app.py

import gradio as gr
import os
import torch
import commons
import utils
from models import SynthesizerTrn
import numpy as np
import json
import shutil
import logging
import random
import re
from huggingface_hub import snapshot_download

# --- 1. SETUP LOGGING ---
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
logger = logging.getLogger(__name__)

# --- 2. ROMAJI CONVERTER ---
try:
    import pykakasi
    kks = pykakasi.kakasi()
    def to_romaji(text):
        if not text or text == "None": return ""
        try:
            result = kks.convert(str(text))
            return "".join([item['hepburn'].capitalize() for item in result])
        except:
            return str(text)
except:
    def to_romaji(text): return str(text)

# --- 3. CLEAN HARDCODED INFO ---
CLEAN_INFO = {
  "0": {"title": "Sanoba Witch & Senren Banka", "example": "こんにちは。", "type": "vits"},
  "1": {"title": "Hamidashi Creative", "example": "こんにちは。", "type": "vits"},
  "2": {"title": "Cafe Stella & Shinigami no Chou", "example": "こんにちは。", "type": "vits"},
  "3": {"title": "Yosuga no Sora", "example": "こんにちは。", "type": "vits"},
  "4": {"title": "Bishoujo Mangekyou", "example": "こんにちは。", "type": "vits"},
  "5": {"title": "Nene & Nanami Pack (Multi)", "example": "[JA]こんにちは。[JA]", "type": "vits"},
  "6": {"title": "The Fox Waiting for You", "example": "안녕하세요.", "type": "vits"},
  "7": {"title": "Galgame Characters Pack (13)", "example": "こんにちは。", "type": "vits"},
  "8": {"title": "Zero no Tsukaima", "example": "こんにちは。", "type": "vits"},
  "9": {"title": "Zero no Tsukaima (VC Mode)", "example": "", "type": "soft-vits-vc"},
  "10": {"title": "Toaru Majutsu no Index (VC)", "example": "", "type": "soft-vits-vc"},
  "11": {"title": "Shiki Natsume (VC Mode)", "example": "", "type": "soft-vits-vc"},
  "12": {"title": "DRACU-RIOT!", "example": "こんにちは。", "type": "vits"},
  "13": {"title": "To LOVE-Ru Series", "example": "こんにちは。", "type": "vits"},
  "14": {"title": "CJKS Multi-Language", "example": "[JA]こんにちは。[JA]", "type": "vits"},
  "15": {"title": "Voistock Mega Pack (2891 Chars)", "example": "[JA]こんにちは。[JA]", "type": "vits"},
  "16": {"title": "Shanghainese Dialect", "example": "侬好！", "type": "vits"},
  "17": {"title": "Chinese Dialects Pack", "example": "[SH]侬好！[SH]", "type": "vits"},
  "18": {"title": "Umamusume: Pretty Derby", "example": "こんにちは。", "type": "vits"},
  "19": {"title": "Princess Connect! Re:Dive", "example": "[JA]こんにちは。[JA]", "type": "vits"},
  "20": {"title": "Magia Record (Madoka Magica)", "example": "こんにちは。", "type": "vits"}
}

# --- 4. DOWNLOAD ASSETS ---
REPO_ID = "Plana-Archive/Plana-TTS"
LOCAL_ROOT = "saved_model"

def download_assets():
    os.makedirs(LOCAL_ROOT, exist_ok=True)
    if not os.path.exists(os.path.join(LOCAL_ROOT, "0")):
        try:
            logger.info("Downloading Assets...")
            snapshot_download(repo_id=REPO_ID, local_dir="temp_dir", allow_patterns=["MOE-TTS/saved_model/*"])
            src_path = os.path.join("temp_dir", "MOE-TTS", "saved_model")
            if os.path.exists(src_path):
                shutil.copytree(src_path, LOCAL_ROOT, dirs_exist_ok=True)
                shutil.rmtree("temp_dir")
        except Exception as e:
            logger.error(f"Download error: {e}")

download_assets()

# --- 5. ENGINE LOAD MODEL ---
loaded_models = {}

def clean_config(conf):
    if isinstance(conf, dict): return {str(k): clean_config(v) for k, v in conf.items()}
    elif isinstance(conf, list): return [clean_config(i) for i in conf]
    return conf

def get_vits_model(m_id):
    mid = str(m_id)
    if mid in loaded_models: return loaded_models[mid]
    try:
        p = os.path.join(LOCAL_ROOT, mid)
        cfg_p = os.path.join(p, "config.json")
        if not os.path.exists(cfg_p): return None
        hps = utils.get_hparams_from_file(cfg_p)
        m_params = clean_config(hps.model.__dict__ if hasattr(hps.model, '__dict__') else dict(hps.model))
        net = SynthesizerTrn(len(hps.symbols), hps.data.filter_length // 2 + 1, hps.train.segment_size // hps.data.hop_length, n_speakers=hps.data.n_speakers, **m_params)
        utils.load_checkpoint(os.path.join(p, "model.pth"), net, None)
        net.eval()
        raw_spks = hps.speakers if hasattr(hps, 'speakers') else [f"Character {i}" for i in range(hps.data.n_speakers)]
        
        display_spks = []
        original_spks = [] 
        for s in raw_spks:
            romaji_name = to_romaji(s)
            if romaji_name and romaji_name.lower() != "none":
                display_spks.append(romaji_name)
                original_spks.append(s)
                
        loaded_models[mid] = (hps, net, display_spks, original_spks)
        return loaded_models[mid]
    except Exception as e:
        logger.error(f"Load Error {mid}: {e}")
        return None

def tts_execute(m_id, text, speaker_romaji, speed):
    data = get_vits_model(m_id)
    if not data: return "❌ Model Loading...", None
    hps, net, display_spks, _ = data
    if not speaker_romaji: 
        if display_spks:
            speaker_romaji = display_spks[0]
        else:
            return "❌ No Speaker Selected", None
    try:
        sid = display_spks.index(speaker_romaji)
        from text import text_to_sequence
        clean_text = re.sub(r'\[[A-Z]{2}\]', '', text)
        cleaners = hps.data.text_cleaners if hasattr(hps.data, 'text_cleaners') else ['japanese_cleaners']
        seq = text_to_sequence(clean_text, hps.symbols, cleaners)
        if hps.data.add_blank: seq = commons.intersperse(seq, 0)
        with torch.no_grad():
            audio = net.infer(torch.LongTensor(seq).unsqueeze(0), torch.LongTensor([len(seq)]), sid=torch.LongTensor([sid]), noise_scale=0.667, noise_scale_w=0.8, length_scale=1.0/speed)[0][0,0].data.cpu().float().numpy()
        return f"✅ Done!", (hps.data.sampling_rate, (audio * 32767).astype(np.int16))
    except Exception as e: return f"Error: {e}", None

def get_random_jp():
    return random.choice(["こんにちは！", "お元気ですか？", "先生、お疲れ様です！", "大好きだよ！", "また明日ね。"])

def get_char_info_html(m_id):
    data = get_vits_model(m_id)
    if not data: return ""
    _, _, _, original_names = data
    
    html = f"""<div class="info-content-area" id="info-content-{m_id}" style="display: grid; grid-template-columns: 1fr 1fr; gap: 10px; padding: 15px; background: white; border: 1px solid #f0f4f8; border-top: none; border-radius: 0 0 10px 10px;">"""
    for i, name in enumerate(original_names):
        html += f"""
        <div style="border: 1px solid #f0f4f8; border-radius: 12px; padding: 10px; border-left: 5px solid #1299ff; background: #fff;">
            <div style="font-weight: 800; color: #2d3748; font-size: 14px; margin-bottom: 2px;">{name}</div>
            <div style="color: #a0aec0; font-size: 11px;">Character {i+1}</div>
        </div>
        """
    html += "</div>"
    return html

# --- 6. UI STYLE ---
css = """
:root { 
    --primary-600: #1299ff !important; 
    --accent-600: #1299ff !important;
}

.gradio-container, .gradio-container * {
    --loader-color: #A2D2FF !important;
}
.loading { border-top-color: #A2D2FF !important; }
svg.loading { fill: #A2D2FF !important; }

/* Judul Header */
.ba-header-container {
    border: 1.5px solid #e1e8f0;
    border-radius: 12px;
    padding: 20px 10px;
    margin-bottom: 12px;
    background: white;
    text-align: center;
}
.ba-header-container h1 {
    color: #1299ff !important;
    font-weight: 700 !important;
    font-size: 42px !important;
    margin: 0;
    line-height: 1.1;
}
.ba-header-container p {
    color: #8a99af;
    font-size: 11px;
    font-weight: 800;
    letter-spacing: 2px;
    margin: 8px 0 0 0;
    text-transform: uppercase;
}

/* System Status Box */
.status-container {
    border: 1.5px solid #e1e8f0;
    border-radius: 12px;
    padding: 15px 22px;
    margin-bottom: 20px;
    background: white;
}
.status-title {
    color: #1299ff !important; /* Biru sesuai permintaan */
    font-weight: 800;
    font-size: 16px;
    margin-bottom: 8px;
}
.status-item {
    display: flex;
    align-items: center;
    gap: 10px;
    font-size: 14px;
    font-weight: 700;
    margin-bottom: 4px;
}
.text-dark-gray {
    color: #4a5568 !important; /* Abu-abu Hitam */
}
.text-green-bold {
    color: #28a745 !important; /* Hijau Tebal */
    font-weight: 900 !important;
    text-shadow: none !important; /* Tanpa bayangan */
}
.text-blue-status {
    color: #1299ff !important; /* Biru */
}
.status-dot {
    width: 8px;
    height: 8px;
    background-color: #1299ff;
    border-radius: 50%;
}
.pulse-dot {
    width: 8px;
    height: 8px;
    background-color: #1299ff;
    border-radius: 50%;
    position: relative;
}
.pulse-dot::after {
    content: "";
    position: absolute;
    width: 100%;
    height: 100%;
    background-color: #1299ff;
    border-radius: 50%;
    animation: pulse 2.5s infinite;
}
@keyframes pulse {
    0% { transform: scale(1); opacity: 0.8; }
    100% { transform: scale(3); opacity: 0; }
}

.slim-card { max-width: 480px; margin: 0 auto; background: transparent; padding: 10px; }
.tabs > .tab-nav {
    display: flex !important;
    overflow-x: auto !important;
    white-space: nowrap !important;
    flex-wrap: nowrap !important;
}
.tabs > .tab-nav button { flex: 0 0 auto !important; }
.scroll-box { height: 200px; overflow-y: auto; border: 1px solid #f0f4f8; border-radius: 12px; padding: 10px; background: #fafbfc; margin-bottom: 10px; }
.char-btn { background: white !important; border: 1px solid #e2e8f0 !important; border-left: 5px solid #1299ff !important; text-align: left !important; padding: 8px !important; font-size: 12px !important; margin-bottom: 4px !important; width: 100%; color: #4a5568 !important; }
.char-btn:hover { background: #f0f7ff !important; border-color: #1299ff !important; }

.warning-card { background: #fff9f0; border: 2px dashed #f5a623; border-radius: 10px; padding: 12px; margin-bottom: 15px; text-align: center; }
.warning-title { color: #f5a623; font-weight: 800; font-size: 13px; margin-bottom: 4px; display: flex; align-items: center; justify-content: center; gap: 5px; }
.warning-text { color: #855d1a; font-size: 11px; line-height: 1.5; font-weight: 600; }

.jp-btn { background: #f8fafc !important; border: 1px solid #cbd5e1 !important; color: #475569 !important; font-weight: 700 !important; border-radius: 10px !important; margin-bottom: 10px; font-size: 12px !important; width: 100%; }
.gen-btn { background: #1299ff !important; color: white !important; font-weight: 700 !important; border-radius: 12px !important; height: 45px !important; width: 100%; border: none !important; cursor: pointer; transition: 0.3s; }
.info-header-custom { background: #1299ff !important; color: white !important; border: none !important; border-radius: 8px 8px 0 0 !important; padding: 12px 15px !important; width: 100% !important; cursor: pointer; display: flex !important; justify-content: center !important; align-items: center !important; font-weight: 800 !important; font-size: 14px !important; margin-top: 15px; gap: 15px; }
.credit-footer { margin-top: 25px; padding: 15px; background: white; border-radius: 12px; text-align: center; border: 1px solid #eef2f6; border-bottom: 4px solid #1299ff; color: #94a3b8; font-weight: 700; font-size: 12px; letter-spacing: 2px; }
"""

with gr.Blocks(css=css) as demo:
    with gr.Column(elem_classes="slim-card"):
        # Header Blue Archive
        gr.HTML("""
            <div class="ba-header-container">
                <h1>Library Anime</h1>
                <p>🍂 Style-Bert-VITS2 🍂</p>
            </div>
            
            <div class="status-container">
                <div class="status-title">System Status</div>
                <div class="status-item">
                    <div class="status-dot"></div> 
                    <span class="text-dark-gray">Model :</span> 
                    <span class="text-green-bold">&nbsp;LOADED ✅</span>
                </div>
                <div class="status-item">
                    <div class="pulse-dot"></div> 
                    <span class="text-dark-gray">System :</span> 
                    <span class="text-blue-status">&nbsp;Online</span>
                </div>
            </div>
        """)
        
        with gr.Tabs(elem_classes="tabs"):
            for m_id in sorted(CLEAN_INFO.keys(), key=int):
                with gr.Tab(f"Model {m_id}"):
                    gr.Markdown(f"### 📂 {CLEAN_INFO[m_id]['title']}")
                    
                    m_data = get_vits_model(m_id)
                    chars = m_data[2] if m_data else []
                    
                    m_p = os.path.join(LOCAL_ROOT, str(m_id))
                    cov = next((os.path.join(m_p, f"cover.{e}") for e in ['png','jpg','jpeg','webp'] if os.path.exists(os.path.join(m_p, f"cover.{e}"))), None)
                    if cov: gr.Image(cov, show_label=False, interactive=False, height=140)
                    
                    sel_name = gr.State("")
                    char_display = gr.Markdown("📍 *Silakan pilih karakter...*")
                    
                    gr.HTML("<p style='font-weight:800; font-size:11px; color:#8a99af; margin-bottom:8px;'>CHARACTER LIST (ROMAJI)</p>")
                    
                    with gr.Column(elem_classes="scroll-box"):
                        if chars:
                            for name in chars:
                                btn = gr.Button(f"👤 {name}", elem_classes="char-btn")
                                btn.click(fn=lambda n=name: (n, f"📍 Selected: **{n}**"), outputs=[sel_name, char_display])
                    
                    gr.HTML("""
                        <div class="warning-card">
                            <div class="warning-title">🔖 PERINGATAN MINNA 🔖</div>
                            <div class="warning-text">
                                Model 19 is not working. Cara pakai VITS ini klik aja character lalu masukkan input text dan Generate Voice! Done ✨
                            </div>
                        </div>
                    """)
                    
                    ex_text = re.sub(r'\[[A-Z]{2}\]', '', CLEAN_INFO[m_id].get("example", "こんにちは。"))
                    txt_in = gr.TextArea(label="Input Text", value=ex_text, lines=3)
                    gr.Button("🎲 INPUTS RANDOM TEXT 🎲", elem_classes="jp-btn").click(get_random_jp, outputs=[txt_in])
                    
                    spd = gr.Slider(0.5, 2.0, 1.0, step=0.1, label="Speed Audio")
                    btn_gen = gr.Button("🎐 GENERATE VOICE 🎐", elem_classes="gen-btn")
                    aud_out = gr.Audio(label="Voice Output")
                    
                    gr.HTML(f"""
                        <button onclick="
                            const content = document.getElementById('info-content-{m_id}');
                            content.style.display = (content.style.display === 'none') ? 'grid' : 'none';
                        " class="info-header-custom">
                            📑 Character Information 📑 &nbsp; ▼
                        </button>
                    """)
                    gr.HTML(get_char_info_html(m_id))
                    
                    status_log = gr.Textbox(visible=False)
                    btn_gen.click(fn=tts_execute, inputs=[gr.State(m_id), txt_in, sel_name, spd], outputs=[status_log, aud_out])

        gr.HTML("""<div class="credit-footer">🌥️ CREATED BY MUTSUMI 🌥️</div>""")

if __name__ == "__main__":
    demo.launch()