app.py

import os
import random
import uuid
import base64
import json
import re
import gradio as gr
import numpy as np
import pandas as pd
import torch
from PIL import Image
from datetime import datetime
from diffusers import StableDiffusionXLPipeline, EulerAncestralDiscreteScheduler
import anthropic

# ============================================================
# === GLOBALS & DATA STORAGE FILES
# ============================================================
LIKES_CACHE_FILE = "likes_cache.json"
LOG_CACHE_FILE = "log_cache.json"
QUOTE_CACHE_FILE = "quotes_cache.json"

STATIC_URL_PREFIX = "https://huggingface.co/spaces/awacke1/dalle-3-xl-lora-v2/file="

# Initialize caches / load from JSON
def load_json(file):
    if os.path.exists(file):
        with open(file, 'r', encoding='utf-8') as f:
            return json.load(f)
    return {}

def save_json(file, data):
    with open(file, 'w', encoding='utf-8') as f:
        json.dump(data, f, indent=4)

likes_cache = load_json(LIKES_CACHE_FILE) or {}
chat_logs = load_json(LOG_CACHE_FILE) if os.path.exists(LOG_CACHE_FILE) else []
quotes = load_json(QUOTE_CACHE_FILE) if os.path.exists(QUOTE_CACHE_FILE) else []

# DataFrame for images
image_metadata = pd.DataFrame(columns=['Filename','Prompt','Likes','Dislikes','Hearts','Created'])

# ============================================================
# === ANTHROPIC CLIENT (Claude)
# ============================================================
anthropic_api_key = os.environ.get("ANTHROPIC_API_KEY", None)
claude_client = anthropic.Anthropic(api_key=anthropic_api_key) if anthropic_api_key else None

# ============================================================
# === IMAGE PIPELINE
# ============================================================
pipe = None
if torch.cuda.is_available():
    pipe = StableDiffusionXLPipeline.from_pretrained(
        "fluently/Fluently-XL-v4",
        torch_dtype=torch.float16,
        use_safetensors=True,
    )
    pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
    pipe.load_lora_weights("ehristoforu/dalle-3-xl-v2", weight_name="dalle-3-xl-lora-v2.safetensors", adapter_name="dalle")
    pipe.set_adapters("dalle")
    pipe.to("cuda")

MAX_SEED = np.iinfo(np.int32).max

# ============================================================
# === HELPER FUNCTIONS
# ============================================================
def randomize_seed_fn(seed: int, randomize_seed: bool):
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
    return int(seed)

def sanitize_prompt(prompt):
    return re.sub(r'[^\w\s-]', '', prompt.lower())[:50]

def save_image_locally(img, prompt):
    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
    safe_prompt = sanitize_prompt(prompt)
    filename = f"{timestamp}_{safe_prompt}.png"
    img.save(filename)
    if filename not in likes_cache:
        likes_cache[filename] = {'likes': 0, 'dislikes': 0, 'hearts': 0}
        save_json(LIKES_CACHE_FILE, likes_cache)
    global image_metadata
    new_row = {
        'Filename': filename,
        'Prompt': prompt,
        'Likes': 0,
        'Dislikes': 0,
        'Hearts': 0,
        'Created': str(datetime.now())
    }
    image_metadata = pd.concat([image_metadata, pd.DataFrame([new_row])], ignore_index=True)
    return filename

def log_input_output(user_input, model_output, link=""):
    global chat_logs
    chat_logs.append({
        "timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
        "input": user_input,
        "output": model_output,
        "file_link": link
    })
    save_json(LOG_CACHE_FILE, chat_logs)

def generate_image(
    prompt, negative_prompt, use_negative_prompt, seed, width, height, guidance_scale, randomize_seed
):
    if pipe is None:
        return ["No GPU available, cannot generate images."], 0, [], [], []

    seed = randomize_seed_fn(seed, randomize_seed)
    if not use_negative_prompt:
        negative_prompt = ""

    images = pipe(
        prompt=prompt,
        negative_prompt=negative_prompt,
        width=width,
        height=height,
        guidance_scale=guidance_scale,
        num_inference_steps=20,
        num_images_per_prompt=1,
        cross_attention_kwargs={"scale": 0.65},
        output_type="pil",
    ).images

    filenames = []
    for img in images:
        fname = save_image_locally(img, prompt)
        filenames.append(fname)

    links = [f"{STATIC_URL_PREFIX}{f}" for f in filenames]
    # Log the generation
    log_input_output(user_input=prompt, model_output="(image generated)", link=", ".join(links))

    # Return Gradio objects
    return filenames, seed, links, get_image_gallery(), image_metadata.values.tolist()

def get_image_gallery():
    return [
        (row["Filename"], f"{row['Filename']}\nPrompt: {row['Prompt']}\n👍 {row['Likes']} 👎 {row['Dislikes']} ❤️ {row['Hearts']}")
        for _, row in image_metadata.iterrows()
        if os.path.exists(row["Filename"])
    ]

def vote_image(filename, vote_type):
    if filename and filename in likes_cache:
        likes_cache[filename][vote_type] += 1
        save_json(LIKES_CACHE_FILE, likes_cache)
        idx = image_metadata.index[image_metadata['Filename'] == filename]
        if not idx.empty:
            image_metadata.at[idx, vote_type.capitalize()] = image_metadata.at[idx, vote_type.capitalize()] + 1
    return get_image_gallery(), image_metadata.values.tolist()

def delete_image(filename):
    if filename and os.path.exists(filename):
        os.remove(filename)
    if filename in likes_cache:
        del likes_cache[filename]
        save_json(LIKES_CACHE_FILE, likes_cache)
    global image_metadata
    image_metadata = image_metadata[image_metadata['Filename'] != filename]
    return get_image_gallery(), image_metadata.values.tolist()

def delete_all_images():
    global image_metadata, likes_cache
    for f in image_metadata["Filename"].tolist():
        if os.path.exists(f):
            os.remove(f)
    image_metadata = pd.DataFrame(columns=['Filename','Prompt','Likes','Dislikes','Hearts','Created'])
    likes_cache.clear()
    save_json(LIKES_CACHE_FILE, likes_cache)
    return get_image_gallery(), image_metadata.values.tolist()

# === QUOTES Demo (Optional) ===
def add_quote(q):
    if q.strip():
        quotes.append({
            "text": q,
            "likes": 0,
            "created": datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        })
        save_json(QUOTE_CACHE_FILE, quotes)
    return [[idx, itm["text"], itm["likes"], itm["created"]] for idx, itm in enumerate(quotes)]

def like_quote(idx):
    if 0 <= idx < len(quotes):
        quotes[idx]["likes"] += 1
    save_json(QUOTE_CACHE_FILE, quotes)
    return [[i, itm["text"], itm["likes"], itm["created"]] for i, itm in enumerate(quotes)]

# === CLAUDE Chat ===
def chat_claude(user_message):
    if not claude_client:
        return "No Anthropic API key configured."
    if not user_message.strip():
        return "Empty message."
    resp = claude_client.messages.create(
        model="claude-3-sonnet-20240229",
        max_tokens=1000,
        messages=[{"role": "user", "content": user_message}],
    )
    text = resp.content[0].text
    log_input_output(user_input=user_message, model_output=text, link="")
    return text

# === Refresh gallery + DF
def refresh_gallery_and_df():
    return gr.update(value=get_image_gallery()), gr.update(value=image_metadata.values.tolist())

# ============================================================
# === BUILD GRADIO UI
# ============================================================
DESCRIPTION = """# 🎨 ArtForge & Claude Chat
Generate AI art, chat with Claude, log everything, and vote on images.
"""

examples = [
    "Futuristic cityscape in neon lighting",
    "Cute cat wearing a wizard hat",
    "Surreal landscape with floating islands",
]

with gr.Blocks(css=".gradio-container {max-width: 1024px !important}") as demo:
    gr.Markdown(DESCRIPTION)

    with gr.Tab("Generate Images"):
        with gr.Row():
            prompt = gr.Text(label="Prompt", max_lines=1)
            run_button = gr.Button("Run")
        result = gr.Gallery(label="Result", columns=1, preview=True)
        use_negative_prompt = gr.Checkbox(label="Use negative prompt", value=True)
        negative_prompt = gr.Text(
            label="Negative prompt",
            lines=3,
            value="(deformed, distorted:1.3), poorly drawn, bad anatomy",
        )
        seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0)
        randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
        width = gr.Slider(label="Width", minimum=512, maximum=2048, step=64, value=1024)
        height = gr.Slider(label="Height", minimum=512, maximum=2048, step=64, value=1024)
        guidance_scale = gr.Slider(label="Guidance Scale", minimum=1, maximum=20, step=0.5, value=7)
        run_button.click(
            fn=generate_image,
            inputs=[prompt, negative_prompt, use_negative_prompt, seed, width, height, guidance_scale, randomize_seed],
            outputs=[result, seed, gr.HTML(visible=False), gr.Gallery(), gr.Dataframe()],
            api_name="run"
        )
        gr.Examples(examples=examples, inputs=prompt)

    with gr.Tab("Chat with Claude"):
        claude_input = gr.Textbox(label="Your Message")
        claude_output = gr.Textbox(label="Claude's Reply", lines=4)
        send_claude = gr.Button("Send to Claude")
        send_claude.click(chat_claude, inputs=claude_input, outputs=claude_output)

    with gr.Tab("Logs & Management"):
        with gr.Accordion("All Logs", open=False):
            logs_data = gr.Dataframe(
                value=pd.DataFrame(chat_logs),
                label="Input/Output Logs",
                interactive=False,
                wrap=True
            )

    with gr.Tab("Gallery & Voting"):
        image_gallery = gr.Gallery(label="Generated Images", columns=4)
        metadata_df = gr.Dataframe(
            label="Image Metadata",
            headers=["Filename", "Prompt", "Likes", "Dislikes", "Hearts", "Created"],
            interactive=False
        )
        selected_image = gr.State()
        with gr.Row():
            like_button = gr.Button("👍 Like")
            dislike_button = gr.Button("👎 Dislike")
            heart_button = gr.Button("❤️ Heart")
            delete_image_button = gr.Button("🗑️ Delete Image")
            delete_all_button = gr.Button("🗑️ Delete All")
        image_gallery.select(fn=lambda evt: evt, inputs=[], outputs=[selected_image])
        like_button.click(fn=lambda x: vote_image(x, 'likes'), inputs=selected_image, outputs=[image_gallery, metadata_df])
        dislike_button.click(fn=lambda x: vote_image(x, 'dislikes'), inputs=selected_image, outputs=[image_gallery, metadata_df])
        heart_button.click(fn=lambda x: vote_image(x, 'hearts'), inputs=selected_image, outputs=[image_gallery, metadata_df])
        delete_image_button.click(fn=delete_image, inputs=selected_image, outputs=[image_gallery, metadata_df])
        delete_all_button.click(fn=delete_all_images, outputs=[image_gallery, metadata_df])

    with gr.Tab("Quotes (Optional)"):
        quote_input = gr.Textbox(label="Enter a quote")
        add_q_button = gr.Button("Add Quote")
        quote_df = gr.Dataframe(value=[(idx, q['text'], q['likes'], q['created']) for idx,q in enumerate(quotes)],
                                headers=["Index","Text","Likes","Created"], interactive=False)
        selected_quote = gr.Number(label="Index to Like")
        like_q_button = gr.Button("Like Quote")
        add_q_button.click(fn=add_quote, inputs=quote_input, outputs=quote_df)
        like_q_button.click(fn=like_quote, inputs=selected_quote, outputs=quote_df)

    demo.load(fn=refresh_gallery_and_df, outputs=[image_gallery, metadata_df])

if __name__ == "__main__":
    demo.queue(max_size=20).launch()