Base41 encoding can be viewed as a simplified form of Ascii85 encoding. It is simple to code and understand, safe to embed in strings (like JSON or most programming language's literals), uses minimal resources and is reasonably efficient (has 50% overhead), making it a good candidate for use in embedded or otherwise constrained systems / applications.
See Base41 Encoding Specification for a detailed treatment of this encoding. Постоји и Спецификација на српском.
This repository has implementations in several languages. All code is under the liberal MIT license. Reference implementation is in C. Other language implementations are references for their perspective languages.
To reduce code size and potential configuration issues (i.e. encoder using a different alphabet than the decoder) and increase performance, the Base41 Specification fixes a particular Alphabet.
Of course, that Alphabet is not good for all purposes. So, a variant of Base41 is to "bring your own Alphabet" (BYOA) of 41 ASCII characters. In the following sections there are some examples thereof.
Botta and Cavagnino published a paper in 2022 which uses the BYOA variant, with Alphabet selected so that Base41 encoded string can be part of an URL. This was the first inspiration for having a BYOA variant of Base41 spec.
They also define the encoding to not be "byte oriented" but
"bit-oriented". That is, data need not have 8 x N
bits. Interesting, but complicates things for a somewhat rare use-case
(well, I guess for them it was not so rare).
They were kind enough to mention this code repo as "apriori art" of sorts. It's an interesting read, with a different focus and presentation style than our specification. Here's the full reference to the paper:
Botta, Marco, and Davide Cavagnino. "Base41: A proposal for printable encoding of bit strings." Engineering Reports (2022): e12606.
Philippe Majerus (github user PhMajerus) proposes an Alphabet with
characters used in alphanumeric QR Codes and which can be part of an
URL. To be precise:
0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ$*-.:Thus one can encode an URL with some binary data in an alphanumeric QR code. Caveats:
$) and asterisk (*) are URL sub-delimiters, thus
reserved mostly because of file naming conventions, but should be OK
for most servers which are aware that URL paths are not the same as
file system paths.:) is a delimiter, could interfere with the server's path
processing, so some care should be taken.One of the above can be replaced with + if that helps HTTP server's
URL processing.
AFAIK, I invented this encoding. If I haven't, great, let the glory go to the other guy.
But, if I did invent it, I hereby put it to public domain as of 2014-05-06 (the actual date of invention in ISO8601 format). That should make any patent claims for it invalid. Don't laugh, I've seen software patents that are much more silly.
I did not invent any of the BYOA Alphabets mentioned above.
No, I was not 41 when I invented it. :)