From 8cd08f4f02de72f6c02cc07ec1e1ab3103081dd4 Mon Sep 17 00:00:00 2001
From: pengzhanbo <volodymyr@foxmail.com>
Date: Thu, 12 Feb 2026 00:59:35 +0800
Subject: [PATCH] feat(plugin-search): add language-based text segmentation
 support for search (#838)

---
 .../src/client/components/SearchBox.vue        | 18 +++++++++++++-----
 .../src/node/prepareSearchIndex.ts             | 12 +++++++++---
 2 files changed, 22 insertions(+), 8 deletions(-)

diff --git a/plugins/plugin-search/src/client/components/SearchBox.vue b/plugins/plugin-search/src/client/components/SearchBox.vue
index 0c8ec01e..d9abd519 100644
--- a/plugins/plugin-search/src/client/components/SearchBox.vue
+++ b/plugins/plugin-search/src/client/components/SearchBox.vue
@@ -25,7 +25,7 @@ import {
   toRef,
   watch,
 } from 'vue'
-import { useRouteLocale, useRouter, withBase } from 'vuepress/client'
+import { usePageLang, useRouteLocale, useRouter, withBase } from 'vuepress/client'
 import { useLocale, useSearchIndex } from '../composables/index.js'
 import { LRUCache } from '../utils/index.js'
 import BackIcon from './icons/BackIcon.vue'
@@ -43,6 +43,7 @@ const emit = defineEmits<{
 
 const routeLocale = useRouteLocale()
 const locale = useLocale(toRef(() => locales))
+const lang = usePageLang()
 
 const el = shallowRef<HTMLElement>()
 const resultsEl = shallowRef<HTMLElement>()
@@ -59,8 +60,14 @@ const { activate } = useFocusTrap(el, {
   immediate: true,
 })
 
-const searchIndex = computedAsync(async () =>
-  markRaw(
+const searchIndex = computedAsync(async () => {
+  let tokenize: ((str: string) => string[]) | undefined
+  if (typeof Intl.Segmenter !== 'undefined') {
+    const segmenter = new Intl.Segmenter(lang.value, { granularity: 'word' })
+    tokenize = str => Array.from(segmenter.segment(str)).map(s => s.segment)
+  }
+
+  return markRaw(
     MiniSearch.loadJSON<Result>(
       (await searchIndexData.value[routeLocale.value]?.())?.default,
       {
@@ -70,13 +77,14 @@ const searchIndex = computedAsync(async () =>
           fuzzy: 0.2,
           prefix: true,
           boost: { title: 4, text: 2, titles: 1 },
+          tokenize,
         },
         ...options.miniSearch?.searchOptions,
         ...options.miniSearch?.options,
       },
     ),
-  ),
-)
+  )
+})
 
 const disableQueryPersistence = computed(() =>
   options?.disableQueryPersistence === true,
diff --git a/plugins/plugin-search/src/node/prepareSearchIndex.ts b/plugins/plugin-search/src/node/prepareSearchIndex.ts
index 44964dcc..3042e3bc 100644
--- a/plugins/plugin-search/src/node/prepareSearchIndex.ts
+++ b/plugins/plugin-search/src/node/prepareSearchIndex.ts
@@ -21,12 +21,16 @@ const SEARCH_INDEX_DIR = 'internal/minisearchIndex/'
 const indexByLocales = new Map<string, MiniSearch<IndexObject>>()
 const indexCache = new Map<string, IndexObject[]>()
 
-function getIndexByLocale(locale: string, options: SearchIndexOptions['searchOptions']) {
+function getIndexByLocale(locale: string, lang: string, options: SearchIndexOptions['searchOptions']) {
+  const segmenter = new Intl.Segmenter(lang, { granularity: 'word' })
   let index = indexByLocales.get(locale)
   if (!index) {
     index = new MiniSearch<IndexObject>({
       fields: ['title', 'titles', 'text'],
       storeFields: ['title', 'titles'],
+      tokenize(text) {
+        return Array.from(segmenter.segment(text)).map(s => s.segment)
+      },
       ...options.miniSearch?.options,
     })
     indexByLocales.set(locale, index)
@@ -93,7 +97,8 @@ export async function onSearchIndexRemoved(
     const page = app.pages.find(p => p.filePathRelative?.endsWith(filepath))!
     const fileId = page.path
     const locale = page.pathLocale
-    const index = getIndexByLocale(locale, searchOptions)
+    const lang = page.lang
+    const index = getIndexByLocale(locale, lang, searchOptions)
     const cache = getIndexCache(fileId)
     if (cache && cache.length)
       index.removeAll(cache)
@@ -131,7 +136,8 @@ async function indexFile(page: Page, options: SearchIndexOptions['searchOptions'
   // get file metadata
   const fileId = page.path
   const locale = page.pathLocale
-  const index = getIndexByLocale(locale, options)
+  const lang = page.lang
+  const index = getIndexByLocale(locale, lang, options)
   const cache = getIndexCache(fileId)
   // retrieve file and split into "sections"
   const html = `<h1><a href="#"><span>${page.frontmatter.title || page.title}</span></a></h1>