@smartput/timezone

Time zones as data: the named zones, every zone written as an offset from UTC, and the parser that reads GMT+3. No runtime dependency at all.

import { OFFSET_ZONES, parseOffsetZone, ZONES, zoneSymbol } from "@smartput/timezone";

ZONES["Asia/Tokyo"];               // { aliases: ["tokyo", "jst", "japan"], symbol: "JST" }
OFFSET_ZONES["+05:30"];            // { aliases: [], symbol: "UTC+05:30" }
parseOffsetZone("gmt+5:30 stand"); // { zone: "+05:30", length: 8 }
zoneSymbol("+05:30");              // "UTC+05:30"

It is a package of its own so a zone picker costs a zone picker. These are tables and one regex; @smartput/datetime is those plus chrono-node and temporal-polyfill, which are several times the size of the engine. A form field offering a list of zones, or validating that a user typed a real one, needs none of that — so the dependency runs from the consumer inwards, the same argument that keeps @smartput/city out of @smartput/country.

Every key is a string Temporal accepts as a time zone id, which is what lets a consumer register the two tables as units of an opaque kind directly. @smartput/datetime does exactly that, and adds nothing to them.

Aliases are single words, because @smartput/core's alias index is keyed by one segmented word: nyc can be an alias and "new york" cannot. The offset zones carry no aliases at all — gmt+3 lexes as three tokens, so no lookup could reach it and parseOffsetZone is the door instead.

Offsets run every quarter hour from -12:00 to +14:00, which is the range and the granularity real zones are kept on.

Full documentation: Dates and time zones.