RE: XMLSAX procedure parameter length

["Larry Ducie" <larry_ducie@xxxxxxxxxxx>]

Hi Rick,

A couple of points:

1) Standard XML parsers do not encode/decode Base64 encoded data.

As you are probably aware, Base64 encoding is used primarily for 
transferring non-text data accross networks. Basically, when you encode data 
as Base64 data you take three bytes (24 bits) and convert it to four bytes 
(32 bits), with each new byte having only the first 6 bits occupied (6 * 4 = 
24 bits). The resultant data is comprised of more bytes than the original, 
but each byte contains a character that exists in all variants of ascii and 
EBCIDIC. Think about it - 6-bits only contains 64 possible values (hence 
Base64). If you include 26 lowercase letters, 26 uppercase letters and 10 
digits - we're already up to 62! Using such a small range of universal 
characters makes it easy to transfer the data accross any network. This was 
originally devised for passing non-text data in e-mails - the ubiqutous MIME 
format.

The link this has with parsing XML is the basic fact that the data will be 
interpreted as text data by the parser, not binary data. As all parsers work 
with text-only data formats, encountering a non-text string would cause a 
fata error. Once parsing is complete you can decode the string from Base64 
to the original binary format.

2) If you're getting strange values in your Base64 encoded data (such as "[" 
changing to "1/2" characters then there is a translation problem or a 
viewing problem. It may be a good idea to post the code you use to open the 
file, together with the CCSID of your file and the job that the open is run 
within. I think you should check the CCSID of your data, you're file, your 
open, your application, AND your interactive job/display device. I've 
included the last one because we work on a lot of data from Europe, encoded 
in a variety of CCSID values, and every week we get somebody looking at 
CCSID 297 data from a CCSID 285 workstation device and declaring that the 
data contains some wrong characters. :-)

3) Remember that parsing the Base64 encoded data into a program variable (or 
onto a pointer) is the first step. You need to decode it to retrieve the 
original binary object. The resultant text from the parsed XML should be 
identical to the data contained within the XML - even if you've translated 
the data. It's only when you decode the data that you get your binary data 
back. The only exceptions to this is when the data is URL encoded or Entity 
References are used - but you should never get a string like & in a 
Base64 encoded object. :-)

Cheers

Larry Ducie