intlayerPrune

intlayerPrune is a Vite plugin that removes unused content fields from compiled dictionary JSON files during a production build. Instead of shipping every field from every dictionary, only the fields that are actually accessed in your source code are included in the bundle.

The plugin is already included and configured automatically when you use intlayer(). You only need to register it manually if you are composing the plugin stack yourself.

Usage

As part of intlayer() (recommended)

Enable pruning through your Intlayer config and the main plugin handles everything:

// intlayer.config.tsimport { defineConfig } from "intlayer";export default defineConfig({  build: {    optimize: true, // enables both prune and minify  },});

// vite.config.tsimport { defineConfig } from "vite";import { intlayer } from "vite-intlayer";export default defineConfig({  plugins: [intlayer()],});

Standalone

If you are composing the plugin stack manually, intlayerPrune and intlayerMinify share a PruneContext object that must be created once and passed to both:

// vite.config.tsimport { defineConfig } from "vite";import { intlayerPrune, intlayerMinify } from "vite-intlayer";import { createPruneContext } from "@intlayer/babel";import { getConfiguration } from "@intlayer/config/node";const intlayerConfig = getConfiguration();const pruneContext = createPruneContext();export default defineConfig({  plugins: [    intlayerPrune(intlayerConfig, pruneContext),    intlayerMinify(intlayerConfig, pruneContext), // optional, reads from same context  ],});

How it works

1. Usage analysis (buildStart)

During buildStart, the intlayerOptimize plugin (also part of intlayer()) scans every component source file listed in build.filesList. For each useIntlayer('key') or getIntlayer('key') call, it records exactly which fields are accessed, e.g.:

const { title, description } = useIntlayer("myDict");// records: myDict → { title, description }

This builds pruneContext.fieldUsageMap before any transform calls run.

2. JSON pruning (transform, enforce: 'pre')

When Vite processes a compiled dictionary JSON file, intlayerPrune intercepts it before Vite's built-in JSON → ESM conversion. It reads the field-usage map from pruneContext and removes any content field that is not in the recorded usage set.

Two content shapes are supported:

• Static dictionaries — { nodeType: "translation", translation: { en: {...}, fr: {...} } }. Fields are pruned per-locale inside translation.
• Dynamic (per-locale) dictionaries — flat { fieldA: ..., fieldB: ... }. Fields are pruned at the top level.

3. Edge cases

If a dictionary's content structure cannot be recognised (e.g. an unusual nested shape), it is added to pruneContext.dictionariesWithEdgeCases and left untouched. A warning is logged. intlayerMinify also skips these dictionaries.

4. Field-rename map

When pruning succeeds, intlayerPrune also writes pruneContext.dictionaryKeyToFieldRenameMap — a mapping from original field names to short aliases. intlayerMinify reads this map to rename fields in the output JSON, and intlayerOptimize's Babel rename pass updates property accesses in source files accordingly.

Activation conditions

intlayerPrune is active only when all of the following are true:

1. The Vite command is build.
2. build.optimize is true (or undefined, which defaults to true for builds).
3. build.purge is true in your Intlayer config.

It is automatically disabled when editor.enabled is true because the editor needs the full dictionary content.