What's the best candidate padding char for url-safe and filename-safe base64?

Question

The padding char for the official base64 is '=', which might need to be percent-encoded when used in a URL. I'm trying to find the best padding char so that my encoded string can be both url safe (I'll be using the encoded string as parameter value, such as id=encodedString) AND filename safe (I'll be using the encoded string directly as filename).

Dot ('.') is a popular candidate, it's url safe but it's not exactly filename safe: Windows won't allow a file name which ends with a trailing dot.

'!' seems to be a viable choice, although I googled and I've never seen anybody using it as the padding char. Any ideas? Thanks!

Update: I replaced "+" with "-" (minus) and replaced "/" with "_" (underscore) in my customized base64 encoding already, so '-' or '_' is not available for the padding char any more.

score 4 · Answer 1 · edited Feb 12 '14 at 10:43

4

The best solution (I've spent last month working on this problem with an email sending website) is to not use padding character (=) at all

The only reason why padding character is there is because of "lazy" decoders. You can extremely easy add missing = -> just do %4 on text and subtract the number you get from 4 and that is how many = you need to add in string end. Here is C# code:

    var pad = 4 - (text.Length % 4);
    if (pad < 4)
        text = text.PadRight(text.Length + pad, '=');

Also, most people who do this are interested in replacing + and / with other URL safe character... I propose:

replace with - / replace with _

DO NOT USE . as it can produce crazy results on different systems / web servers (for example on IIS Base64 encoded string can't end with . or IIS will search for the file)

edited Feb 12 '14 at 10:43

Young Bob

733
3
9

answered Jun 01 '12 at 22:41

nikib3ro

20,366
24
120
181

1

Not quite correct, the result of the padding is not divisible by 4. Need to append (4-pad) number of '=' characters not pad number. Otherwise neat solution. – Young Bob Feb 11 '14 at 13:51
@YoungBob I've edited answer, please edit it if you can propose better code. Thanks! – nikib3ro Feb 11 '14 at 18:16
In Python you can do simply `text + (-len(text) % 4) * "="` but this way of calculating it cannot be used in other languages (the % operator in Python gives a positive value for a positive divisor). – Tronic Nov 10 '21 at 14:15

score 2 · Answer 2 · answered Mar 18 '09 at 19:26

2

The RFC 2396 unreserved characters in URIs are:

"-" | "_" | "." | "!" | "~" | "*" | "'" | "(" | ")"

It's worth pointing out, though, that the Microsoft article also says "Do not assume case sensitivity." Perhaps you should just stick with base 16 or 32?

answered Mar 18 '09 at 19:26

Miles

31,360
7
64
74

I would avoid . ~ * ' ( ) for filenames though – cobbal Mar 18 '09 at 19:28
Thanks Miles, seems like '!' is good for both url and file name. Maybe I'll go with base32 with '!' as the padding char. – SamS Mar 18 '09 at 19:41

score 1 · Answer 3 · answered Dec 06 '10 at 14:54

1

The Wikipedia article states;

a modified Base64 for URL variant exists, where no padding '=' will be used

answered Dec 06 '10 at 14:54

bob

11
1

score 0 · Answer 4 · answered Mar 18 '09 at 19:22

0

I would go with '-' or '_'
They're URL and file safe, and they looks more or less like padding

answered Mar 18 '09 at 19:22

cobbal

69,903
20
143
156

Sorry I should have mentioned this: I replaced "+" with "-" (minus) and replaced "/" with "_" (underscore) in my customized base64 encoding already, so '-' or '_' is not available for the padding char any more. – SamS Mar 18 '09 at 19:31

What's the best candidate padding char for url-safe and filename-safe base64?

4 Answers4

Linked