Fix llms.txt and llms-full.txt language variants filtering (#4179)

packages/gitbook/src/components/SpaceLayout/categorizeVariants.ts

@@ -1,20 +1,17 @@
 import { languages } from '@/intl/translations';
 import type { GitBookSiteContext } from '@/lib/context';
+import { getSiteSpaceLanguages, normalizeLanguage } from '@/lib/sites';
 
 /**
  * Categorize the variants of the space into generic and translation variants.
  */
 export function categorizeVariants(context: GitBookSiteContext) {
     const { siteSpace, visibleSiteSpaces: siteSpaces } = context;
-    const normalizeLanguage = (language: string | undefined) =>
-        language === undefined ? languages.en.locale : language;
 
     const currentLanguage = normalizeLanguage(context.locale);
 
     // Get all languages of the variants.
-    const variantLanguages = [
-        ...new Set(siteSpaces.map((space) => normalizeLanguage(space.space.language))),
-    ];
+    const variantLanguages = getSiteSpaceLanguages(siteSpaces);
 
     // We show the language picker when there are at least 2 distinct languages.
     // Spaces without an explicit language are treated as English, matching runtime defaults.

packages/gitbook/src/lib/sites.test.ts

@@ -0,0 +1,31 @@
+import { describe, expect, it } from 'bun:test';
+import type { SiteSpace } from '@gitbook/api';
+import { TranslationLanguage } from '@gitbook/api';
+
+import { filterSiteSpacesByLocale } from './sites';
+
+function makeSiteSpace(language: TranslationLanguage | undefined): SiteSpace {
+    return { space: { language } } as unknown as SiteSpace;
+}
+
+describe('filterSiteSpacesByLocale', () => {
+    it('returns all spaces on a single-language site', () => {
+        const spaces = [makeSiteSpace(undefined), makeSiteSpace(undefined)];
+        expect(filterSiteSpacesByLocale(spaces, TranslationLanguage.En)).toEqual(spaces);
+    });
+
+    it('filters by locale on a multi-language site', () => {
+        const en = makeSiteSpace(TranslationLanguage.En);
+        const fr = makeSiteSpace(TranslationLanguage.Fr);
+        expect(filterSiteSpacesByLocale([en, fr], TranslationLanguage.Fr)).toEqual([fr]);
+    });
+
+    it('treats undefined language as English', () => {
+        const undefinedLanguage = makeSiteSpace(undefined);
+        const en = makeSiteSpace(TranslationLanguage.En);
+        expect(filterSiteSpacesByLocale([undefinedLanguage, en], TranslationLanguage.En)).toEqual([
+            undefinedLanguage,
+            en,
+        ]);
+    });
+});

packages/gitbook/src/lib/sites.ts

@@ -1,3 +1,4 @@
+import { languages } from '@/intl/translations';
 import type { GitBookSiteContext } from '@/lib/context';
 import type {
     LocalizedString,
@@ -39,6 +40,21 @@ export function getSiteStructureSections(
         : [];
 }
 
+/**
+ * Normalize a space language to a locale string.
+ * Spaces without an explicit language are treated as English, matching runtime defaults.
+ */
+export function normalizeLanguage(language: string | undefined): string {
+    return language === undefined ? languages.en.locale : language;
+}
+
+/**
+ * Return the distinct normalized languages across a set of site spaces.
+ */
+export function getSiteSpaceLanguages(siteSpaces: SiteSpace[]): string[] {
+    return [...new Set(siteSpaces.map((space) => normalizeLanguage(space.space.language)))];
+}
+
 /**
  * Filter site spaces to only include those matching the given locale
  */
@@ -46,7 +62,15 @@ export function filterSiteSpacesByLocale(
     siteSpaces: SiteSpace[],
     locale: TranslationLanguage | undefined
 ): SiteSpace[] {
-    return siteSpaces.filter((siteSpace) => siteSpace.space.language === locale);
+    const variantLanguages = getSiteSpaceLanguages(siteSpaces);
+    if (variantLanguages.length <= 1) {
+        return siteSpaces;
+    }
+
+    const normalizedLocale = normalizeLanguage(locale);
+    return siteSpaces.filter(
+        (siteSpace) => normalizeLanguage(siteSpace.space.language) === normalizedLocale
+    );
 }
 
 /*