Update generator.py

generator.py

@@ -0,0 +1,136 @@
+# generator.py
+import os, random, json
+from datetime import datetime
+from gtts import gTTS
+from moviepy.editor import AudioFileClip, ImageClip, CompositeVideoClip, TextClip, CompositeAudioClip
+from PIL import Image, ImageDraw, ImageFilter, ImageFont
+
+# Folders (ensure these exist in repo)
+OUTPUT_IMAGES = "outputs/images"
+OUTPUT_AUDIO  = "outputs/audio"
+OUTPUT_VIDEOS = "outputs/videos"
+BG_DIR = "backgrounds"
+MUSIC_DIR = "music"
+
+for d in (OUTPUT_IMAGES, OUTPUT_AUDIO, OUTPUT_VIDEOS, BG_DIR, MUSIC_DIR):
+    os.makedirs(d, exist_ok=True)
+
+# helpers
+def next_index(folder, prefix, ext):
+    files = [f for f in os.listdir(folder) if f.startswith(prefix) and f.endswith(ext)]
+    if not files: return 1
+    nums = []
+    for f in files:
+        try:
+            # expects prefix_N.ext
+            part = f[len(prefix) + 1: -len(ext)]
+            nums.append(int(part))
+        except:
+            pass
+    return max(nums, default=0) + 1
+
+def generate_placeholder_image(prompt, out_path, size=(720,1280)):
+    h = abs(hash(prompt))
+    r = 90 + (h % 120)
+    g = 70 + ((h >> 8) % 140)
+    b = 100 + ((h >> 16) % 120)
+    img = Image.new("RGB", size, (r, g, b))
+    draw = ImageDraw.Draw(img)
+    # soft shapes
+    for i in range(5):
+        shape_color = ((r + i*10) % 256, (g + i*15) % 256, (b + i*20) % 256)
+        xy = [(random.randint(0,size[0]), random.randint(0,size[1])) for _ in range(3)]
+        draw.polygon(xy, fill=shape_color)
+    img = img.filter(ImageFilter.GaussianBlur(radius=6))
+    try:
+        font = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf", 28)
+    except:
+        font = ImageFont.load_default()
+    caption = prompt if len(prompt) <= 60 else prompt[:57] + "..."
+    w,h_text = draw.textsize(caption, font=font)
+    draw.text(((size[0]-w)/2, size[1]*0.85), caption, font=font, fill=(255,255,255))
+    img.save(out_path)
+    return out_path
+
+def create_tts(text, lang="en"):
+    idx = next_index(OUTPUT_AUDIO, "voice", ".mp3")
+    out = os.path.join(OUTPUT_AUDIO, f"voice_{idx}.mp3")
+    tts = gTTS(text=text, lang=lang)
+    tts.save(out)
+    return out
+
+def pick_random_background():
+    files = [f for f in os.listdir(BG_DIR) if f.lower().endswith((".jpg", ".jpeg", ".png"))]
+    if not files: return None
+    return os.path.join(BG_DIR, random.choice(files))
+
+def pick_random_music():
+    files = [f for f in os.listdir(MUSIC_DIR) if f.lower().endswith(".mp3")]
+    if not files: return None
+    return os.path.join(MUSIC_DIR, random.choice(files))
+
+def add_looping_music(narration_path, music_path):
+    narration = AudioFileClip(narration_path)
+    if not music_path or not os.path.exists(music_path):
+        return narration
+    music = AudioFileClip(music_path).volumex(0.25)
+    reps = int(narration.duration // music.duration) + 2
+    # create one long clip by concatenation
+    clips = [music] * reps
+    # MoviePy: CompositeAudioClip won't concatenate; we use CompositeAudioClip of differently-started tracks
+    from moviepy.audio.AudioClip import concatenate_audioclips
+    looped = concatenate_audioclips(clips).subclip(0, narration.duration)
+    looped = looped.audio_fadeout(2)
+    final = CompositeAudioClip([looped, narration.set_start(0)])
+    return final
+
+def build_vertical_video(text, image_path, narration_path, music_path, size=(720,1280)):
+    narration = AudioFileClip(narration_path)
+    audio_final = add_looping_music(narration_path, music_path)
+
+    dur = narration.duration
+    bg = ImageClip(image_path).set_duration(dur).resize(size)
+
+    txt = TextClip(text, fontsize=36, color='white', method='caption', size=(int(size[0]*0.85), None)).set_duration(dur).set_position(('center','center'))
+
+    video = CompositeVideoClip([bg, txt])
+    video = video.set_audio(audio_final)
+
+    vid_idx = next_index(OUTPUT_VIDEOS, "affirmation", ".mp4")
+    out = os.path.join(OUTPUT_VIDEOS, f"affirmation_{vid_idx}.mp4")
+    video.write_videofile(out, fps=24, codec='libx264', audio_codec='aac')
+    return out
+
+def generate_one(text, prompt=None, chosen_bg=None, chosen_music=None, lang='en'):
+    # pick or create image
+    if chosen_bg and os.path.exists(chosen_bg):
+        img_path = chosen_bg
+    else:
+        bg = pick_random_background()
+        if bg:
+            img_path = bg
+        else:
+            idx = next_index(OUTPUT_IMAGES, "bg", ".png")
+            img_path = os.path.join(OUTPUT_IMAGES, f"bg_{idx}.png")
+            generate_placeholder_image(prompt or text, img_path, size=(720,1280))
+
+    voice_path = create_tts(text, lang)
+    # pick music priority: chosen_music path > random > None
+    music_path = chosen_music if (chosen_music and os.path.exists(chosen_music)) else pick_random_music()
+
+    print(f"🎨 Background used: {img_path}")
+    print(f"🎵 Music used:      {music_path}")
+
+    video_path = build_vertical_video(text, img_path, voice_path, music_path)
+    # return paths
+    return {
+        "video": video_path,
+        "audio": voice_path,
+        "image": img_path,
+        "music": music_path,
+        "timestamp": datetime.now().isoformat()
+    }
+
+# quick test (uncomment to run quick test)
+# if __name__ == '__main__':
+#     print(generate_one("You are capable of amazing things", prompt="sunrise over mountains"))