feat: add spaces.

huggingface-cn · Mar 11, 2024 · 5527bcd · 5527bcd
1 parent 4af5501
commit 5527bcd
Showing 1 changed file with 29 additions and 7 deletions.
diff --git a/markdownUrls.html b/markdownUrls.html
@@ -2,26 +2,42 @@
 <html lang="en">
 <head>
     <meta charset="UTF-8">
-    <title>Markdown Link Extractor with WeChat Link Exclusion</title>
+    <title>Markdown Formatter and Link Extractor</title>
     <style>
         #inputMarkdown, #outputMarkdown {
             width: 90%;
-            height: 150px;
+            height: 200px;
             margin: 10px;
         }
     </style>
 </head>
 <body>
 
-<h2>Markdown Link Extractor with WeChat Link Exclusion</h2>
+<h2>Markdown Formatter and Link Extractor</h2>
 <textarea id="inputMarkdown" placeholder="输入Markdown内容..."></textarea><br>
-<button onclick="extractAndAppendLinks()">提取并追加链接</button><br>
+<button onclick="formatAndExtractLinks()">格式化并提取链接</button><br>
 <textarea id="outputMarkdown" readonly placeholder="处理后的结果..."></textarea>
 
 <script>
-    function extractAndAppendLinks() {
+    function formatAndExtractLinks() {
         let inputText = document.getElementById('inputMarkdown').value;
         let output = document.getElementById('outputMarkdown');
+
+        // 保留代码块内容
+        let codeBlocks = [];
+        let codeBlockRegex = /```[\s\S]*?```/g;
+        inputText = inputText.replace(codeBlockRegex, match => {
+            codeBlocks.push(match);
+            return '\0'; // 使用空字符作为占位符
+        });
+
+        // 自动在数字和中文、中英文之间加入空格
+        inputText = inputText.replace(/([\u4e00-\u9fa5])([A-Za-z0-9])/g, '$1 $2');
+        inputText = inputText.replace(/([A-Za-z0-9])([\u4e00-\u9fa5])/g, '$1 $2');
+
+        // 转换全角括号并添加空格
+        inputText = inputText.replace(/（/g, '( ');
+        inputText = inputText.replace(/）/g, ' )');
 
         // 分段处理，以换行符分割文本
         let paragraphs = inputText.split('\n\n');
@@ -51,8 +67,14 @@ <h2>Markdown Link Extractor with WeChat Link Exclusion</h2>
             return processedText;
         });
 
-        // 将处理后的段落重新拼接成一个字符串，并显示在输出框中
-        output.value = processedParagraphs.join('\n\n');
+        // 将处理后的段落重新拼接成一个字符串
+        let formattedText = processedParagraphs.join('\n\n');
+
+        // 恢复代码块
+        formattedText = formattedText.replace(/\0/g, () => codeBlocks.shift());
+
+        // 显示在输出框中
+        output.value = formattedText;
     }
 </script>