From 47bf7364dbebb89acecddf653dc09e8f3970dc7a Mon Sep 17 00:00:00 2001
From: Mark Kuhn <kuhn.mark@outlook.com>
Date: Mon, 15 Aug 2022 10:25:57 -0700
Subject: [PATCH] add function to strip nonascii and stringify numbers (#84)

* Add function to strip non-ascii characters and stringify numbers

Co-authored-by: Mark Kuhn <kuhnmar@amazon.com>
---
 .../logger/metrics_context.py                 | 20 ++++----
 .../serializers/log_serializer.py             |  4 ++
 tests/serializer/test_log_serializer.py       | 50 +++++++++++++++++++
 3 files changed, 64 insertions(+), 10 deletions(-)

diff --git a/aws_embedded_metrics/logger/metrics_context.py b/aws_embedded_metrics/logger/metrics_context.py
index 47ed1dd..3f46894 100644
--- a/aws_embedded_metrics/logger/metrics_context.py
+++ b/aws_embedded_metrics/logger/metrics_context.py
@@ -58,30 +58,30 @@ def put_metric(self, key: str, value: float, unit: str = None) -> None:
             self.metrics[key] = Metric(value, unit)
 
     @staticmethod
-    def validate_dimension_set(dimensions: Dict[str, str]) -> None:
+    def validate_dimension_set(dimension_set: Dict[str, str]) -> None:
         """
         Validates dimension set length is not more than MAX_DIMENSION_SET_SIZE
         """
-        if len(dimensions) > MAX_DIMENSION_SET_SIZE:
+        if len(dimension_set) > MAX_DIMENSION_SET_SIZE:
             raise DimensionSetExceededError(
                 f"Maximum number of dimensions per dimension set allowed are {MAX_DIMENSION_SET_SIZE}")
 
-    def put_dimensions(self, dimensions: Dict[str, str]) -> None:
+    def put_dimensions(self, dimension_set: Dict[str, str]) -> None:
         """
         Adds dimensions to the context.
         ```
         context.put_dimensions({ "k1": "v1", "k2": "v2" })
         ```
         """
-        if dimensions is None:
+        if dimension_set is None:
             # TODO add ability to define failure strategy
             return
 
-        self.validate_dimension_set(dimensions)
+        self.validate_dimension_set(dimension_set)
 
-        self.dimensions.append(dimensions)
+        self.dimensions.append(dimension_set)
 
-    def set_dimensions(self, dimensionSets: List[Dict[str, str]]) -> None:
+    def set_dimensions(self, dimension_sets: List[Dict[str, str]]) -> None:
         """
         Overwrite all dimensions.
         ```
@@ -92,10 +92,10 @@ def set_dimensions(self, dimensionSets: List[Dict[str, str]]) -> None:
         """
         self.should_use_default_dimensions = False
 
-        for dimensionSet in dimensionSets:
-            self.validate_dimension_set(dimensionSet)
+        for dimension_set in dimension_sets:
+            self.validate_dimension_set(dimension_set)
 
-        self.dimensions = dimensionSets
+        self.dimensions = dimension_sets
 
     def set_default_dimensions(self, default_dimensions: Dict) -> None:
         """
diff --git a/aws_embedded_metrics/serializers/log_serializer.py b/aws_embedded_metrics/serializers/log_serializer.py
index 156e868..09e93fa 100644
--- a/aws_embedded_metrics/serializers/log_serializer.py
+++ b/aws_embedded_metrics/serializers/log_serializer.py
@@ -31,6 +31,10 @@ def serialize(context: MetricsContext) -> List[str]:
         dimensions_properties: Dict[str, str] = {}
 
         for dimension_set in context.get_dimensions():
+
+            # Stringify numerical values and strip non-ascii characters from dimension set values
+            dimension_set = {k: str(v).encode('ascii', 'ignore').decode('ascii') for k, v in dimension_set.items()}
+
             keys = list(dimension_set.keys())
             if len(keys) > MAX_DIMENSION_SET_SIZE:
                 err_msg = (f"Maximum number of dimensions per dimension set allowed are {MAX_DIMENSION_SET_SIZE}. "
diff --git a/tests/serializer/test_log_serializer.py b/tests/serializer/test_log_serializer.py
index e892922..650c4d0 100644
--- a/tests/serializer/test_log_serializer.py
+++ b/tests/serializer/test_log_serializer.py
@@ -32,6 +32,56 @@ def test_serialize_dimensions():
     assert_json_equality(result_json, expected)
 
 
+def test_serialize_dimensions_with_numeric_value():
+    # arrange
+    expected_key = fake.word()
+    value = fake.random.randrange(0, 100)
+    expected_value = str(value)
+
+    dimension = {}
+    dimension[expected_key] = value
+
+    expected_dimension = {}
+    expected_dimension[expected_key] = expected_value
+
+    expected = {**get_empty_payload(), **expected_dimension}
+    expected["_aws"]["CloudWatchMetrics"][0]["Dimensions"].append([expected_key])
+
+    context = get_context()
+    context.put_dimensions(dimension)
+
+    # act
+    result_json = serializer.serialize(context)[0]
+
+    # assert
+    assert_json_equality(result_json, expected)
+
+
+def test_serialize_dimenions_with_non_ascii_values():
+    # arrange
+    expected_key = fake.word()
+    value = "asciié🤔"
+    expected_value = "ascii"
+
+    dimension = {}
+    dimension[expected_key] = value
+
+    expected_dimension = {}
+    expected_dimension[expected_key] = expected_value
+
+    expected = {**get_empty_payload(), **expected_dimension}
+    expected["_aws"]["CloudWatchMetrics"][0]["Dimensions"].append([expected_key])
+
+    context = get_context()
+    context.put_dimensions(dimension)
+
+    # act
+    result_json = serializer.serialize(context)[0]
+
+    # assert
+    assert_json_equality(result_json, expected)
+
+
 def test_serialize_properties():
     # arrange
     expected_key = fake.word()