@vuepress-plume/plugin-search 1.0.0-rc.214 → 1.0.0-rc.215

package/dist/client/composables/index.js

@@ -61,5 +61,4 @@ if (__VUEPRESS_DEV__ && (import.meta.webpackHot || import.meta.hot))
   __VUE_HMR_RUNTIME__.updateSearchIndex = (data) => {
     searchIndexData.value = data
   }
-
 export {loadSearchIndexJSON, useLocale, useSearchIndex}

package/dist/client/config.js

@@ -15,5 +15,4 @@ var config_default = defineClientConfig({
     )
   },
 })
-
 export {config_default as default}

package/dist/node/index.d.ts

@@ -20,9 +20,6 @@ interface SearchIndexOptions {
  *
  * 为 VuePress 应用中的所有页面准备搜索索引。
  *
- * This function indexes all pages concurrently, creating separate MiniSearch
- * instances for each locale, and writes the indexes to temporary files.
- *
  * 此函数并发索引所有页面，为每个语言创建独立的 MiniSearch 实例，
  * 并将索引写入临时文件。
  *

package/dist/node/index.js

@@ -485,5 +485,4 @@ function searchPlugin({locales = {}, isSearchable, ...searchOptions} = {}) {
     },
   })
 }
-
 export {prepareSearchIndex, searchPlugin}

package/dist/shared/index.d.ts

@@ -1,5 +1,27 @@
 import {Options} from 'minisearch'
 import {LocaleConfig, Page} from 'vuepress/core'
+/**
+ * Create a tokenizer shared by the node (index build) and client (query) sides.
+ *
+ * 创建由 Node 端（构建索引）与客户端（查询）共享的分词器。
+ *
+ * MiniSearch requires the tokenizer used to build the index to be identical to
+ * the one used at query time. Defining it in one place guarantees both sides
+ * use the same strategy: `Intl.Segmenter` when available, otherwise the same
+ * deterministic whitespace fallback. Blank and punctuation-only tokens are
+ * discarded on both sides, so they never pollute the index or the query.
+ *
+ * MiniSearch 要求构建索引与查询时使用完全一致的分词器。集中定义可保证两端策略一致：
+ * 支持 `Intl.Segmenter` 时使用它分词，否则统一退化为确定性的空白符分词。
+ * 两端都会丢弃空白与纯标点词元，避免污染索引与查询。
+ *
+ * @param lang - Language code for word segmentation / 用于分词的语言代码
+ * @returns Tokenizer that splits a string into tokens / 将字符串切分为词元的函数
+ * @example
+ * const tokenize = createTokenizer('en')
+ * tokenize('Hello world') // ['Hello', 'world'] with Intl.Segmenter
+ */
+export declare function createTokenizer(lang: string): (text: string) => string[]
 /**
  * Locale options for search UI strings.
  *
@@ -59,28 +81,6 @@ export interface SearchLocaleOptions {
  * 搜索框的语言配置类型。
  */
 export type SearchBoxLocales = LocaleConfig<SearchLocaleOptions>
-/**
- * Create a tokenizer shared by the node (index build) and client (query) sides.
- *
- * 创建由 Node 端（构建索引）与客户端（查询）共享的分词器。
- *
- * MiniSearch requires the tokenizer used to build the index to be identical to
- * the one used at query time. Defining it in one place guarantees both sides
- * use the same strategy: `Intl.Segmenter` when available, otherwise the same
- * deterministic whitespace fallback. Blank and punctuation-only tokens are
- * discarded on both sides, so they never pollute the index or the query.
- *
- * MiniSearch 要求构建索引与查询时使用完全一致的分词器。集中定义可保证两端策略一致：
- * 支持 `Intl.Segmenter` 时使用它分词，否则统一退化为确定性的空白符分词。
- * 两端都会丢弃空白与纯标点词元，避免污染索引与查询。
- *
- * @param lang - Language code for word segmentation / 用于分词的语言代码
- * @returns Tokenizer that splits a string into tokens / 将字符串切分为词元的函数
- * @example
- * const tokenize = createTokenizer('en')
- * tokenize('Hello world') // ['Hello', 'world'] with Intl.Segmenter
- */
-export declare function createTokenizer(lang: string): (text: string) => string[]
 /**
  * Options for the search plugin.
  *

package/dist/shared/index.js

@@ -1,16 +1,32 @@
+const SPACE_OR_PUNCTUATION_RE = /[\n\r\p{Z}\p{P}]+/u
+const UNSPACED_RUN_RE =
+  /[\p{scx=Han}\p{scx=Hiragana}\p{scx=Katakana}]+|[^\p{scx=Han}\p{scx=Hiragana}\p{scx=Katakana}]+/gu
+const UNSPACED_START_RE = /^[\p{scx=Han}\p{scx=Hiragana}\p{scx=Katakana}]/u
+const NO_SPACE_AND_PUNCTUATION_RE_RE = /\p{L}|\p{N}/u
+
 function isWordToken(token) {
-  return /\p{L}|\p{N}/u.test(token)
+  return NO_SPACE_AND_PUNCTUATION_RE_RE.test(token)
 }
 
 function createTokenizer(lang) {
   if (typeof Intl !== 'undefined' && typeof Intl.Segmenter === 'function') {
     const segmenter = new Intl.Segmenter(lang, {granularity: 'word'})
-    return (text) =>
-      Array.from(segmenter.segment(text))
-        .map((segment) => segment.segment)
+    return (text) => {
+      return text
+        .split(SPACE_OR_PUNCTUATION_RE)
+        .flatMap(
+          (term) =>
+            term
+              .match(UNSPACED_RUN_RE)
+              ?.flatMap((run) =>
+                UNSPACED_START_RE.test(run)
+                  ? Array.from(segmenter.segment(run), (part) => part.segment)
+                  : run,
+              ) ?? term,
+        )
         .filter(isWordToken)
+    }
   }
-  return (text) => text.split(/\s+/).filter(isWordToken)
+  return (text) => text.split(SPACE_OR_PUNCTUATION_RE).filter(isWordToken)
 }
-
 export {createTokenizer}

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@vuepress-plume/plugin-search",
   "type": "module",
-  "version": "1.0.0-rc.214",
+  "version": "1.0.0-rc.215",
   "description": "The Plugin for VuePress 2 - local search",
   "author": "pengzhanbo <volodymyr@foxmail.com>",
   "license": "MIT",
@@ -44,12 +44,12 @@
     "@vueuse/core": "^15.0.0",
     "@vueuse/integrations": "^15.0.0",
     "chokidar": "5.0.0",
-    "focus-trap": "^8.2.3",
+    "focus-trap": "^8.2.4",
     "mark.js": "^8.11.1",
     "minisearch": "^7.2.0",
-    "p-map": "^7.0.8",
-    "sanitize-html": "^2.18.0",
-    "vue": "^3.5.43",
+    "p-map": "^7.1.1",
+    "sanitize-html": "^2.18.1",
+    "vue": "^3.5.44",
     "vue-router": "^5.4.0"
   },
   "publishConfig": {