fix: case-insensitive decode for base16/base32/base36 (multibase spec)

[spokodev]

Problem

The multibase registry marks several codecs as case-insensitive: base16, base16upper, base32, base32upper, base32pad, base32padupper, base32hex, base32hexupper, base32hexpad, base32hexpadupper, base36, base36upper.

The multibase spec test suite (tests/README.md) describes case_insensitivity.csv as:

case_insensitivity.csv | hello world | Differently cased than expected, must decode without errors

On the current code these decoders throw on differently cased input. Every line below encodes hello world (hex 68656c6c6f20776f726c64):

import { base16 } from 'multiformats/bases/base16'
import { base32 } from 'multiformats/bases/base32'
import { base36 } from 'multiformats/bases/base36'

base16.decode('f68656c6c6f20776F726C64')   // throws: Non-base16 character
base32.decode('bnbswy3dpeB3W64TMMQ')        // throws: Non-base32 character
base36.decode('kfUvrsIvVnfRbjWaJo')         // throws: Non-base36 character

All 12 lines of the spec's case_insensitivity.csv fail this way. The other three official fixtures (basic, leading_zero, two_leading_zeros) already pass, so this is purely the case-insensitivity gap. The repo's own test/test-multibase-spec.spec.ts carries the basic fixtures inline but omits case_insensitivity.csv, which hid the gap.

This matters for interop: a multibase string produced by a case-insensitive-tolerant implementation (for example go-multibase output that was upper/lower-folded in transit) does not decode in JS today.

References

• multibase tests/case_insensitivity.csv fixture ("must decode without errors").
• multibase README.md registry table marks these codecs case-insensitive.
• RFC 4648 section 3.4 (base16/base32): decoders accept either case.
• go-multibase routes base16 through hex.DecodeString (case-insensitive) and base32 through case-insensitive decoders, and multibase_test.go tampers case for case-insensitive codecs and requires decode to succeed.

Fix

Thread an optional caseInsensitive flag from the rfc4648 and baseX factories. When set, the decode lookup table maps both the lower and upper case of each alphabet character to the same index. The flag is enabled only for the spec-designated codecs above.

Scope notes:

• Only decode becomes case-insensitive. Encode output is unchanged and stays canonical (lowercase or uppercase per codec).
• Case-sensitive codecs (base58btc, base58flickr, base64, base64url, base64pad, base256emoji, identity, base2, base8, base10) keep the default false and are unaffected. base32z is not in the case-insensitive registry and is left case-sensitive.

Tests

Added a case insensitivity block to test/test-multibase-spec.spec.ts using the exact 12 fixture strings from the spec's case_insensitivity.csv, asserting each decodes to hello world. These tests fail on master and pass with this change. The full node suite is green (478 passing), encode output is byte-identical, and case-sensitive codecs still reject mixed case.

[rvagg]

yeah, legit, and matches go-multibase

[github-actions]

🎉 This PR is included in version 14.0.1 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀