fix:JSON serialization problem

nano_graphrag/_op.py

@@ -1,5 +1,5 @@
 import re
-import json
+import json5
 import asyncio
 import tiktoken
 from typing import Union
@@ -680,7 +680,7 @@ async def _find_most_related_community_from_entities(
     for node_d in node_datas:
         if "clusters" not in node_d:
             continue
-        related_communities.extend(json.loads(node_d["clusters"]))
+        related_communities.extend(json5.loads(node_d["clusters"]))
     related_community_dup_keys = [
         str(dp["cluster"])
         for dp in related_communities

nano_graphrag/_storage/gdb_networkx.py

@@ -1,5 +1,5 @@
 import html
-import json
+import json5
 import os
 from collections import defaultdict
 from dataclasses import dataclass
@@ -154,7 +154,7 @@ async def community_schema(self) -> dict[str, SingleCommunitySchema]:
         for node_id, node_data in self._graph.nodes(data=True):
             if "clusters" not in node_data:
                 continue
-            clusters = json.loads(node_data["clusters"])
+            clusters = json5.loads(node_data["clusters"])
             this_node_edges = self._graph.edges(node_id)
 
             for cluster in clusters:
@@ -195,7 +195,7 @@ async def community_schema(self) -> dict[str, SingleCommunitySchema]:
 
     def _cluster_data_to_subgraphs(self, cluster_data: dict[str, list[dict[str, str]]]):
         for node_id, clusters in cluster_data.items():
-            self._graph.nodes[node_id]["clusters"] = json.dumps(clusters)
+            self._graph.nodes[node_id]["clusters"] = json5.dumps(clusters)
 
     async def _leiden_clustering(self):
         from graspologic.partition import hierarchical_leiden

nano_graphrag/_utils.py

@@ -1,5 +1,6 @@
 import asyncio
 import html
+import json5
 import json
 import logging
 import os
@@ -151,14 +152,14 @@ def compute_mdhash_id(content, prefix: str = ""):
 
 def write_json(json_obj, file_name):
     with open(file_name, "w", encoding="utf-8") as f:
-        json.dump(json_obj, f, indent=2, ensure_ascii=False)
+        json5.dump(json_obj, f, indent=2, ensure_ascii=False)
 
 
 def load_json(file_name):
     if not os.path.exists(file_name):
         return None
     with open(file_name, encoding="utf-8") as f:
-        return json.load(f)
+        return json5.load(f)
 
 
 # it's dirty to type, so it's a good way to have fun