fix: Ensure encoding errors handled during SQL obfuscation for Trilogy

instrumentation/mysql2/test/opentelemetry/instrumentation/mysql2/instrumentation_test.rb

@@ -172,6 +172,19 @@
         _(span.attributes['net.peer.name']).must_equal host.to_s
         _(span.attributes['net.peer.port']).must_equal port.to_s
       end
+
+      it 'encodes invalid byte sequences for db.statement' do
+        # \255 is off-limits https://en.wikipedia.org/wiki/UTF-8#Codepage_layout
+        sql = "SELECT * from users where users.id = 1 and users.email = '[email protected]\255'"
+        obfuscated_sql = 'SELECT * from users where users.id = ? and users.email = ?'
+
+        expect do
+          client.query(sql)
+        end.must_raise Mysql2::Error
+
+        _(span.name).must_equal 'mysql'
+        _(span.attributes['db.statement']).must_equal obfuscated_sql
+      end
     end
 
     describe 'when db_statement set as omit' do

instrumentation/trilogy/lib/opentelemetry/instrumentation/trilogy/patches/client.rb

@@ -87,10 +87,13 @@ def obfuscate_sql(sql)
             if sql.size > 2000
               'SQL query too large to remove sensitive data ...'
             else
-              obfuscated = sql.gsub(FULL_SQL_REGEXP, '?')
+              obfuscated = OpenTelemetry::Common::Utilities.utf8_encode(sql, binary: true)
+              obfuscated = obfuscated.gsub(FULL_SQL_REGEXP, '?')
               obfuscated = 'Failed to obfuscate SQL query - quote characters remained after obfuscation' if detect_unmatched_pairs(obfuscated)
               obfuscated
             end
+          rescue StandardError
+            'OpenTelemetry error: failed to obfuscate sql'
           end
 
           def detect_unmatched_pairs(obfuscated)
@@ -140,6 +143,7 @@ def extract_statement_type(sql)
             QUERY_NAME_RE.match(sql) { |match| match[1].downcase } unless sql.nil?
           rescue StandardError => e
             OpenTelemetry.logger.error("Error extracting sql statement type: #{e.message}")
+            nil
           end
         end
       end

instrumentation/trilogy/test/opentelemetry/instrumentation/trilogy/instrumentation_test.rb

@@ -20,6 +20,7 @@
       port: port,
       username: username,
       password: password,
+      database: database,
       ssl: false
     }
   end
@@ -230,6 +231,19 @@
         _(span.name).must_equal 'select'
         _(span.attributes[OpenTelemetry::SemanticConventions::Trace::DB_STATEMENT]).must_equal obfuscated_sql
       end
+
+      it 'encodes invalid byte sequences for db.statement' do
+        # \255 is off-limits https://en.wikipedia.org/wiki/UTF-8#Codepage_layout
+        sql = "SELECT * from users where users.id = 1 and users.email = '[email protected]\255'"
+        obfuscated_sql = 'SELECT * from users where users.id = ? and users.email = ?'
+
+        expect do
+          client.query(sql)
+        end.must_raise Trilogy::Error
+
+        _(span.name).must_equal 'mysql'
+        _(span.attributes[OpenTelemetry::SemanticConventions::Trace::DB_STATEMENT]).must_equal obfuscated_sql
+      end
     end
 
     describe 'when db_statement is set to omit' do
@@ -245,5 +259,64 @@
         _(span.attributes[OpenTelemetry::SemanticConventions::Trace::DB_STATEMENT]).must_be_nil
       end
     end
+
+    describe 'when db_statement is configured via environment variable' do
+      describe 'when db_statement set as omit' do
+        it 'omits db.statement attribute' do
+          OpenTelemetry::TestHelpers.with_env('OTEL_RUBY_INSTRUMENTATION_TRILOGY_CONFIG_OPTS' => 'db_statement=omit;') do
+            instrumentation.instance_variable_set(:@installed, false)
+            instrumentation.install
+            sql = "SELECT * from users where users.id = 1 and users.email = '[email protected]'"
+            expect do
+              client.query(sql)
+            end.must_raise Trilogy::Error
+
+            _(span.attributes['db.system']).must_equal 'mysql'
+            _(span.name).must_equal 'select'
+            _(span.attributes[OpenTelemetry::SemanticConventions::Trace::DB_STATEMENT]).must_be_nil
+          end
+        end
+      end
+
+      describe 'when db_statement set as obfuscate' do
+        it 'obfuscates SQL parameters in db.statement' do
+          OpenTelemetry::TestHelpers.with_env('OTEL_RUBY_INSTRUMENTATION_TRILOGY_CONFIG_OPTS' => 'db_statement=obfuscate;') do
+            instrumentation.instance_variable_set(:@installed, false)
+            instrumentation.install
+
+            sql = "SELECT * from users where users.id = 1 and users.email = '[email protected]'"
+            obfuscated_sql = 'SELECT * from users where users.id = ? and users.email = ?'
+            expect do
+              client.query(sql)
+            end.must_raise Trilogy::Error
+
+            _(span.attributes['db.system']).must_equal 'mysql'
+            _(span.name).must_equal 'select'
+            _(span.attributes['db.statement']).must_equal obfuscated_sql
+          end
+        end
+      end
+
+      describe 'when db_statement is set differently than local config' do
+        let(:config) { { db_statement: :omit } }
+
+        it 'overrides local config and obfuscates SQL parameters in db.statement' do
+          OpenTelemetry::TestHelpers.with_env('OTEL_RUBY_INSTRUMENTATION_TRILOGY_CONFIG_OPTS' => 'db_statement=obfuscate') do
+            instrumentation.instance_variable_set(:@installed, false)
+            instrumentation.install
+
+            sql = "SELECT * from users where users.id = 1 and users.email = '[email protected]'"
+            obfuscated_sql = 'SELECT * from users where users.id = ? and users.email = ?'
+            expect do
+              client.query(sql)
+            end.must_raise Trilogy::Error
+
+            _(span.attributes['db.system']).must_equal 'mysql'
+            _(span.name).must_equal 'select'
+            _(span.attributes['db.statement']).must_equal obfuscated_sql
+          end
+        end
+      end
+    end
   end
 end