fix: fix type annotation

Author: ChenZiHong-Gavin

graphgen/graphgen.py

@@ -119,7 +119,6 @@ async def insert(self):
         # TODO: configurable whether to use coreference resolution
 
         # Step 2: Split chunks and filter existing ones
-        inserting_chunks = {}
         assert isinstance(data, list) and isinstance(data[0], dict)
         new_docs = {
             compute_content_hash(doc["content"], prefix="doc-"): {

graphgen/operators/split/split_chunks.py

@@ -3,7 +3,11 @@
 
 from tqdm.asyncio import tqdm as tqdm_async
 
-from graphgen.models import ChineseRecursiveTextSplitter, RecursiveCharacterSplitter
+from graphgen.models import (
+    ChineseRecursiveTextSplitter,
+    RecursiveCharacterSplitter,
+    Tokenizer,
+)
 from graphgen.utils import compute_content_hash, detect_main_language
 
 _MAPPING = {
@@ -35,7 +39,7 @@ async def chunk_documents(
     new_docs: dict,
     chunk_size: int = 1024,
     chunk_overlap: int = 100,
-    tokenizer_instance=None,
+    tokenizer_instance: Tokenizer = None,
     progress_bar=None,
 ) -> dict:
     inserting_chunks = {}