From: Dave Burt <dave@...>
Date: 2005-08-12T00:26:10+09:00
Subject: Re: java strings read and write

Karsten Meier wrote...
> Now I like to create a file with "native java strings", in this case
> strings that can be read with DataInputStream.readUTF() method.
> I hope that this simplifies the code and improves the speed.
>
> I don't need a complete Ruby/Java Integration, I just want to write
> these java strings to a file (or to stdout as a cgi-script).
> Are there any libraries for this?

The way I read the Java API, the format is simply the UTF-8-encoded string 
preceded by its length in bytes as a 16-bit integer.

Except that (from 
http://java.sun.com/j2se/1.5.0/docs/api/java/io/DataInput.html):
* The null byte '\u0000' is encoded in 2-byte format rather than 1-byte, so 
that the encoded strings never have embedded nulls.
* Only the 1-byte, 2-byte, and 3-byte formats are used.
* Supplementary characters are represented in the form of surrogate pairs.

So, in Ruby:
# untested code (I'm on Windows with no iconv)
require 'iconv'
class String
  def to_java_utf
    # Convert the string to UTF-8
    result = Iconv.new('iso-8859-1', 'utf-8').iconv(self)
    # Re-encode null characters in two-byte format
    result.gsub!("\0", "\300\200")
    # The string's size must fit in 2 bytes, so the string must be less than 
64kb
    fail if result.size >= 65536
    # Prepend string length as short integer in network (big-endian) byte 
order
    result.insert(0, [converted_text.size].pack("n"))
  end
end
text = "Test string\n\0with embedded newline and 0"
p text.to_java_utf  #=> "\000(Test string\n\300\200with embedded newline and 
0"

Cheers,
Dave