tmtm/mysql_unicode_collation.rb

## mysql_unicode_collation.rb
#!/usr/local/bin/ruby
# -*- coding: utf-8 -*-

require 'cgi'

filename, = ARGV         # path of file `ctype-uca.c'
ctype_uca = File.read(filename)

ctype_uca =~ /^uchar uca_length\[256\]={\n(.*?)};$/m or raise 'uca_length not found'
uca_length = $1.gsub(/\s/,'').split(/,/).map(&:to_i)

uca = Hash.new{|h,k| h[k] = []}
ctype_uca.scan(/^uint16 page([0-9A-F]+)data\[\]= {.*?\n(.*?)};$/m) do |page, data|
  page = page.hex
  data.gsub(/\s/,'').split(/,/).each_slice(uca_length[page]).each_with_index do |data, i|
    data.reject!{|d| d == '0x0000'}
    next if data.empty?
    data = data.join(',')
    code = page * 256 + i
    uca[data].push code
  end
end

puts <<EOS
<html>
<head>
<style type="text/css">
body {font-family: monospace}
table {border-collapse: collapse; empty-cells: show}
th,td {border: 1px solid}
</style>
<title>MySQL 5.5.11 unicode_ci で同一視される文字</title>
</head>
<body>
<h1>MySQL 5.5.11 unicode_ci で同一視される文字</h1>
EOS

uca.sort_by{|k,v| v}.each do |data, code_list|
  next if code_list.size == 1
  row1 = ''
  row2 = ''
  puts '<table>'
  code_list.each do |code|
    uni = "%04X" % code
    chr = eval("\"\\u#{uni}\"")
    row1.concat "<td>#{uni}</td>"
    row2.concat "<td>#{CGI.escapeHTML chr}</td>"
  end
  puts '<tr><th>Unicode</th>', row1, '</tr>'
  puts '<tr><th>Character</th>', row2, '</tr>'
  puts '</table><br>'
end

puts <<EOS
</body>
</html>
EOS
	#!/usr/local/bin/ruby
	# -- coding: utf-8 --

	require 'cgi'

	filename, = ARGV # path of file `ctype-uca.c'
	ctype_uca = File.read(filename)

	ctype_uca =~ /^uchar uca_length\[256\]={\n(.*?)};$/m or raise 'uca_length not found'
	uca_length = $1.gsub(/\s/,'').split(/,/).map(&:to_i)

	uca = Hash.new{\|h,k\| h[k] = []}
	ctype_uca.scan(/^uint16 page([0-9A-F]+)data\[\]= {.?\n(.?)};$/m) do \|page, data\|
	page = page.hex
	data.gsub(/\s/,'').split(/,/).each_slice(uca_length[page]).each_with_index do \|data, i\|
	data.reject!{\|d\| d == '0x0000'}
	next if data.empty?
	data = data.join(',')
	code = page * 256 + i
	uca[data].push code
	end
	end

	puts <<EOS
	<html>
	<head>
	<style type="text/css">
	body {font-family: monospace}
	table {border-collapse: collapse; empty-cells: show}
	th,td {border: 1px solid}
	</style>
	<title>MySQL 5.5.11 unicode_ci で同一視される文字</title>
	</head>
	<body>
	<h1>MySQL 5.5.11 unicode_ci で同一視される文字</h1>
	EOS

	uca.sort_by{\|k,v\| v}.each do \|data, code_list\|
	next if code_list.size == 1
	row1 = ''
	row2 = ''
	puts '<table>'
	code_list.each do \|code\|
	uni = "%04X" % code
	chr = eval("\"\\u#{uni}\"")
	row1.concat "<td>#{uni}</td>"
	row2.concat "<td>#{CGI.escapeHTML chr}</td>"
	end
	puts '<tr><th>Unicode</th>', row1, '</tr>'
	puts '<tr><th>Character</th>', row2, '</tr>'
	puts '</table><br>'
	end

	puts <<EOS
	</body>
	</html>
	EOS