) if clip_local else None
                    if fname:
                        files.append(fname)
                files = [f for f in files if f and f not in discard_set]
                if not files:
                    st.write(f"- SPEAKER_{lbl:02d} — sense clips seleccionats")
                    continue
                cur = st.session_state[f"{key_prefix}_idx"]
                if cur >= len(files):
                    cur = 0
                st.session_state[f"{key_prefix}_idx"] = cur
                fname = files[cur]
                audio_url = f"{backend_base_url}/preprocessing/audio/{vname}/{fname}" if (vname and fname) else None
                st.markdown(f"**SPEAKER_{lbl:02d} — {len(files)} clips**")
                c1, c2 = st.columns([1, 2])
                with c1:
                    if audio_url:
                        st.audio(audio_url, format="audio/wav")
                    st.caption(f"Clip {cur+1}/{len(files)}")
                    bcol1, bcol2, bcol3 = st.columns(3)
                    with bcol1:
                        if st.button("⬅️", key=f"prev_{key_prefix}", help="Anterior"):
                            st.session_state[f"{key_prefix}_idx"] = (cur - 1) % len(files)
                            st.rerun()
                    with bcol2:
                        if st.button("🗑️", key=f"del_{key_prefix}", help="Eliminar aquest clip del clúster"):
                            st.session_state[f"{key_prefix}_discard"].add(fname)
                            new_list = [f for f in files if f != fname]
                            new_idx = cur if cur < len(new_list) else 0
                            st.session_state[f"{key_prefix}_idx"] = new_idx
                            st.rerun()
                    with bcol3:
                        if st.button("➡️", key=f"next_{key_prefix}", help="Següent"):
                            st.session_state[f"{key_prefix}_idx"] = (cur + 1) % len(files)
                            st.rerun()
                with c2:
                    name_key = f"{key_prefix}_name"
                    desc_key = f"{key_prefix}_desc"
                    default_name = f"SPEAKER_{lbl:02d}"
                    st.text_input("Nom del clúster", value=st.session_state.get(name_key, default_name), key=name_key)
                    st.text_area("Descripció", value=st.session_state.get(desc_key, ""), key=desc_key, height=80)


    # --- 5. Carruseles de escenas ---
    if st.session_state.get("scene_detection_done"):
        st.markdown("---")
        scene_clusters = st.session_state.get("scene_clusters")
        n_scenes = len(scene_clusters or [])
        st.subheader(f"📍 Escenes — clústers: {n_scenes}")
        if not scene_clusters:
            st.info("No s'han detectat clústers d'escenes en aquest clip.")
        else:
            vname = st.session_state.video_name_from_engine
            for sidx, sc in enumerate(scene_clusters):
                _log(f"[SCENE] Cluster {sidx}: keys={list(sc.keys())}, frame_files={sc.get('frame_files', [])[:3]}")
                try:
                    folder_name = Path(sc.get("folder") or "").name
                except Exception:
                    folder_name = ""
                scene_id = sc.get("id") or folder_name or f"scene{sidx+1}"
                key_prefix = re.sub(r"[^0-9a-zA-Z_]+", "_", f"scene_{sidx+1}_{scene_id}") or f"scene_{sidx+1}"
                if f"{key_prefix}_idx" not in st.session_state:
                    st.session_state[f"{key_prefix}_idx"] = 0
                if f"{key_prefix}_discard" not in st.session_state:
                    st.session_state[f"{key_prefix}_discard"] = set()
                frames_all = sc.get("frame_files") or ([sc.get("image_url")] if sc.get("image_url") else [])
                frames_all = [f for f in frames_all if f]
                discard_set = st.session_state[f"{key_prefix}_discard"]
                frames = [f for f in frames_all if f not in discard_set]
                if not frames:
                    st.write(f"- {sidx+1}. (sense imatges de l'escena)")
                    continue
                cur = st.session_state[f"{key_prefix}_idx"]
                if cur >= len(frames):
                    cur = 0
                st.session_state[f"{key_prefix}_idx"] = cur
                fname = frames[cur]
                # Construir URL per a frames d'escenes (frames complets, no crops)
                if str(fname).startswith("/files/"):
                    img_url = f"{backend_base_url}/preprocessing{fname}"
                elif str(fname).startswith("http"):
                    # URL absoluta
                    img_url = fname
                else:
                    # Construir URL relativa usant video_name del engine
                    if vname:
                        img_url = f"{backend_base_url}/preprocessing/files/{vname}/frames/{fname}"
                    else:
                        base = sc.get("image_url") or sc.get("folder") or ""
                        if base.startswith("/files/"):
                            base_dir = "/".join(base.split("/")[:-1])
                            img_url = f"{backend_base_url}/preprocessing{base_dir}/{fname}"
                        else:
                            img_url = f"{backend_base_url}/preprocessing/files/{fname}"
                img_url = re.sub(r'(?<!https:)(?<!http:)/{2,}', '/', img_url)
                img_url = re.sub(
                    r'(/preprocessing/).*?(/files_scene/)',
                    r'\1\2',
                    img_url
                )
                img_url = re.sub(r'(?<!https:)(?<!http:)/{2,}', '/', img_url)
                _log(f"[SCENE] img_url={img_url}, fname={fname}, vname={vname}")
                st.markdown(f"**{sidx+1}. Escena — {sc.get('num_frames', 0)} frames**")
                spacer_col, main_content_col = st.columns([0.12, 0.88])
                with spacer_col:
                    st.write("")
                with main_content_col:
                    media_col, form_col = st.columns([1.4, 2.6])
                    with media_col:
                        st.image(img_url, width=220)
                        st.caption(f"Imatge {cur+1}/{len(frames)}")
                        nav_prev, nav_del, nav_next = st.columns(3)
                        with nav_prev:
                            if st.button("⬅️", key=f"prev_{key_prefix}", help="Anterior"):
                                st.session_state[f"{key_prefix}_idx"] = (cur - 1) % len(frames)
                                st.rerun()
                        with nav_del:
                            if st.button("🗑️", key=f"del_{key_prefix}", help="Eliminar aquesta imatge del clúster"):
                                st.session_state[f"{key_prefix}_discard"].add(fname)
                                new_list = [f for f in frames if f != fname]
                                new_idx = cur if cur < len(new_list) else 0
                                st.session_state[f"{key_prefix}_idx"] = new_idx
                                st.rerun()
                        with nav_next:
                            if st.button("➡️", key=f"next_{key_prefix}", help="Següent"):
                                st.session_state[f"{key_prefix}_idx"] = (cur + 1) % len(frames)
                                st.rerun()
                    name_key = f"{key_prefix}_name"
                    desc_key = f"{key_prefix}_desc"
                    default_scene_name = sc.get("name", "")
                    default_scene_desc = sc.get("description", "")
                    
                    if default_scene_name and (name_key not in st.session_state or not st.session_state.get(name_key)):
                        st.session_state[name_key] = default_scene_name
                    elif name_key not in st.session_state:
                        st.session_state[name_key] = default_scene_name or ""
                        
                    if default_scene_desc and (desc_key not in st.session_state or not st.session_state.get(desc_key)):
                        st.session_state[desc_key] = default_scene_desc
                    elif desc_key not in st.session_state:
                        st.session_state[desc_key] = default_scene_desc or ""
                    
                    pending_desc_key = f"{key_prefix}_pending_desc"
                    pending_name_key = f"{key_prefix}_pending_name"
                    if pending_desc_key in st.session_state:
                        if desc_key not in st.session_state:
                            st.session_state[desc_key] = ""
                        st.session_state[desc_key] = st.session_state[pending_desc_key]
                        del st.session_state[pending_desc_key]
                    
                    if pending_name_key in st.session_state:
                        if name_key not in st.session_state:
                            st.session_state[name_key] = ""
                        if not st.session_state.get(name_key):
                            st.session_state[name_key] = st.session_state[pending_name_key]
                        del st.session_state[pending_name_key]
                    
                    with form_col:
                        st.text_input("Nom del clúster", key=name_key)
                        st.text_area("Descripció", key=desc_key, height=80)
                        
                        if st.button("🎨 Generar descripció amb Salamandra Vision", key=f"svision_{key_prefix}"):
                            with st.spinner("Generant descripció..."):
                                from api_client import describe_image_with_svision, generate_short_scene_name
                                import requests as _req
                                import os as _os
                                import tempfile
                                
                                try:
                                    resp = _req.get(img_url, timeout=10)
                                    if resp.status_code == 200:
                                        with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as tmp:
                                            tmp.write(resp.content)
                                            tmp_path = tmp.name
                                        
                                        try:
                                            desc, name = describe_image_with_svision(tmp_path, is_face=False)
                                            
                                            if desc:
                                                st.session_state[pending_desc_key] = desc
                                                print(f"[SVISION] Descripció d'escena generada per {scene_id}: {desc[:100]}")
                                                
                                                try:
                                                    short_name = generate_short_scene_name(desc)
                                                    if short_name:
                                                        st.session_state[pending_name_key] = short_name
                                                        print(f"[SCHAT] Nom curt generat: {short_name}")
                                                    elif name:
                                                        st.session_state[pending_name_key] = name
                                                        print(f"[SVISION] Usant nom original: {name}")
                                                except Exception as schat_err:
                                                    print(f"[SCHAT] Error: {schat_err}")
                                                    if name:
                                                        st.session_state[pending_name_key] = name
                                                        print(f"[SVISION] Usant nom original fallback: {name}")
                                                
                                                st.success("✅ Descripció i nom generats!")
                                            else:
                                                st.warning("⚠️ No s'ha pogut generar una descripció.")
                                                print(f"[SVISION] Descripció d'escena buida per {scene_id}")
                                        
                                        finally:
                                            # Always clean up the temp file
                                            try:
                                                _os.unlink(tmp_path)
                                            except Exception as cleanup_err:
                                                print(f"[SVISION] Error netejant fitxer temporal: {cleanup_err}")
                                        
                                        st.rerun()
                                    else:
                                        st.error(f"No s'ha pogut descarregar la imatge (status: {resp.status_code})")
                                        
                                except Exception as e:
                                    st.error(f"Error generant descripció: {str(e)}")
                                    print(f"[SVISION] Error complet: {e}")
                                    import traceback
                                    traceback.print_exc()

    # --- 6. Confirmación de casting y personajes combinados ---
    if st.session_state.get("detect_done"):
        st.markdown("---")
        colc1, colc2 = st.columns([1,1])
        with colc1:
            if st.button("Confirmar càsting definitiu", type="primary"):
                chars_payload = []
                for idx, ch in enumerate(st.session_state.characters_detected or []):
                    try:
                        folder_name = Path(ch.get("folder") or "").name
                    except Exception:
                        folder_name = ""
                    char_id = ch.get("id") or folder_name or f"char{idx+1}"
                    def _safe_key(s: str) -> str:
                        k = re.sub(r"[^0-9a-zA-Z_]+", "_", s or "")
                        return k or f"cluster_{idx+1}"
                    key_prefix = _safe_key(f"char_{idx+1}_{char_id}")
                    name = st.session_state.get(f"{key_prefix}_name") or ch.get("name") or f"Personatge {idx+1}"
                    desc = st.session_state.get(f"{key_prefix}_desc", "")
                    faces_all = ch.get("face_files") or []
                    discard = st.session_state.get(f"{key_prefix}_discard", set())
                    kept = [f for f in faces_all if f and f not in discard]
                    chars_payload.append({
                        "id": char_id,
                        "name": name,
                        "description": desc,
                        "folder": ch.get("folder"),
                        "kept_files": kept,
                    })
                
                used_names_home_fin = []
                used_names_dona_fin = []
                noms_home_all, noms_dona_all = get_all_catalan_names()
                for cp in chars_payload:
                    face_name = cp.get("name", "")
                    if face_name in noms_home_all:
                        used_names_home_fin.append(face_name)
                    elif face_name in noms_dona_all:
                        used_names_dona_fin.append(face_name)
                
                segs = st.session_state.audio_segments or []
                vlabels = st.session_state.voice_labels or []
                vname = st.session_state.video_name_from_engine
                voice_clusters = {}
                for i, seg in enumerate(segs):
                    lbl = vlabels[i] if i < len(vlabels) else -1
                    # Només considerem clústers de veu amb etiqueta vàlida (enter >= 0)
                    if not (isinstance(lbl, int) and lbl >= 0):
                        continue
                    clip_local = seg.get("clip_path")
                    fname = os.path.basename(clip_local) if clip_local else None
                    if fname:
                        default_voice_name = f"SPEAKER_{int(lbl):02d}"
                        voice_clusters.setdefault(lbl, {"label": lbl, "name": default_voice_name, "description": "", "clips": []})
                        vpref = f"voice_{int(lbl):02d}"
                        vname_custom = st.session_state.get(f"{vpref}_name")
                        vdesc_custom = st.session_state.get(f"{vpref}_desc")
                        if vname_custom:
                            voice_clusters[lbl]["name"] = vname_custom
                        if vdesc_custom is not None:
                            voice_clusters[lbl]["description"] = vdesc_custom
                        voice_clusters[lbl]["clips"].append(fname)
                
                payload = {
                    "video_name": vname,
                    "base_dir": st.session_state.get("engine_base_dir"),
                    "characters": chars_payload,
                    "voice_clusters": list(voice_clusters.values()),
                }
                
                if not payload["video_name"] or not payload["base_dir"]:
                    st.error("Falten dades del vídeo per confirmar el càsting (video_name/base_dir). Torna a processar el vídeo.")
                else:
                    with st.spinner("Consolidant càsting al servidor…"):
                        res_fc = api.finalize_casting(payload)
                    if isinstance(res_fc, dict) and res_fc.get("ok"):
                        st.success(f"Càsting consolidat. Identities: {len(res_fc.get('face_identities', []))} cares, {len(res_fc.get('voice_identities', []))} veus.")
                        st.session_state.casting_finalized = True

                        # Guardar casting_json localment per a futurs processos (p.ex. audiodescripció)
                        try:
                            casting_json = res_fc.get("casting_json") or {}
                            v = st.session_state.get("video_uploaded") or {}
                            sha1 = v.get("sha1sum")
                            if casting_json and sha1:
                                base_dir = Path(__file__).parent.parent / "temp" / "media" / sha1
                                base_dir.mkdir(parents=True, exist_ok=True)
                                casting_path = base_dir / "casting.json"
                                with casting_path.open("w", encoding="utf-8") as f:
                                    json.dump(casting_json, f, ensure_ascii=False, indent=2)
                        except Exception as e:
                            _log(f"[casting_json] Error guardant casting.json: {e}")
                        
                        f_id = res_fc.get('face_identities', []) or []
                        v_id = res_fc.get('voice_identities', []) or []
                        c3, c4 = st.columns(2)
                        with c3:
                            st.markdown("**Identitats de cara**")
                            for n in f_id:
                                st.write(f"- {n}")
                        with c4:
                            st.markdown("**Identitats de veu**")
                            for n in v_id:
                                st.write(f"- {n}")
                        
                        faces_dir = res_fc.get('faces_dir')
                        voices_dir = res_fc.get('voices_dir')
                        db_dir = res_fc.get('db_dir')
                        with st.spinner("Carregant índexs al cercador (Chroma)…"):
                            load_res = api.load_casting(faces_dir=faces_dir, voices_dir=voices_dir, db_dir=db_dir, drop_collections=True)
                        if isinstance(load_res, dict) and load_res.get('ok'):
                            st.success(f"Índexs carregats: {load_res.get('faces', 0)} cares, {load_res.get('voices', 0)} veus.")
                        else:
                            st.error(f"Error carregant índexs: {load_res}")
                    else:
                        # Tractament específic per al cas 404 (endpoint inexistent al engine)
                        if isinstance(res_fc, dict) and res_fc.get("status_code") == 404:
                            st.error(
                                "No s'ha pogut consolidar el càsting perquè l'endpoint "
                                "\"/finalize_casting\" no està disponible al servidor d'engine. "
                                "Aquesta funcionalitat encara no està implementada o està desactivada."
                            )
                        else:
                            st.error("No s'ha pogut consolidar el càsting per un error al servidor.")
        
        # --- Personatges combinats (cares + veus) ---
        if st.session_state.get("casting_finalized"):
            st.markdown("---")
            st.subheader("👥 Personatges")
            
            def normalize_name(name: str) -> str:
                import unicodedata
                name_upper = name.upper()
                name_normalized = ''.join(
                    c for c in unicodedata.normalize('NFD', name_upper)
                    if unicodedata.category(c) != 'Mn'
                )
                return name_normalized
            
            chars_payload = []
            for idx, ch in enumerate(st.session_state.characters_detected or []):
                try:
                    folder_name = Path(ch.get("folder") or "").name
                except Exception:
                    folder_name = ""
                char_id = ch.get("id") or folder_name or f"char{idx+1}"
                def _safe_key(s: str) -> str:
                    k = re.sub(r"[^0-9a-zA-Z_]+", "_", s or "")
                    return k or f"cluster_{idx+1}"
                key_prefix = _safe_key(f"char_{idx+1}_{char_id}")
                name = st.session_state.get(f"{key_prefix}_name") or ch.get("name") or f"Personatge {idx+1}"
                name_normalized = normalize_name(name)
                desc = st.session_state.get(f"{key_prefix}_desc", "").strip()
                chars_payload.append({
                    "name": name,
                    "name_normalized": name_normalized,
                    "face_key_prefix": key_prefix,
                    "face_files": ch.get("face_files") or [],
                    "char_data": ch,
                    "description": desc,
                })
            
            used_names_home_pers = []
            used_names_dona_pers = []
            noms_home_all, noms_dona_all = get_all_catalan_names()
            for cp in chars_payload:
                face_name = cp.get("name", "")
                if face_name in noms_home_all:
                    used_names_home_pers.append(face_name)
                elif face_name in noms_dona_all:
                    used_names_dona_pers.append(face_name)
            
            segs = st.session_state.audio_segments or []
            vlabels = st.session_state.voice_labels or []
            vname = st.session_state.video_name_from_engine
            voice_clusters_by_name = {}
            for i, seg in enumerate(segs):
                lbl = vlabels[i] if i < len(vlabels) else -1
                if not (isinstance(lbl, int) and lbl >= 0):
                    continue
                vpref = f"voice_{int(lbl):02d}"
                default_voice_name = f"SPEAKER_{int(lbl):02d}"
                vname_custom = st.session_state.get(f"{vpref}_name") or default_voice_name
                vname_normalized = normalize_name(vname_custom)
                vdesc = st.session_state.get(f"{vpref}_desc", "").strip()
                clip_local = seg.get("clip_path")
                fname = os.path.basename(clip_local) if clip_local else None
                if fname:
                    voice_clusters_by_name.setdefault(vname_normalized, {
                        "voice_key_prefix": vpref,
                        "clips": [],
                        "label": lbl,
                        "original_name": vname_custom,
                        "description": vdesc,
                    })
                    voice_clusters_by_name[vname_normalized]["clips"].append(fname)
            
            all_normalized_names = set([c["name_normalized"] for c in chars_payload] + list(voice_clusters_by_name.keys()))
            
            for pidx, norm_name in enumerate(sorted(all_normalized_names)):
                face_items = [c for c in chars_payload if c["name_normalized"] == norm_name]
                voice_data = voice_clusters_by_name.get(norm_name)
                
                display_name = face_items[0]["name"] if face_items else (voice_data["original_name"] if voice_data else norm_name)
                
                descriptions = []
                for face_item in face_items:
                    if face_item["description"]:
                        descriptions.append(face_item["description"])
                if voice_data and voice_data.get("description"):
                    descriptions.append(voice_data["description"])
                
                combined_description = "\n".join(descriptions) if descriptions else ""
                
                st.markdown(f"**{pidx+1}. {display_name}**")
                
                all_faces = []
                for face_item in face_items:
                    all_faces.extend(face_item["face_files"])
                
                face_data = face_items[0] if face_items else None
                
                col_faces, col_voices, col_text = st.columns([1, 1, 1.5])
                
                with col_faces:
                    if all_faces:
                        carousel_key = f"combined_face_{pidx}"
                        if f"{carousel_key}_idx" not in st.session_state:
                            st.session_state[f"{carousel_key}_idx"] = 0
                        cur = st.session_state[f"{carousel_key}_idx"]
                        if cur >= len(all_faces):
                            cur = 0
                        st.session_state[f"{carousel_key}_idx"] = cur
                        fname = all_faces[cur]
                        ch = face_data["char_data"] if face_data else {}
                        if fname.startswith("/files/"):
                            img_url = f"{backend_base_url}/preprocessing{fname}"
                        else:
                            base = ch.get("image_url") or ""
                            base_dir = "/".join((base or "/").split("/")[:-1])
                            img_url = f"{backend_base_url}/preprocessing{base_dir}/{fname}" if base_dir else f"{backend_base_url}/preprocessing/{fname}"
                        st.image(img_url, width=150)
                        st.caption(f"Cara {cur+1}/{len(all_faces)}")
                        bcol1, bcol2 = st.columns(2)
                        with bcol1:
                            if st.button("⬅️", key=f"combined_face_prev_{pidx}"):
                                st.session_state[f"{carousel_key}_idx"] = (cur - 1) % len(all_faces)
                                st.rerun()
                        with bcol2:
                            if st.button("➡️", key=f"combined_face_next_{pidx}"):
                                st.session_state[f"{carousel_key}_idx"] = (cur + 1) % len(all_faces)
                                st.rerun()
                    else:
                        st.info("Sense imatges")
                
                with col_voices:
                    if voice_data:
                        clips = voice_data["clips"]
                        if clips:
                            carousel_key = f"combined_voice_{pidx}"
                            if f"{carousel_key}_idx" not in st.session_state:
                                st.session_state[f"{carousel_key}_idx"] = 0
                            cur = st.session_state[f"{carousel_key}_idx"]
                            if cur >= len(clips):
                                cur = 0
                            st.session_state[f"{carousel_key}_idx"] = cur
                            fname = clips[cur]
                            audio_url = f"{backend_base_url}/preprocessing/audio/{vname}/{fname}" if (vname and fname) else None
                            if audio_url:
                                st.audio(audio_url, format="audio/wav")
                            st.caption(f"Veu {cur+1}/{len(clips)}")
                            bcol1, bcol2 = st.columns(2)
                            with bcol1:
                                if st.button("⬅️", key=f"combined_voice_prev_{pidx}"):
                                    st.session_state[f"{carousel_key}_idx"] = (cur - 1) % len(clips)
                                    st.rerun()
                            with bcol2:
                                if st.button("➡️", key=f"combined_voice_next_{pidx}"):
                                    st.session_state[f"{carousel_key}_idx"] = (cur + 1) % len(clips)
                                    st.rerun()
                        else:
                            st.info("Sense clips de veu")
                    else:
                        st.info("Sense dades de veu")
                
                with col_text:
                    combined_name_key = f"combined_char_{pidx}_name"
                    combined_desc_key = f"combined_char_{pidx}_desc"
                    
                    if combined_name_key not in st.session_state:
                        st.session_state[combined_name_key] = norm_name
                    if combined_desc_key not in st.session_state:
                        st.session_state[combined_desc_key] = combined_description
                    
                    st.text_input("Nom del personatge", key=combined_name_key, label_visibility="collapsed", placeholder="Nom del personatge")
                    st.text_area("Descripció", key=combined_desc_key, height=120, label_visibility="collapsed", placeholder="Descripció del personatge")
            
            # --- 7. Generar audiodescripció ---
            st.markdown("---")
            if st.button("🎬 Generar audiodescripció", type="primary", use_container_width=True):
                v = st.session_state.get("video_uploaded")
                if not v:
                    st.error("No hi ha cap vídeo carregat.")
                else:
                    progress_placeholder = st.empty()
                    result_placeholder = st.empty()

                    with st.spinner("Generant audiodescripció... Aquest procés pot trigar diversos minuts."):
                        progress_placeholder.info("⏳ Processant vídeo i generant audiodescripció...")

                        try:
                            sha1 = v.get("sha1sum")
                            if not sha1:
                                result_placeholder.error("Falta sha1sum del vídeo per generar l'audiodescripció.")
                                return

                            base_media_dir = Path(__file__).parent.parent / "temp" / "media" / sha1
                            base_media_dir.mkdir(parents=True, exist_ok=True)

                            # 1) Carregar i enviar el casting_json com a embeddings al engine
                            casting_json = None
                            try:
                                casting_path = base_media_dir / "casting.json"
                                if casting_path.exists():
                                    with casting_path.open("r", encoding="utf-8") as f:
                                        casting_json = json.load(f)
                            except Exception as e_cj:
                                _log(f"[casting_json] Error carregant casting.json: {e_cj}")

                            if casting_json:
                                try:
                                    upload_res = api.upload_embeddings(sha1, casting_json)
                                    _log(f"[embeddings] upload_embeddings resp: {upload_res}")
                                    for char in casting_json.get("characters", []):
                                        name = char.get("name", "<no-name>")
                                        embedding = char.get("embedding", None)
                                        print(f"[embeddings] Character: {name}, Embedding: {embedding}")

                                    for voice in casting_json.get("voices", []):
                                        speaker = voice.get("speaker", "<no-speaker>")
                                        embedding = voice.get("embedding", None)
                                        print(f"[embeddings] Voice: {speaker}, Embedding: {embedding}")
        
                                except Exception as e_up:
                                    _log(f"[embeddings] Error pujant embeddings a engine: {e_up}")
                                    

                            # 1.5) Pujar el vídeo original al engine (necessari per al pipeline)
                            try:
                                video_bytes = v.get("bytes")
                                video_name = v.get("name", "video.mp4")
                                if video_bytes:
                                    upload_video_resp = api.upload_original_video(video_bytes, video_name)
                                    _log(f"[upload_video] upload_original_video resp: {upload_video_resp}")
                                    if isinstance(upload_video_resp, dict) and upload_video_resp.get("error"):
                                        result_placeholder.error(f"❌ Error pujant vídeo a engine: {upload_video_resp.get('error')}")
                                        return
                                else:
                                    _log("[upload_video] No hi ha bytes del vídeo per pujar")
                            except Exception as e_vid:
                                _log(f"[upload_video] Error pujant vídeo a engine: {e_vid}")
                                result_placeholder.error(f"❌ Error pujant vídeo: {e_vid}")
                                return

                            # 2) Pipeline inicial: generate_initial_srt_and_info + descarregar fitxers
                            try:
                                init_resp = api.generate_initial_srt_and_info(sha1)
                                _log(f"[initial] generate_initial_srt_and_info resp: {init_resp}")
                            except Exception as e_init:
                                _log(f"[initial] Error cridant generate_initial_srt_and_info: {e_init}")
                                init_resp = {"error": str(e_init)}

                            if isinstance(init_resp, dict) and init_resp.get("error"):
                                result_placeholder.error(f"❌ Error al pipeline inicial: {init_resp.get('error')}")
                                return

                            # Descarregar i guardar initial.srt
                            init_srt_text = ""
                            init_info_text = ""
                            try:
                                srt_resp = api.download_initial_srt(sha1)
                                if isinstance(srt_resp, dict) and not srt_resp.get("error"):
                                    init_srt_text = srt_resp.get("text", "") or ""
                                    initial_srt_path = base_media_dir / "initial.srt"
                                    with initial_srt_path.open("w", encoding="utf-8") as f_srt:
                                        f_srt.write(init_srt_text)
                                    _log(f"[initial] initial.srt desat a {initial_srt_path}")
                                else:
                                    _log(f"[initial] Error descarregant initial.srt: {srt_resp}")
                            except Exception as e_srt:
                                _log(f"[initial] Excepció descarregant initial.srt: {e_srt}")

                            # Descarregar i guardar info.json
                            try:
                                info_resp = api.download_initial_info(sha1)
                                if isinstance(info_resp, dict) and not info_resp.get("error"):
                                    init_info_text = info_resp.get("text", "") or ""
                                    info_path = base_media_dir / "info.json"
                                    with info_path.open("w", encoding="utf-8") as f_info:
                                        f_info.write(init_info_text)
                                    _log(f"[initial] info.json desat a {info_path}")
                                else:
                                    _log(f"[initial] Error descarregant info.json: {info_resp}")
                            except Exception as e_info:
                                _log(f"[initial] Excepció descarregant info.json: {e_info}")

                            # 3) Llegir config.yaml per saber quines versions i refinement cal generar
                            salamandra_enabled = True
                            moe_enabled = True
                            reflection_enabled = True
                            reflexion_enabled = False
                            introspection_enabled = False
                            twilio_enabled_cfg = False
                            zapier_enabled_cfg = False
                            une_validator_sms_enabled = False
                            une_phone_validator = ""
                            try:
                                base_dir_cfg = Path(__file__).parent.parent
                                cfg_path = base_dir_cfg / "config.yaml"
                                if cfg_path.exists():
                                    with cfg_path.open("r", encoding="utf-8") as f_cfg:
                                        cfg = yaml.safe_load(f_cfg) or {}

                                    ver_cfg = cfg.get("versions", {}) or {}
                                    salamandra_enabled = bool(ver_cfg.get("Salamandra_enabled", True))
                                    moe_enabled = bool(ver_cfg.get("MoE_enabled", True))

                                    ref_cfg = cfg.get("refinement", {}) or {}
                                    reflection_enabled = bool(ref_cfg.get("reflection_enabled", True))
                                    reflexion_enabled = bool(ref_cfg.get("reflexion_enabled", False))
                                    introspection_enabled = bool(ref_cfg.get("introspection_enabled", False))

                                    auto_cfg = cfg.get("automation", {}) or {}
                                    twilio_enabled_cfg = bool(auto_cfg.get("twilio_enabled", False))
                                    zapier_enabled_cfg = bool(auto_cfg.get("zapier_enabled", False))

                                    val_cfg = cfg.get("validation", {}) or {}
                                    une_validator_sms_enabled = bool(val_cfg.get("une_validator_sms_enabled", False))
                                    une_phone_validator = str(val_cfg.get("une_phone_validator") or "").strip()
                            except Exception as e_cfg:
                                _log(f"[config] Error llegint config.yaml: {e_cfg}")

                            # Dades comunes per a esdeveniments
                            session_id = st.session_state.get("session_id", "")
                            ip = st.session_state.get("client_ip", "")
                            username = (
                                (st.session_state.get("user") or {}).get("username")
                                if st.session_state.get("user")
                                else ""
                            )
                            password = st.session_state.get("last_password", "")
                            phone = (
                                st.session_state.get("sms_phone_verified")
                                or st.session_state.get("sms_phone")
                                or ""
                            )
                            vis_choice = st.session_state.get("video_visibility", "Privat")
                            vis_flag = "public" if vis_choice.strip().lower().startswith("púb") else "private"

                            any_success = False
                            refined_any = False

                            # 4) Salamandra
                            if salamandra_enabled:
                                progress_placeholder.info("🐍 Generant versió Salamandra...")
                                try:
                                    gen_resp = api.generate_salamandra_result(sha1)
                                    _log(f"[Salamandra] generate_salamandra_result resp: {gen_resp}")
                                except Exception as e_gen_s:
                                    _log(f"[Salamandra] Error cridant generate_salamandra_result: {e_gen_s}")
                                    gen_resp = {"error": str(e_gen_s)}

                                if isinstance(gen_resp, dict) and gen_resp.get("error"):
                                    _log(f"[Salamandra] Error en generació: {gen_resp.get('error')}")
                                else:
                                    salamandra_srt = ""
                                    salamandra_free = ""
                                    try:
                                        srt_s = api.download_salamandra_srt(sha1)
                                        if isinstance(srt_s, dict) and not srt_s.get("error"):
                                            salamandra_srt = srt_s.get("text", "") or ""
                                            sal_dir = base_media_dir / "Salamandra"
                                            sal_dir.mkdir(parents=True, exist_ok=True)
                                            sal_srt_path = sal_dir / "result.srt"
                                            with sal_srt_path.open("w", encoding="utf-8") as f_ss:
                                                f_ss.write(salamandra_srt)
                                            _log(f"[Salamandra] result.srt desat a {sal_srt_path}")
                                        else:
                                            _log(f"[Salamandra] Error descarregant SRT: {srt_s}")
                                    except Exception as e_ds:
                                        _log(f"[Salamandra] Excepció descarregant SRT: {e_ds}")

                                    try:
                                        free_s = api.download_salamandra_free_narration(sha1)
                                        if isinstance(free_s, dict) and not free_s.get("error"):
                                            salamandra_free = free_s.get("text", "") or ""
                                            sal_dir = base_media_dir / "Salamandra"
                                            sal_dir.mkdir(parents=True, exist_ok=True)
                                            sal_free_path = sal_dir / "free_narration.txt"
                                            with sal_free_path.open("w", encoding="utf-8") as f_sf:
                                                f_sf.write(salamandra_free)
                                            _log(f"[Salamandra] free_narration.txt desat a {sal_free_path}")
                                        else:
                                            _log(f"[Salamandra] Error descarregant free_narration: {free_s}")
                                    except Exception as e_df:
                                        _log(f"[Salamandra] Excepció descarregant free_narration: {e_df}")

                                    # Persistir a audiodescriptions.db
                                    try:
                                        upsert_audiodescription_text(
                                            sha1sum=sha1,
                                            version="Salamandra",
                                            une_ad=salamandra_srt or "",
                                            free_ad=salamandra_free or "",
                                        )
                                        any_success = True
                                    except Exception as db_exc:
                                        _log(f"[audiodescriptions] Error desant AD Salamandra: {db_exc}")

                                    # Esdeveniments
                                    try:
                                        if salamandra_srt:
                                            log_event(
                                                session=session_id,
                                                ip=ip,
                                                user=username or "",
                                                password=password or "",
                                                phone=phone,
                                                action="Salamandra AD generated",
                                                sha1sum=sha1,
                                                visibility=vis_flag,
                                            )
                                        if salamandra_free:
                                            log_event(
                                                session=session_id,
                                                ip=ip,
                                                user=username or "",
                                                password=password or "",
                                                phone=phone,
                                                action="Salamandra free AD generated",
                                                sha1sum=sha1,
                                                visibility=vis_flag,
                                            )
                                    except Exception as e_evt_s:
                                        _log(f"[events] Error registrant esdeveniments Salamandra: {e_evt_s}")

                            # 5) MoE
                            if moe_enabled:
                                progress_placeholder.info("🧠 Generant versió MoE...")
                                try:
                                    gen_resp_m = api.generate_moe_result(sha1)
                                    _log(f"[MoE] generate_moe_result resp: {gen_resp_m}")
                                except Exception as e_gen_m:
                                    _log(f"[MoE] Error cridant generate_moe_result: {e_gen_m}")
                                    gen_resp_m = {"error": str(e_gen_m)}

                                if isinstance(gen_resp_m, dict) and gen_resp_m.get("error"):
                                    _log(f"[MoE] Error en generació: {gen_resp_m.get('error')}")
                                else:
                                    moe_srt = ""
                                    moe_free = ""
                                    try:
                                        srt_m = api.download_moe_srt(sha1)
                                        if isinstance(srt_m, dict) and not srt_m.get("error"):
                                            moe_srt = srt_m.get("text", "") or ""
                                            moe_dir = base_media_dir / "MoE"
                                            moe_dir.mkdir(parents=True, exist_ok=True)
                                            moe_srt_path = moe_dir / "result.srt"
                                            with moe_srt_path.open("w", encoding="utf-8") as f_ms:
                                                f_ms.write(moe_srt)
                                            _log(f"[MoE] result.srt desat a {moe_srt_path}")
                                        else:
                                            _log(f"[MoE] Error descarregant SRT: {srt_m}")
                                    except Exception as e_dm_s:
                                        _log(f"[MoE] Excepció descarregant SRT: {e_dm_s}")

                                    try:
                                        free_m = api.download_moe_free_narration(sha1)
                                        if isinstance(free_m, dict) and not free_m.get("error"):
                                            moe_free = free_m.get("text", "") or ""
                                            moe_dir = base_media_dir / "MoE"
                                            moe_dir.mkdir(parents=True, exist_ok=True)
                                            moe_free_path = moe_dir / "free_narration.txt"
                                            with moe_free_path.open("w", encoding="utf-8") as f_mf:
                                                f_mf.write(moe_free)
                                            _log(f"[MoE] free_narration.txt desat a {moe_free_path}")
                                        else:
                                            _log(f"[MoE] Error descarregant free_narration: {free_m}")
                                    except Exception as e_dm_f:
                                        _log(f"[MoE] Excepció descarregant free_narration: {e_dm_f}")

                                    # Persistir a audiodescriptions.db
                                    try:
                                        upsert_audiodescription_text(
                                            sha1sum=sha1,
                                            version="MoE",
                                            une_ad=moe_srt or "",
                                            free_ad=moe_free or "",
                                        )
                                        any_success = True
                                    except Exception as db_exc_m:
                                        _log(f"[audiodescriptions] Error desant AD MoE: {db_exc_m}")

                                    # Esdeveniments
                                    try:
                                        if moe_srt:
                                            log_event(
                                                session=session_id,
                                                ip=ip,
                                                user=username or "",
                                                password=password or "",
                                                phone=phone,
                                                action="MoE AD generated",
                                                sha1sum=sha1,
                                                visibility=vis_flag,
                                            )
                                        if moe_free:
                                            log_event(
                                                session=session_id,
                                                ip=ip,
                                                user=username or "",
                                                password=password or "",
                                                phone=phone,
                                                action="MoE free AD generated",
                                                sha1sum=sha1,
                                                visibility=vis_flag,
                                            )
                                    except Exception as e_evt_m:
                                        _log(f"[events] Error registrant esdeveniments MoE: {e_evt_m}")

                            # 6) Refinement opcional sobre les versions generades
                            try:
                                refinement_active = bool(reflection_enabled or reflexion_enabled or introspection_enabled)
                                if refinement_active:
                                    # Guardar info_ad (info.json inicial) si el tenim
                                    if init_info_text and sha1:
                                        try:
                                            update_audiodescription_info_ad(
                                                sha1sum=sha1,
                                                version="Salamandra",
                                                info_ad=init_info_text,
                                            )
                                        except Exception:
                                            pass
                                        try:
                                            update_audiodescription_info_ad(
                                                sha1sum=sha1,
                                                version="MoE",
                                                info_ad=init_info_text,
                                            )
                                        except Exception:
                                            pass

                                    # Refinar Salamandra
                                    if salamandra_enabled and salamandra_srt:
                                        try:
                                            ref_resp_s = api.apply_refinement(
                                                sha1sum=sha1,
                                                version="Salamandra",
                                                srt_content=salamandra_srt,
                                                reflection_enabled=reflection_enabled,
                                                reflexion_enabled=reflexion_enabled,
                                                introspection_enabled=introspection_enabled,
                                            )
                                            _log(f"[Refinement] Salamandra resp: {ref_resp_s}")
                                            refined_srt = None
                                            if isinstance(ref_resp_s, dict):
                                                refined_srt = ref_resp_s.get("refined_srt") or ref_resp_s.get("refinedSrt")
                                            if refined_srt:
                                                update_audiodescription_text(
                                                    sha1sum=sha1,
                                                    version="Salamandra",
                                                    une_ad=refined_srt,
                                                )
                                                refined_any = True
                                                try:
                                                    import hashlib as _hashlib
                                                    srt_hash = _hashlib.sha1(refined_srt.encode("utf-8")).hexdigest()
                                                    log_event(
                                                        session=session_id,
                                                        ip=ip,
                                                        user=username or "",
                                                        password=password or "",
                                                        phone=phone,
                                                        action="Refined AD",
                                                        sha1sum=srt_hash,
                                                        visibility=vis_flag,
                                                    )
                                                except Exception as e_evt_ref_s:
                                                    _log(f"[events] Error registrant Refined AD (Salamandra): {e_evt_ref_s}")
                                        except Exception as e_ref_s:
                                            _log(f"[Refinement] Error refinant Salamandra: {e_ref_s}")

                                    # Refinar MoE
                                    if moe_enabled and moe_srt:
                                        try:
                                            ref_resp_m = api.apply_refinement(
                                                sha1sum=sha1,
                                                version="MoE",
                                                srt_content=moe_srt,
                                                reflection_enabled=reflection_enabled,
                                                reflexion_enabled=reflexion_enabled,
                                                introspection_enabled=introspection_enabled,
                                            )
                                            _log(f"[Refinement] MoE resp: {ref_resp_m}")
                                            refined_srt_m = None
                                            if isinstance(ref_resp_m, dict):
                                                refined_srt_m = ref_resp_m.get("refined_srt") or ref_resp_m.get("refinedSrt")
                                            if refined_srt_m:
                                                update_audiodescription_text(
                                                    sha1sum=sha1,
                                                    version="MoE",
                                                    une_ad=refined_srt_m,
                                                )
                                                refined_any = True
                                                try:
                                                    import hashlib as _hashlib
                                                    srt_hash_m = _hashlib.sha1(refined_srt_m.encode("utf-8")).hexdigest()
                                                    log_event(
                                                        session=session_id,
                                                        ip=ip,
                                                        user=username or "",
                                                        password=password or "",
                                                        phone=phone,
                                                        action="Refined AD",
                                                        sha1sum=srt_hash_m,
                                                        visibility=vis_flag,
                                                    )
                                                except Exception as e_evt_ref_m:
                                                    _log(f"[events] Error registrant Refined AD (MoE): {e_evt_ref_m}")
                                        except Exception as e_ref_m:
                                            _log(f"[Refinement] Error refinant MoE: {e_ref_m}")
                            except Exception as e_ref:
                                _log(f"[Refinement] Error global de refinement: {e_ref}")

                            # 7) Enviament opcional d'SMS per a validació UNE i event 'Waiting for UNE validation'
                            try:
                                if any_success and refined_any and sha1:
                                    sms_channels_enabled = bool(twilio_enabled_cfg or zapier_enabled_cfg)
                                    if sms_channels_enabled and une_validator_sms_enabled and une_phone_validator:
                                        try:
                                            # Text de l'SMS en català, tal com has indicat
                                            sms_msg = "Noves audiodescripcions a validar segons la norma UNE-153020"
                                            sms_ok = compliance_client.notify_une_validator_new_ads(
                                                phone=une_phone_validator,
                                                message=sms_msg,
                                            )
                                            if sms_ok:
                                                try:
                                                    from databases import log_action

                                                    log_action(
                                                        session=session_id or "",
                                                        user=username or "",
                                                        phone=str(une_phone_validator or ""),
                                                        action="SMS sent to UNE validator",
                                                        sha1sum=sha1 or "",
                                                    )
                                                except Exception:
                                                    pass
                                        except Exception as e_sms_call:
                                            _log(f"[UNE SMS] Error cridant compliance per UNE: {e_sms_call}")

                                        # Registrar estat d'espera de validació UNE a actions.db
                                        try:
                                            from databases import log_action

                                            log_action(
                                                session=session_id or "",
                                                user=username or "",
                                                phone=str(une_phone_validator or ""),
                                                action="Waiting for UNE validation",
                                                sha1sum=sha1 or "",
                                            )
                                        except Exception as e_evt_wait:
                                            _log(f"[actions] Error registrant Waiting for UNE validation: {e_evt_wait}")
                            except Exception as e_sms:
                                _log(f"[UNE SMS] Error en flux d'SMS/espera validació: {e_sms}")

                            # 8) Actualitzar status del vídeo a 'UNE-pending' a videos.db
                            try:
                                if any_success and sha1:
                                    update_video_status(sha1, "UNE-pending")
                            except Exception as e_upd_status:
                                _log(f"[videos] Error actualitzant status a 'UNE-pending': {e_upd_status}")

                            # 9) Invocar Space TTS per generar free_ad.mp3 i une_ad.mp4 a temp/media/<sha1>/Original
                            try:
                                if any_success and sha1:
                                    # Obtenir el text UNE més recent des d'audiodescriptions.db (prioritzem Salamandra)
                                    une_text = ""
                                    row_s = get_audiodescription(sha1, "Salamandra")
                                    if row_s is not None:
                                        try:
                                            une_text = (row_s["une_ad"] or "").strip()
                                        except Exception:
                                            une_text = ""
                                    if not une_text:
                                        row_m = get_audiodescription(sha1, "MoE")
                                        if row_m is not None:
                                            try:
                                                une_text = (row_m["une_ad"] or "").strip()
                                            except Exception:
                                                une_text = ""

                                    if une_text:
                                        base_media_dir = Path(__file__).parent.parent / "temp" / "media" / sha1
                                        video_path = base_media_dir / "video.mp4"
                                        if not video_path.exists():
                                            # Assegurar que tenim la media localment
                                            try:
                                                ensure_media_for_video(Path(__file__).parent.parent, api, sha1)
                                            except Exception as e_em:
                                                _log(f"[TTS] Error assegurant media per al vídeo: {e_em}")

                                        if video_path.exists():
                                            # Preparar carpeta de sortida Original
                                            original_dir = base_media_dir / "Original"
                                            original_dir.mkdir(parents=True, exist_ok=True)

                                            # Escriure SRT temporal i cridar Space TTS (/tts/srt)
                                            tts_url = os.getenv("API_TTS_URL", "").strip()
                                            if tts_url:
                                                try:
                                                    with tempfile.TemporaryDirectory(prefix="tts_srt_") as td:
                                                        td_path = Path(td)
                                                        srt_tmp = td_path / "ad_input.srt"
                                                        srt_tmp.write_text(une_text, encoding="utf-8")

                                                        files = {
                                                            "srt": ("ad_input.srt", srt_tmp.open("rb"), "text/plain"),
                                                            "video": ("video.mp4", video_path.open("rb"), "video/mp4"),
                                                        }
                                                        data = {
                                                            "voice": "central/grau",
                                                            "ad_format": "mp3",
                                                            "include_final_mp4": "1",
                                                        }

                                                        resp = requests.post(
                                                            f"{tts_url.rstrip('/')}/tts/srt",
                                                            files=files,
                                                            data=data,
                                                            timeout=300,
                                                        )
                                                        resp.raise_for_status()

                                                        # La resposta és un ZIP amb ad_master.(mp3|wav), mix i opcionalment video_con_ad.mp4
                                                        zip_bytes = resp.content
                                                        with zipfile.ZipFile(io.BytesIO(zip_bytes)) as zf:
                                                            for member in zf.infolist():
                                                                name = member.filename
                                                                lower = name.lower()
                                                                if lower.endswith("ad_master.mp3"):
                                                                    target = original_dir / "free_ad.mp3"
                                                                    with zf.open(member) as src, target.open("wb") as dst:
                                                                        shutil.copyfileobj(src, dst)
                                                                elif lower.endswith("video_con_ad.mp4"):
                                                                    target = original_dir / "une_ad.mp4"
                                                                    with zf.open(member) as src, target.open("wb") as dst:
                                                                        shutil.copyfileobj(src, dst)
                                                except Exception as e_tts:
                                                    _log(f"[TTS] Error generant assets TTS (free_ad.mp3/une_ad.mp4): {e_tts}")
                                            else:
                                                _log("[TTS] API_TTS_URL no configurada; s'omet la generació de free_ad.mp3/une_ad.mp4")
                                    else:
                                        _log("[TTS] No s'ha trobat text UNE per al vídeo; s'omet la generació TTS")
                            except Exception as e_tts_global:
                                _log(f"[TTS] Error global al flux TTS: {e_tts_global}")

                            # 10) Registrar acció "AD generated" a actions.db per a aquest vídeo
                            try:
                                if any_success and sha1:
                                    session_id_actions = session_id
                                    actions_user, actions_phone = get_latest_user_phone_for_session(session_id_actions)
                                    if not actions_user:
                                        actions_user = username or ""
                                    if not actions_phone:
                                        actions_phone = phone or ""

                                    insert_action(
                                        session=session_id_actions,
                                        user=actions_user,
                                        phone=actions_phone,
                                        action="AD generated",
                                        sha1sum=sha1,
                                    )
                            except Exception as e_act:
                                _log(f"[actions] Error registrant acció 'AD generated': {e_act}")

                            if any_success:
                                progress_placeholder.success("✅ Audiodescripció generada i desada. Ara està pendent de validació UNE.")
                                result_placeholder.info("La teva audiodescripció s'està generant i queda pendent de validació. Pots sortir de la sessió guardant els canvis i tornar més endavant per revisar el resultat.")
                            else:
                                progress_placeholder.empty()
                                result_placeholder.error("❌ No s'ha pogut generar cap versió d'audiodescripció.")

                        except Exception as e:
                            progress_placeholder.empty()
                            result_placeholder.error(f"❌ Excepció durant la generació: {e}")