deduplicate sources in local search context (#1468)

volksen · AlonsoGuevara · web-flow · commit b1f2ca785e42 · 2024-12-06T13:05:00.000-06:00
Co-authored-by: Alonso Guevara &lt;alonsog@microsoft.com&gt;
diff --git a/graphrag/query/structured_search/local_search/mixed_context.py b/graphrag/query/structured_search/local_search/mixed_context.py
@@ -334,6 +334,7 @@ def _build_text_unit_context(
                     num_relationships = count_relationships(
                         entity_relationships, selected_unit
                     )
+                    text_unit_ids_set.add(text_id)
                     unit_info_list.append((selected_unit, index, num_relationships))
 
         # sort by entity_order and the number of relationships desc

Original file line number	Diff line number	Diff line change
`@@ -334,6 +334,7 @@ def _build_text_unit_context(`
`334`	`334`	`num_relationships = count_relationships(`
`335`	`335`	`entity_relationships, selected_unit`
`336`	`336`	`)`
	`337`	`+ text_unit_ids_set.add(text_id)`
`337`	`338`	`unit_info_list.append((selected_unit, index, num_relationships))`
`338`	`339`
`339`	`340`	`# sort by entity_order and the number of relationships desc`