class Mysql2::Error

Constants

ENCODE_OPTS

Attributes

errno[RW]
error_number[RW]
server_version[W]
sql_state[R]

Public Class Methods

new(msg, server_version=nil) click to toggle source
Calls superclass method
# File lib/mysql2/error.rb, line 18
def initialize(msg, server_version=nil)
  self.server_version = server_version

  super(clean_message(msg))
end

Public Instance Methods

sql_state=(state) click to toggle source
# File lib/mysql2/error.rb, line 24
def sql_state=(state)
  @sql_state = state.respond_to?(:encode) ? state.encode(ENCODE_OPTS) : state
end

Private Instance Methods

clean_message(message) click to toggle source

In MySQL 5.5+ error messages are always constructed server-side as UTF-8 then returned in the encoding set by the `character_set_results` system variable.

See dev.mysql.com/doc/refman/5.5/en/charset-errors.html for more context.

Before MySQL 5.5 error message template strings are in whatever encoding is associated with the error message language. See dev.mysql.com/doc/refman/5.1/en/error-message-language.html for more information.

The issue is that the user-data inserted in the message could potentially be in any encoding MySQL supports and is insert into the latin1, euckr or koi8r string raw. Meaning there's a high probability the string will be corrupt encoding-wise.

See dev.mysql.com/doc/refman/5.1/en/charset-errors.html for more information.

So in an attempt to make sure the error message string is always in a valid encoding, we'll assume UTF-8 and clean the string of anything that's not a valid UTF-8 character.

Except for if we're on 1.8, where we'll do nothing ;)

Returns a valid UTF-8 string in Ruby 1.9+, the original string on Ruby 1.8

# File lib/mysql2/error.rb, line 57
def clean_message(message)
  return message unless message.respond_to?(:encode)

  if @server_version && @server_version > 50500
    message.encode(ENCODE_OPTS)
  else
    message.encode(Encoding::UTF_8, ENCODE_OPTS)
  end
end