CDATA Sections in XML and JSON Mapping

CDATA (Character Data) sections allow you to include text in XML that would otherwise need escaping. They are enclosed in <![CDATA[...]]> markers and are commonly used to embed HTML, code snippets, or any text with special XML characters.

XML with CDATA:

<article>
  <title>Introduction to HTML</title>
  <content><![CDATA[
    <p>Use <strong>bold</strong> for emphasis.</p>
    <p>Entities like &amp; are written literally here.</p>
  ]]></content>
  <code><![CDATA[
    if (x < 10 && y > 5) {
      return true;
    }
  ]]></code>
</article>

Converted to JSON:

{
  "article": {
    "title": "Introduction to HTML",
    "content": "\n    <p>Use <strong>bold</strong> for emphasis.</p>\n    <p>Entities like &amp; are written literally here.</p>\n  ",
    "code": "\n    if (x < 10 && y > 5) {\n      return true;\n    }\n  "
  }
}

Key points about CDATA:

1. CDATA preserves literal text. Inside a CDATA section, <, >, & are not treated as XML markup. This is why you can embed raw HTML or code without escaping every special character.
2. In JSON, CDATA becomes a plain string. The CDATA wrapper is stripped, and the content becomes a JSON string value. Newlines become \n, quotes become \", and the result is a standard escaped JSON string.
3. The only forbidden sequence inside CDATA is ]]>, which terminates the CDATA section. If your content contains this exact sequence, you must split the CDATA section.

When converting JSON to XML:

If a JSON string contains characters that are special in XML (<, >, &), the converter has two choices:

Strategy	Output for "x < 10"
Entity escaping	<value>x &lt; 10</value>
CDATA wrapping	<value><![CDATA[x < 10]]></value>

CDATA is preferred when:

• The string contains significant amounts of HTML or code
• Readability of the XML matters (entity-escaped HTML is nearly unreadable)
• The content has many special characters that would require extensive escaping

Entity escaping is preferred when:

• The string is short and has few special characters
• You want the XML to be more compact
• The consuming parser does not need to handle CDATA