[monitor opentelemetry exporter] Add support for user id and authId

sdk/monitor/azure-monitor-opentelemetry-exporter/CHANGELOG.md

@@ -3,6 +3,8 @@
 ## 1.0.0b47 (Unreleased)
 
 ### Features Added
+- Add support for user id and authId
+([#44662](https://github.com/Azure/azure-sdk-for-python/pull/44662))
 
 ### Breaking Changes
 

sdk/monitor/azure-monitor-opentelemetry-exporter/azure/monitor/opentelemetry/exporter/export/logs/_exporter.py

@@ -5,20 +5,35 @@
 from typing import Optional, Sequence, Any
 
 from opentelemetry._logs.severity import SeverityNumber
+from opentelemetry.sdk._logs import ReadableLogRecord
+from opentelemetry.sdk._logs.export import LogRecordExporter, LogRecordExportResult
 from opentelemetry.semconv.attributes.exception_attributes import (
     EXCEPTION_ESCAPED,
     EXCEPTION_MESSAGE,
     EXCEPTION_STACKTRACE,
     EXCEPTION_TYPE,
 )
-from opentelemetry.sdk._logs import ReadableLogRecord
-from opentelemetry.sdk._logs.export import LogRecordExporter, LogRecordExportResult
+
+try:
+    from opentelemetry.semconv.logs import (
+        LogRecordAttributes as _SemconvLogRecordAttributes,
+    )
+except ImportError:
+    _SemconvLogRecordAttributes = None
+try:
+    from opentelemetry.semconv._incubating.attributes import (
+        enduser_attributes as _enduser_attributes,
+    )
+except ImportError:
+    _enduser_attributes = None
 
 from azure.monitor.opentelemetry.exporter import _utils
 from azure.monitor.opentelemetry.exporter._constants import (
+    _APPLICATION_INSIGHTS_EVENT_MARKER_ATTRIBUTE,
+    _DEFAULT_LOG_MESSAGE,
     _EXCEPTION_ENVELOPE_NAME,
     _MESSAGE_ENVELOPE_NAME,
-    _DEFAULT_LOG_MESSAGE,
+    _MICROSOFT_CUSTOM_EVENT_NAME,
 )
 from azure.monitor.opentelemetry.exporter._generated.models import (
     ContextTagKeys,
@@ -34,17 +49,24 @@
     ExportResult,
 )
 from azure.monitor.opentelemetry.exporter.export.trace import _utils as trace_utils
-from azure.monitor.opentelemetry.exporter._constants import (
-    _APPLICATION_INSIGHTS_EVENT_MARKER_ATTRIBUTE,
-    _MICROSOFT_CUSTOM_EVENT_NAME,
-)
 from azure.monitor.opentelemetry.exporter.statsbeat._state import (
     get_statsbeat_shutdown,
     get_statsbeat_custom_events_feature_set,
     is_statsbeat_enabled,
     set_statsbeat_custom_events_feature_set,
 )
 
+_ENDUSER_ID_ATTRIBUTE = (
+    getattr(_SemconvLogRecordAttributes, "ENDUSER_ID", None)
+    or getattr(_enduser_attributes, "ENDUSER_ID", None)
+    or "enduser.id"
+)
+_ENDUSER_PSEUDO_ID_ATTRIBUTE = (
+    getattr(_SemconvLogRecordAttributes, "ENDUSER_PSEUDO_ID", None)
+    or getattr(_enduser_attributes, "ENDUSER_PSEUDO_ID", None)
+    or "enduser.pseudo.id"
+)
+
 _logger = logging.getLogger(__name__)
 
 _DEFAULT_SPAN_ID = 0
@@ -123,27 +145,32 @@ def _convert_log_to_envelope(readable_log_record: ReadableLogRecord) -> Telemetr
     log_record = readable_log_record.log_record
     time_stamp = log_record.timestamp if log_record.timestamp is not None else log_record.observed_timestamp
     envelope = _utils._create_telemetry_item(time_stamp)
-    envelope.tags.update(_utils._populate_part_a_fields(readable_log_record.resource))  # type: ignore
-    envelope.tags[ContextTagKeys.AI_OPERATION_ID] = "{:032x}".format(  # type: ignore
-        log_record.trace_id or _DEFAULT_TRACE_ID
-    )
-    envelope.tags[ContextTagKeys.AI_OPERATION_PARENT_ID] = "{:016x}".format(  # type: ignore
+    tags = envelope.tags or {}
+    envelope.tags = tags
+    tags.update(_utils._populate_part_a_fields(readable_log_record.resource))  # type: ignore
+    tags[ContextTagKeys.AI_OPERATION_ID] = "{:032x}".format(log_record.trace_id or _DEFAULT_TRACE_ID)  # type: ignore
+    if log_record.attributes and _ENDUSER_ID_ATTRIBUTE in log_record.attributes:
+        tags[ContextTagKeys.AI_USER_AUTH_USER_ID] = log_record.attributes[_ENDUSER_ID_ATTRIBUTE]
+    if log_record.attributes and _ENDUSER_PSEUDO_ID_ATTRIBUTE in log_record.attributes:
+        tags[ContextTagKeys.AI_USER_ID] = log_record.attributes[_ENDUSER_PSEUDO_ID_ATTRIBUTE]
+
+    tags[ContextTagKeys.AI_OPERATION_PARENT_ID] = "{:016x}".format(  # type: ignore
         log_record.span_id or _DEFAULT_SPAN_ID
     )
     if (
         log_record.attributes
         and ContextTagKeys.AI_OPERATION_NAME in log_record.attributes
         and log_record.attributes[ContextTagKeys.AI_OPERATION_NAME] is not None
     ):
-        envelope.tags[ContextTagKeys.AI_OPERATION_NAME] = log_record.attributes.get(  # type: ignore
+        tags[ContextTagKeys.AI_OPERATION_NAME] = log_record.attributes.get(  # type: ignore
             ContextTagKeys.AI_OPERATION_NAME
         )
     if _utils._is_any_synthetic_source(log_record.attributes):
-        envelope.tags[ContextTagKeys.AI_OPERATION_SYNTHETIC_SOURCE] = "True"  # type: ignore
+        tags[ContextTagKeys.AI_OPERATION_SYNTHETIC_SOURCE] = "True"  # type: ignore
     # Special use case: Customers want to be able to set location ip on log records
     location_ip = trace_utils._get_location_ip(log_record.attributes)
     if location_ip:
-        envelope.tags[ContextTagKeys.AI_LOCATION_IP] = location_ip  # type: ignore
+        tags[ContextTagKeys.AI_LOCATION_IP] = location_ip  # type: ignore
     properties = _utils._filter_custom_properties(
         log_record.attributes, lambda key, val: not _is_ignored_attribute(key)  # type: ignore
     )
@@ -236,7 +263,7 @@ def _map_body_to_message(log_body: Any) -> str:
 
     try:
         return json.dumps(log_body)[:32768]
-    except:  # pylint: disable=bare-except
+    except Exception:  # pylint: disable=broad-except
         return str(log_body)[:32768]
 
 
@@ -252,6 +279,8 @@ def _is_ignored_attribute(key: str) -> bool:
         EXCEPTION_ESCAPED,
         _APPLICATION_INSIGHTS_EVENT_MARKER_ATTRIBUTE,
         _MICROSOFT_CUSTOM_EVENT_NAME,
+        _ENDUSER_ID_ATTRIBUTE,
+        _ENDUSER_PSEUDO_ID_ATTRIBUTE,
     )
 )
 

sdk/monitor/azure-monitor-opentelemetry-exporter/azure/monitor/opentelemetry/exporter/export/trace/_exporter.py

@@ -14,6 +14,13 @@
 )
 from opentelemetry.semconv.trace import DbSystemValues, SpanAttributes
 from opentelemetry.semconv._incubating.attributes import gen_ai_attributes
+
+try:
+    from opentelemetry.semconv._incubating.attributes import (
+        enduser_attributes as _enduser_attributes,
+    )
+except ImportError:
+    _enduser_attributes = None
 from opentelemetry.sdk.resources import Resource
 from opentelemetry.sdk.trace import ReadableSpan
 from opentelemetry.sdk.trace.export import SpanExporter, SpanExportResult
@@ -56,6 +63,17 @@
 
 __all__ = ["AzureMonitorTraceExporter"]
 
+_ENDUSER_ID_ATTRIBUTE = (
+    getattr(SpanAttributes, "ENDUSER_ID", None)
+    or (getattr(_enduser_attributes, "ENDUSER_ID", None) if _enduser_attributes is not None else None)
+    or "enduser.id"
+)
+_ENDUSER_PSEUDO_ID_ATTRIBUTE = (
+    getattr(SpanAttributes, "ENDUSER_PSEUDO_ID", None)
+    or (getattr(_enduser_attributes, "ENDUSER_PSEUDO_ID", None) if _enduser_attributes is not None else None)
+    or "enduser.pseudo.id"
+)
+
 _STANDARD_OPENTELEMETRY_ATTRIBUTE_PREFIXES = [
     "http.",
     "db.",
@@ -107,9 +125,7 @@ def __init__(self, **kwargs: Any):
         self._tracer_provider = kwargs.pop("tracer_provider", None)
         super().__init__(**kwargs)
 
-    def export(
-        self, spans: Sequence[ReadableSpan], **kwargs: Any  # pylint: disable=unused-argument
-    ) -> SpanExportResult:
+    def export(self, spans: Sequence[ReadableSpan], **_kwargs: Any) -> SpanExportResult:
         """Export span data.
 
         :param spans: Open Telemetry Spans to export.
@@ -222,8 +238,10 @@ def _convert_span_to_envelope(span: ReadableSpan) -> TelemetryItem:
     envelope = _utils._create_telemetry_item(start_time)
     envelope.tags.update(_utils._populate_part_a_fields(span.resource))
     envelope.tags[ContextTagKeys.AI_OPERATION_ID] = "{:032x}".format(span.context.trace_id)
-    if SpanAttributes.ENDUSER_ID in span.attributes:
-        envelope.tags[ContextTagKeys.AI_USER_ID] = span.attributes[SpanAttributes.ENDUSER_ID]
+    if _ENDUSER_ID_ATTRIBUTE in span.attributes:
+        envelope.tags[ContextTagKeys.AI_USER_AUTH_USER_ID] = span.attributes[_ENDUSER_ID_ATTRIBUTE]
+    if _ENDUSER_PSEUDO_ID_ATTRIBUTE in span.attributes:
+        envelope.tags[ContextTagKeys.AI_USER_ID] = span.attributes[_ENDUSER_PSEUDO_ID_ATTRIBUTE]
     if _utils._is_any_synthetic_source(span.attributes):
         envelope.tags[ContextTagKeys.AI_OPERATION_SYNTHETIC_SOURCE] = "True"
     if span.parent and span.parent.span_id:

sdk/monitor/azure-monitor-opentelemetry-exporter/tests/logs/test_logs.py

@@ -80,7 +80,26 @@ def setUpClass(cls):
                 severity_text="WARNING",
                 severity_number=SeverityNumber.WARN,
                 body="Test message",
-                attributes={"test": "attribute", "ai.operation.name": "TestOperationName"},
+                attributes={
+                    "test": "attribute",
+                    "ai.operation.name": "TestOperationName",
+                },
+            ),
+            resource=Resource.create(attributes={"asd": "test_resource"}),
+            instrumentation_scope=InstrumentationScope("test_name"),
+        )
+        cls._log_data_user_fields = _logs.ReadWriteLogRecord(
+            LogRecord(
+                timestamp=1646865018558419456,
+                context=ctx,
+                severity_text="WARNING",
+                severity_number=SeverityNumber.WARN,
+                body="Test message",
+                attributes={
+                    "test": "attribute",
+                    "enduser.id": "test-auth",
+                    "enduser.pseudo.id": "test-user",
+                },
             ),
             resource=Resource.create(attributes={"asd": "test_resource"}),
             instrumentation_scope=InstrumentationScope("test_name"),
@@ -92,7 +111,10 @@ def setUpClass(cls):
                 severity_text="WARNING",
                 severity_number=SeverityNumber.WARN,
                 body="",
-                attributes={"test": "attribute", "ai.operation.name": "TestOperationName"},
+                attributes={
+                    "test": "attribute",
+                    "ai.operation.name": "TestOperationName",
+                },
             ),
             resource=Resource.create(attributes={"asd": "test_resource"}),
             instrumentation_scope=InstrumentationScope("test_name"),
@@ -116,7 +138,10 @@ def setUpClass(cls):
                 severity_text="WARNING",
                 severity_number=SeverityNumber.WARN,
                 body={"foo": {"bar": "baz", "qux": 42}},
-                attributes={"test": "attribute", "ai.operation.name": "TestOperationName"},
+                attributes={
+                    "test": "attribute",
+                    "ai.operation.name": "TestOperationName",
+                },
             ),
             resource=Resource.create(attributes={"asd": "test_resource"}),
             instrumentation_scope=InstrumentationScope("test_name"),
@@ -246,7 +271,12 @@ def setUpClass(cls):
                 severity_text="EXCEPTION",
                 severity_number=SeverityNumber.FATAL,
                 body="test exception",
-                attributes={"test": "attribute", EXCEPTION_TYPE: "", EXCEPTION_MESSAGE: "", EXCEPTION_STACKTRACE: ""},
+                attributes={
+                    "test": "attribute",
+                    EXCEPTION_TYPE: "",
+                    EXCEPTION_MESSAGE: "",
+                    EXCEPTION_STACKTRACE: "",
+                },
             ),
             resource=Resource.create(attributes={"asd": "test_resource"}),
             instrumentation_scope=InstrumentationScope("test_name"),
@@ -258,7 +288,12 @@ def setUpClass(cls):
                 severity_text="EXCEPTION",
                 severity_number=SeverityNumber.FATAL,
                 body="",
-                attributes={"test": "attribute", EXCEPTION_TYPE: "", EXCEPTION_MESSAGE: "", EXCEPTION_STACKTRACE: ""},
+                attributes={
+                    "test": "attribute",
+                    EXCEPTION_TYPE: "",
+                    EXCEPTION_MESSAGE: "",
+                    EXCEPTION_STACKTRACE: "",
+                },
             ),
             resource=Resource.create(attributes={"asd": "test_resource"}),
             instrumentation_scope=InstrumentationScope("test_name"),
@@ -354,26 +389,44 @@ def test_log_to_envelope_partA(self):
         self.assertEqual(envelope.instrumentation_key, "1234abcd-5678-4efa-8abc-1234567890ab")
         self.assertIsNotNone(envelope.tags)
         self.assertEqual(
-            envelope.tags.get(ContextTagKeys.AI_DEVICE_ID), azure_monitor_context[ContextTagKeys.AI_DEVICE_ID]
+            envelope.tags.get(ContextTagKeys.AI_DEVICE_ID),
+            azure_monitor_context[ContextTagKeys.AI_DEVICE_ID],
         )
         self.assertEqual(
-            envelope.tags.get(ContextTagKeys.AI_DEVICE_LOCALE), azure_monitor_context[ContextTagKeys.AI_DEVICE_LOCALE]
+            envelope.tags.get(ContextTagKeys.AI_DEVICE_LOCALE),
+            azure_monitor_context[ContextTagKeys.AI_DEVICE_LOCALE],
         )
         self.assertEqual(
-            envelope.tags.get(ContextTagKeys.AI_DEVICE_TYPE), azure_monitor_context[ContextTagKeys.AI_DEVICE_TYPE]
+            envelope.tags.get(ContextTagKeys.AI_DEVICE_TYPE),
+            azure_monitor_context[ContextTagKeys.AI_DEVICE_TYPE],
         )
         self.assertEqual(
             envelope.tags.get(ContextTagKeys.AI_INTERNAL_SDK_VERSION),
             azure_monitor_context[ContextTagKeys.AI_INTERNAL_SDK_VERSION],
         )
 
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE), "testServiceNamespace.testServiceName")
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE_INSTANCE), "testServiceInstanceId")
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_INTERNAL_NODE_NAME), "testServiceInstanceId")
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE),
+            "testServiceNamespace.testServiceName",
+        )
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE_INSTANCE),
+            "testServiceInstanceId",
+        )
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_INTERNAL_NODE_NAME),
+            "testServiceInstanceId",
+        )
         trace_id = self._log_data.log_record.trace_id
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_OPERATION_ID), "{:032x}".format(trace_id))
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_OPERATION_ID),
+            "{:032x}".format(trace_id),
+        )
         span_id = self._log_data.log_record.span_id
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_OPERATION_PARENT_ID), "{:016x}".format(span_id))
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_OPERATION_PARENT_ID),
+            "{:016x}".format(span_id),
+        )
         self._log_data.resource = old_resource
         self.assertEqual(envelope.tags.get(ContextTagKeys.AI_OPERATION_NAME), "TestOperationName")
 
@@ -402,6 +455,15 @@ def test_log_to_envelope_log(self):
         self.assertEqual(envelope.data.base_data.severity_level, 2)
         self.assertEqual(envelope.data.base_data.properties["test"], "attribute")
 
+    def test_log_to_envelope_user_fields(self):
+        exporter = self._exporter
+        envelope = exporter._log_to_envelope(self._log_data_user_fields)
+
+        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_USER_AUTH_USER_ID), "test-auth")
+        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_USER_ID), "test-user")
+        self.assertNotIn("enduser.id", envelope.data.base_data.properties)
+        self.assertNotIn("enduser.pseudo.id", envelope.data.base_data.properties)
+
     def test_log_to_envelope_log_none(self):
         exporter = self._exporter
         envelope = exporter._log_to_envelope(self._log_data_none)
@@ -429,7 +491,10 @@ def test_log_to_envelope_log_complex_body(self):
         envelope = exporter._log_to_envelope(self._log_data_complex_body)
         self.assertEqual(envelope.name, "Microsoft.ApplicationInsights.Message")
         self.assertEqual(envelope.data.base_type, "MessageData")
-        self.assertEqual(envelope.data.base_data.message, json.dumps(self._log_data_complex_body.log_record.body))
+        self.assertEqual(
+            envelope.data.base_data.message,
+            json.dumps(self._log_data_complex_body.log_record.body),
+        )
         self.assertEqual(envelope.tags.get(ContextTagKeys.AI_OPERATION_NAME), "TestOperationName")
 
     def test_log_to_envelope_log_complex_body_not_serializeable(self):
@@ -438,7 +503,8 @@ def test_log_to_envelope_log_complex_body_not_serializeable(self):
         self.assertEqual(envelope.name, "Microsoft.ApplicationInsights.Message")
         self.assertEqual(envelope.data.base_type, "MessageData")
         self.assertEqual(
-            envelope.data.base_data.message, str(self._log_data_complex_body_not_serializeable.log_record.body)
+            envelope.data.base_data.message,
+            str(self._log_data_complex_body_not_serializeable.log_record.body),
         )
 
     def test_log_to_envelope_exception_with_string_message(self):
@@ -593,8 +659,14 @@ def test_log_to_envelope_synthetic_source(self):
         envelope = exporter._log_to_envelope(log_data)
 
         self.assertEqual(envelope.tags.get(ContextTagKeys.AI_OPERATION_SYNTHETIC_SOURCE), "True")
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE), "testServiceNamespace.testServiceName")
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE_INSTANCE), "testServiceInstanceId")
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE),
+            "testServiceNamespace.testServiceName",
+        )
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE_INSTANCE),
+            "testServiceInstanceId",
+        )
 
     def test_log_to_envelope_synthetic_load_always_on(self):
         exporter = self._exporter
@@ -631,8 +703,14 @@ def test_log_to_envelope_synthetic_load_always_on(self):
         envelope = exporter._log_to_envelope(log_data)
 
         self.assertEqual(envelope.tags.get(ContextTagKeys.AI_OPERATION_SYNTHETIC_SOURCE), "True")
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE), "testServiceNamespace.testServiceName")
-        self.assertEqual(envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE_INSTANCE), "testServiceInstanceId")
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE),
+            "testServiceNamespace.testServiceName",
+        )
+        self.assertEqual(
+            envelope.tags.get(ContextTagKeys.AI_CLOUD_ROLE_INSTANCE),
+            "testServiceInstanceId",
+        )
 
 
 class TestAzureLogExporterWithDisabledStorage(TestAzureLogExporter):