What's new in Humanizer 4

Humanizer 4 gives applications clearer control over localized text, durations, data units, numbers, and identifiers. This page highlights the changes that are most useful in application code. For compatibility work, follow the Humanizer 4 upgrade guide instead.

Install one package and reach every supported culture

The Humanizer package now contains the runtime, generated locale data, and analyzers. There is no separate locale-package graph to assemble. The current inventory covers 102 culture entries, and every culture on the supported list implements every localized feature that applies to it.

For example, Swahili number words ship with the same package:

var swahili = CultureInfo.GetCultureInfo("sw");
Console.WriteLine(21.ToWords(swahili)); // ishirini na moja

See package selection for the dependency layout and supported cultures for the generated inventory. Use Configurator.IsCultureSupported before selecting an application fallback.

Make durations fit the interface

Duration formatting can use localized compact symbols, retain sub-second values as fractional seconds, and select an authored grammatical case. Humanizer can also parse invariant TimeSpan text and compact input such as 1h 30m 15.25s. Configurator.TimeSpanHumanizeStrategy provides one configuration point when an application needs a custom duration policy.

Compact output uses the same precision and unit bounds as ordinary humanization:

var duration = new TimeSpan(0, 1, 2, 3, 4);
Console.WriteLine(duration.HumanizeToSymbols(
    precision: 4,
    culture: CultureInfo.GetCultureInfo("en-US")));
// 1h, 2min, 3s, 4ms

The durations and ages guide covers symbols, fractional seconds, grammatical case, parsing, and custom strategies.

Choose exactly what a byte unit means

Byte-size operations can now select DecimalSi or BinaryIec explicitly. Decimal SI uses powers of 1000 and symbols such as MB; binary IEC uses powers of 1024 and symbols such as MiB. The choice is available for parsing, formatting, composite output, and transfer rates. Existing behavior remains available through Legacy.

The same stored byte count can therefore be labeled without ambiguity:

var size = ByteSize.FromBytes(1_000_000);
var culture = CultureInfo.GetCultureInfo("en-US");
Console.WriteLine(size.Format(ByteSizeUnitSystem.DecimalSi, formatProvider: culture)); // 1 MB
Console.WriteLine(size.Format(ByteSizeUnitSystem.BinaryIec, formatProvider: culture)); // 976.56 KiB

See byte sizes and transfer rates for unit-system parsing, exact composite output, full words, and rates.

Read and write richer numbers

Every built-in number-word locale now covers the full signed long range. Number-word input includes localized decimal phrases with preserved scale, and the wider number surface adds common fractions, 64-bit numeric ordinals, Indian scale styles, Chinese financial numerals, and locale-owned metric scale words.

Spoken decimal input retains meaningful trailing zeros:

var value = "one point five zero".ToDecimalNumber(
    CultureInfo.GetCultureInfo("en-US"));
Console.WriteLine(value.ToString(CultureInfo.InvariantCulture)); // 1.50

Start with writing numbers and ordinals or parsing number words, then use the metric numeral guide when the output needs a compact scale.

Preserve the text that carries meaning

Identifier conversion is less eager to discard information. Humanize preserves ampersands, registered acronyms can supply their canonical casing, and Underscore(preserveCase: true) inserts boundaries without forcing the result to lowercase. Dots are recognized as identifier boundaries, while decimal points survive title casing.

For example, an acronym can remain uppercase while the word boundary becomes explicit:

Console.WriteLine("HTMLParser".Underscore(preserveCase: true)); // HTML_Parser

See strings and casing for the full set of identifier transformations and acronym configuration.

The GitHub releases page is the canonical home for the complete Humanizer 4 release notes and contributor list when the release is published.