@localeguard/react-analyzer

Static analysis for LocaleGuard that catches hardcoded UI text in React/TypeScript source — strings that never make it into the locale files at all. Powered by the TypeScript compiler API; no project compilation required.

It reports two issue types:

• hardcoded-string — literal JSX text, e.g. <Button>Create Cluster</Button>
• hardcoded-attribute — literal aria-label, title, alt, or placeholder, e.g. <button aria-label="Close dialog">

Design: low false positives

Only literal values are flagged. Anything already dynamic is left alone:

| Code | Flagged? | | --- | --- | | <h1>Cluster Manager</h1> | ✅ hardcoded-string | | <p>{t("app.title")}</p> | ❌ already localized | | <Trans i18nKey="x">Welcome</Trans> | ❌ inside a translation component | | <span>100%</span> | ❌ no letters | | <button aria-label="Close"> | ✅ hardcoded-attribute | | <button aria-label={t("close")}> | ❌ already localized | | <img alt="" /> | ❌ decorative empty alt |

Translation component names (default Trans) are configurable via translationComponents in your LocaleGuard config.

Usage

This package is composed automatically by the localeguard CLI — running localeguard check includes source-code analysis using your config's include, ignore, and translationComponents. Disable it with localeguard check --no-code.

It can also be used directly:

import { analyzeProject } from "@localeguard/react-analyzer";

const issues = analyzeProject(
  { include: ["src/**/*.{ts,tsx}"], translationComponents: ["Trans"] },
  { rootDir: process.cwd() },
);