Monthly Archives: September 2009

database column limits and utf8 strings

Wolf and I fixed a bug today where we needed to truncate a string of text that we use internally to annotate the database.  Now, the annotation is just for our reference, so we limit it to 50 bytes — that’s bytes, mind you, not characters, even though the PostgreSQL database will tell you it is “character varying(50)”

We use unicode internally, specifically UTF8, which is a fabulous and widely used standard.  However, it does have a challenging property where a character may be 1-4 bytes long. We were frustrated with what we thought ought to be a simple problem of truncating a string so that it would be no more than 50 bytes.  The tricky part, of course, was that the 49th byte might actually fall in the middle of a character.

To solve the problem, we added a method to the Rails Multibyte::Chars class, which is part of ActiveSupport.  For those who speak Ruby, Rails and RSpec, below is the solution we came up with (first the spec, then the implementation).

The solution we came up with was borrowed from the private translate_offset method. The key interesting part is that you can discover whether you’ve chopped up a string in the middle of a character by calling chunk.unpack(‘U*’) — the unpack method on String in Ruby will throw an exception when you ask it to interpret the UTF-8 characters as unsigned integers with the “U” directive.

describe "Chars#limit_bytes" do
  it 'should return "" on ""' do
    "".mb_chars.limit_bytes(0).should == ""
    "".mb_chars.limit_bytes(1).should == ""
  end

  it 'should truncate single byte character strings as expected' do
    a = "abcd"
    a.mb_chars.limit_bytes(0).should == ''
    a.mb_chars.limit_bytes(1).should == 'a'
    a.mb_chars.limit_bytes(50).should == 'abcd'
  end

  it 'should truncate multi-byte character strings at character boundaries' do
    k = "こんいちわ"
    k.mb_chars.limit_bytes(0).should == ''
    k.mb_chars.limit_bytes(1).should == ''
    k.mb_chars.limit_bytes(3).should == 'こ'
    k.mb_chars.limit_bytes(4).should == 'こ'
    k.mb_chars.limit_bytes(5).should == 'こ'
    k.mb_chars.limit_bytes(6).should == 'こん'
    k.mb_chars.limit_bytes(7).should == 'こん'
    k.mb_chars.limit_bytes(50).should == 'こんいちわ'
  end
end

module ActiveSupport #:nodoc:
  module Multibyte #:nodoc:
    class Chars
      def limit_bytes(limit)
        limit -= 1 while !valid_boundary?(limit)
        s = @wrapped_string.slice(0,limit)
        s.mb_chars
      end

      def valid_boundary?(length)
        chunk = @wrapped_string.slice(0,length)
        begin
          chunk.unpack('U*')
          true
        rescue
          false
        end
      end
    end
  end
end

We’ve written this up as a Lighthouse ticket in case the Rails folk want to add it to the platform or if other people developing multi-lingual database apps run into the same challenge and look there.

The above code can be used under the MIT License:

The MIT License (MIT)

Copyright (c) 2009 Mightyverse, Inc.

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.