Locales

Morrow ships 81 built-in locales for formatting, parsing, humanize, and dehumanize. Apart from Simplified and Traditional Chinese, their text matches Arrow 1.4.0. The Chinese locales keep Morrow's 1.0 output.

from morrow import Locale, Morrow

var dt = Morrow(2026, 1, 1, 15)
print(dt.format("dddd Do MMMM YYYY", locale="fr"))   # jeudi 1er janvier 2026
print(dt.shift(hours=-5).humanize(dt, locale="ru"))  # 5 часов назад
print(dt.dehumanize("vor 2 Tagen", locale="de"))
print(Morrow.get("1 января 2026", "D MMMM YYYY", locale="ru"))

Every locale argument accepts a name or a Locale value. Names ignore case and accept _ or -, so pt_BR, pt-br, and PT-BR all load the same locale. An unknown name raises unsupported locale.

Reusing a locale

Passing a name loads that locale's data on every call. In a loop, load the Locale once and pass the value instead:

var french = Locale("fr")
for day in Morrow.iter_range("day", start, limit=365):
    print(day.format("dddd D MMMM", locale=french))

Locale.available() returns the 81 canonical names. Locale.aliases("ar") returns every accepted name for a locale.

Custom locales

Locale fields are public. Start from a built-in locale and change what you need:

var pirate = Locale("en")
pirate.name = "en-pirate"
pirate.month_names[0] = "Janarrr"
pirate.meridians = ["ay", "pee", "AY", "PEE"]
pirate.set_timeframe("hours", "{0} bells")
pirate.past = "{0} back"

print(Morrow(2026, 1, 1, 15).format("MMMM D A", locale=pirate))  # Janarrr 1 PEE

Name lists start at index 0, so month_names[0] is January and day_names[0] is Monday. meridians holds am, pm, AM, and PM, in that order. ordinals can override the built-in day ordinals 1–31.

Timeframes use a {0} placeholder for the count. For keyed forms, set a plural_rule and pass matching keys:

plural_rule	Keys
slavic	singular, dual, plural
arabic, hebrew	2, ten, higher
double	double, higher
maltese	dual, plural
czech	zero, past, future or future-singular/future-paucal
future_if_positive, past_if_negative, icelandic	past, future

var keys: List[String] = ["singular", "dual", "plural"]
var forms: List[String] = ["{0} dzień", "{0} dni", "{0} dni"]
pirate.plural_rule = "slavic"
pirate.set_timeframe("days", keys, forms)

Locale.describe(frame, delta) and describe_multi(frames, deltas) format one or more signed timeframes directly, such as describe("hours", -2).

Built-in locales

Name	Language	Name	Language
af	Afrikaans	lo	Lao
am	Amharic	lt	Lithuanian
ar	Arabic	lv	Latvian
ar-iq	Levant Arabic	mk	Macedonian
ar-ma	Moroccan Arabic	mk-latn	Macedonian (Latin)
ar-mr	Mauritanian Arabic	ml	Malayalam
ar-tn	Algerian/Tunisian Arabic	mr	Marathi
az	Azerbaijani	ms	Malay
be	Belarusian	mt	Maltese
bg	Bulgarian	nb	Norwegian Bokmål
bn	Bengali	ne	Nepali
ca	Catalan	nl	Dutch
cs	Czech	nn	Norwegian Nynorsk
da	Danish	or	Odia
de	German	pl	Polish
de-at	Austrian German	pt	Portuguese
de-ch	Swiss German	pt-br	Brazilian Portuguese
ee	Estonian	rm	Romansh
el	Greek	ro	Romanian
en	English	ru	Russian
eo	Esperanto	se	Sami
es	Spanish	si	Sinhala
eu	Basque	sk	Slovak
fa	Persian	sl	Slovenian
fi	Finnish	sq	Albanian
fr	French	sr	Serbian
fr-ca	Canadian French	sv	Swedish
he	Hebrew	sw	Swahili
hi	Hindi	ta	Tamil
hr	Croatian	th	Thai
hu	Hungarian	tl	Tagalog
hy	Armenian	tr	Turkish
id	Indonesian	ua	Ukrainian
is	Icelandic	ur	Urdu
it	Italian	uz	Uzbek
ja	Japanese	vi	Vietnamese
ka	Georgian	zh-cn	Simplified Chinese
kk	Kazakh	zh-hk	Hong Kong Chinese
ko	Korean	zh-tw	Traditional Chinese
la	Latin	zu	Zulu
lb	Luxembourgish

Region aliases follow Arrow, for example en-gb, es-es, ar-eg, and sr-rs. zh, zh-hans, and zh-hant select the Chinese locales.

Differences from Arrow

• Thai and Lao format years in the Buddhist era (+543) and parse them back to Gregorian years. For YY, 00–99 map to 2500–2599 BE.
• Automatic humanize falls back to days in the 16 locales without week text, where Arrow raises. Explicit week or quarter granularity still raises not translated in locales without that text.
• dehumanize works in every built-in locale, including the 18 that Arrow rejects, such as Arabic, Hebrew, Korean, and Thai. It reads the text humanize produces. Icelandic "nokkrar sekúndur" ("a few seconds") has no count and reads as one second.
• Do also parses the ordinals that format writes, such as German 3. and Korean 세번째.
• Multi-unit Hebrew text takes its direction from the whole distance, not from the last unit.
• Amharic future hours use the hour word; Arrow 1.4.0 repeats the seconds word there.

English and Chinese dehumanize keep their 1.0 grammar.