Percent-encoding (RFC 3986 "quote") / percent-decoding ("unquote").

liburlparser exposes path()/query()/params() as the raw (still percent-encoded) substrings of the URL - same as the WHATWG URL standard and every URL library that follows it (including ada). None of them decode automatically, on purpose: it can be lossy/ambiguous to do so implicitly. But unlike Python's urllib.parse (quote/unquote) or ada's internal (and) ada::unicode::percent_encode/percent_decode, liburlparser did not expose any public helper to do this decoding explicitly - so callers had to reach for a second library just to read "café" out of "caf%C3%A9". This closes that gap, implemented from scratch (no external dependency): a 256-bit membership bitmap gives O(1) "does this byte need encoding" checks, and both functions do a single allocation sized to a conservative upper bound.