From 83cb3bbc9f2d9dd775749f8667aa04bc52d66f50 Mon Sep 17 00:00:00 2001
From: Boqi Chen <boqi.chen@mail.mcgill.ca>
Date: Mon, 17 Nov 2025 15:36:55 -0500
Subject: [PATCH] adapt arxiv search to new response format

---
 src/sherpa_ai/tools.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/src/sherpa_ai/tools.py b/src/sherpa_ai/tools.py
index 6297c1e9..c23b09c7 100644
--- a/src/sherpa_ai/tools.py
+++ b/src/sherpa_ai/tools.py
@@ -126,6 +126,10 @@ def _run(
         summaries = re.findall(summary_pattern, xml_content, re.DOTALL)
         title_pattern = r"<title>(.*?)</title>"
         titles = re.findall(title_pattern, xml_content, re.DOTALL)
+
+        # Note that arXiv may return on extra title for the search feed itself
+        titles = titles[len(titles) - len(summaries):]  # align titles and summaries
+
         id_pattern = r"<id>(.*?)</id>"
         ids = re.findall(id_pattern, xml_content, re.DOTALL)