From d204bd3b61033ad2956085304b79c3241c3a71c0 Mon Sep 17 00:00:00 2001
From: fabgpt-coder <276921870+fabgpt-coder@users.noreply.github.com>
Date: Sat, 6 Jun 2026 21:34:45 +0200
Subject: [PATCH] Guard against empty input in chunk_markdown()

The function currently proceeds to split an empty string into sections, which results in a single empty chunk that is then filtered out by the `min_chars` check. This triggers unnecessary regex compilation and list processing for no gain. Adding an early return for empty input improves performance and prevents the creation of transient empty state.
---
 agent/rag/chunker.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/agent/rag/chunker.py b/agent/rag/chunker.py
index 75c524f..36d5020 100644
--- a/agent/rag/chunker.py
+++ b/agent/rag/chunker.py
@@ -75,6 +75,9 @@ def chunk_markdown(
     min_chars: int = 200,
 ) -> list[Chunk]:
     """Heading-aware chunker. Returns chunks in document order."""
+    if not text or not text.strip():
+        return []
+    
     sections = _split_sections(text)
     chunks: list[Chunk] = []
     stack: list[str] = [""] * 7  # index by heading level