fix: raise a clear error for empty inputs in retrieval and answer evaluators

haystack/components/evaluators/answer_exact_match.py

@@ -53,6 +53,9 @@ def run(self, ground_truth_answers: list[str], predicted_answers: list[str]) ->
             - `score` - A number from 0.0 to 1.0 that represents the proportion of questions where any predicted
                          answer matched one of the ground truth answers.
         """
+        if len(ground_truth_answers) == 0 or len(predicted_answers) == 0:
+            raise ValueError("ground_truth_answers and predicted_answers must be provided.")
+
         if not len(ground_truth_answers) == len(predicted_answers):
             raise ValueError("The length of ground_truth_answers and predicted_answers must be the same.")
 

haystack/components/evaluators/document_map.py

@@ -108,6 +108,10 @@ def run(
             - `individual_scores` - A list of numbers from 0.0 to 1.0 that represents how high retrieved documents
                 are ranked.
         """
+        if len(ground_truth_documents) == 0 or len(retrieved_documents) == 0:
+            msg = "ground_truth_documents and retrieved_documents must be provided."
+            raise ValueError(msg)
+
         if len(ground_truth_documents) != len(retrieved_documents):
             msg = "The length of ground_truth_documents and retrieved_documents must be the same."
             raise ValueError(msg)

haystack/components/evaluators/document_mrr.py

@@ -106,6 +106,10 @@ def run(
             - `individual_scores` - A list of numbers from 0.0 to 1.0 that represents how high the first retrieved
                 document is ranked.
         """
+        if len(ground_truth_documents) == 0 or len(retrieved_documents) == 0:
+            msg = "ground_truth_documents and retrieved_documents must be provided."
+            raise ValueError(msg)
+
         if len(ground_truth_documents) != len(retrieved_documents):
             msg = "The length of ground_truth_documents and retrieved_documents must be the same."
             raise ValueError(msg)

haystack/components/evaluators/document_recall.py

@@ -172,6 +172,10 @@ def run(
             - `individual_scores` - A list of numbers from 0.0 to 1.0 that represents the proportion of matching
                 documents retrieved. If the mode is `single_hit`, the individual scores are 0 or 1.
         """
+        if len(ground_truth_documents) == 0 or len(retrieved_documents) == 0:
+            msg = "ground_truth_documents and retrieved_documents must be provided."
+            raise ValueError(msg)
+
         if len(ground_truth_documents) != len(retrieved_documents):
             msg = "The length of ground_truth_documents and retrieved_documents must be the same."
             raise ValueError(msg)

releasenotes/notes/evaluators-empty-input-value-error-c2e8b41d7a56f309.yaml

@@ -0,0 +1,7 @@
+---
+fixes:
+  - |
+    ``AnswerExactMatchEvaluator``, ``DocumentMAPEvaluator``, ``DocumentMRREvaluator`` and
+    ``DocumentRecallEvaluator`` now raise a descriptive ``ValueError`` when called with
+    empty input lists, instead of failing with an opaque ``ZeroDivisionError``. This
+    matches the existing behavior of ``DocumentNDCGEvaluator``.

test/components/evaluators/test_answer_exact_match.py

@@ -71,3 +71,12 @@ def test_run_with_different_lengths():
 
     with pytest.raises(ValueError):
         evaluator.run(ground_truth_answers=["Berlin", "Paris"], predicted_answers=["Berlin"])
+
+
+def test_run_with_empty_inputs():
+    evaluator = AnswerExactMatchEvaluator()
+
+    # Empty (equal-length) inputs previously fell through the length check and
+    # crashed with a bare ZeroDivisionError when averaging over zero items.
+    with pytest.raises(ValueError, match="must be provided"):
+        evaluator.run(ground_truth_answers=[], predicted_answers=[])

test/components/evaluators/test_document_map.py

@@ -147,3 +147,12 @@ def test_run_with_different_lengths():
             ground_truth_documents=[[Document(content="Berlin")], [Document(content="Paris")]],
             retrieved_documents=[[Document(content="Berlin")]],
         )
+
+
+def test_run_with_empty_inputs():
+    evaluator = DocumentMAPEvaluator()
+
+    # Empty (equal-length) inputs previously fell through the length check and
+    # crashed with a bare ZeroDivisionError when averaging over zero items.
+    with pytest.raises(ValueError, match="must be provided"):
+        evaluator.run(ground_truth_documents=[], retrieved_documents=[])

test/components/evaluators/test_document_mrr.py

@@ -150,3 +150,12 @@ def test_run_with_different_lengths():
             ground_truth_documents=[[Document(content="Berlin")], [Document(content="Paris")]],
             retrieved_documents=[[Document(content="Berlin")]],
         )
+
+
+def test_run_with_empty_inputs():
+    evaluator = DocumentMRREvaluator()
+
+    # Empty (equal-length) inputs previously fell through the length check and
+    # crashed with a bare ZeroDivisionError when averaging over zero items.
+    with pytest.raises(ValueError, match="must be provided"):
+        evaluator.run(ground_truth_documents=[], retrieved_documents=[])

test/components/evaluators/test_document_recall.py

@@ -68,6 +68,15 @@ def test_run_with_nested_meta_comparison():
     assert result == {"individual_scores": [0.5], "score": 0.5}
 
 
+def test_run_with_empty_inputs():
+    evaluator = DocumentRecallEvaluator()
+
+    # Empty (equal-length) inputs previously fell through the length check and
+    # crashed with a bare ZeroDivisionError when averaging over zero items.
+    with pytest.raises(ValueError, match="must be provided"):
+        evaluator.run(ground_truth_documents=[], retrieved_documents=[])
+
+
 class TestDocumentRecallEvaluatorSingleHit:
     @pytest.fixture
     def evaluator(self):