Update app.py

app.py

@@ -1,64 +1,93 @@
+import os
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
+import tempfile
+import shutil
+import pandas as pd
+from PIL import Image
+from preprocess import convert_pdf_to_images, preprocess_image
+from llm_utils import load_image, generate_response
 
+# Global temporary directory for image processing
+temp_dir = tempfile.mkdtemp()
+
+pdf_image_map = {}  # Map from PDF to its image list
+image_preview_map = {}  # Map from image name to full path
+
+
+def extract_images_from_pdfs(pdf_files):
+    global pdf_image_map, image_preview_map
+    pdf_image_map.clear()
+    image_preview_map.clear()
+
+    previews = []
+
+    for pdf in pdf_files:
+        image_paths = convert_pdf_to_images(pdf.name, temp_dir)
+        pdf_image_map[pdf.name] = image_paths
+
+        for img_path in image_paths:
+            img_name = os.path.basename(img_path)
+            image_preview_map[img_name] = img_path
+            previews.append((img_name, img_path))
+
+    # Return preview (tuples: filename, image_path)
+    return [img for _, img in previews]
+
+
+def process_selected_images(selected_images, output_excel_name):
+    results = []
+
+    for img_name in selected_images:
+        img_path = image_preview_map.get(img_name)
+        if img_path is None:
+            continue
+
+        processed = preprocess_image(img_path)
+        processed_path = os.path.join(temp_dir, f"processed_{img_name}")
+        Image.fromarray(processed).save(processed_path)
+
+        # Run LLM
+        pixel_values = load_image(processed_path)
+        response = generate_response(pixel_values)
+
+        # Clean filename to find source pdf
+        base_name = img_name.rsplit("_page_", 1)[0] + ".pdf"
+        results.append({"Source PDF": base_name, "Page Image": img_name, "LLM Output": response})
+
+    df = pd.DataFrame(results)
+    output_excel = os.path.join(temp_dir, output_excel_name)
+    df.to_excel(output_excel, index=False)
+    return output_excel
+
+
+def reset_all():
+    shutil.rmtree(temp_dir)
+    os.makedirs(temp_dir, exist_ok=True)
+
+
+with gr.Blocks() as demo:
+    gr.Markdown("## 🧠 PDF → Image → LLM → Excel Output")
+
+    with gr.Row():
+        pdf_input = gr.File(file_types=[".pdf"], file_count="multiple", label="📎 Upload multiple PDF files")
+        extract_btn = gr.Button("🔍 Extract Images")
+
+    gallery = gr.Gallery(label="🖼️ Choose images to process", columns=4, allow_preview=True, interactive=True, show_label=True).style(grid=4)
+    selected_images = gr.CheckboxGroup(choices=[], label="Select image filenames for processing")
+
+    with gr.Row():
+        output_name = gr.Textbox(label="📄 Output Excel filename", value="output.xlsx")
+        generate_btn = gr.Button("🚀 Generate Excel")
+
+    excel_output = gr.File(label="📥 Download Excel")
+
+    def update_gallery(pdf_files):
+        previews = extract_images_from_pdfs(pdf_files)
+        choices = list(image_preview_map.keys())
+        return gr.update(value=previews), gr.update(choices=choices, value=[])
+
+    extract_btn.click(update_gallery, inputs=[pdf_input], outputs=[gallery, selected_images])
+    generate_btn.click(fn=process_selected_images, inputs=[selected_images, output_name], outputs=[excel_output])
 
 if __name__ == "__main__":
     demo.launch()