app.py

import gradio as gr
import edge_tts
import asyncio
import re
import os
import uuid
from pydub import AudioSegment

# ---------------------------------------------------------
# 1. إعدادات المشاعر
# ---------------------------------------------------------
EMOTION_PRESETS = {
    "محايد": {"rate": "+0%", "pitch": "+0Hz"},
    "سعيد 😊": {"rate": "+15%", "pitch": "+10Hz"},
    "حزين 😢": {"rate": "-15%", "pitch": "-10Hz"},
    "غاضب 😠": {"rate": "+20%", "pitch": "+5Hz"},
    "هادئ 😌": {"rate": "-10%", "pitch": "+0Hz"},
    "متحمس 🎉": {"rate": "+30%", "pitch": "+15Hz"}
}

# ---------------------------------------------------------
# 2. وظائف المساعدة والتنظيف
# ---------------------------------------------------------
def sanitize_text(text, ignored_chars_str):
    """
    [ACEE] ENHANCED DYNAMIC SANITIZATION
    تنظيف النص من الرموز التي يحددها المستخدم بالإضافة إلى الإيموجي.
    """
    if not text:
        return ""
    
    # 1. تحليل قائمة الرموز المتجاهلة من المدخلات
    # مثال المدخلات: "*, #, $" -> ["*", "#", "$"]
    if ignored_chars_str:
        # تقسيم النص بالفواصل أو المسافات
        chars_to_remove = [char.strip() for char in ignored_chars_str.split(',') if char.strip()]
        
        for char in chars_to_remove:
            # استخدام replace العادي للأحرف البسيطة لتجنب مشاكل Regex الخاصة
            text = text.replace(char, "")
    
    # 2. إزالة الإيموجي (نطاق Unicode واسع)
    emoji_pattern = re.compile("["
        u"\U0001F600-\U0001F64F"  # Emoticons
        u"\U0001F300-\U0001F5FF"  # Symbols & Pictographs
        u"\U0001F680-\U0001F6FF"  # Transport & Map Symbols
        u"\U0001F1E0-\U0001F1FF"  # Flags (iOS)
        u"\U00002700-\U000027BF"  # Dingbats
        u"\u2600-\u26FF"          # Misc symbols
        u"\u2700-\u27BF"          # Dingbats
        "]+", flags=re.UNICODE)
    
    text = emoji_pattern.sub(r'', text)
    return text.strip()

async def generate_segment(text, voice, rate, pitch, filename):
    """توليد جزء صوتي واحد"""
    if not text or not text.strip():
        return False
    try:
        communicate = edge_tts.Communicate(text, voice, rate=rate, pitch=pitch)
        await communicate.save(filename)
        return True
    except Exception as e:
        print(f"Error generating segment '{text}': {e}")
        return False

# ---------------------------------------------------------
# 3. المنطق الرئيسي
# ---------------------------------------------------------
async def text_to_speech_edge(text, character, emotion, use_advanced, manual_rate, manual_pitch, ignored_chars):
    # [ACEE] 1. Sanitize Input using the dynamic list
    text = sanitize_text(text, ignored_chars)
    
    if not text or text.strip() == "":
        return None

    # [ACEE] 2. Unique Session ID
    session_id = str(uuid.uuid4())

    # إعداد الصوت
    voice_map = {
        "رجل (مصري)": "ar-EG-ShakirNeural",
        "سيدة (مصرية)": "ar-EG-SalmaNeural",
        "طفل (محاكاة)": "ar-EG-SalmaNeural",
        "رجل (سعودي)": "ar-SA-HamedNeural",
        "سيدة (سعودية)": "ar-SA-ZariyahNeural"
    }
    voice = voice_map.get(character, "ar-EG-SalmaNeural")

    # إعداد Rate & Pitch
    if use_advanced:
        rate_percent = int((manual_rate - 1.0) * 100)
        rate = f"{rate_percent:+d}%"
        pitch = manual_pitch
    else:
        preset = EMOTION_PRESETS.get(emotion, EMOTION_PRESETS["محايد"])
        rate = preset["rate"]
        pitch = preset["pitch"]
    
    if character == "طفل (محاكاة)":
        pitch = "+50Hz"

    # ------------------------------------------------------------------
    # 🔥 REGEX MAGIC: التعرف على كل الأنماط
    # ------------------------------------------------------------------
    pattern = r"(\[pause:[\d\.]+\]|<break\s+time=[\"'][\d\.]+s[\"']\s*/>|\.\.\.)"
    
    parts = re.split(pattern, text)
    
    segments_to_generate = []
    temp_files_cleanup = []
    timeline = []

    print(f"Session {session_id}: Detected {len(parts)} parts")

    for i, part in enumerate(parts):
        part = part.strip()
        if not part:
            continue

        pause_duration = 0.0
        is_pause = False

        # --- تحليل النمط ---
        
        # الحالة 1: [pause:2.5]
        if part.startswith("[pause:") and part.endswith("]"):
            try:
                val = part.split(":")[1].strip("]")
                pause_duration = float(val)
                is_pause = True
            except:
                pass

        # الحالة 2: <break time="4s"/>
        elif part.startswith("<break") and "time=" in part:
            try:
                match = re.search(r"time=[\"']([\d\.]+)s[\"']", part)
                if match:
                    pause_duration = float(match.group(1))
                    is_pause = True
            except:
                pass
        
        # الحالة 3: ... (ثلاث نقاط)
        elif part == "...":
            pause_duration = 1.0 
            is_pause = True

        # --- جدولة المهام ---
        if is_pause:
            if pause_duration > 0:
                timeline.append({'type': 'silence', 'duration': pause_duration})
        else:
            temp_filename = f"temp_{session_id}_{i}.mp3"
            temp_files_cleanup.append(temp_filename)
            timeline.append({'type': 'audio', 'file': temp_filename})
            segments_to_generate.append(
                generate_segment(part, voice, rate, pitch, temp_filename)
            )

    # معالجة متوازية
    if segments_to_generate:
        await asyncio.gather(*segments_to_generate)

    # الدمج
    final_audio = AudioSegment.empty()
    
    for item in timeline:
        if item['type'] == 'silence':
            final_audio += AudioSegment.silent(duration=item['duration'] * 1000)
        
        elif item['type'] == 'audio':
            try:
                if os.path.exists(item['file']):
                    segment = AudioSegment.from_mp3(item['file'])
                    final_audio += segment
                else:
                    print(f"Warning: Missing file {item['file']}")
            except Exception as e:
                print(f"Merge Error: {e}")

    output_file = f"output_{session_id}.mp3"
    try:
        final_audio.export(output_file, format="mp3")
    except:
        return None

    # تنظيف
    for f in temp_files_cleanup:
        if os.path.exists(f):
            try: os.remove(f)
            except: pass

    return output_file

# ---------------------------------------------------------
# 4. الواجهة
# ---------------------------------------------------------
css = """
footer {visibility: hidden}
.gradio-container {direction: rtl}
"""

with gr.Blocks(title="🎙️ TTS Pro Max", theme=gr.themes.Soft(), css=css) as demo:
    gr.Markdown("# 🎙️ TTS Pro: دعم كامل للتوقفات")
    
    with gr.Row():
        with gr.Column(scale=2):
            text_input = gr.Textbox(
                label="📝 النص",
                lines=5,
                placeholder="اكتب نصك هنا... استخدم [pause:N] للتوقف",
                text_align="right"
            )
            
            # [ACEE] New Field for Ignored Characters
            ignored_chars_input = gr.Textbox(
                label="🚫 أحرف للتجاهل (افصل بينها بفاصلة)",
                value="*, #, _",
                placeholder="مثال: *, #, $",
                text_align="right"
            )
            
            with gr.Row():
                voice_selector = gr.Radio(
                    ["رجل (مصري)", "سيدة (مصرية)", "طفل (محاكاة)", "رجل (سعودي)", "سيدة (سعودية)"],
                    label="الصوت", value="رجل (سعودي)"
                )
                emotion_selector = gr.Radio(
                    list(EMOTION_PRESETS.keys()),
                    label="الشعور", value="محايد"
                )
            
            with gr.Accordion("⚙️ إعدادات", open=True): 
                use_advanced = gr.Checkbox(label="يدوي", value=True) 
                manual_rate = gr.Slider(0.5, 2.0, 0.7, label="السرعة") 
                manual_pitch = gr.Textbox("+10Hz", label="Pitch") 

            btn = gr.Button("🎬 توليد", variant="primary")

        with gr.Column(scale=1):
            audio_out = gr.Audio(label="النتيجة")
            gr.Markdown("""
            ### ℹ️ مميزات النظام:
            - **قائمة التجاهل**: يمكنك الآن تحديد الرموز التي تريد حذفها (مثل `*`, `#`) من الحقل الجديد.
            - **توقفات دقيقة**: دعم `[pause:N]`، `<break time="Ns"/>`، و `...`.
            """)

    btn.click(
        text_to_speech_edge, 
        inputs=[text_input, voice_selector, emotion_selector, use_advanced, manual_rate, manual_pitch, ignored_chars_input], 
        outputs=audio_out
    )

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