fix: clarify single-page toc detection

pageindex/page_index.py

@@ -103,10 +103,15 @@ async def check_title_appearance_in_start_concurrent(structure, page_list, model
 
 def toc_detector_single_page(content, model=None):
     prompt = f"""
-    Your job is to detect if there is a table of content provided in the given text.
+    Your job is to detect if there is a table of content provided in the given single-page text.
 
     Given text: {content}
 
+    Guidance for single-page documents:
+    - Text with actual document content, including policies, rules, regulations, memos, or numbered sections, is not a table of contents just because it is structured.
+    - A true table of contents primarily lists references to content elsewhere in the document, usually as navigation entries for later sections or pages.
+    - If the given page contains the content itself instead of references to other content, answer "no".
+
     return the following JSON format:
     {{
         "thinking": <why do you think there is a table of content in the given text>
@@ -1151,4 +1156,4 @@ def validate_and_truncate_physical_indices(toc_with_page_number, page_list_lengt
     if truncated_items:
         print(f"Truncated {len(truncated_items)} TOC items that exceeded document length")
 
-    return toc_with_page_number
+    return toc_with_page_number

tests/test_toc_detector.py

@@ -0,0 +1,32 @@
+import importlib
+
+
+page_index = importlib.import_module("pageindex.page_index")
+
+
+def test_single_page_toc_prompt_distinguishes_content_from_toc(monkeypatch):
+    captured = {}
+
+    def fake_completion(model, prompt):
+        captured["model"] = model
+        captured["prompt"] = prompt
+        return '{"thinking": "structured content, not a toc", "toc_detected": "no"}'
+
+    monkeypatch.setattr(page_index, "llm_completion", fake_completion)
+
+    result = page_index.toc_detector_single_page(
+        "1. Scope\nThis policy applies to all staff.\n\n"
+        "2. Requirements\nUsers must rotate secrets.\n\n"
+        "3. Exceptions\nExceptions require approval.",
+        model="test-model",
+    )
+
+    assert result == "no"
+    assert captured["model"] == "test-model"
+
+    prompt = captured["prompt"].lower()
+    assert "single-page" in prompt
+    assert "actual document content" in prompt
+    assert "numbered sections" in prompt
+    assert "true table of contents" in prompt
+    assert "references to content elsewhere" in prompt