app.py

import os
import gradio as gr
import subprocess
import cv2
from PIL import Image
import google.generativeai as genai
import base64

# ========== SETUP ==========
def set_api_key(api_key):
    os.environ["GOOGLE_API_KEY"] = api_key
    genai.configure(api_key=api_key)


# ========== TEXT TO IMAGE ==========
def text_to_image(api_key, prompt, image_file, resolution):
    try:
        set_api_key(api_key)
        model = genai.GenerativeModel("imagen-3.0")

        if image_file:
            final_prompt = {"text": prompt, "image": Image.open(image_file)}
        else:
            final_prompt = prompt

        result = model.generate_content(final_prompt)
        if not result.candidates:
            return None, "❌ Gagal: tiada kandidat dihasilkan"

        image_bytes = result.candidates[0].content.parts[0].inline_data.data
        fname = "output_img.png"
        with open(fname, "wb") as f:
            f.write(base64.b64decode(image_bytes))

        return fname, "✅ Gambar berjaya dihasilkan!"
    except Exception as e:
        return None, f"❌ Error: {str(e)}"


# ========== SHORT VIDEO ==========
def text_to_video(api_key, prompt, image_file, duration, aspect_ratio, resolution):
    try:
        set_api_key(api_key)
        model = genai.GenerativeModel("veo-3")

        # aspect ratio & res
        if aspect_ratio == "9:16":
            res = {"width": 720 if resolution == "720p" else 1080,
                   "height": 1280 if resolution == "720p" else 1920}
        else:
            res = {"width": 1280 if resolution == "720p" else 1920,
                   "height": 720 if resolution == "720p" else 1080}

        if image_file:
            final_prompt = {"text": prompt, "image": Image.open(image_file)}
        else:
            final_prompt = prompt

        result = model.generate_content({
            "text": prompt,
            "config": {
                "duration_seconds": int(duration),
                "fps": 12,
                "video": {"width": res["width"], "height": res["height"]}
            }
        })

        if not result.candidates:
            return None, "❌ Gagal: tiada kandidat video dihasilkan"

        video_bytes = result.candidates[0].content.parts[0].inline_data.data
        fname = "output_vid.mp4"
        with open(fname, "wb") as f:
            f.write(base64.b64decode(video_bytes))

        return fname, "✅ Video berhasil dibuat!"
    except Exception as e:
        return None, f"❌ Error: {str(e)}"


# ========== LONG VIDEO ==========
def auto_segment_long_video(api_key, prompt, start_img, num_segments, duration_each, aspect_ratio, resolution):
    try:
        set_api_key(api_key)
        model = genai.GenerativeModel("veo-3")

        if aspect_ratio == "9:16":
            res = {"width": 720 if resolution == "720p" else 1080,
                   "height": 1280 if resolution == "720p" else 1920}
        else:
            res = {"width": 1280 if resolution == "720p" else 1920,
                   "height": 720 if resolution == "720p" else 1080}

        words = prompt.split()
        chunk_size = max(1, len(words) // num_segments)
        seg_prompts = [" ".join(words[i:i+chunk_size]) for i in range(0, len(words), chunk_size)][:num_segments]

        carry_image = Image.open(start_img) if start_img else None
        segment_files = []

        for idx, text in enumerate(seg_prompts):
            if carry_image:
                final_prompt = {"text": text, "image": carry_image}
            else:
                final_prompt = text

            result = model.generate_content({
                "text": text,
                "config": {
                    "duration_seconds": int(duration_each),
                    "fps": 12,
                    "video": {"width": res["width"], "height": res["height"]}
                }
            })

            if not result.candidates:
                return None, f"❌ Gagal buat segmen {idx}"

            video_bytes = result.candidates[0].content.parts[0].inline_data.data
            fname = f"seg_{idx}.mp4"
            with open(fname, "wb") as f:
                f.write(base64.b64decode(video_bytes))
            segment_files.append(fname)

            cap = cv2.VideoCapture(fname)
            cap.set(cv2.CAP_PROP_POS_FRAMES, cap.get(cv2.CAP_PROP_FRAME_COUNT) - 1)
            ret, frame = cap.read()
            cap.release()
            if ret:
                framefile = f"carry_{idx}.jpg"
                cv2.imwrite(framefile, frame)
                carry_image = Image.open(framefile)

        # gabung semua segmen
        list_file = "segments.txt"
        with open(list_file, "w") as f:
            for seg in segment_files:
                f.write(f"file '{seg}'\n")

        final_out = "final_long.mp4"
        subprocess.run(
            ["ffmpeg", "-y", "-f", "concat", "-safe", "0", "-i", list_file, "-c", "copy", final_out],
            check=True
        )

        return final_out, "✅ Video panjang berhasil dibuat!"
    except Exception as e:
        return None, f"❌ Error: {str(e)}"


# ========== UI ==========
with gr.Blocks(css="""
body {background: linear-gradient(160deg,#0f0c29,#302b63,#24243e);}
.gradio-container {color:white;}
h1,h2,h3,h4,label {color:white !important;}
button {background: linear-gradient(90deg,#6a11cb,#2575fc) !important; color:white !important;}
""") as demo:

    gr.HTML("<h1 style='text-align:center'>🎥 Penjana Video VEO</h1>")

    api_key = gr.Textbox(label="🔑 Masukkan Kunci API Gemini anda", type="password")

    with gr.Tabs():
        with gr.Tab("🎬 Video Pendek"):
            prompt_vid = gr.Textbox(label="Prompt", placeholder="Tulis ide video...", lines=4)
            vid_file = gr.File(label="Imej Rujukan (Opsional)", file_types=[".jpg",".png"], type="filepath")
            duration = gr.Slider(3, 8, value=5, label="Tempoh (detik)")
            aspect_ratio = gr.Dropdown(choices=["16:9", "9:16"], value="16:9", label="Nisbah Aspek")
            resolution = gr.Dropdown(choices=["720p","1080p"], value="720p", label="Resolusi")
            btn_vid = gr.Button("Hasilkan Video Pendek")
            out_vid = gr.Video()
            msg_vid = gr.Label()
            btn_vid.click(fn=text_to_video,
                          inputs=[api_key, prompt_vid, vid_file, duration, aspect_ratio, resolution],
                          outputs=[out_vid, msg_vid])

        with gr.Tab("🎞️ Video Panjang"):
            full_prompt = gr.Textbox(label="Prompt Narasi Panjang", placeholder="Tulis cerita panjang...", lines=6)
            start_img = gr.File(label="Imej Awal (Opsional)", file_types=[".jpg",".png"], type="filepath")
            num_segments = gr.Slider(1, 10, value=3, step=1, label="Jumlah Segmen")
            dur_each = gr.Slider(3, 8, value=5, step=1, label="Durasi/Segmen (detik)")
            aspect_ratio_long = gr.Dropdown(choices=["16:9", "9:16"], value="16:9", label="Nisbah Aspek")
            resolution_long = gr.Dropdown(choices=["720p","1080p"], value="720p", label="Resolusi")
            btn_long = gr.Button("Hasilkan Video Panjang")
            out_long = gr.Video()
            msg_long = gr.Label()
            btn_long.click(fn=auto_segment_long_video,
                           inputs=[api_key, full_prompt, start_img, num_segments, dur_each,
                                   aspect_ratio_long, resolution_long],
                           outputs=[out_long, msg_long])

        with gr.Tab("🖼️ Imej"):
            prompt_img = gr.Textbox(label="Prompt", placeholder="Tulis ide gambar...", lines=4)
            img_file = gr.File(label="Imej Rujukan (Opsional)", file_types=[".jpg",".png"], type="filepath")
            resolution_img = gr.Dropdown(choices=["720p","1080p"], value="720p", label="Resolusi (dummy)")
            btn_img = gr.Button("Hasilkan Imej")
            out_img = gr.Image()
            msg_img = gr.Label()
            btn_img.click(fn=text_to_image,
                          inputs=[api_key, prompt_img, img_file, resolution_img],
                          outputs=[out_img, msg_img])

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