Dedupe conflict document pairs

tests/test_conflicts.py

@@ -67,6 +67,13 @@ def test_chunks_from_one_document_are_not_a_conflict_with_themselves():
     assert detect_conflicts([_hit("hr-policy"), _hit("hr-policy")]) == []
 
 
+def test_duplicate_chunks_from_two_documents_report_one_conflict_pair():
+    hits = [_hit("hr-policy"), _hit("hr-policy"), _hit("finance-policy"), _hit("finance-policy")]
+    conflicts = detect_conflicts(hits)
+    assert len(conflicts) == 1
+    assert {conflicts[0].doc_a, conflicts[0].doc_b} == {"hr-policy", "finance-policy"}
+
+
 def test_a_single_source_of_truth_produces_no_conflict():
     assert detect_conflicts([_hit("hr-policy")]) == []
 

vaultrag/conflicts.py

@@ -48,12 +48,17 @@ def detect_conflicts(hits: list[Hit]) -> list[Conflict]:
     both answering the same question. That is worth a flag even when they happen to agree.
     """
     out: list[Conflict] = []
+    seen_pairs: set[frozenset[str]] = set()
     official = [h for h in hits if h.is_official]
 
     for i, a in enumerate(official):
         for b in official[i + 1 :]:
             if a.doc_id == b.doc_id:
                 continue
+            pair = frozenset((a.doc_id, b.doc_id))
+            if pair in seen_pairs:
+                continue
+            seen_pairs.add(pair)
             out.append(
                 Conflict(
                     doc_a=a.doc_id,
@@ -70,6 +75,10 @@ def detect_conflicts(hits: list[Hit]) -> list[Conflict]:
     for a in official:
         for b in unofficial:
             if a.doc_id != b.doc_id:
+                pair = frozenset((a.doc_id, b.doc_id))
+                if pair in seen_pairs:
+                    continue
+                seen_pairs.add(pair)
                 out.append(
                     Conflict(
                         doc_a=a.doc_id,