app.py

from share_btn import community_icon_html, loading_icon_html, share_js
import os, subprocess
import torch

# (The setup function and cache download section has been commented out)

import sys
sys.path.append('src/blip')
sys.path.append('clip-interrogator')

import gradio as gr
from clip_interrogator import Config, Interrogator
import io 
from PIL import Image

config = Config()
config.device = 'cuda' if torch.cuda.is_available() else 'cpu'
config.blip_offload = False if torch.cuda.is_available() else True
config.chunk_size = 2048
config.flavor_intermediate_count = 512
config.blip_num_beams = 64

ci = Interrogator(config)

def inference(input_images, mode, best_max_flavors):
    # Process each image in the list and generate prompt results
    prompt_results = []
    for image_bytes in input_images:
        image = Image.open(io.BytesIO(image_bytes)).convert('RGB')
        if mode == 'best':
            prompt_result = ci.interrogate(image, max_flavors=int(best_max_flavors))
        elif mode == 'classic':
            prompt_result = ci.interrogate_classic(image)
        else:
            prompt_result = ci.interrogate_fast(image)
        prompt_results.append((image, prompt_result))  # Use dictionary to set image labels
    
    # Convert prompt_results to text format
    text_results = [f"Image {i+1}: {result[1]}" for i, result in enumerate(prompt_results)]
    return "\n".join(text_results)

title = """
    <div style="text-align: center; max-width: 500px; margin: 0 auto;">
        <div
        style="
            display: inline-flex;
            align-items: center;
            gap: 0.8rem;
            font-size: 1.75rem;
            margin-bottom: 10px;
        "
        >
        <h1 style="font-weight: 600; margin-bottom: 7px;">
            CLIP Interrogator 2.1
        </h1>
        </div>
        <p style="margin-bottom: 10px;font-size: 94%;font-weight: 100;line-height: 1.5em;">
        Want to figure out what a good prompt might be to create new images like an existing one? 
        <br />The CLIP Interrogator is here to get you answers!
        <br />This version is specialized for producing nice prompts for use with Stable Diffusion 2.0 using the ViT-H-14 OpenCLIP model!
        </p>
    </div>
"""

article = """
<div style="text-align: center; max-width: 500px; margin: 0 auto;font-size: 94%;">
    
    <p>
    Server busy? You can also run on <a href="https://colab.research.google.com/github/pharmapsychotic/clip-interrogator/blob/open-clip/clip_interrogator.ipynb">Google Colab</a>
    </p>
    <p>
    Has this been helpful to you? Follow Pharma on twitter 
    <a href="https://twitter.com/pharmapsychotic">@pharmapsychotic</a> 
    and check out more tools at his
    <a href="https://pharmapsychotic.com/tools.html">Ai generative art tools list</a>
    </p>
</div>
"""

css = '''
#col-container {width: width: 80%;; margin-left: auto; margin-right: auto;}
a {text-decoration-line: underline; font-weight: 600;}
.animate-spin {
    animation: spin 1s linear infinite;
}
@keyframes spin {
    from {
        transform: rotate(0deg);
    }
    to {
        transform: rotate(360deg);
    }
}
#share-btn-container {
    display: flex; padding-left: 0.5rem !important; padding-right: 0.5rem !important; background-color: #000000; justify-content: center; align-items: center; border-radius: 9999px !important; width: 13rem;
}
#share-btn {
    all: initial; color: #ffffff;font-weight: 600; cursor:pointer; font-family: 'IBM Plex Sans', sans-serif; margin-left: 0.5rem !important; padding-top: 0.25rem !important; padding-bottom: 0.25rem !important;
}
#share-btn * {
    all: unset;
}
#share-btn-container div:nth-child(-n+2){
    width: auto !important;
    min-height: 0px !important;
}
#share-btn-container .wrap {
    display: none !important;
}
#gallery .caption-label {
    font-size: 15px !important;
    right: 0 !important;
    max-width: 100% !important;
    text-overflow: clip !important;
    white-space: normal !important;
    overflow: auto !important;
    height: 20% !important;
}

#gallery .caption {
    padding: var(--size-2) var(--size-3) !important;
    text-overflow: clip !important;
    white-space: normal !important; /* Allows the text to wrap */
    color: var(--block-label-text-color) !important;
    font-weight: var(--weight-semibold) !important;
    text-align: center !important;
    height: 100% !important;
    font-size: 17px !important;
}

'''

with gr.Blocks(css=css) as block:
    with gr.Column(elem_id="col-container"):
        gr.HTML(title)

        input_image = gr.Files(label = "Inputs", file_count="multiple", type='bytes', elem_id='inputs')
        with gr.Row():
            mode_input = gr.Radio(['best', 'classic', 'fast'], label='Select mode', value='best')
            flavor_input = gr.Slider(minimum=2, maximum=24, step=2, value=4, label='best mode max flavors')
        
        submit_btn = gr.Button("Submit")
        
        # Change from Gallery to Textbox for displaying results
        result_textbox = gr.Textbox(label="Outputs", type="str", readonly=True, elem_id="output-textbox")

        with gr.Group(elem_id="share-btn-container"):
            loading_icon = gr.HTML(loading_icon_html, visible=False)

        gr.HTML(article)
    submit_btn.click(fn=inference, inputs=[input_image,mode_input,flavor_input], outputs=[result_textbox], api_name="clipi2")
    
block.queue(max_size=32,concurrency_count=10).launch(show_api=False)