is-canonical-base64

a regular expression to test wether a string is a valid base64, without ambiguity.

note npm has implemented a (badly working) spam filter which disallowed me to publish is-canonical-base64 so this is published as reversable-base64 and canonical-base64 (I'd much prefer is- at the start however, because the purpose of this module is a boolean test. And "is" turns it into a yes/no name)

Base64 encodes bytes (base 256, 8 bits each) as chars each representing 6 bits. That means a one byte buffer necessarily becomes 2 chars, except only the top 2 bits of the last char are used (and the string is then padded with an == to take the spaces those chars would be)

It's recommended that implementations encode zeros here, but many implementations (including node.js's) accept non-zeros here, but just ignore them.

This can be a problem for signed data, because if the signature signs base64, you must retain exactly the encoded representation (including any ambiguity). So, it's good to enforce that ambigious representations are rejected.

usage

simple

var isCanonicalBase64 = require('canonical-base64')() //note, returns a function
isCanonicalBase64.test(string)

advanced: you may pass in a prefix and suffix, which is combined into the regular expression.

var isCanonicalBase64 = require('canonical-base64')('@', '(?:\.ed25519)')
isCanonicalBase64.test(string)

License

MIT