fix(logs): skip malformed utf-8 messages

Author: solnic

sentry-ruby/lib/sentry/client.rb

@@ -4,6 +4,7 @@
 require "sentry/log_event"
 require "sentry/log_event_buffer"
 require "sentry/utils/uuid"
+require "sentry/utils/encoding_helper"
 
 module Sentry
   class Client
@@ -194,8 +195,11 @@ def event_from_log(message, level:, **options)
 
       attributes = options.reject { |k, _| k == :level || k == :severity || k == :origin }
       origin = options[:origin]
+      body = Utils::EncodingHelper.safe_utf_8_string(message)
 
-      LogEvent.new(level: level, body: message, attributes: attributes, origin: origin)
+      return unless body
+
+      LogEvent.new(level: level, body: body, attributes: attributes, origin: origin)
     end
 
     # Initializes an Event object with the given Transaction object.

sentry-ruby/lib/sentry/utils/encoding_helper.rb

@@ -3,6 +3,8 @@
 module Sentry
   module Utils
     module EncodingHelper
+      EMPTY_STRING = ""
+
       def self.encode_to_utf_8(value)
         if value.encoding != Encoding::UTF_8 && value.respond_to?(:force_encoding)
           value = value.dup.force_encoding(Encoding::UTF_8)
@@ -17,6 +19,10 @@ def self.valid_utf_8?(value)
 
         value.dup.force_encoding(Encoding::UTF_8).valid_encoding?
       end
+
+      def self.safe_utf_8_string(value)
+        valid_utf_8?(value) && value
+      end
     end
   end
 end

sentry-ruby/spec/sentry/structured_logger_spec.rb

@@ -95,6 +95,13 @@
           expect(log_event[:attributes]["sentry.message.parameter.name"]).to eql({ value: "Jane", type: "string" })
           expect(log_event[:attributes]["sentry.message.parameter.day"]).to eql({ value: "Monday", type: "string" })
         end
+
+        it "doesn't choke on malformed UTF-8 strings" do
+          malformed_string = "Hello World\x92".dup.force_encoding("UTF-8")
+          Sentry.logger.public_send(level, malformed_string, user_id: 123)
+
+          expect(sentry_logs).to be_empty
+        end
       end
     end