When you create an issue via email, the subject field in the issue will be broken if the subject line in the email is encoded with ISO-2022-JP and contains vendor-defined characters such as "①".

Suppose a user is trying to create an issue via email with the subject "①丸数字テスト" and it is encoded with ISO-2022-JP. Actually, the subject field of the issue will be "$B-!(B $B4]?t;z%F%9%H(B".

!{width: 283px; border: 1px solid #ccc;}.broken-subject@2x.png!

This is because the email's subject contains the character "①". The CIRCLED DIGIT ONE character, U+2460 in Unicode, is not defined in ISO-2022-JP. But it is defined in some vendor-extended variants like ISO-2022-JP-MS and widely used in the real world. Probably Ruby would not process "①" in ISO-2022-JP and raises @Encoding::UndefinedConversionError@.

The undefined character "①" should be replaced with the replacement character instead of breaking the whole subject. Please see ":undef" and ":replace" option of String#encode in http://ruby-doc.org/core-2.5.1/String.html#method-i-encode.

The attached patch is a test to detect the problem.

$ ruby test/unit/mail_handler_test.rb
    .
  (snip)
    .
Encoding conversion failed "\xAD\xA1" to UTF-8 in conversion from ISO-2022-JP to stateless-ISO-2022-JP to EUC-JP to UTF-8
Encoding conversion failed "\xAD\xA1" to UTF-8 in conversion from ISO-2022-JP to stateless-ISO-2022-JP to EUC-JP to UTF-8
Encoding conversion failed "\xAD\xA1" to UTF-8 in conversion from ISO-2022-JP to stateless-ISO-2022-JP to EUC-JP to UTF-8
Encoding conversion failed "\xAD\xA1" to UTF-8 in conversion from ISO-2022-JP to stateless-ISO-2022-JP to EUC-JP to UTF-8
F

Failure:
MailHandlerTest#test_add_issue_with_iso_2022_jp_ms_subject [test/unit/mail_handler_test.rb:753]:
Expected /丸数字テスト/ to match "\e$B-!\e(B \e$B4]?t;z%F%9%H\e(B".


bin/rails test test/unit/mail_handler_test.rb:742