#!/usr/bin/env ruby

# Takes the MySQL error messages from sql/share/errmsg-utf8.txt, locates the
# provided error message type (for example, ER_HOST_NOT_PRIVILEGED), then
# creates XML snippets for each locale to be used in Recog.  Note that this
# cannot be used as-is to generate mysql_errors.xml, or oftentimes even parts
# -- it merely spits out XML snippets that you can start with; many will still
# need to be modified by hand.

require 'builder'
require 'open-uri'
require 'securerandom'

def generate_recog(error_name, locale, error_message)
  xml = Builder::XmlMarkup.new(target: STDOUT, indent: 2)
  xml.fingerprint(pattern: error_message) do
    xml.description "Oracle MySQL error #{error_name} (#{locale})"
    xml.example(error_message)
    xml.param(pos: 0, name: 'service.vendor', value: 'Oracle')
    xml.param(pos: 0, name: 'service.family', value: 'MySQL')
    xml.param(pos: 0, name: 'service.product', value: 'MySQL')
  end
end

unless ARGV.size == 2
  fail "Usage: #{$PROGRAM_NAME} <path/URI for errmsg-utf8.txt> <error name>"
end

path = ARGV.first
error_name = ARGV.last

lines = IO.readlines(open(path))

fail "Nothing read from #{path}" if lines.empty?

unless (error_start = lines.find_index { |line| line.strip =~ /^#{error_name}(?:\s+\S+)?$/ })
  fail "Unable to find #{error_name} in #{path}"
end

locale_map = {}
lines.slice(error_start + 1, lines.size).each do |line|
  if /^\s+(?<locale>\S+)\s+"(?<error_message>.*)",?$/ =~ line
    locale_map[locale] = error_message
  else
    break
  end
end

# Many of the error messages contain format strings.  This can be problematic
# in that they need to be removed or otherwise handled as part of the 'pattern'
# attribute and appropriately filled in in any example elements.  So simply try
# a rough count of the possible format strings and warn the user so that they
# can deal with it.
format_count = locale_map.values.map { |error_message| error_message.scan(/%/).size }.inject(&:+)
unless format_count == 0
  warn("#{format_count} possible format strings found -- you'll need to deal with this")
end

Hash[locale_map.sort].map do |locale, error_message|
  generate_recog(error_name, locale, error_message)
end