feat: enhance FlexSearch encoding for CJK support (#553)

- Added support for CJK (Chinese, Japanese, Korean) languages in FlexSearch encoding.
- Introduced `isCJK` function to detect language and select appropriate encoding method.
- Implemented `encodeCJK` and `encodeDefault` functions for different tokenization strategies.

assets/js/flexsearch.js

@@ -195,8 +195,19 @@ document.addEventListener("DOMContentLoaded", function () {
    */
   async function preloadIndex() {
     const tokenize = '{{- site.Params.search.flexsearch.tokenize | default  "forward" -}}';
+
+    const isCJK = () => {
+      const lang = document.documentElement.lang || "en";
+      return lang.startsWith("zh") || lang.startsWith("ja") || lang.startsWith("ko");
+    }
+
+    const encodeCJK = (str) => str.replace(/[\x00-\x7F]/g, "").split("");
+    const encodeDefault = (str) => (""+str).toLocaleLowerCase().split(/[\p{Z}\p{S}\p{P}\p{C}]+/u);
+    const encodeFunction = isCJK() ? encodeCJK : encodeDefault;
+
     window.pageIndex = new FlexSearch.Document({
       tokenize,
+      encode: encodeFunction,
       cache: 100,
       document: {
         id: 'id',
@@ -207,6 +218,7 @@ document.addEventListener("DOMContentLoaded", function () {
 
     window.sectionIndex = new FlexSearch.Document({
       tokenize,
+      encode: encodeFunction,
       cache: 100,
       document: {
         id: 'id',