UTF-8 encode extract_statement sql

open-telemetry · Oct 10, 2023 · b8b2605 · b8b2605
1 parent 5fa5039
commit b8b2605
Show file tree

Hide file tree

Showing 5 changed files with 24 additions and 3 deletions.
diff --git a/helpers/mysql/lib/opentelemetry/helpers/mysql.rb b/helpers/mysql/lib/opentelemetry/helpers/mysql.rb
@@ -3,6 +3,7 @@
 # Copyright The OpenTelemetry Authors
 #
 # SPDX-License-Identifier: Apache-2.0module OpenTelemetry
+require 'opentelemetry-common'
 
 module OpenTelemetry
   module Helpers
@@ -36,7 +37,7 @@ class MySQL
         'create table'
       ].freeze
 
-      QUERY_NAME_REGEX = Regexp.new("(#{QUERY_NAMES.join('|')})", Regexp::IGNORECASE)
+      QUERY_NAME_REGEX = Regexp.new("\\b(#{QUERY_NAMES.join('|')})\\b", Regexp::IGNORECASE)
 
       # This is a span naming utility intended for use in MySQL database
       # adapter instrumentation.
@@ -62,6 +63,8 @@ def self.database_span_name(sql, operation, database_name, config)
 
       # @api private
       def self.extract_statement_type(sql)
+        sql = OpenTelemetry::Common::Utilities.utf8_encode(sql, binary: true)
+
         QUERY_NAME_REGEX.match(sql) { |match| match[1].downcase } unless sql.nil?
       rescue StandardError => e
         OpenTelemetry.handle_error(message: 'Error extracting SQL statement type', exception: e)

diff --git a/helpers/mysql/opentelemetry-helpers-mysql.gemspec b/helpers/mysql/opentelemetry-helpers-mysql.gemspec
@@ -26,6 +26,7 @@ Gem::Specification.new do |spec|
   spec.required_ruby_version = '>= 3.0'
 
   spec.add_dependency 'opentelemetry-api', '~> 1.0'
+  spec.add_dependency 'opentelemetry-common', '~> 0.20'
 
   spec.add_development_dependency 'bundler', '~> 2.4'
   spec.add_development_dependency 'minitest', '~> 5.0'

diff --git a/helpers/mysql/test/helpers/mysql_test.rb b/helpers/mysql/test/helpers/mysql_test.rb
@@ -93,6 +93,23 @@
       end
     end
 
+    describe 'when sql contains invalid byte sequences' do
+      let(:sql) { "SELECT * from users where users.id = 1 and users.email = 'test@test.com\255'" }
+
+      it 'extracts the statement' do
+        assert_equal('select', extract_statement_type)
+      end
+    end
+
+    describe 'when sql contains unknown query statement' do
+      let(:sql) { 'DESELECT 1' }
+
+      # nil sets the span name to 'mysql'
+      it 'returns nil' do
+        assert_nil(extract_statement_type)
+      end
+    end
+
     describe 'when sql has marginalia-style prepended comments' do
       let(:sql) do
         '/*application:web,controller:blob,action:show,correlation_id:01EZVMR923313VV44ZJDJ7PMEZ,' \

diff --git a/instrumentation/mysql2/test/opentelemetry/instrumentation/mysql2/instrumentation_test.rb b/instrumentation/mysql2/test/opentelemetry/instrumentation/mysql2/instrumentation_test.rb
@@ -182,7 +182,7 @@
           client.query(sql)
         end.must_raise Mysql2::Error
 
-        _(span.name).must_equal 'mysql'
+        _(span.name).must_equal 'select'
         _(span.attributes['db.statement']).must_equal obfuscated_sql
       end
 

diff --git a/instrumentation/trilogy/test/opentelemetry/instrumentation/trilogy/instrumentation_test.rb b/instrumentation/trilogy/test/opentelemetry/instrumentation/trilogy/instrumentation_test.rb
@@ -285,7 +285,7 @@
           client.query(sql)
         end.must_raise Trilogy::Error
 
-        _(span.name).must_equal 'mysql'
+        _(span.name).must_equal 'select'
         _(span.attributes[OpenTelemetry::SemanticConventions::Trace::DB_STATEMENT]).must_equal obfuscated_sql
       end