legesher/language-packs

Legesher Language Packs

Legesher's language packs 📦 are the localization layer for code; the fixed vocabulary of a programming language in other natural languages.

Every reserved word within a programming language's grammar (keywords, builtins, exceptions, error messages) is compiled into a language pack that, when utilized by Legesher, allows code to be in any language and it compiles all the same.

Release	Languages supported	Target	Versions	License	legesher-i18n	Commit
0.4.1	51	python	3.10-3.14	Apache-2.0	0.23.0	05ec1a05413d

Languages

Pack	Language	Native name	Status	Package
am	Amharic	አማርኛ	experimental	0.0.3
ar	Arabic	العربية	experimental	0.0.5
bn	Bengali	বাংলা	experimental	0.0.3
bs	Bosnian	Bosanski	experimental	0.0.3
ca	Catalan	Català	experimental	0.0.3
cs	Czech	Čeština	experimental	0.0.3
cy	Welsh	Cymraeg	experimental	0.0.3
da	Danish	Dansk	experimental	0.0.4
de	German	Deutsch	experimental	0.0.4
el	Greek	Ελληνικά	experimental	0.0.5
en	English	English	experimental	0.0.4
es	Spanish	Español	experimental	0.0.5
fa	Persian	فارسی	experimental	0.0.3
fi	Finnish	Suomi	experimental	0.0.5
fr	French	Français	experimental	0.0.4
ga	Irish	Gaeilge	experimental	0.0.3
gd	Scottish Gaelic	Gàidhlig	experimental	0.0.3
gl	Galician	Galego	experimental	0.0.3
gu	Gujarati	ગુજરાતી	experimental	0.0.3
he	Hebrew	עברית	experimental	0.0.5
hi	Hindi	हिन्दी	experimental	0.0.5
hr	Croatian	Hrvatski	experimental	0.0.3
hu	Hungarian	Magyar	experimental	0.0.5
id	Indonesian	Bahasa Indonesia	experimental	0.0.4
it	Italian	Italiano	experimental	0.0.5
ja	Japanese	日本語	experimental	0.0.4
kn	Kannada	ಕನ್ನಡ	experimental	0.0.3
ko	Korean	한국어	experimental	0.0.5
lv	Latvian	Latviešu	experimental	0.0.3
mk	Macedonian	Македонски	experimental	0.0.3
ml	Malayalam	മലയാളം	experimental	0.0.3
mn	Mongolian	Монгол	experimental	0.0.3
ms	Malay	Bahasa Melayu	experimental	0.0.4
my	Burmese	မြန်မာ	experimental	0.0.3
nl	Dutch	Nederlands	experimental	0.0.4
no	Norwegian	Norsk	experimental	0.0.5
pl	Polish	Polski	experimental	0.0.5
pt	Portuguese	Português	experimental	0.0.5
ro	Romanian	Română	experimental	0.0.3
ru	Russian	Русский	experimental	0.0.5
si	Sinhala	සිංහල	experimental	0.0.3
sr	Serbian	Српски	experimental	0.0.3
su	Sundanese	Basa Sunda	experimental	0.0.4
sv	Swedish	Svenska	experimental	0.0.5
te	Telugu	తెలుగు	experimental	0.0.3
th	Thai	ไทย	experimental	0.0.5
tr	Turkish	Türkçe	experimental	0.0.5
ur	Urdu	اردو	experimental	0.0.4
uz	Uzbek	O'zbek	experimental	0.0.3
vi	Vietnamese	Tiếng Việt	experimental	0.0.5
zh	Chinese	中文	experimental	0.0.6

What a pack is made of

Target

Each language pack is created based on the requirements of a programming language, its target, which sets the categories it carries, the reserved words in each, and the forms a rendering is allowed to take. The specific versions of a programming language a particular reserved word belongs to are noted as well, so the language pack grows with the progression of a programming language.

Categories

Category	Subcategories	Terms	What it holds
keywords	keyword	39	The words the grammar reserves. A rendering cannot collide with anything else in the vocabulary. All are legal identifiers except except*, whose renderings carry the marker through.
builtins	builtin_func, builtin_type, special_constant	73	Functions, types and constants available without importing anything: print, dict, Ellipsis.
exceptions	exception	70	The error hierarchy, from BaseException down.
implicit_params	implicit_param	2	self and cls. Conventional parameter names that Legesher treats deterministically, not words the grammar reserves.
symbols	--	33	What a symbol is called: _ is underscore in English. Prose names, not identifiers.
variable_conventions	--	15	The idiomatic short name for a variable role: the loop counter, the throwaway, the accumulator.
messages	20 message families	54	Error-message templates with named {placeholders}, and the traceback scaffolding that frames them.

Maturity

Stage	What it means
experimental	Machine-drafted, then curated by maintainers and passed through the mechanical gates: identifier validity, cross-section collision checks, Unicode normalization, and a regeneration check proving the packs match their source of truth. No native speaker has signed off.
reviewed	A native speaker of the language has reviewed the pack and signed off on it.
official	Certified review plus term screening.

The dataset

Config	One row per	Answers
locales	natural language	who is this language?
pl_terminology	(target, category, source term)	what vocabulary does the target carry, and on which versions?
vocabulary	(locale, target, version, category, source term)	what is it called in this language?
packs	(locale, target)	how far along is this pairing?

pl_terminology

The programming language's own terminology. No natural language appears in this config.

• target, category, source_term: the key
• subcategory: the finer kind, as in the table above
• source_text: the English surface, where source_term is not itself that surface. symbols carry the name, so paren_open is opening parenthesis. variable_conventions carry the identifier, so accumulator is total. messages carry the template. Null on keywords, builtins, exceptions and implicit_params, where source_term is the English. A message with no text carries "", which is a different fact from null: AssertionError on a bare assert genuinely has no message
• glyph: the character a symbol names, so paren_open is (. Null on every other category, and null on the symbols with no printable form: indentation, a line break, a space
• pl_versions: the target versions the entry is valid for

Availability lives here, once: except* carries ["3.11", ...], and a term in two categories carries two rows with their own sets, so type is a builtin on every version and a keyword only from 3.12.

This config is also the coverage checklist. A pl_terminology entry with no matching vocabulary rows for a locale is a gap in that pack, stated by the data rather than discovered downstream: the symbols, variable_conventions and messages categories ship their source-side inventory ahead of any locale's renderings for exactly that reason.

vocabulary

The renderings, one row per (locale, target, version, category, source term).

• locale: join into locales; target, category, source_term: join into pl_terminology
• pl_version: rows are given resolved, so each states what you would get on that version of the target
• translation: the rendering Legesher ships
• defined_at: the version whose pack file defines this rendering. A row inherits when defined_at differs from pl_version. This is a fact about the rendering, not the term: Sinhala redefines terms at 3.13 to conjunct forms, so its defined_at differs from every other language's on the same term

category is part of the key on purpose: type is rendered once as a keyword and once as a builtin, and those renderings may differ.

locales

Natural-language identity: names (language_name, native_name), identity tags (iso639_2, bcp47), script properties (rtl, has_case_distinction).

locale is a pack id, not a language tag. Where the two differ, locale is the shorter form: no is BCP 47 nb, zh is zh-Hans, sr is sr-Cyrl. Join to anything keyed by language tag on bcp47, or you drop exactly the rows that differ.

packs

The shipping unit, one row per (locale, target) pairing.

• package_version: the semver of this pairing's own distributable. A row reading ar / python / 0.0.5 means these are the words in legesher-i18n-python-ar==0.0.5. Each pairing versions on its own clock
• tier: the maturity of exactly this pairing
• pl_versions: the target versions this pack covers

JSON

json/<locale>/<target>.json is the loadable form: the translations alone, with the pack's package_version and a pl_version_range string naming the versions it covers. Values are the latest version's resolved rendering over the union of terms across the range; the parquet configs carry the full per-version structure the JSON leaves out.

Normalization

Python normalizes identifiers to NFKC when it parses them (PEP 3131), so a rendering whose own NFKC form differs from it is typed one way and resolved another.

23 renderings in this release are NFKC-unstable, all in Thai (th), all through U+0E33 THAI CHARACTER SARA AM. If you compare a translation value against a name taken from a running interpreter, normalize both to NFKC first.

Packages

Legesher uses this dataset as the language packs for coding in other natural languages. Each language ships as its own distribution, legesher-i18n-python-<locale>, so Arabic is legesher-i18n-python-ar, and they are loaded by legesher-i18n, which resolves a pack against the interpreter you are running.

Citation

@misc{legesher_language_packs,
  title        = {Legesher Language Packs: programming vocabulary rendered in natural languages},
  version      = {0.4.1},
  author       = {Legesher},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/legesher/language-packs}}
}