app.py

# FILE: app.py
import gradio as gr
import os
import asyncio
from openai import AsyncOpenAI
from functools import partial 
import datetime 

from ui_layout import (
    create_main_input_components, create_speaker_config_components,
    create_action_and_output_components, create_examples_ui,
    TTS_MODELS_AVAILABLE, MODEL_DEFAULT_ENV, APP_AVAILABLE_VOICES, 
    DEFAULT_GLOBAL_VOICE, VIBE_CHOICES, DEFAULT_VIBE, PREDEFINED_VIBES
)

from event_handlers import (
    handle_script_processing, handle_calculate_cost,
    handle_speaker_config_method_visibility_change, 
    handle_tts_model_change, 
    handle_load_refresh_per_speaker_ui_trigger,
    handle_dynamic_accordion_input_change, # For controls inside @gr.render
    get_speakers_from_script
)

OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
NSFW_API_URL_TEMPLATE = os.getenv("NSFW_API_URL_TEMPLATE")
MODEL_DEFAULT_FROM_ENV = os.getenv("MODEL_DEFAULT", MODEL_DEFAULT_ENV)
EFFECTIVE_MODEL_DEFAULT = MODEL_DEFAULT_FROM_ENV if MODEL_DEFAULT_FROM_ENV in TTS_MODELS_AVAILABLE else MODEL_DEFAULT_ENV
async_openai_client = None

if not OPENAI_API_KEY:
    try:
        from huggingface_hub import HfApi
        api = HfApi()
        space_id = os.getenv("SPACE_ID") 
        if space_id:
            secrets = api.get_space_secrets(repo_id=space_id)
            OPENAI_API_KEY = secrets.get("OPENAI_API_KEY")
            NSFW_API_URL_TEMPLATE = secrets.get("NSFW_API_URL_TEMPLATE", NSFW_API_URL_TEMPLATE)
            MODEL_DEFAULT_FROM_HUB = secrets.get("MODEL_DEFAULT", EFFECTIVE_MODEL_DEFAULT)
            EFFECTIVE_MODEL_DEFAULT = MODEL_DEFAULT_FROM_HUB if MODEL_DEFAULT_FROM_HUB in TTS_MODELS_AVAILABLE else EFFECTIVE_MODEL_DEFAULT
            print("Loaded secrets from Hugging Face Hub.")
    except Exception as e:
        print(f"Could not retrieve secrets from Hugging Face Hub: {e}. OPENAI_API_KEY might be missing.")

if OPENAI_API_KEY:
    async_openai_client = AsyncOpenAI(api_key=OPENAI_API_KEY)
else:
    print("CRITICAL ERROR: OPENAI_API_KEY secret is not set. The application will not function properly.")


with gr.Blocks(theme=gr.themes.Soft(), elem_id="main_blocks_ui") as demo:
    gr.Markdown("# Dialogue Script to Speech (OpenAI TTS) - Using @gr.render")
    if not OPENAI_API_KEY or not async_openai_client:
        gr.Markdown("<h3 style='color:red;'>⚠️ Warning: OPENAI_API_KEY not set or invalid. Audio generation will fail. Please configure it in your Space settings.</h3>")

    speaker_configs_state = gr.State({}) 

    (script_input, tts_model_dropdown, pause_input,
     global_speed_input, global_instructions_input) = create_main_input_components(EFFECTIVE_MODEL_DEFAULT)

    (speaker_config_method_dropdown, single_voice_group, global_voice_dropdown,
     detailed_per_speaker_ui_group_container,
     load_per_speaker_ui_button) = create_speaker_config_components()

    (calculate_cost_button, generate_button, cost_output,
     individual_lines_zip_output, merged_dialogue_mp3_output,
     status_output) = create_action_and_output_components()

    with detailed_per_speaker_ui_group_container: # Define @gr.render inside this column
        @gr.render(
            inputs=[script_input, speaker_configs_state, tts_model_dropdown],
            triggers=[load_per_speaker_ui_button.click, tts_model_dropdown.change]
        )
        def render_dynamic_speaker_ui(current_script_text: str, current_speaker_configs: dict, current_tts_model: str):
            """Defines and re-renders the dynamic UI for per-speaker configuration."""
            print(f"DEBUG: @gr.render CALLED. Model: {current_tts_model}. Script: '{current_script_text[:30]}...'. State Keys: {list(current_speaker_configs.keys()) if isinstance(current_speaker_configs,dict) else 'Not a dict'}")
            
            unique_speakers = get_speakers_from_script(current_script_text)

            if not unique_speakers:
                gr.Markdown("<p style='color: #888; margin-top:10px;'>Enter a script with speaker tags (e.g., `[SpeakerName] Text`) and click 'Load/Refresh' to see per-speaker settings.</p>")
                return

            for speaker_idx, speaker_name in enumerate(unique_speakers):
                # Ensure current_speaker_configs is a valid dictionary
                if not isinstance(current_speaker_configs, dict):
                    current_speaker_configs = {} 
                
                speaker_specific_config = current_speaker_configs.get(speaker_name, {})
                accordion_elem_id = f"accordion_spk_{speaker_idx}_{speaker_name.replace(' ','_').lower()}"

                with gr.Accordion(f"Settings for Speaker: {speaker_name}", open=False, elem_id=accordion_elem_id):
                    gr.Markdown(f"Configure voice for **{speaker_name}** using **{current_tts_model}** model.")

                    default_voice = speaker_specific_config.get("voice", DEFAULT_GLOBAL_VOICE)
                    voice_dd_elem_id = f"voice_dd_spk_{speaker_idx}"
                    voice_dropdown = gr.Dropdown(
                        APP_AVAILABLE_VOICES, 
                        value=default_voice, 
                        label="Voice",
                        elem_id=voice_dd_elem_id
                    )
                    voice_dropdown.change(
                        fn=partial(handle_dynamic_accordion_input_change, speaker_name=speaker_name, config_key="voice"),
                        inputs=[voice_dropdown, speaker_configs_state], # Pass current state to handler
                        outputs=[speaker_configs_state]
                    )

                    if current_tts_model in ["tts-1", "tts-1-hd"]:
                        default_speed = float(speaker_specific_config.get("speed", 1.0))
                        speed_slider_elem_id = f"speed_slider_spk_{speaker_idx}"
                        speed_slider = gr.Slider(
                            minimum=0.25, maximum=4.0, value=default_speed, step=0.05, 
                            label="Speed", elem_id=speed_slider_elem_id
                        )
                        speed_slider.change(
                            fn=partial(handle_dynamic_accordion_input_change, speaker_name=speaker_name, config_key="speed"),
                            inputs=[speed_slider, speaker_configs_state],
                            outputs=[speaker_configs_state]
                        )
                    elif current_tts_model == "gpt-4o-mini-tts":
                        default_vibe = speaker_specific_config.get("vibe", DEFAULT_VIBE)
                        vibe_dd_elem_id = f"vibe_dd_spk_{speaker_idx}"
                        vibe_dropdown = gr.Dropdown(
                            VIBE_CHOICES, value=default_vibe, label="Vibe/Emotion", elem_id=vibe_dd_elem_id
                        )
                        
                        default_custom_instructions = speaker_specific_config.get("custom_instructions", "")
                        custom_instr_tb_elem_id = f"custom_instr_tb_spk_{speaker_idx}"
                        custom_instructions_textbox = gr.Textbox(
                            label="Custom Instructions",
                            value=default_custom_instructions,
                            placeholder="e.g., Speak with a slightly hesitant tone.",
                            lines=2,
                            visible=(default_vibe == "Custom..."), # Initial visibility
                            elem_id=custom_instr_tb_elem_id
                        )

                        vibe_dropdown.change(
                            fn=partial(handle_dynamic_accordion_input_change, speaker_name=speaker_name, config_key="vibe"),
                            inputs=[vibe_dropdown, speaker_configs_state],
                            outputs=[speaker_configs_state]
                        ).then( # Chain another update for visibility
                            fn=lambda vibe_val: gr.update(visible=(vibe_val == "Custom...")),
                            inputs=[vibe_dropdown],
                            outputs=[custom_instructions_textbox]
                        )
                        
                        custom_instructions_textbox.change(
                            fn=partial(handle_dynamic_accordion_input_change, speaker_name=speaker_name, config_key="custom_instructions"),
                            inputs=[custom_instructions_textbox, speaker_configs_state],
                            outputs=[speaker_configs_state]
                        )

    # --- Event Listeners for other UI elements ---
    tts_model_dropdown.change(
        fn=handle_tts_model_change,
        inputs=[tts_model_dropdown, speaker_configs_state],
        outputs=[global_speed_input, global_instructions_input, speaker_configs_state]
    )

    speaker_config_method_dropdown.change(
        fn=handle_speaker_config_method_visibility_change,
        inputs=[speaker_config_method_dropdown],
        outputs=[single_voice_group, detailed_per_speaker_ui_group_container]
    )
    
    load_per_speaker_ui_button.click(
        fn=handle_load_refresh_per_speaker_ui_trigger,
        inputs=[script_input, speaker_configs_state, tts_model_dropdown], 
        outputs=[speaker_configs_state] 
    )
    
    calculate_cost_button.click(
        fn=handle_calculate_cost,
        inputs=[script_input, tts_model_dropdown],
        outputs=[cost_output]
    )
    
    generate_button_fn = partial(handle_script_processing, OPENAI_API_KEY, async_openai_client, NSFW_API_URL_TEMPLATE)
    generate_button.click(
        fn=generate_button_fn,
        inputs=[
            script_input, tts_model_dropdown, pause_input,
            speaker_config_method_dropdown, global_voice_dropdown,
            speaker_configs_state, 
            global_speed_input, global_instructions_input
        ],
        outputs=[individual_lines_zip_output, merged_dialogue_mp3_output, status_output]
    )

    example_inputs_list = [
        script_input, tts_model_dropdown, pause_input,
        speaker_config_method_dropdown, global_voice_dropdown,
        speaker_configs_state, 
        global_speed_input, global_instructions_input
    ]
    example_outputs_list = [individual_lines_zip_output, merged_dialogue_mp3_output, status_output]
    example_process_fn = partial(handle_script_processing, OPENAI_API_KEY, async_openai_client, NSFW_API_URL_TEMPLATE)
    
    _ = create_examples_ui(
        inputs_for_examples=example_inputs_list,
        process_fn=example_process_fn if OPENAI_API_KEY else None,
        outputs_for_examples=example_outputs_list if OPENAI_API_KEY else None
    )

if __name__ == "__main__":
    if os.name == 'nt':
        asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())
    demo.queue().launch(debug=True, share=False)