diff --git a/cognee/tasks/chunks/chunk_by_paragraph.py b/cognee/tasks/chunks/chunk_by_paragraph.py index c6abd5b9d..a05451bd3 100644 --- a/cognee/tasks/chunks/chunk_by_paragraph.py +++ b/cognee/tasks/chunks/chunk_by_paragraph.py @@ -67,7 +67,7 @@ def chunk_by_paragraph(data: str, paragraph_length: int = 1024, batch_paragraphs "chunk_id": uuid5(NAMESPACE_OID, current_chunk), "paragraph_ids": paragraph_ids, "chunk_index": chunk_index, - "cut_type": last_cut_type + "cut_type": "sentence_cut" if last_cut_type == "word" else last_cut_type } diff --git a/cognee/tests/unit/processing/chunks/chunk_by_paragraph_test.py b/cognee/tests/unit/processing/chunks/chunk_by_paragraph_test.py index 55eeb7fae..3ddc6f4f5 100644 --- a/cognee/tests/unit/processing/chunks/chunk_by_paragraph_test.py +++ b/cognee/tests/unit/processing/chunks/chunk_by_paragraph_test.py @@ -32,7 +32,7 @@ GROUND_TRUTH = { { "text": "\nThird paragraph is cut and is missing the dot at the end", "word_count": 12, - "cut_type": "word", + "cut_type": "sentence_cut", }, ], }