fix(graphrag): count the findings dropped from evidences

Review asked for visibility on the new filter. Log a debug-level count of the
findings it omits, without their content, and hold it with an assertion so
removing the log fails the test.

Left the rest of index.py unformatted on purpose. ruff format rewrites a regex
list and two logging calls 200 lines away, and that churn does not belong here.

Author: marmar9615-cloud

rag/graphrag/general/index.py

@@ -936,13 +936,17 @@ async def save_community_checkpoint(checkpoint_key: str, payload):
 
     chunks = []
     for stru, rep in zip(community_structure, community_reports):
+        # A finding may be a plain string rather than a dict, which the schema check allows
+        # because it only requires ("findings", list). CommunityReportsExtractor
+        # ._get_text_output handles both shapes. A string finding carries no separate
+        # explanation, and its text still reaches content_ltks through the report.
+        dict_findings = [f for f in stru["findings"] if isinstance(f, dict)]
+        skipped_findings = len(stru["findings"]) - len(dict_findings)
+        if skipped_findings:
+            logging.debug("Omitted %d non-dict findings from community evidences", skipped_findings)
         obj = {
             "report": rep,
-            # A finding may be a plain string rather than a dict, which the schema check
-            # allows because it only requires ("findings", list). CommunityReportsExtractor
-            # ._get_text_output handles both shapes. A string finding carries no separate
-            # explanation, and its text still reaches content_ltks through the report.
-            "evidences": "\n".join([f.get("explanation", "") for f in stru["findings"] if isinstance(f, dict)]),
+            "evidences": "\n".join([f.get("explanation", "") for f in dict_findings]),
         }
         # Deterministic id derived from (kb_id, community title) so reruns of
         # extract_community produce stable ids.  Combined with insert-then-

test/unit_test/rag/graphrag/test_community_chunk_findings.py

@@ -21,6 +21,7 @@
 """
 
 import json
+import logging
 from types import SimpleNamespace
 from unittest.mock import MagicMock
 
@@ -73,13 +74,15 @@ async def fake_insert(chunks, *_args, **_kwargs):
 class TestCommunityChunkFindings:
     @pytest.mark.p2
     @pytest.mark.asyncio
-    async def test_string_findings_do_not_break_chunk_building(self, monkeypatch):
+    async def test_string_findings_do_not_break_chunk_building(self, monkeypatch, caplog):
         structure = [{"title": "Community", "weight": 1.0, "entities": ["A"], "findings": ["Alpha matters", _DICT_FINDING]}]
 
-        chunks = await _run_extract_community(monkeypatch, structure, ["report text"])
+        with caplog.at_level(logging.DEBUG, logger="root"):
+            chunks = await _run_extract_community(monkeypatch, structure, ["report text"])
 
         assert len(chunks) == 1
         assert json.loads(chunks[0]["content_with_weight"])["evidences"] == "Because of beta"
+        assert "Omitted 1 non-dict findings" in caplog.text
 
     @pytest.mark.p2
     @pytest.mark.asyncio