Merge pull request #30 from smori1983/matcher-for-japanese-characters

Matcher for Japanese characters
smori1983 · Sep 18, 2023 · 4d583b8 · 4d583b8
2 parents bb2dde0 + 2c79566
commit 4d583b8
Show file tree

Hide file tree

Showing 4 changed files with 137 additions and 3 deletions.
diff --git a/src/components/SearchBoxFlexSearch.vue b/src/components/SearchBoxFlexSearch.vue
@@ -51,6 +51,7 @@
 import { Document } from 'flexsearch'
 import data from '@dynamic/vuepress-plugin-flexsearch/data'
 import excerpt from './excerpt'
+import matcher from './matcher';
 import ngram from '../tokenizer/ngram'
 
 /* global FLEX_SEARCH_HOTKEYS */
@@ -159,6 +160,7 @@ export default {
     setUpFlexSearchDocument () {
       for (const locale in data) {
         const doc = new Document({
+          matcher: matcher,
           tokenize: 'forward',
           id: 'key',
           index: [

diff --git a/src/components/excerpt.js b/src/components/excerpt.js
@@ -1,11 +1,13 @@
 const ExcerptHtml = require('./excerpt-html');
 const TextProcessing = require('./text-processing');
 
+const matcher = require('./matcher');
+
 /**
  * @typedef {Object} ExcerptOption
  * @property {number} [aroundLength]
- * @property {string} [head]
- * @property {string} [tail]
+ * @property {string} [headText]
+ * @property {string} [tailText]
  */
 
 /**
@@ -22,7 +24,7 @@ const create = (content, query, option) => {
   } = option || {};
 
   const contentLowerCase = content.toLowerCase();
-  const queryLowerCase = query.toLowerCase();
+  const queryLowerCase = addMatcher(query).toLowerCase();
 
   const textProcessing = new TextProcessing();
 
@@ -46,6 +48,22 @@ const create = (content, query, option) => {
   return content.slice(0, aroundLength) + (content.length > aroundLength ? tailText : '');
 };
 
+/**
+ * @param {string} query
+ * @return {string}
+ */
+const addMatcher = (query) => {
+  let converted1 = query;
+  let converted2 = query;
+
+  Object.keys(matcher).forEach((key) => {
+    converted1 = converted1.replaceAll(key, matcher[key]);
+    converted2 = converted2.replaceAll(matcher[key], key);
+  })
+
+  return [query, converted1, converted2].join(' ');
+};
+
 /**
  * @param {string} excerpt
  * @param {string[]} queries

diff --git a/src/components/matcher.js b/src/components/matcher.js
@@ -0,0 +1,95 @@
+module.exports = {
+  '1': '１',
+  '2': '２',
+  '3': '３',
+  '4': '４',
+  '5': '５',
+  '6': '６',
+  '7': '７',
+  '8': '８',
+  '9': '９',
+  '0': '０',
+
+  'あ': 'ア',
+  'い': 'イ',
+  'う': 'ウ',
+  'え': 'エ',
+  'お': 'オ',
+  'か': 'カ',
+  'き': 'キ',
+  'く': 'ク',
+  'け': 'ケ',
+  'こ': 'コ',
+  'さ': 'サ',
+  'し': 'シ',
+  'す': 'ス',
+  'せ': 'セ',
+  'そ': 'ソ',
+  'た': 'タ',
+  'ち': 'チ',
+  'つ': 'ツ',
+  'て': 'テ',
+  'と': 'ト',
+  'な': 'ナ',
+  'に': 'ニ',
+  'ぬ': 'ヌ',
+  'ね': 'ネ',
+  'の': 'ノ',
+  'は': 'ハ',
+  'ひ': 'ヒ',
+  'ふ': 'フ',
+  'へ': 'ヘ',
+  'ほ': 'ホ',
+  'ま': 'マ',
+  'み': 'ミ',
+  'む': 'ム',
+  'め': 'メ',
+  'も': 'モ',
+  'や': 'ヤ',
+  'ゆ': 'ユ',
+  'よ': 'ヨ',
+  'ら': 'ラ',
+  'り': 'リ',
+  'る': 'ル',
+  'れ': 'レ',
+  'ろ': 'ロ',
+  'わ': 'ワ',
+  'を': 'ヲ',
+  'ん': 'ン',
+
+  'が': 'ガ',
+  'ぎ': 'ギ',
+  'ぐ': 'グ',
+  'げ': 'ゲ',
+  'ご': 'ゴ',
+  'ざ': 'ザ',
+  'じ': 'ジ',
+  'ず': 'ズ',
+  'ぜ': 'ゼ',
+  'ぞ': 'ゾ',
+  'だ': 'ダ',
+  'ぢ': 'ヂ',
+  'づ': 'ヅ',
+  'で': 'デ',
+  'ど': 'ド',
+  'ば': 'バ',
+  'び': 'ビ',
+  'ぶ': 'ブ',
+  'べ': 'ベ',
+  'ぼ': 'ボ',
+
+  'ぱ': 'パ',
+  'ぴ': 'ピ',
+  'ぷ': 'プ',
+  'ぺ': 'ペ',
+  'ぽ': 'ポ',
+
+  'ぁ': 'ァ',
+  'ぃ': 'ィ',
+  'ぅ': 'ゥ',
+  'ぇ': 'ェ',
+  'ぉ': 'ォ',
+  'ゃ': 'ャ',
+  'ゅ': 'ュ',
+  'ょ': 'ョ',
+};
diff --git a/test/components/excerpt.js b/test/components/excerpt.js
@@ -84,6 +84,25 @@ describe('components', () => {
         'weigh ...',
       ],
 
+      // matcher
+      [
+        '2023',
+        {},
+        '２０２３年',
+        '<strong>２０２３</strong>年',
+      ],
+      [
+        '２０２３',
+        {},
+        '2023年',
+        '<strong>2023</strong>年',
+      ],
+      [
+        'いぬ ねこ',
+        {},
+        'イヌとネコ',
+        '<strong>イヌ</strong>と<strong>ネコ</strong>',
+      ],
     ];
 
     dataSet.forEach(([query, option, content, highlighted]) => {