Disallow duplicate dimension sets

Author: Stephen-Bao

aws_embedded_metrics/logger/metrics_context.py

@@ -17,7 +17,7 @@
 from aws_embedded_metrics.constants import MAX_DIMENSION_SET_SIZE
 from aws_embedded_metrics.exceptions import DimensionSetExceededError
 from aws_embedded_metrics.logger.metric import Metric
-from typing import List, Dict, Any
+from typing import List, Dict, Any, Set
 
 
 class MetricsContext(object):
@@ -79,6 +79,12 @@ def put_dimensions(self, dimensions: Dict[str, str]) -> None:
 
         self.validate_dimension_set(dimensions)
 
+        # Duplicate dimension sets are removed before being added to the end of the collection.
+        # This ensures only latest dimension value is used as a target member on the root EMF node.
+        # This operation is O(n^2), but acceptable given sets are capped at 30 dimensions
+        incoming_keys: Set = set(dimensions.keys())
+        self.dimensions = list(filter(lambda dim: (set(dim.keys()) != incoming_keys), self.dimensions))
+
         self.dimensions.append(dimensions)
 
     def set_dimensions(self, dimensionSets: List[Dict[str, str]]) -> None:

tests/logger/test_metrics_context.py

@@ -54,6 +54,77 @@ def test_put_dimension_adds_to_dimensions():
     assert context.dimensions == [dimension_set]
 
 
+def test_put_dimensions_accept_multiple_unique_dimensions():
+    # arrange
+    context = MetricsContext()
+    dimension1 = {fake.word(): fake.word()}
+    dimension2 = {fake.word(): fake.word()}
+
+    # act
+    context.put_dimensions(dimension1)
+    context.put_dimensions(dimension2)
+
+    # assert
+    assert len(context.get_dimensions()) == 2
+    assert context.get_dimensions()[0] == dimension1
+    assert context.get_dimensions()[1] == dimension2
+
+
+def test_put_dimension_prevent_duplicate_dimensions():
+    # arrange
+    context = MetricsContext()
+    pair1 = [fake.word(), fake.word()]
+    pair2 = [fake.word(), fake.word()]
+
+    dimension1 = {pair1[0]: pair1[1]}
+    dimension2 = {pair2[0]: pair2[1]}
+    dimension3 = {pair1[0]: pair1[1], pair2[0]: pair2[1]}
+
+    # act
+    context.put_dimensions(dimension1)
+    context.put_dimensions(dimension2)
+    context.put_dimensions(dimension1)
+    context.put_dimensions(dimension3)
+    context.put_dimensions(dimension2)
+    context.put_dimensions(dimension3)
+
+    # assert
+    assert len(context.get_dimensions()) == 3
+    assert context.get_dimensions()[0] == dimension1
+    assert context.get_dimensions()[1] == dimension2
+    assert context.get_dimensions()[2] == dimension3
+
+
+def test_put_dimension_sort_duplicate_dimensions():
+    # arrange
+    context = MetricsContext()
+    key1 = fake.word()
+    key2 = fake.word()
+    val1 = fake.word()
+    val2 = fake.word()
+
+    dimension1 = {key1: val1}
+    dimension2 = {key2: val2}
+    dimension3 = {key1: val2}
+    dimension4 = {key2: val1}
+    dimension5 = {key1: val1, key2: val2}
+    dimension6 = {key1: val2, key2: val1}
+
+    # act
+    context.put_dimensions(dimension1)
+    context.put_dimensions(dimension2)
+    context.put_dimensions(dimension5)
+    context.put_dimensions(dimension3)
+    context.put_dimensions(dimension4)
+    context.put_dimensions(dimension6)
+
+    # assert
+    assert len(context.get_dimensions()) == 3
+    assert context.get_dimensions()[0] == dimension3
+    assert context.get_dimensions()[1] == dimension4
+    assert context.get_dimensions()[2] == dimension6
+
+
 def test_get_dimensions_returns_only_custom_dimensions_if_no_default_dimensions_not_set():
     # arrange
     context = MetricsContext()