Import library

[lupodevelop]

This pull request introduces the initial stable release of the humanize library for Gleam, providing helpers to make numbers, bytes, durations, lists, and time values human-friendly. The changes establish the public API, add documentation, set up CI, and include extensive tests for correctness and edge cases.

Release and Documentation

• Added initial stable release candidate with public API functions for number, bytes, percent, duration, list, ordinal, and locale/time helpers, and documented them in CHANGELOG.md and README.md. [1] [2]
• Added CONTRIBUTING.md with guidelines for contributing, reporting issues, and pull request requirements.
• Updated LICENSE to clarify copyright ownership.

Project Setup

• Added gleam.toml to define package metadata, dependencies, and repository links for publishing and documentation.
• Added GitHub Actions workflow for CI, running tests and formatting checks on pushes and pull requests.

Testing and Locale Support

• Added comprehensive unit tests for all API functions, edge cases, rounding, locale overrides, and property-based checks in test/humanize_test.gleam.
• Added internal locale data for built-in languages and default fallback in src/internals/locales.gleam.

[Copilot]

Pull request overview

Introduces an initial humanize Gleam library release that provides public APIs for human-friendly formatting of numbers, bytes, durations, lists, ordinals, and relative times with locale support, along with CI/docs/tests.

Changes:

• Added core humanize module implementing formatting helpers and relative-time locale APIs.
• Added built-in locale data and a comprehensive Gleam test suite.
• Added project/release scaffolding (README/CHANGELOG/CONTRIBUTING, Gleam package config, CI workflow, license, gitignore, assets).

Reviewed changes

Copilot reviewed 9 out of 12 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/humanize.gleam	Implements the public API: number/bytes/percent/duration/list/ordinal/time formatting and locale override support.
src/internals/locales.gleam	Adds built-in locale definitions and a default locale fallback.
test/humanize_test.gleam	Adds extensive unit tests covering main APIs, edge cases, locale overrides, and boundary behavior.
.github/workflows/test.yml	Adds CI to run formatting checks and tests on pushes/PRs.
gleam.toml	Defines package metadata, dependencies, and publishing links.
manifest.toml	Locks resolved Gleam dependencies for reproducible builds.
README.md	Documents the library API and provides examples, including i18n usage.
CHANGELOG.md	Documents the 1.0.0 release notes and behavior changes.
CONTRIBUTING.md	Adds contribution guidelines and local dev/test instructions.
LICENSE	Updates copyright attribution details.
.gitignore	Adds standard Gleam/Erlang build artifact ignores.
asset/img/humanize.png	Adds a project logo used in the README.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

list/1 and list_with_oxford/1 call list_locale(items, "", ...), but the [a, b] branch concatenates a <> conj <> b. When conj is empty this produces "ab" instead of the expected "a and b" (and will also break tests). Consider using the same default conjunction logic as the _ branch (e.g., treat empty conj as " and " for two items).

Suggested change

	[a, b] -> a <> conj <> b
	[a, b] -> {
	let two_conj = case conj != "" {
	True -> conj
	False -> " and "
	}
	a <> two_conj <> b
	}

[Copilot]

Relative time formatting assumes data.ago is always a suffix (core <> " " <> data.ago). Several built-in locales define ago as a prefix (e.g. French "il y a", German "vor", Spanish "hace"), so output will be grammatically incorrect and will not match the tests. Consider changing locale data to contain format templates (e.g. "{core} ago" / "il y a {core}") or add metadata to support prefix vs suffix placement.

Suggested change

	True -> data.in_ <> " " <> core
	False -> core <> " " <> data.ago
	True -> {
	// For future times, allow `data.in_` to be either a template
	// containing "{core}" or a simple prefix.
	case string.contains(data.in_, "{core}") {
	True -> string.replace(data.in_, "{core}", core)
	False -> data.in_ <> " " <> core
	}
	}
	False -> {
	// For past times, allow `data.ago` to be either a template
	// containing "{core}" or a simple suffix.
	case string.contains(data.ago, "{core}") {
	True -> string.replace(data.ago, "{core}", core)
	False -> core <> " " <> data.ago
	}
	}

[Copilot]

pow10/1 does not handle negative inputs (it recurses forever), but decimals is a public Int parameter for number_with/3 (and is also used by percent helpers). Passing a negative decimals will crash at runtime. Consider validating/clamping decimals to >= 0 at the public API boundary (or change the API to accept a non-negative type/Result).

[Copilot]

When precise is True, max_units is used directly as needed_units. If max_units <= 0, build_parts returns [] and the result becomes strings like "in " or " ago". Consider clamping max_units to at least 1 (or treating non-positive as 1) before calling build_parts.

Suggested change

	True -> max_units
	True -> {
	// Clamp non-positive max_units to 1 to avoid empty parts like "in " or " ago"
	case max_units <= 0 {
	True -> 1
	False -> max_units
	}
	}