Update app.py

app.py

@@ -124,12 +124,9 @@ def index_from_url(url: str) -> Tuple[str, str]:
     Download a PDF from URL and index it.
     Returns: (status_message, saved_pdf_path)
     """
-    tmp_dir = tempfile.mkdtemp(prefix="colpali_")
-    local_path = os.path.join(tmp_dir, "document.pdf")
-    urlretrieve(url, local_path)
-    status = index_from_path(local_path)
-    return status, local_path
-
+    images = load_dataset(url, "corpus", split="test")["image"]
+    status = index_gpu(images)
+    return status, "path"
 
 def query_gpt(query: str, retrieved_images: list[tuple[Image.Image, str]]) -> str:
     """Calls OpenAI's GPT model with the query and image data."""
@@ -596,41 +593,41 @@ def build_ui():
         )
 
         # ---- Tab 1: Index & Preview
-        # with gr.Tab("1) Index & Preview"):
-        #     with gr.Row():
-        #         with gr.Column(scale=1):
-        #             pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
-        #             index_btn = gr.Button("📥 Index Uploaded PDF", variant="secondary")
-        #             url_box = gr.Textbox(
-        #                 label="Or index from URL",
-        #                 placeholder="https://example.com/file.pdf",
-        #                 value="",
-        #             )
-        #             index_url_btn = gr.Button("🌐 Load From URL", variant="secondary")
-        #             status_box = gr.Textbox(label="Status", interactive=False)
-        #         with gr.Column(scale=2):
-        #             pdf_view = PDF(label="PDF Preview")
-
-        #     # wiring
-        #     def handle_upload(file):
-        #         global current_pdf_path
-        #         if file is None:
-        #             return "Please upload a PDF.", None
-        #         path = getattr(file, "name", file)
-        #         status = index_from_path(path)
-        #         current_pdf_path = path
-        #         return status, path
-
-        #     def handle_url(url: str):
-        #         global current_pdf_path
-        #         if not url or not url.lower().endswith(".pdf"):
-        #             return "Please provide a direct PDF URL ending in .pdf", None
-        #         status, path = index_from_url(url)
-        #         current_pdf_path = path
-        #         return status, path
-
-        #     index_btn.click(handle_upload, inputs=[pdf_input], outputs=[status_box, pdf_view])
-        #     index_url_btn.click(handle_url, inputs=[url_box], outputs=[status_box, pdf_view])
+        with gr.Tab("1) Index & Preview"):
+            with gr.Row():
+                with gr.Column(scale=1):
+                    pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
+                    index_btn = gr.Button("📥 Index Uploaded PDF", variant="secondary")
+                    url_box = gr.Textbox(
+                        label="Or index from HF Dataset",
+                        placeholder="manu/test-dataset",
+                        value="",
+                    )
+                    index_url_btn = gr.Button("🌐 Load From HF Datset", variant="secondary")
+                    status_box = gr.Textbox(label="Status", interactive=False)
+                with gr.Column(scale=2):
+                    pdf_view = PDF(label="PDF Preview")
+
+            # wiring
+            def handle_upload(file):
+                global current_pdf_path
+                if file is None:
+                    return "Please upload a PDF.", None
+                path = getattr(file, "name", file)
+                status = index_from_path(path)
+                current_pdf_path = path
+                return status, path
+
+            def handle_url(url: str):
+                global current_pdf_path
+                if not url or not url.lower().endswith(".pdf"):
+                    return "Please provide a direct PDF URL ending in .pdf", None
+                status, path = index_from_url(url)
+                current_pdf_path = path
+                return status, path
+
+            index_btn.click(handle_upload, inputs=[pdf_input], outputs=[status_box, pdf_view])
+            index_url_btn.click(handle_url, inputs=[url_box], outputs=[status_box, pdf_view])
 
         # ---- Tab 2: Ask (Direct — returns indices)
         with gr.Tab("2) Direct Search"):
@@ -716,9 +713,6 @@ def build_ui():
 
 if __name__ == "__main__":
     demo = build_ui()
-    images = load_dataset("vidore/esg_reports_human_labeled_v2", "corpus", split="test")["image"]
-    print("Indexing")
-    print(index_gpu(images))
     # mcp_server=True exposes this app's MCP endpoint at /gradio_api/mcp/
     # We keep the MCP server available, but the agent never uses MCP to pass images.
     demo.queue(max_size=5).launch(debug=True, mcp_server=True)