Update the app

app.py

@@ -1,9 +1,12 @@
 import gradio as gr
 import os
 import shutil
+import datetime
 from typing import List, Optional
-import spaces
 
+# ──────────────────────────────────────────────────────────────────────────────
+# Import project‑specific helpers — unchanged from initial version
+# ──────────────────────────────────────────────────────────────────────────────
 from scripts.generate_scripts import generate_script, generate_title, generate_description
 from scripts.generate_voice import generate_voice
 from scripts.get_footage import get_video_montage_from_folder
@@ -15,49 +18,51 @@ from scripts.generate_subtitles import (
 )
 
 # ──────────────────────────────────────────────────────────────────────────────
-# Constants & helper utils
+# Constants & utilities
 # ──────────────────────────────────────────────────────────────────────────────
-
 WORDS_PER_SECOND = 2.3  # ≃ 140 wpm
+ASSETS_DIRS = (
+    "./assets/audio",
+    "./assets/backgrounds",
+    "./assets/output",
+    "./assets/video_music",
+)
 
+for d in ASSETS_DIRS:
+    os.makedirs(d, exist_ok=True)
 
 def safe_copy(src: str, dst: str) -> str:
+    """Copy src → dst unless they are the same file, returns destination path."""
     if os.path.abspath(src) == os.path.abspath(dst):
         return src
     shutil.copy(src, dst)
     return dst
 
+# Wrapper util to timestamp generated files so different runs don't overwrite each other
+
+def timestamped_filename(prefix: str, ext: str) -> str:
+    ts = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
+    return os.path.join("./assets/output", f"{prefix}_{ts}.{ext}")
+
 # ──────────────────────────────────────────────────────────────────────────────
-# Core processing pipeline
+# Independent functional endpoints (Gradio callbacks)
 # ──────────────────────────────────────────────────────────────────────────────
-def process_video(
+
+def cb_generate_script(
     context: str,
     instruction: str,
     target_duration: int,
     script_mode: str,
     custom_script: Optional[str],
-    lum: float,
-    contrast: float,
-    gamma: float,
-    add_subs: bool,
-    accumulated_videos: List[str] | None = None,
-    user_music: Optional[str] = None,
-    show_progress_bar: bool = True,
 ):
-    """Build the final video using user‑defined visual parameters (brightness, contrast, gamma)."""
-
-    if not accumulated_videos:
-        raise ValueError("❌ Please upload at least one background video (.mp4) before generating.")
+    """Generate (or accept) a script + title/description."""
 
     approx_words = int(target_duration * WORDS_PER_SECOND)
 
-    # --- 1. Script (AI or custom) ---
     if script_mode == "Use my script":
         if not custom_script or not custom_script.strip():
-            raise ValueError("❌ You selected 'Use my script' but the script field is empty!")
+            raise gr.Error("❌ You selected 'Use my script' but the script field is empty!")
         script = custom_script.strip()
-        title = generate_title(script)
-        description = generate_description(script)
     else:
         prompt = (
             f"You are a video creation expert. Here is the context: {context.strip()}\n"
@@ -65,155 +70,224 @@ def process_video(
             f"🔴 Strict target duration: {target_duration}s — ≈ {approx_words} words (must be respected)."
         )
         script = generate_script(prompt)
-        title = generate_title(script)
-        description = generate_description(script)
 
-    # --- 2. Prepare folders ---
-    for folder in ("./assets/audio", "./assets/backgrounds", "./assets/output"):
-        os.makedirs(folder, exist_ok=True)
+    title = generate_title(script)
+    description = generate_description(script)
+    return script, title, description, script  # last return for state update
+
+
+def cb_generate_voice(script: str):
+    """TTS generation from a given script, returns path to MP3."""
+    if not script or not script.strip():
+        raise gr.Error("❌ Script text is empty – generate or paste a script first.")
+
+    voice_path = timestamped_filename("voice", "mp3")
+    generate_voice(script, voice_path)
+    return voice_path, voice_path  # second value updates state
 
-    voice_path = "./assets/audio/voice.mp3"
-    final_no_subs = "./assets/output/final_video.mp4"
-    final_with_subs = "./assets/output/final_video_subtitles.mp4"
 
-    # --- 3. Copy videos ---
+def accumulate_files(new: List[str], state: List[str] | None):
+    """Append only new valid MP4 files to state."""
+    state = state or []
+    for f in new or []:
+        if (
+            isinstance(f, str)
+            and os.path.isfile(f)
+            and f.lower().endswith(".mp4")
+            and f not in state
+        ):
+            state.append(f)
+    return state
+
+
+def cb_create_montage(
+    accumulated_videos: List[str],
+    voice_path: str,
+    lum: float,
+    contrast: float,
+    gamma: float,
+    show_bar: bool,
+):
+    """Create the background‑video montage synced to the narration audio."""
+
+    if not accumulated_videos:
+        raise gr.Error("❌ Please upload at least one background video (.mp4) before generating the montage.")
+    if not voice_path or not os.path.isfile(voice_path):
+        raise gr.Error("❌ A narration audio file (.mp3) is required – generate or upload one first.")
+
+    # Clean previous backgrounds, then copy new ones
     for f in os.listdir("./assets/backgrounds"):
         if f.lower().endswith(".mp4"):
             os.remove(os.path.join("./assets/backgrounds", f))
     for idx, v in enumerate(accumulated_videos):
-        if not os.path.isfile(v) or not v.lower().endswith(".mp4"):
-            raise ValueError(f"❌ Invalid file: {v}")
         safe_copy(v, os.path.join("./assets/backgrounds", f"video_{idx:03d}.mp4"))
 
-    # --- 4. AI voice ---
-    generate_voice(script, voice_path)
-
-    # --- 5. Video montage ---
-    music_path = user_music if user_music and os.path.isfile(user_music) else None
-    out_no_audio = get_video_montage_from_folder(
+    montage_path = timestamped_filename("montage", "mp4")
+    _ = get_video_montage_from_folder(
         folder_path="./assets/backgrounds",
         audio_path=voice_path,
         output_dir="./assets/video_music",
         lum=lum,
         contrast=contrast,
         gamma=gamma,
-        show_progress_bar=show_progress_bar,
+        show_progress_bar=show_bar,
     )
+    # get_video_montage_from_folder already saves the file – we just need its path
+    # It returns the path, so capture it
+    montage_path = _
+    return montage_path, montage_path
 
-    # --- 6. Mixing & subtitles ---
-    edit_video(out_no_audio, voice_path, music_path, final_no_subs)
 
-    if add_subs:
-        segments = transcribe_audio_to_subs(voice_path)
-        subs = chunk_text_by_words(segments, max_words=3)
-        add_subtitles_to_video(final_no_subs, subs, final_with_subs)
-        return script, title, description, final_with_subs
-    else:
-        return script, title, description, final_no_subs
+def cb_mix_audio(
+    montage_path: str,
+    voice_path: str,
+    music_file: Optional[str] = None,
+):
+    """Combine montage video, voice audio, and optional background music."""
+    if not montage_path or not os.path.isfile(montage_path):
+        raise gr.Error("❌ Please generate a montage video first.")
+    if not voice_path or not os.path.isfile(voice_path):
+        raise gr.Error("❌ Narration audio missing – generate or upload it.")
+
+    music_path = music_file if music_file and os.path.isfile(music_file) else None
+    final_no_subs = timestamped_filename("final_no_subs", "mp4")
+    edit_video(montage_path, voice_path, music_path, final_no_subs)
+    return final_no_subs, final_no_subs
+
+
+def cb_add_subtitles(final_no_subs: str, voice_path: str):
+    """Overlay dynamic subtitles on the mixed video."""
+    if not final_no_subs or not os.path.isfile(final_no_subs):
+        raise gr.Error("❌ Mixed video not found – run the 'Mix Audio/Video' step first.")
+    if not voice_path or not os.path.isfile(voice_path):
+        raise gr.Error("❌ Narration audio missing.")
+
+    segments = transcribe_audio_to_subs(voice_path)
+    subs = chunk_text_by_words(segments, max_words=3)
+    final_with_subs = timestamped_filename("final_with_subs", "mp4")
+    add_subtitles_to_video(final_no_subs, subs, final_with_subs)
+    return final_with_subs
 
 # ──────────────────────────────────────────────────────────────────────────────
-# Upload helper
+# Gradio UI – one tab per function
 # ──────────────────────────────────────────────────────────────────────────────
 
-def accumulate_files(new: List[str], state: List[str] | None):
-    state = state or []
-    for f in new or []:
-        if isinstance(f, str) and os.path.isfile(f) and f.lower().endswith(".mp4") and f not in state:
-            state.append(f)
-    return state
+demo = gr.Blocks(theme="gradio/soft")
 
-# ──────────────────────────────────────────────────────────────────────────────
-# Gradio UI
-# ──────────────────────────────────────────────────────────────────────────────
+with demo:
+    gr.Markdown("# 🎬 Modular AI Video Toolkit")
+    gr.Markdown(
+        "Each tab exposes **one single processing step** so you can mix & match them as you like. 💡"
+    )
 
-with gr.Blocks(theme="gradio/soft") as demo:
-    gr.Markdown("# 🎬 AI Video Generator — Advanced Controls")
+    # Shared state across tabs
+    script_state = gr.State("")
+    voice_state = gr.State("")
+    montage_state = gr.State("")
+    final_no_subs_state = gr.State("")
 
-    # ------------------- Parameters -------------------
-    with gr.Tab("🛠️ Settings"):
+    # ───────────────────────── Script generation ─────────────────────────
+    with gr.Tab("1️⃣ Generate Script"):
         with gr.Row():
-            context_input = gr.Textbox(label="🧠 Context", lines=4)
-            instruction_input = gr.Textbox(label="🎯 Instruction", lines=4)
-
+            context_in = gr.Textbox(label="🧠 Context", lines=4)
+            instruction_in = gr.Textbox(label="🎯 Instruction", lines=4)
         duration_slider = gr.Slider(5, 120, 1, 60, label="⏱️ Target duration (s)")
-
         script_mode = gr.Radio([
             "Generate script with AI",
             "Use my script",
         ], value="Generate script with AI", label="Script mode")
+        custom_script_in = gr.Textbox(label="✍️ My script", lines=8, interactive=False)
 
-        custom_script_input = gr.Textbox(label="✍️ My script", lines=8, interactive=False)
-
-        def toggle_script_input(mode):
+        def _toggle(mode):
             return gr.update(interactive=(mode == "Use my script"))
 
-        script_mode.change(toggle_script_input, inputs=script_mode, outputs=custom_script_input)
+        script_mode.change(_toggle, inputs=script_mode, outputs=custom_script_in)
 
-        with gr.Accordion("🎨 Video Settings (brightness/contrast/gamma)", open=False):
-            lum_slider = gr.Slider(0, 20, 6, step=0.5, label="Brightness (0–20)")
-            contrast_slider = gr.Slider(0.5, 2.0, 1.0, step=0.05, label="Contrast (0.5–2.0)")
-            gamma_slider = gr.Slider(0.5, 2.0, 1.0, step=0.05, label="Gamma (0.5–2.0)")
+        gen_script_btn = gr.Button("📝 Create Script", variant="primary")
+        script_out = gr.Textbox(label="Script", lines=8, interactive=False)
+        title_out = gr.Textbox(label="Title", lines=1, interactive=False)
+        desc_out = gr.Textbox(label="Description", lines=3, interactive=False)
 
-        with gr.Row():
-            add_subs_checkbox = gr.Checkbox(label="Add dynamic subtitles", value=True)
+        gen_script_btn.click(
+            cb_generate_script,
+            [context_in, instruction_in, duration_slider, script_mode, custom_script_in],
+            [script_out, title_out, desc_out, script_state],
+        )
 
-        with gr.Row():
-            show_bar = gr.Checkbox(label="Show progress bar", value=True)
+    # ───────────────────────── Voice generation ─────────────────────────
+    with gr.Tab("2️⃣ Generate Voice"):
+        script_in_voice = gr.Textbox(label="Script (paste or use from previous step)", lines=8)
+        gen_voice_btn = gr.Button("🔈 Synthesize Voice", variant="primary")
+        voice_audio = gr.Audio(label="Generated voice", interactive=False)
+
+        gen_voice_btn.click(
+            cb_generate_voice,
+            inputs=[script_in_voice],
+            outputs=[voice_audio, voice_state],
+        )
+        # Auto‑populate script textbox with state when it updates
+        script_state.change(lambda s: s, script_state, script_in_voice, queue=False)
 
-        # Upload videos
+    # ───────────────────────── Montage creation ─────────────────────────
+    with gr.Tab("3️⃣ Create Montage"):
         videos_dropzone = gr.Files(label="🎞️ Background videos (MP4)", file_types=[".mp4"], type="filepath")
         videos_state = gr.State([])
-        video_list_display = gr.Textbox(label="✅ Selected videos", interactive=False, lines=4)
         videos_dropzone.upload(accumulate_files, [videos_dropzone, videos_state], videos_state, queue=False)
-        videos_state.change(lambda s: "\n".join(os.path.basename(f) for f in s), videos_state, video_list_display, queue=False)
-
+        videos_display = gr.Textbox(label="Selected videos", interactive=False)
+        videos_state.change(lambda s: "\n".join(os.path.basename(f) for f in s), videos_state, videos_display, queue=False)
 
-        user_music = gr.File(label="🎵 Background music (MP3, optional)", file_types=[".mp3"], type="filepath")
+        with gr.Accordion("🎨 Visual settings", open=False):
+            lum_slider = gr.Slider(0, 20, 6, step=0.5, label="Brightness (0–20)")
+            contrast_slider = gr.Slider(0.5, 2.0, 1.0, step=0.05, label="Contrast (0.5–2.0)")
+            gamma_slider = gr.Slider(0.5, 2.0, 1.0, step=0.05, label="Gamma (0.5–2.0)")
+        show_bar = gr.Checkbox(label="Show progress bar", value=True)
 
-        generate_btn = gr.Button("🚀 Generate the video", variant="primary")
+        create_montage_btn = gr.Button("🎞️ Build Montage", variant="primary")
+        montage_video = gr.Video(label="Montage Preview")
 
-    with gr.Tab("📤 Results"):
-        video_output = gr.Video(label="🎬 Generated Video")
+        create_montage_btn.click(
+            cb_create_montage,
+            [videos_state, voice_state, lum_slider, contrast_slider, gamma_slider, show_bar],
+            [montage_video, montage_state],
+        )
 
-        # Script + copy button
-        script_output = gr.Textbox(label="📝 Script", lines=6, interactive=False)
-        copy_script_btn = gr.Button("📋 Copy")
-        copy_script_btn.click(
-            None,
-            inputs=[script_output],
-            outputs=None,
-            js="(text) => navigator.clipboard.writeText(text)"
+    # ───────────────────────── Mixing (voice + music) ─────────────────────────
+    with gr.Tab("4️⃣ Mix Audio / Video"):
+        voice_in = gr.File(label="Narration MP3 (optional – leave empty to use state)", file_types=[".mp3"], type="filepath")
+        montage_in = gr.File(label="Montage MP4 (optional – leave empty to use state)", file_types=[".mp4"], type="filepath")
+        music_in = gr.File(label="Background music (MP3 – optional)", file_types=[".mp3"], type="filepath")
+
+        def _use_state(file, state):
+            return file if file else state
+
+        mix_btn = gr.Button("🎚️ Mix", variant="primary")
+        final_no_subs_vid = gr.Video(label="Mixed video (no subtitles)")
+
+        mix_btn.click(
+            lambda montage, voice, music, montage_state_val, voice_state_val: cb_mix_audio(
+                _use_state(montage, montage_state_val),
+                _use_state(voice, voice_state_val),
+                music,
+            ),
+            [montage_in, voice_in, music_in, montage_state, voice_state],
+            [final_no_subs_vid, final_no_subs_state],
         )
 
-        # Title + copy button
-        title_output = gr.Textbox(label="🎬 Title", lines=1, interactive=False)
-        copy_title_btn = gr.Button("📋 Copy")
-        copy_title_btn.click(None, inputs=title_output, outputs=None, js="(text) => {navigator.clipboard.writeText(text);}")
-
-        # Description + copy button
-        desc_output = gr.Textbox(label="📄 Description", lines=3, interactive=False)
-        copy_desc_btn = gr.Button("📋 Copy")
-        copy_desc_btn.click(None, inputs=desc_output, outputs=None, js="(text) => {navigator.clipboard.writeText(text);}")
-
-    # ------------------- Generation Callback -------------------
-    generate_btn.click(
-        fn=process_video,
-        inputs=[
-            context_input,
-            instruction_input,
-            duration_slider,
-            script_mode,
-            custom_script_input,
-            lum_slider,
-            contrast_slider,
-            gamma_slider,
-            add_subs_checkbox,
-            videos_state,
-            user_music,
-            show_bar,
-        ],
-        outputs=[script_output, title_output, desc_output, video_output],
-    )
+    # ───────────────────────── Subtitles ─────────────────────────
+    with gr.Tab("5️⃣ Add Subtitles"):
+        video_in_sub = gr.File(label="Video MP4 (optional – defaults to last mixed video)", type="filepath", file_types=[".mp4"])
+        voice_in_sub = gr.File(label="Narration MP3 (optional – defaults to last generated voice)", type="filepath", file_types=[".mp3"])
+        add_subs_btn = gr.Button("🔤 Add Subtitles", variant="primary")
+        final_subs_video = gr.Video(label="Final video with subtitles")
+
+        add_subs_btn.click(
+            lambda v_in, a_in, v_state, a_state: cb_add_subtitles(
+                v_in if v_in else v_state,
+                a_in if a_in else a_state,
+            ),
+            [video_in_sub, voice_in_sub, final_no_subs_state, voice_state],
+            final_subs_video,
+        )
 
+    # Startup
     demo.launch()
-