🚧 fetch_file

app.py

@@ -22,6 +22,7 @@ from tools import (
     fetch_raw_html,
     extract_links,
     extract_text_from_pdf,
+    fetch_file,
 )
 
 # (Keep Constants as is)
@@ -84,6 +85,7 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
                 fetch_raw_html,
                 extract_links,
                 extract_text_from_pdf,
+                fetch_file,
             ],
             verbosity_level=2,
             additional_authorized_imports=authorized_imports,

tools.py

@@ -6,6 +6,7 @@ import torch
 import spaces
 from transformers import pipeline
 from pdfminer.high_level import extract_text
+import os
 
 
 @tool
@@ -291,3 +292,33 @@ def extract_text_from_pdf(pdf_path: str) -> str:
         return text or ""
     except Exception:
         return ""
+
+
+@tool
+def fetch_file(url: str, save_path: str) -> str:
+    """
+    Download a file from a URL and save it locally, creating directories if needed.
+
+    Args:
+        url (str): Direct link to the file (e.g., a PDF).
+        save_path (str): Local file path where content will be saved
+                         (e.g., "input/paper.pdf").
+
+    Returns:
+        str: The local save_path if download and save succeeded;
+             empty string if an error occurred.
+    """
+    try:
+        parent_dir = os.path.dirname(save_path)
+        if parent_dir:
+            os.makedirs(parent_dir, exist_ok=True)
+
+        resp = requests.get(url, timeout=20)
+        resp.raise_for_status()
+
+        with open(save_path, "wb") as f:
+            f.write(resp.content)
+
+        return save_path
+    except Exception:
+        return ""