fix - make sure to pass a deep copy of merged_graph to summarizer as it overwrites it

Author: ksachdeva

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "langchain-graphrag"
-version = "0.0.5"
+version = "0.0.6"
 description = "Implementation of GraphRAG (https://arxiv.org/pdf/2404.16130)"
 authors = [{ name = "Kapil Sachdeva", email = "[email protected]" }]
 dependencies = [

src/langchain_graphrag/indexing/graph_generation/generator.py

@@ -1,3 +1,5 @@
+from copy import deepcopy
+
 import networkx as nx
 import pandas as pd
 
@@ -20,5 +22,7 @@ def __init__(
     def run(self, text_units: pd.DataFrame) -> tuple[nx.Graph, nx.Graph]:
         er_graphs = self._er_extractor.invoke(text_units)
         er_merged_graph = self._graphs_merger(er_graphs)
-        er_summarized_graph = self._er_description_summarizer.invoke(er_merged_graph)
+        er_summarized_graph = self._er_description_summarizer.invoke(
+            deepcopy(er_merged_graph)
+        )
         return er_merged_graph, er_summarized_graph