From 10e8690890406285ad02ab7d073f49f8cbffae87 Mon Sep 17 00:00:00 2001 From: Wang Qi Date: Mon, 1 Jun 2026 13:05:54 +0800 Subject: [PATCH] GraphRAG - NER - spacy - fix spacy extraction (#14783) Fix spacy extraction --- rag/graphrag/general/index.py | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/rag/graphrag/general/index.py b/rag/graphrag/general/index.py index 98e7ab8fd3..265ea07598 100644 --- a/rag/graphrag/general/index.py +++ b/rag/graphrag/general/index.py @@ -331,12 +331,13 @@ async def run_graphrag_for_kb( callback(msg=f"[GraphRAG] chunk_list returned {len(raw_chunks)} raw chunks for doc:{doc_id}") + contents = [content for chunk in raw_chunks if (content := chunk.get("content_with_weight", "")) +] # For NER-based extractionm, no need to batch extract entity and relation if _select_extractor_type(graphrag_config) == "ner": - return raw_chunks + return contents - for d in raw_chunks: - content = d["content_with_weight"] + for content in contents: if num_tokens_from_string(current_chunk + content) < batch_chunk_token_size: current_chunk += content else: