feat(pymongo): introduce db.operation, refactor db.statement

This change updates the opentelemetry-instrumentation-pymongo to
better align with OpenTelemetry semantic conventions and
the behavior of other language instrumentations (e.g. NodeJS, Ruby, etc.).

see https://github.com/open-telemetry/opentelemetry-js-contrib/blob/main/plugins/node/opentelemetry-instrumentation-mongodb/src/instrumentation.ts

The key changes are:
 - The `db.operation` attribute is now used to record the MongoDB command name (e.g., find, insert).
 - The `db.statement` attribute is no longer populated by default. It is only recorded if the `capture_statement = True` option is passed to the instrumentor.
 - The span name has been made more specific, changing from `{db_name}.{command}` to `{db_name}.{collection}.{command}` when a collection is present.

This provides more consistent and conventional telemetry data for MongoDB operations.

Author: smoke

instrumentation/opentelemetry-instrumentation-pymongo/src/opentelemetry/instrumentation/pymongo/__init__.py

@@ -136,8 +136,7 @@ def started(self, event: monitoring.CommandStartedEvent):
         if not self.is_enabled or not is_instrumentation_enabled():
             return
         command_name = event.command_name
-        span_name = f"{event.database_name}.{command_name}"
-        statement = self._get_statement_by_command_name(command_name, event)
+        span_name = _get_span_name(event)
         collection = event.command.get(event.command_name)
 
         try:
@@ -147,7 +146,12 @@ def started(self, event: monitoring.CommandStartedEvent):
                     SpanAttributes.DB_SYSTEM, DbSystemValues.MONGODB.value
                 )
                 span.set_attribute(SpanAttributes.DB_NAME, event.database_name)
-                span.set_attribute(SpanAttributes.DB_STATEMENT, statement)
+                span.set_attribute(SpanAttributes.DB_OPERATION, command_name)
+                if self.capture_statement:
+                    span.set_attribute(
+                        SpanAttributes.DB_STATEMENT,
+                        _get_statement(event),
+                    )
                 if collection:
                     span.set_attribute(
                         SpanAttributes.DB_MONGODB_COLLECTION, collection
@@ -210,16 +214,21 @@ def failed(self, event: monitoring.CommandFailedEvent):
     def _pop_span(self, event: CommandEvent) -> Span | None:
         return self._span_dict.pop(_get_span_dict_key(event), None)
 
-    def _get_statement_by_command_name(
-        self, command_name: str, event: CommandEvent
-    ) -> str:
-        statement = command_name
-        command_attribute = COMMAND_TO_ATTRIBUTE_MAPPING.get(command_name)
-        command = event.command.get(command_attribute)
-        if command and self.capture_statement:
-            statement += " " + str(command)
-        return statement
 
+def _get_span_name(event: CommandEvent) -> str:
+    """Get the span name for a given pymongo event."""
+    command_name = event.command_name
+    collection = event.command.get(command_name)
+    if collection:
+        return f"{event.database_name}.{collection}.{command_name}"
+    return f"{event.database_name}.{command_name}"
+
+def _get_statement(event: CommandEvent) -> str:
+    """Get the statement for a given pymongo event."""
+    command_name = event.command_name
+    command_attribute = COMMAND_TO_ATTRIBUTE_MAPPING.get(command_name)
+    command = event.command.get(command_attribute)
+    return f"{command}"
 
 def _get_span_dict_key(
     event: CommandEvent,

instrumentation/opentelemetry-instrumentation-pymongo/tests/test_pymongo.py

@@ -62,7 +62,8 @@ def test_started(self):
         self.assertEqual(
             span.attributes[SpanAttributes.DB_NAME], "database_name"
         )
-        self.assertEqual(span.attributes[SpanAttributes.DB_STATEMENT], "find")
+        self.assertEqual(span.attributes[SpanAttributes.DB_OPERATION], "find")
+        self.assertNotIn(SpanAttributes.DB_STATEMENT, span.attributes)
         self.assertEqual(
             span.attributes[SpanAttributes.NET_PEER_NAME], "test.com"
         )
@@ -210,7 +211,10 @@ def test_capture_statement_getmore(self):
 
         self.assertEqual(
             span.attributes[SpanAttributes.DB_STATEMENT],
-            "getMore test_collection",
+            "test_collection",
+        )
+        self.assertEqual(
+            span.attributes[SpanAttributes.DB_OPERATION], "getMore"
         )
 
     def test_capture_statement_aggregate(self):
@@ -232,10 +236,13 @@ def test_capture_statement_aggregate(self):
         self.assertEqual(len(spans_list), 1)
         span = spans_list[0]
 
-        expected_statement = f"aggregate {pipeline}"
+        expected_statement = f"{pipeline}"
         self.assertEqual(
             span.attributes[SpanAttributes.DB_STATEMENT], expected_statement
         )
+        self.assertEqual(
+            span.attributes[SpanAttributes.DB_OPERATION], "aggregate"
+        )
 
     def test_capture_statement_disabled_getmore(self):
         command_attrs = {
@@ -253,9 +260,11 @@ def test_capture_statement_disabled_getmore(self):
         span = spans_list[0]
 
         self.assertEqual(
-            span.attributes[SpanAttributes.DB_STATEMENT], "getMore"
+            span.attributes[SpanAttributes.DB_OPERATION], "getMore"
         )
 
+        self.assertNotIn(SpanAttributes.DB_STATEMENT, span.attributes)
+
     def test_capture_statement_disabled_aggregate(self):
         pipeline = [{"$match": {"status": "active"}}]
         command_attrs = {
@@ -273,9 +282,11 @@ def test_capture_statement_disabled_aggregate(self):
         span = spans_list[0]
 
         self.assertEqual(
-            span.attributes[SpanAttributes.DB_STATEMENT], "aggregate"
+            span.attributes[SpanAttributes.DB_OPERATION], "aggregate"
         )
 
+        self.assertNotIn(SpanAttributes.DB_STATEMENT, span.attributes)
+
 
 class MockCommand:
     def __init__(self, command_attrs):