From bdaadd79f9aed8a8de3eb54231f63d531bc223c4 Mon Sep 17 00:00:00 2001
From: Johannes Wesch <johannes.wesch@aleph-alpha.com>
Date: Tue, 6 Feb 2024 09:28:03 +0100
Subject: [PATCH] added break for recursive summary when number of partial
 summary doesn't change

---
 src/intelligence_layer/core/chunk.py | 4 +---
 tests/core/test_chunk.py             | 5 +++--
 2 files changed, 4 insertions(+), 5 deletions(-)

diff --git a/src/intelligence_layer/core/chunk.py b/src/intelligence_layer/core/chunk.py
index 0bdd6b566..92b3d8c2f 100644
--- a/src/intelligence_layer/core/chunk.py
+++ b/src/intelligence_layer/core/chunk.py
@@ -96,9 +96,7 @@ def __init__(
                     overlap_length_tokens, max_tokens_per_chunk
                 )
             )
-        self.chunk_task = ChunkTask(
-            client, model, overlap_length_tokens // 2
-        )
+        self.chunk_task = ChunkTask(client, model, overlap_length_tokens // 2)
         self.tokenizer = client.tokenizer(model)
         self.max_tokens_per_chunk = max_tokens_per_chunk
         self.overlap_length_tokens = overlap_length_tokens
diff --git a/tests/core/test_chunk.py b/tests/core/test_chunk.py
index b2c6ca1bc..888e5d912 100644
--- a/tests/core/test_chunk.py
+++ b/tests/core/test_chunk.py
@@ -37,14 +37,15 @@ def test_overlapped_chunking(
         print(first)
 
         assert (
-            len(first) <= MAX_TOKENS + 2
+            len(first)
+            <= MAX_TOKENS + 2
             # `+2` because re-tokenizing the chunk can add a few extra tokens at
             # the beginning or end of each chunk. This is a hack.
         )
         next = output_tokenized[chunk_index + 1].tokens
 
         found = False
-        for offset in range(len(next)-OVERLAP//2):
+        for offset in range(len(next) - OVERLAP // 2):
             if first[-OVERLAP // 2 :] != next[offset : offset + OVERLAP // 2]:
                 continue
             found = True