Fix span attribute value truncation (#173)

Author: aabmass

opentelemetry-exporter-gcp-trace/CHANGELOG.md

@@ -5,6 +5,9 @@
 - Add optional resource attributes to trace spans with regex
   ([#145](https://github.com/GoogleCloudPlatform/opentelemetry-operations-python/pull/145))
 - Upgrade `google-cloud-trace` dependency to version 1.1 or newer.
+  ([#170](https://github.com/GoogleCloudPlatform/opentelemetry-operations-python/pull/170))
+- Fix span attribute value truncation
+  ([#173](https://github.com/GoogleCloudPlatform/opentelemetry-operations-python/pull/173))
 
 ## Version 1.0.0
 

opentelemetry-exporter-gcp-trace/src/opentelemetry/exporter/cloud_trace/__init__.py

@@ -89,6 +89,8 @@
 MAX_EVENT_ATTRS = 4
 MAX_LINK_ATTRS = 32
 MAX_SPAN_ATTRS = 32
+MAX_ATTR_KEY_BYTES = 128
+MAX_ATTR_VAL_BYTES = 16 * 1024  # 16 kilobytes
 
 
 class CloudTraceSpanExporter(SpanExporter):
@@ -437,7 +439,7 @@ def _extract_attributes(
     )  # type: BoundedDict[str, trace_types.AttributeValue]
     invalid_value_dropped_count = 0
     for ot_key, ot_value in attrs.items() if attrs else []:
-        key = _truncate_str(ot_key, 128)[0]
+        key = _truncate_str(ot_key, MAX_ATTR_KEY_BYTES)[0]
         if key in LABELS_MAPPING:  # pylint: disable=consider-using-get
             key = LABELS_MAPPING[key]
         value = _format_attribute_value(ot_value)
@@ -485,14 +487,16 @@ def _format_attribute_value(
         value_type = "int_value"
     elif isinstance(value, str):
         value_type = "string_value"
-        value = _get_truncatable_str_object(value, 256)
+        value = _get_truncatable_str_object(value, MAX_ATTR_VAL_BYTES)
     elif isinstance(value, float):
         value_type = "string_value"
-        value = _get_truncatable_str_object("{:0.4f}".format(value), 256)
+        value = _get_truncatable_str_object(
+            "{:0.4f}".format(value), MAX_ATTR_VAL_BYTES
+        )
     elif isinstance(value, SequenceABC):
         value_type = "string_value"
         value = _get_truncatable_str_object(
-            ",".join(str(x) for x in value), 256
+            ",".join(str(x) for x in value), MAX_ATTR_VAL_BYTES
         )
     else:
         logger.warning(

opentelemetry-exporter-gcp-trace/tests/test_cloud_trace_exporter.py

@@ -96,6 +96,8 @@ def setUpClass(cls):
         cls.example_span_id = "95bb5edabd45950f"
         cls.example_time_in_ns = 1589919268850900051
         cls.example_time_stamp = _get_time_from_ns(cls.example_time_in_ns)
+        cls.str_20kb = "a" * 20 * 1024
+        cls.str_16kb = "a" * 16 * 1024
         cls.str_300 = "a" * 300
         cls.str_256 = "a" * 256
         cls.str_128 = "a" * 128
@@ -326,11 +328,24 @@ def test_agent_attribute_priority(self):
         )
 
     def test_attribute_value_truncation(self):
+        # shouldn't truncate
         self.assertEqual(
             _format_attribute_value(self.str_300),
             AttributeValue(
                 string_value=TruncatableString(
-                    value=self.str_256, truncated_byte_count=300 - 256
+                    value=self.str_300,
+                    truncated_byte_count=0,
+                )
+            ),
+        )
+
+        # huge string should truncate
+        self.assertEqual(
+            _format_attribute_value(self.str_20kb),
+            AttributeValue(
+                string_value=TruncatableString(
+                    value=self.str_16kb,
+                    truncated_byte_count=(20 - 16) * 1024,
                 )
             ),
         )