Add new processing rule.

src/components/article/article_body/ArticleBody.astro

@@ -2,25 +2,32 @@
 // get rendered ArticleBodyContent HTML
 const html = await Astro.slots.render("default");
 
-// If some lines are not started with ASCII character,
-// join them to their previous line.
-// This is important.
-// Because I add one newline character to wrap Chinese.
+// rule:
+//   1. If a line belongs to the area where the line break should be kept.
+//   2. If a line ends with [a-zA-Z] add a space to the end of the line.
+//   3. If a line does not start in ASCII chars, join it to the previous line.
+// This is important in using one newline character to wrap Chinese.
 const arr = html.split("\n");
 let articleHTMLFinal = arr[0];
-// Whether the current line belongs to the area where
-// the line break should be kept.
 let remainIntactArea = false;
 for (let i = 1; i < arr.length; i++) {
+  // The first line of the HTML string is a table-of-content head recognized by
+  // remark-toc so it can be ignored.
+  if (arr[i].charAt(arr[i].length - 1).match(/[a-zA-Z]/) !== null) {
+    arr[i] += " ";
+  }
+  // Check if the current line belongs to some block area.
   if (arr[i].match(/^(<pre|<code|<blockquote|<table)/) !== null) {
     remainIntactArea = true;
   } else {
     remainIntactArea = false;
   }
   if (
-    // If the first character is not ascii character,
-    // or the final character of previous line is not character.
-    // AND
+    // (
+    //   If the first character is not ascii character,
+    //   OR
+    //   the final character of previous line is not ascii character.
+    // ) AND
     // Current line should not belong to area that remains intact.
     (arr[i].charAt(0).match(/[ -~]/) === null ||
       arr[i - 1].charAt(arr[i - 1].length - 1).match(/[ -~]/) === null) &&