@financica/nace-codes
v3.0.0
Published
TypeScript library for NACE and NACEBEL economic activity classification codes
Maintainers
Readme
nace-codes
A TypeScript library for working with NACE and NACEBEL economic activity classification codes.
Features
- Full NACE Rev. 2.1 Support: Complete hierarchy of European economic activity classifications
- NACEBEL Extension: Support for Belgian-specific 6-digit extensions
- Multi-language: Descriptions in 24 languages (NACE) and 4 languages (NACEBEL), each loadable on its own
- Flexible Code Lookup: Search by various formats (e.g., "70.20", "7020", "702", "A")
- Hierarchical Navigation: Traverse parent-child relationships in the classification tree
- Rich Metadata: Access includes/excludes rules and explanatory notes
- Type-Safe: Full TypeScript support with strict typing
- Pay For What You Load: English ships in the default bundle; each other language is a separate module you opt into. Translations are the bulk of this dataset, so one locale costs roughly one locale.
- Tree-Shakeable: Core NACE and the Belgian NACEBEL extension are separate entry points with disjoint data. Importing NACE never pulls Belgian data into your bundle.
- Zero Dependencies: No external runtime dependencies
Installation
npm install @financica/nace-codesEntry Points
| Import | Contents |
| ------------------------------------------- | ----------------------------------------------------- |
| @financica/nace-codes | Core NACE Rev. 2.1 (EU) — the NACE class, English |
| @financica/nace-codes/nacebel | Belgian NACEBEL extension — the NACEBEL class |
| @financica/nace-codes/lang/<code> | NACE heading translations for one language |
| @financica/nace-codes/nacebel/lang/<code> | NACEBEL explanatory notes for one language (fr/nl/de) |
Only reach for nace-codes/nacebel when you need the Belgian national codes; it
bundles the NACE core data plus the Belgian delta.
Approximate gzipped cost, so you can see where the weight is:
| Module | gzipped |
| --------------------- | ------- |
| core NACE (English) | ~194 KB |
| lang/<code> | ~15 KB |
| NACEBEL (English) | ~470 KB |
| nacebel/lang/<code> | ~275 KB |
Quick Start
import { NACE } from "@financica/nace-codes";
import { NACEBEL } from "@financica/nace-codes/nacebel";
// Initialize the NACE classifier
const nace = new NACE();
// Look up a code
const code = nace.getCode("70.20");
console.log(code.description.en); // "Management consultancy activities"
// Other languages are opt-in — see "Languages" below
console.log(code.description.fr); // undefined
// Navigate hierarchy
const parent = nace.getParent("70.20");
console.log(parent?.code); // "70.2"
const children = nace.getChildren("70");
console.log(children.map((c) => c.code)); // ["70.1", "70.2"]
// Access metadata
const details = nace.getCode("01.11");
console.log(details.includes); // Description of what this code includes
console.log(details.excludes); // Description of what this code excludesLanguages
The 24 EU translations are ~95% of this dataset, so only English ships in the default bundle. Every other language is a module you import explicitly and hand to the constructor:
import { NACE } from "@financica/nace-codes";
import da from "@financica/nace-codes/lang/da";
const nace = new NACE({ languages: [da] });
nace.getCode("01.11").description.da; // "Dyrkning af korn ..."
nace.search("dyrkning", { language: "da" }); // matches Danish textOnly the languages you pass are populated. description.fr is undefined
without the fr pack, and search({ language: "fr" }) returns no results —
there is nothing to match against.
Load a locale on demand with a dynamic import:
const pack = await import(`@financica/nace-codes/lang/${locale}`);
const nace = new NACE({ languages: [pack.default] });Language packs are resolved when the code map is built, so pass them at construction. To switch locale at runtime, construct a new instance.
Languages with NACEBEL
NACEBEL has two independent translated fields:
nationalTitles(nl/fr/de/en) — the official Belgian titles. These ship eagerly innace-codes/nacebel, need no pack, and are what NACEBELsearch()matches for nl/fr/de.explanatoryNote— the long Markdown notes, by far the largest field. English ships in the bundle; fr/nl/de come fromnacebel/lang/<code>.
description (the inherited EU heading) still comes from lang/<code>, so a
French consumer who wants both fields passes both packs:
import { NACEBEL } from "@financica/nace-codes/nacebel";
import fr from "@financica/nace-codes/lang/fr";
import frNotes from "@financica/nace-codes/nacebel/lang/fr";
const nacebel = new NACEBEL({ languages: [fr, frNotes] });API Reference
NACE Class
constructor(options?: NACEOptions)
Initialize a new NACE classifier instance.
interface NACEOptions {
preload?: boolean; // Load all data on initialization (default: false)
languages?: readonly LanguagePack[]; // Translations to layer on (default: none)
}Note that preload only controls when the bundled data is parsed and indexed.
It has no effect on bundle size; that is what the language modules are for.
getCode(code: string): NACECode | null
Retrieve information about a specific NACE code.
const code = nace.getCode("70.20");
// Accepts various formats: "70.20", "7020", "702", "70", "M", etc.getParent(code: string): NACECode | null
Get the parent code in the hierarchy.
const parent = nace.getParent("70.20"); // Returns code 70.2getChildren(code: string): NACECode[]
Get all direct children of a code.
const children = nace.getChildren("70"); // Returns codes 70.1 and 70.2getAncestors(code: string): NACECode[]
Get all ancestors up to the top level.
const ancestors = nace.getAncestors("70.20");
// Returns: [70.2, 70, M] (from immediate parent to section)getDescendants(code: string): NACECode[]
Get all descendants recursively.
const descendants = nace.getDescendants("70");
// Returns all codes under 70, including 70.1, 70.10, 70.2, 70.21, 70.22getSiblings(code: string): NACECode[]
Get all codes at the same level with the same parent.
const siblings = nace.getSiblings("70.21");
// Returns: [70.22] (other codes under 70.2)search(query: string, options?: SearchOptions): NACECode[]
Search for codes by description.
interface SearchOptions {
language?: Language; // Language to search in (default: 'en')
fuzzy?: boolean; // Enable fuzzy matching (default: false)
limit?: number; // Maximum results (default: 10)
}
const results = nace.search("consultant", { language: "en", fuzzy: true });getAllCodes(level?: number): NACECode[]
Get all codes, optionally filtered by level.
const sections = nace.getAllCodes(1); // All section codes (A-U)
const divisions = nace.getAllCodes(2); // All 2-digit divisionsgetLevel(code: string): number
Get the hierarchical level of a code.
nace.getLevel("A"); // Returns: 1 (section)
nace.getLevel("01"); // Returns: 2 (division)
nace.getLevel("01.1"); // Returns: 3 (group)
nace.getLevel("01.11"); // Returns: 4 (class)NACEBEL Class
Extends NACE with Belgian-specific codes up to 6 digits.
Additional Methods
All NACE methods are available, plus:
getBelgianExtensions(naceCode: string): NACEBELCode[]
Get Belgian-specific extensions for a NACE code.
const extensions = nacebel.getBelgianExtensions("01.11");
// Returns 6-digit Belgian extensions like 01.11001, 01.11002, etc.Types
interface NACECode {
code: string;
level: number;
description: LanguageDescriptions;
parent?: string;
includes?: string;
includesAlso?: string;
excludes?: string;
implementationRule?: string;
}
interface NACEBELCode extends NACECode {
version: string;
nationalTitles: {
nl: string;
fr: string;
de: string;
en: string;
};
}
// `en` is always present; every other language appears only when its pack is loaded
type LanguageDescriptions = { en: string } & Partial<Record<OptionalLanguage, string>>;
// What `lang/<code>` and `nacebel/lang/<code>` default-export
interface LanguagePack {
language: OptionalLanguage;
descriptions?: Record<string, string>;
explanatoryNotes?: Record<string, string>;
}
type Language =
| "en"
| "fr"
| "de"
| "nl"
| "es"
| "it"
| "pt"
| "bg"
| "cs"
| "da"
| "el"
| "et"
| "fi"
| "ga"
| "hr"
| "hu"
| "lt"
| "lv"
| "mt"
| "pl"
| "ro"
| "sk"
| "sl"
| "sv";
type OptionalLanguage = Exclude<Language, "en">;The LANGUAGES, OPTIONAL_LANGUAGES, and BUNDLED_LANGUAGE constants are
exported too, if you need to enumerate what exists at runtime.
Code Format Normalization
The library automatically normalizes various input formats:
"70.20"→"7020"(internal format)"7020"→"7020"(already normalized)"702"→"702"(3-digit group)"70"→"70"(2-digit division)"M"→"M"(section)"m"→"M"(case insensitive for sections)
Examples
Finding Related Activities
// Find all consultancy-related activities
const consultancy = nace.search("consultancy");
consultancy.forEach((code) => {
console.log(`${code.code}: ${code.description.en}`);
});Building a Classification Tree
function printTree(code: string, indent = 0): void {
const node = nace.getCode(code);
if (!node) return;
console.log(" ".repeat(indent) + `${node.code}: ${node.description.en}`);
const children = nace.getChildren(code);
children.forEach((child) => printTree(child.code, indent + 2));
}
printTree("M"); // Print entire "Professional, scientific and technical activities" treeValidating Codes
function isValidNACE(code: string): boolean {
return nace.getCode(code) !== null;
}
function isValidNACEBEL(code: string): boolean {
return nacebel.getCode(code) !== null;
}Multi-language Support
import de from "@financica/nace-codes/lang/de";
import fr from "@financica/nace-codes/lang/fr";
import nl from "@financica/nace-codes/lang/nl";
const nace = new NACE({ languages: [fr, de, nl] });
const code = nace.getCode("70.20");
for (const lang of ["en", "fr", "de", "nl"] as const) {
console.log(`${lang}: ${code.description[lang]}`);
}Working with Inclusions and Exclusions
const code = nace.getCode("01.13");
console.log("This class includes:");
console.log(code.includes);
console.log("\nThis class also includes:");
console.log(code.includesAlso);
console.log("\nThis class excludes:");
console.log(code.excludes);Migrating from 2.x
Translations no longer ship in the default bundle. If you only read
description.en, or you only use NACEBEL's nationalTitles, nothing changes.
Otherwise, note that TypeScript will not flag this for you: the non-English
fields were already optional, so the break shows up at runtime as undefined
descriptions and empty search() results. Audit for:
description.<lang>for any language other thanensearch(query, { language })with a non-enlanguage — except NACEBEL nl/fr/de, which matchnationalTitlesand keep workingexplanatoryNote.<lang>for fr/nl/de on NACEBEL codes
Each is fixed by importing the matching pack and passing it to the constructor:
- const nace = new NACE();
+ import fr from "@financica/nace-codes/lang/fr";
+ const nace = new NACE({ languages: [fr] });Data Sources
This library uses official classification data from:
- NACE Rev. 2.1: European Union statistical classification
- NACEBEL 2025: Belgian national extension of NACE
Development
This repo uses Bun and the oxc toolchain: oxlint, oxfmt, and tsdown (rolldown-powered bundler).
bun install
bun run generate:data # regenerate src/generated/* from data/
bun run build # generate data + bundle with tsdown
bun run test # vitest
bun run typecheck # tsc --noEmit
bun run lint # oxlint
bun run format # oxfmt --write
bun run ci # typecheck + lint + test + buildThe classification data lives in data/ as TSV. generate:data compiles it
into src/generated/*.ts, which is committed. Core NACE data is emitted to
naceData.ts; each national extension (e.g. NACEBEL) is emitted as a separate
delta module so the entry points stay tree-shakeable. Translations are split out
per language into generated/lang/*.ts and generated/nacebel/lang/*.ts, one
build entry each.
src/languageList.ts is the single source of truth for the language axis: the
generator, the tsdown entry map, and the Language type all derive from it, so
adding a language means editing one list.
License
MIT
Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
Support
For issues and questions, please use the GitHub issue tracker.
