base28 v1.1.1
base28
Represent your integer database ids as 6-character base-28 strings – an ideal format for URLs.
- Maximum supported input is 0x0fffffff, i.e. 268,435,455.
- Distributable size is 2.7kb, minified and gzipped.
- Zero runtime dependencies.
Base28 Alphabet
This package's base28 encoding comprises of 8 numbers and 20 characters:
BCDFGHJKMN
PQRSTVWXYZ
23456789
or
bcdfghjkmn
pqrstvwxyz
23456789These numbers and letters have been picked according to the criteria:
- Encoded ids should typically look like ids - not words. This is achieved by omitting the vowels.
- It should easy to visually distinguish between all characters - which is why
1and0have been omitted - which can be confused with 'I' and 'O'. - It should be easy to communicate the characters through voice - thus lower and upper case characters can be used interchangeably.
Encoding
It's often undesireable for sequential ids to be readily identified as such. To avoid this, this encoding obfuscates your ids by applying a reversible pseudo-random looking mapping during encoding and decoding. For example:
0 -> GXBSXJ -> 0
1 -> PNPV8H -> 1
2 -> M3BXDH -> 2
3 -> TSP3GS -> 3While this won't prevent any dedicated attackers from iterating through your ids, it will at least provide some protection against casual voyeurs.
Prior Art
- The obfuscation algorithm is inspired by the answers to Stack Overflow question.
- The base28 encoder is based on cbschuld/uuid-base58.
Contributing
Pull requests would be appreciated for:
If you want to work with inputs greater than 268,435,455, it should be possible to extend this to work with 7 or 8-character strings.
Configuration of the obfuscator. In particular, it should be possible to pick configure the XOR and ROT numbers through a single large integer.
Much of the size of this library is due to a bundled, pared-down version of Google's BigInt polyfill. Exporting a second distributable with native BigInt would be a great pull request.
License
Copyright 2022, James K Nelson. Available for use under MIT License.