tessera-ctl/apps/web/src/messages/umlaut-guard.spec.ts

import { describe, expect, it } from 'vitest';
import de from './de.json';
import en from './en.json';
import { UMLAUT_ALLOWLIST, UMLAUT_REPLACEMENTS } from './umlaut-dictionary';

/**
 * umlaut-guard — regression guard for the de.json umlaut correction.
 *
 * This spec reads exclusively the parsed JSON content of `de.json` /
 * `en.json` and never greps repo-wide over source files. Grepping
 * source files would trigger on `umlaut-dictionary.ts` itself, since it
 * necessarily contains the wrong substitute spellings as its map keys
 * (e.g. `'fuer'`, `'loeschen'`).
 *
 * The second test below is the actual regression guard: it fails on any
 * NEWLY introduced substitute spelling (`ae`/`oe`/`ue`/`ss` token not on
 * the allowlist) without pinning today's wording — it pins vocabulary,
 * not phrasing. Fixing a failure is a one-line addition to either
 * `UMLAUT_REPLACEMENTS` (if it is a substitute spelling) or
 * `UMLAUT_ALLOWLIST` (if it is correct German), as instructed by the
 * assertion message itself.
 */

const UMLAUT_TOKEN_RE = /[A-Za-zÄÖÜäöüß]+/g;
// Case-insensitive, sonst rutscht jede grossgeschriebene Ersatzschreibung durch:
// "Aenderungen" beginnt mit "Ae", nicht mit "ae". Genau daran sind beim ersten
// Durchgang "Aenderung", "Aenderungen" und "Oeffnen" vorbeigekommen — darunter
// zwei gut sichtbare Schaltflaechen.
const SUSPECT_RE = /(ae|oe|ue|ss)/i;

/** Recursively flattens a nested message object into dot-joined leaf key paths. */
function flattenKeys(obj: unknown, prefix = ''): string[] {
  if (obj === null || typeof obj !== 'object') {
    return [prefix];
  }
  return Object.entries(obj as Record<string, unknown>).flatMap(([key, value]) =>
    flattenKeys(value, prefix ? `${prefix}.${key}` : key),
  );
}

/**
 * Walks every leaf string value in `obj`, skipping values that contain
 * `@` (e.g. the `testToPlaceholder` example email address), and invokes
 * `visit` with each whole-word token found plus its dot-joined key path.
 */
function walkTokens(obj: unknown, visit: (token: string, path: string) => void, prefix = ''): void {
  if (obj === null || typeof obj !== 'object') {
    return;
  }
  for (const [key, value] of Object.entries(obj as Record<string, unknown>)) {
    const path = prefix ? `${prefix}.${key}` : key;
    if (typeof value === 'string') {
      if (value.includes('@')) {
        continue;
      }
      for (const token of value.match(UMLAUT_TOKEN_RE) ?? []) {
        visit(token, path);
      }
    } else if (value && typeof value === 'object') {
      walkTokens(value, visit, path);
    }
  }
}

describe('de.json umlaut regression guard', () => {
  it('contains no token from UMLAUT_REPLACEMENTS (known substitute spelling)', () => {
    const hits: string[] = [];
    walkTokens(de, (token, path) => {
      if (Object.prototype.hasOwnProperty.call(UMLAUT_REPLACEMENTS, token)) {
        hits.push(`${path}: "${token}" should be "${UMLAUT_REPLACEMENTS[token]}"`);
      }
    });
    expect(hits, `Substitute spellings found in de.json:\n${hits.join('\n')}`).toEqual([]);
  });

  it('flags any new ae/oe/ue/ss token not on UMLAUT_ALLOWLIST', () => {
    const allowlist = new Set(UMLAUT_ALLOWLIST);
    const hits: string[] = [];
    walkTokens(de, (token, path) => {
      if (SUSPECT_RE.test(token) && !allowlist.has(token)) {
        hits.push(
          `${path}: "${token}" is a new word not on UMLAUT_ALLOWLIST. ` +
            `Add it to UMLAUT_REPLACEMENTS in umlaut-dictionary.ts if it is a substitute ` +
            `spelling, or to UMLAUT_ALLOWLIST if it is already correct German.`,
        );
      }
    });
    expect(hits, `New ae/oe/ue/ss words found in de.json:\n${hits.join('\n')}`).toEqual([]);
  });

  it('has an identical (recursively-flattened) key set in de and en', () => {
    const deKeys = flattenKeys(de).sort();
    const enKeys = flattenKeys(en).sort();

    const missingInEn = deKeys.filter((key) => !enKeys.includes(key));
    const missingInDe = enKeys.filter((key) => !deKeys.includes(key));

    expect(missingInEn, `Keys present in de.json but missing in en.json: ${missingInEn.join(', ')}`).toEqual([]);
    expect(missingInDe, `Keys present in en.json but missing in de.json: ${missingInDe.join(', ')}`).toEqual([]);
    expect(deKeys).toEqual(enKeys);
  });
});