Update app.py

app.py

@@ -1,152 +1,204 @@
-import os, subprocess
+import os
+import sys
+import subprocess
 import torch
 
+
+def run(cmd):
+    print(f"\nRunning: {' '.join(cmd)}")
+    subprocess.run(cmd, check=True)
+
+
 def setup():
     install_cmds = [
-        ['pip', 'install', 'ftfy', 'gradio', 'regex', 'tqdm', 'transformers==4.21.2', 'timm', 'fairscale', 'requests'],
-        ['pip', 'install', 'open_clip_torch'],
-        ['pip', 'install', '-e', 'git+https://github.com/pharmapsychotic/BLIP.git@lib#egg=blip'],
-        ['git', 'clone', '-b', 'open-clip', 'https://github.com/pharmapsychotic/clip-interrogator.git']
+        ["pip", "install", "setuptools<82", "wheel"],
+        [
+            "pip", "install",
+            "ftfy",
+            "regex",
+            "tqdm",
+            "transformers==4.21.2",
+            "timm",
+            "fairscale",
+            "requests",
+        ],
+        [
+            "pip", "install",
+            "--no-build-isolation",
+            "-e",
+            "git+https://github.com/pharmapsychotic/BLIP.git@lib#egg=blip",
+        ],
     ]
+
     for cmd in install_cmds:
-        print(subprocess.run(cmd, stdout=subprocess.PIPE).stdout.decode('utf-8'))
+        run(cmd)
+
+    if not os.path.exists("clip-interrogator"):
+        run([
+            "git", "clone",
+            "-b", "open-clip",
+            "https://github.com/pharmapsychotic/clip-interrogator.git",
+        ])
+
 
 setup()
 
-# download cache files
+
 print("Download preprocessed cache files...")
+
 CACHE_URLS = [
-    'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_artists.pkl',
-    'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_flavors.pkl',
-    'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_mediums.pkl',
-    'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_movements.pkl',
-    'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_trendings.pkl',
+    "https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_artists.pkl",
+    "https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_flavors.pkl",
+    "https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_mediums.pkl",
+    "https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_movements.pkl",
+    "https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_trendings.pkl",
 ]
-os.makedirs('cache', exist_ok=True)
+
+os.makedirs("cache", exist_ok=True)
+
 for url in CACHE_URLS:
-    print(subprocess.run(['wget', url, '-P', 'cache'], stdout=subprocess.PIPE).stdout.decode('utf-8'))
+    subprocess.run(["wget", "-nc", url, "-P", "cache"], check=False)
 
-import sys
-sys.path.append('src/blip')
-sys.path.append('clip-interrogator')
+
+sys.path.append("src/blip")
+sys.path.append("clip-interrogator")
 
 import gradio as gr
 from clip_interrogator import Config, Interrogator
 
+
 config = Config()
-config.device = 'cuda' if torch.cuda.is_available() else 'cpu'
+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(image, mode, best_max_flavors):
-    """
-    Generate a descriptive prompt from an input image using different interrogation modes.
-
-    Args:
-        image: A PIL Image object representing the input image to be analyzed.
-        mode: A string specifying the interrogation mode to use. 
-              Can be one of ['best', 'classic', 'fast']:
-              - 'best': Produces a prompt using the 'best' interrogation mode with max flavors control.
-              - 'classic': Uses the classic interrogation method.
-              - 'fast': Uses a faster but less detailed interrogation method.
-        best_max_flavors: An integer controlling the maximum number of flavor descriptors 
-                          when using 'best' mode (ignored in other modes).
-
-    Returns:
-        A string containing the generated prompt describing the image.
-    """
-    
-    
-    image = image.convert('RGB')
-    if mode == 'best':
-        
-        prompt_result = ci.interrogate(image, max_flavors=int(best_max_flavors))
-        
-        print("mode best: " + prompt_result)
-        
+    if image is None:
+        return "Please upload an image."
+
+    image = image.convert("RGB")
+
+    if mode == "best":
+        prompt_result = ci.interrogate(
+            image,
+            max_flavors=int(best_max_flavors),
+        )
+        print("mode best:", prompt_result)
         return prompt_result
-    
-    elif mode == 'classic':
-        
+
+    if mode == "classic":
         prompt_result = ci.interrogate_classic(image)
-        
-        print("mode classic: " + prompt_result)
-        
-        return prompt_result
-    
-    else:
-        
-        prompt_result = ci.interrogate_fast(image)
-        
-        print("mode fast: " + prompt_result)
-        
+        print("mode classic:", prompt_result)
         return prompt_result
 
+    prompt_result = ci.interrogate_fast(image)
+    print("mode fast:", prompt_result)
+    return prompt_result
+
+
 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;
-        "
-        >
+<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? 
+    </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>
+    </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>
+        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>
+        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 {max-width: 700px; margin-left: auto; margin-right: auto;}
-a {text-decoration-line: underline; font-weight: 600;}
-'''
+css = """
+#col-container {
+    max-width: 700px;
+    margin-left: auto;
+    margin-right: auto;
+}
+
+a {
+    text-decoration-line: underline;
+    font-weight: 600;
+}
+"""
+
 
 with gr.Blocks(css=css) as demo:
     with gr.Column(elem_id="col-container"):
         gr.HTML(title)
 
-        input_image = gr.Image(type='pil', elem_id="input-img")
+        input_image = gr.Image(type="pil", elem_id="input-img")
+
         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')
-        
+            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")
-        
-        output_text = gr.Textbox(label="Description Output", elem_id="output-txt")
 
-        examples=[['27E894C4-9375-48A1-A95D-CB2425416B4B.png', "best",4], ['DB362F56-BA98-4CA1-A999-A25AA94B723B.png',"fast",4]]
-        ex = gr.Examples(examples=examples, fn=inference, inputs=[input_image, mode_input, flavor_input], outputs=[output_text], cache_examples=False, run_on_click=True)
-        
+        output_text = gr.Textbox(
+            label="Description Output",
+            elem_id="output-txt",
+        )
+
+        examples = [
+            ["27E894C4-9375-48A1-A95D-CB2425416B4B.png", "best", 4],
+            ["DB362F56-BA98-4CA1-A999-A25AA94B723B.png", "fast", 4],
+        ]
+
+        gr.Examples(
+            examples=examples,
+            fn=inference,
+            inputs=[input_image, mode_input, flavor_input],
+            outputs=[output_text],
+            cache_examples=False,
+            run_on_click=True,
+        )
+
         gr.HTML(article)
 
-    submit_btn.click(fn=inference, inputs=[input_image,mode_input,flavor_input], outputs=[output_text], api_name="clipi2")
-    
-demo.queue(max_size=32).launch(show_api=True, ssr_mode=False, mcp_server=True)
+    submit_btn.click(
+        fn=inference,
+        inputs=[input_image, mode_input, flavor_input],
+        outputs=[output_text],
+        api_name="clipi2",
+    )
+
+
+demo.queue(max_size=32).launch(
+    footer_links=["api"],
+    ssr_mode=False,
+    mcp_server=True,
+)