heckle

A meme case conversion library.

heckle exists to facilitate lulz with meme cases, such as spONgEBoB caSE and BILLY MAYS MODE. It is intended to be unicode-aware, internally consistent-ish, and reasonably useless.

Cases

sPoNGeBoB CAsE

Randomly alternates letter casing with no more than 3 consecutive same-case characters. If you want more than 3 characters, too bad, it's hardcoded.

Also it's not really random because the RNG is always seeded with 42. That's the answer to life, the universe, and everything, and also the answer to meme case RNG seeding.

use heckle::ToSpongebobCase;
assert_eq!("hello world".to_spongebob_case(), "HELlo wORLd");

BILLY MAYS MODE

Uppercases every character that has a case, dropping non-alphabetic, non-whitespace characters. BUT WAIT! you might say; well, too bad, no more.

use heckle::ToBillyMaysMode;
assert_eq!("wait, there's more!!!".to_billy_mays_mode(), "WAIT THERES MORE");

Word boundaries

Unlike the popular and useful heck library, heckle does not care about word boundaries. In BILLY MAYS MODE, non-alphabetic, non-whitespace characters are dropped entirely. In Spongebob Case, only alphabetic characters with a real case mapping participate in the run counter — everything else passes through unchanged.

Performance

Benchmarks compare heckle against heck as a baseline. heck is a mature, widely-used case-conversion library with a similar char-iteration / String-building profile, making it a reasonable reference point even though the two libraries perform different operations.

Methodology

Benchmarks are implemented with Criterion.rs (benches/conversions.rs). Each conversion is run against five representative inputs:

Label	Description	Length
short	"hello, world!"	13 chars
medium	pangram with trailing punctuation	90 chars
punctuation_heavy	punctuation interspersed throughout every word	69 chars
long	pangram repeated 222×	~9,990 chars
multiline	multi-sentence, multi-line block repeated 50×	~3,400 chars

Three heck conversions are included as baselines: to_snake_case, to_upper_camel_case, and to_shouty_snake_case.

To reproduce:

cargo bench

Results

Machine: MacBook M1 Pro. Results will vary on other hardware. The relative ordering between conversions is expected to be stable, but absolute numbers and ratios may differ on x86 or under different memory pressure.

Median wall-clock time per call:

Input	spongebob_case	billy_mays_mode	heck snake	heck upper_camel	heck shouty_snake
short	175 ns	112 ns	183 ns	187 ns	195 ns
medium	1,124 ns	709 ns	1,281 ns	1,237 ns	1,298 ns
punctuation_heavy	470 ns	305 ns	739 ns	618 ns	694 ns
long	122 µs	74 µs	129 µs	119 µs	127 µs
multiline	37.4 µs	25.8 µs	41.9 µs	40.9 µs	41.4 µs

Analysis

billy_mays_mode is the fastest conversion across every input, running ~35–40% faster than heck on typical inputs. The reasons are structural: Billy Mays is a simple filter-and-uppercase pass with no word-boundary detection and no case-transition state machine. The pending_space trick (deferring whitespace emission rather than trimming after the fact) means the hot loop is a single branch per character with no backtracking.

spongebob_case is roughly on par with heck, within ~5–10% in either direction. The main extra cost over heck's boundary detection is one gen_bool(0.5) RNG call per alphabetic character. The thread-local SmallRng access is cheap, but not free.

Punctuation-heavy input is where the gap is largest. On punctuation_heavy, billy_mays_mode (305 ns) is 57% faster than heck snake (739 ns), and even spongebob_case (470 ns) beats all three heck variants. heck must track word-boundary state across every punctuation character. Both heckle conversions discard non-alphabetic, non-whitespace characters in a single branch and move on, so punctuation is essentially free.

A Note on AI Usage

This is pure vibe-coded slop. Thanks, Claude!