feat(new tool): Unicode Search

Fix CorentinTh#735

package.json

@@ -41,8 +41,10 @@
     "@tiptap/pm": "2.1.6",
     "@tiptap/starter-kit": "2.1.6",
     "@tiptap/vue-3": "2.0.3",
-    "@types/markdown-it": "^13.0.7",
     "@types/figlet": "^1.5.8",
+    "@types/markdown-it": "^13.0.7",
+    "@types/utf8": "^3.0.3",
+    "@unicode/unicode-15.1.0": "^1.6.0",
     "@vicons/material": "^0.12.0",
     "@vicons/tabler": "^0.12.0",
     "@vueuse/core": "^10.3.0",
@@ -89,6 +91,7 @@
     "ulid": "^2.3.0",
     "unicode-emoji-json": "^0.4.0",
     "unplugin-auto-import": "^0.16.4",
+    "utf8": "^3.0.0",
     "uuid": "^9.0.0",
     "vue": "^3.3.4",
     "vue-i18n": "^9.9.1",

src/tools/index.ts

@@ -2,6 +2,7 @@ import { tool as base64FileConverter } from './base64-file-converter';
 import { tool as base64StringConverter } from './base64-string-converter';
 import { tool as basicAuthGenerator } from './basic-auth-generator';
 import { tool as emailNormalizer } from './email-normalizer';
+import { tool as unicodeSearch } from './unicode-search';
 
 import { tool as asciiTextDrawer } from './ascii-text-drawer';
 
@@ -135,6 +136,7 @@ export const toolsByCategory: ToolCategory[] = [
       httpStatusCodes,
       jsonDiff,
       safelinkDecoder,
+      unicodeSearch,
     ],
   },
   {

src/tools/unicode-search/index.ts

@@ -0,0 +1,12 @@
+import { FileText } from '@vicons/tabler';
+import { defineTool } from '../tool';
+
+export const tool = defineTool({
+  name: 'Unicode Search',
+  path: '/unicode-search',
+  description: 'Search in Unicode Characters',
+  keywords: ['unicode', 'search'],
+  component: () => import('./unicode-search.vue'),
+  icon: FileText,
+  createdAt: new Date('2024-08-15'),
+});

src/tools/unicode-search/unicode-search.vue

@@ -0,0 +1,122 @@
+<script setup lang="ts">
+import unicodeNames from '@unicode/unicode-15.1.0/Names/index.js';
+import unicodeCategories from '@unicode/unicode-15.1.0/General_Category';
+import utf8 from 'utf8';
+
+import { useFuzzySearch } from '@/composable/fuzzySearch';
+import useDebouncedRef from '@/composable/debouncedref';
+
+function toPaddedHex(num: number) {
+  return num.toString(16).padStart(4, '0').toUpperCase();
+}
+
+function toUTF8(codePoint: number) {
+  const utf8String = utf8.encode(String.fromCodePoint(codePoint));
+  return [...utf8String].map(c => `\\x${c.codePointAt(0)?.toString(16).toUpperCase()}`);
+}
+
+const searchQuery = useDebouncedRef('', 500);
+const parsedSearchQuery = computed(() => {
+  const parsedRegex = /^\s*(?:\&#x(?<hex1>[\da-f]+);|\&#(?<dec>\d+);|(?:U\+|\\u)?\s*(?<hex2>[\da-f]+))\s*$/gi;
+  const parsedQuery = parsedRegex.exec(searchQuery.value);
+  if (parsedQuery) {
+    if (parsedQuery.groups?.hex1 || parsedQuery.groups?.hex2) {
+      return `=${toPaddedHex(Number.parseInt(parsedQuery.groups.hex1 || parsedQuery.groups.hex1, 16))}`;
+    }
+    if (parsedQuery.groups?.dec) {
+      return `=${toPaddedHex(Number.parseInt(parsedQuery.groups.dec, 10))}`;
+    }
+  }
+  return searchQuery.value;
+});
+
+const unicodeSearchData = [...unicodeNames].map(([codePoint, characterName]) => {
+  const hex = toPaddedHex(codePoint);
+  return {
+    codePoint,
+    characterName: `${characterName} (U+${hex})`,
+    hex,
+  };
+});
+
+const { searchResult } = useFuzzySearch({
+  search: parsedSearchQuery,
+  data: unicodeSearchData,
+  options: {
+    keys: ['characterName', 'hex'],
+    threshold: 0.2,
+    isCaseSensitive: false,
+    minMatchCharLength: 3,
+    useExtendedSearch: true,
+  },
+});
+</script>
+
+<template>
+  <div mx-auto max-w-2400px important:flex-1>
+    <div flex items-center gap-3>
+      <c-input-text
+        v-model:value="searchQuery"
+        placeholder="Search Unicode by name (e.g. 'zero width') or code point..."
+        mx-auto max-w-600px
+      >
+        <template #prefix>
+          <icon-mdi-search mr-6px color-black op-70 dark:color-white />
+        </template>
+      </c-input-text>
+    </div>
+
+    <div v-if="searchQuery.trim().length > 0">
+      <div
+        v-if="searchResult.length === 0"
+        mt-4
+        text-20px
+        font-bold
+      >
+        No results
+      </div>
+
+      <div v-else>
+        <div mt-4 text-20px font-bold>
+          Search result
+        </div>
+
+        <n-table>
+          <thead>
+            <th>UCOD</th>
+            <th>Display/UTF8</th>
+            <th style="width:30%">
+              Category
+            </th>
+            <th>Html</th>
+            <th style="width:30%">
+              Name
+            </th>
+          </thead>
+          <tbody>
+            <tr v-for="(result, ix) in searchResult" :key="ix">
+              <td>
+                <input-copyable :value="`U+${toPaddedHex(result.codePoint)}`" mb-1 />
+                <!-- //NOSONAR --><n-a :href="`https://unicodeplus.com/U+${toPaddedHex(result.codePoint)}`" target="_blank">
+                  &gt; More info
+                </n-a>
+              </td>
+              <td>
+                <input-copyable :value="String.fromCodePoint(result.codePoint)" mb-1 />
+                <input-copyable :value="toUTF8(result.codePoint)" />
+              </td>
+              <td>
+                <input-copyable :value="unicodeCategories.get(result.codePoint)" />
+              </td>
+              <td>
+                <input-copyable :value="`\&\#x${toPaddedHex(result.codePoint)};`" mb-1 />
+                <input-copyable :value="`\&\#${result.codePoint};`" />
+              </td>
+              <td><input-copyable :value="result.characterName" /></td>
+            </tr>
+          </tbody>
+        </n-table>
+      </div>
+    </div>
+  </div>
+</template>

src/tools/unicode-search/unicode.d.ts

@@ -0,0 +1,9 @@
+declare module '@unicode/unicode-15.1.0/Names/index.js'{
+    const unicode: HashSet<number, string>;
+    export default unicode;
+}
+
+declare module '@unicode/unicode-15.1.0/General_Category'{
+    const unicode: HashSet<number, string>;
+    export default unicode;
+}