camelcase

Convert strings to camelCase or PascalCase — with Unicode-aware case mapping, digit handling, and acronym preservation.

const camelcase = @import("camelcase");

const a = try camelcase.camelCase(allocator, "foo-bar", .{}); // "fooBar"
const b = try camelcase.camelCase(allocator, "XMLHttpRequest", .{}); // "xmlHttpRequest"
const c = try camelcase.camelCase(allocator, "розовый_пушистый", .{}); // "розовыйПушистый"
// Caller owns the returned memory.

Word boundaries are _, ., -, and space — and case transitions: fooIDs → fooIds, FOOBar → fooBar. Leading _ and $ are kept (_foo_bar → _fooBar), whitespace is trimmed, and input that is entirely separators yields an empty string.

Installation

Requires Zig 0.16.0.

zig fetch --save git+https://github.com/thelastbackspace/zig-camelcase#v1.0.0

Then in build.zig:

const camelcase = b.dependency("camelcase", .{ .target = target, .optimize = optimize });
exe.root_module.addImport("camelcase", camelcase.module("camelcase"));

API

camelCase(allocator, input: []const u8, options: Options) ![]u8

Convert one string. Caller owns the returned memory.

camelCaseMany(allocator, inputs: []const []const u8, options: Options) ![]u8

Convert several strings into one identifier, as if joined with -. Empty and whitespace-only elements are dropped, and each element is trimmed first: ["foo", "-bar"] → fooBar.

Options

• pascal_case: bool = false — FooBar instead of fooBar.
• preserve_consecutive_uppercase: bool = false — keep runs of uppercase: foo-BAR → fooBAR.
• capitalize_after_number: bool = true — foo2bar → foo2Bar. When false, numbers do not start a new word (Google Java Style Guide): foo2bar stays foo2bar, and the letter's original case is kept (Textures_3D → textures3D).
• locale: Locale = .default — .turkic applies the Turkish/ Azerbaijani dotted-i rules (lorem-ipsum → loremİpsum).

Notes

• Case mapping uses simple Unicode mappings (Unicode 16.0.0) generated into src/unicode_data.zig. Multi-character mappings such as ß → SS are not applied; the simple mapping (ß → ẞ) is.
• Text is processed by Unicode code point, not UTF-16 unit, which can differ for astral-plane input in rare edge cases.
• Input is expected to be valid UTF-8; invalid bytes are passed through unchanged.

Testing

zig build test

The suite is a port of the upstream project's own test file — about 250 assertions covering separators, acronyms, digits, Unicode (Latin, Cyrillic, CJK, Arabic, Hebrew, emoji), locales, and option combinations. Two upstream tests have no Zig equivalent (a TypeError for non-string input, and mocking of JavaScript Intl internals).

Contributing

PRs welcome — see CONTRIBUTING.md. The CI gate (zig fmt --check, tests, build) must pass.

Credits

Behavior and the test suite follow the npm package camelcase v9.0.0 (MIT, © Sindre Sorhus); the implementation is original, and the Unicode tables are generated from the Unicode Character Database 16.0.0. Zig implementation © 2026 thelastbackspace, MIT — see LICENSE.