From 3367144d502c2e61cb51d3c52885901e7948c39b Mon Sep 17 00:00:00 2001
From: BukeLy <bukely0119@foxmail.com>
Date: Thu, 16 Jul 2026 18:44:01 +0800
Subject: [PATCH] fix(security): validate TOC identity before index updates

---
 pageindex/page_index.py  |  6 ++++++
 tests/test_page_index.py | 32 ++++++++++++++++++++++++++++++++
 2 files changed, 38 insertions(+)
 create mode 100644 tests/test_page_index.py

diff --git a/pageindex/page_index.py b/pageindex/page_index.py
index 235144b..8afe4a6 100644
--- a/pageindex/page_index.py
+++ b/pageindex/page_index.py
@@ -761,6 +761,12 @@ def process_toc_no_page_numbers(toc_content, toc_page_list, page_list,  start_in
             raise ValueError(
                 "LLM returned a different number of TOC entries than expected."
             )
+        if any(
+            (update.get("structure"), update.get("title"))
+            != (current.get("structure"), current.get("title"))
+            for update, current in zip(llm_result, toc_with_page_number)
+        ):
+            raise ValueError("LLM returned reordered or modified TOC entries.")
         valid_indices = _extract_chunk_marker_set(group_text)
         
         for idx, current in enumerate(toc_with_page_number):
diff --git a/tests/test_page_index.py b/tests/test_page_index.py
new file mode 100644
index 0000000..4c3631a
--- /dev/null
+++ b/tests/test_page_index.py
@@ -0,0 +1,32 @@
+import unittest
+from unittest.mock import Mock, patch
+
+from pageindex.page_index import process_toc_no_page_numbers
+
+
+class ProcessTocNoPageNumbersTest(unittest.TestCase):
+    def test_rejects_same_length_reordered_llm_toc(self):
+        toc = [
+            {"structure": "1", "title": "First"},
+            {"structure": "2", "title": "Second"},
+        ]
+        reordered = [
+            {"structure": "2", "title": "Second", "physical_index": "<physical_index_2>"},
+            {"structure": "1", "title": "First", "physical_index": "<physical_index_1>"},
+        ]
+
+        with patch("pageindex.page_index.toc_transformer", return_value=toc), \
+             patch("pageindex.page_index.count_tokens", return_value=1), \
+             patch("pageindex.page_index.page_list_to_group_text", return_value=["<physical_index_1> <physical_index_2>"]), \
+             patch("pageindex.page_index.add_page_number_to_toc", return_value=reordered):
+            with self.assertRaises(ValueError):
+                process_toc_no_page_numbers(
+                    "toc",
+                    [],
+                    [["page one"], ["page two"]],
+                    logger=Mock(),
+                )
+
+
+if __name__ == "__main__":
+    unittest.main()