Description

Javadoc inline tags {@literal} and {@code} with XML special characters (<, >) not handled correctly in metadata and xdoc generation .

There are two bugs:

1. {@literal} is not handled - JavadocMetadataScraperUtil.constructSubTreeText() and SiteUtil.getDescriptionFromJavadocForXdoc() have no logic for LITERAL_INLINE_TAG nodes. The child nodes (including syntax tokens {,@literal, }) are written to the output as plain text instead of being processed. This becomes a breaking issue when the content contains XML special characters like < or >, as they are not escaped.

2. {@code} does not escape < and > - adjustCodeInlineTagChildToHtml() only escapes & -> &, but not < -> < or > -> >. This means {@code <p>} produces <code><p></code> (broken XML) instead of the correct <code>&lt;p&gt;</code>.

How to Reproduce

Add {@literal <} or {@code <p>} to any module's class-level Javadoc and run mvn clean verify. The build will fail because the generated meta XML and/or xdoc XML files contain invalid XML.

For example, changing a module's Javadoc from:

Each paragraph but the first has <p> immediately

to the preferred inline tag form:

Each paragraph but the first has {@code <p>} immediately

causes the meta XML generator to produce <code><p></code> instead of <code>&lt;p&gt;</code>, resulting in malformed XML because the <p> is not escaped and gets interpreted as an actual HTML tag.

Actual Behavior

Currently, the generators produce broken output:

• {@literal <} -> {@literal <} (raw structural tokens leak into the output as plain text)
• {@literal >} -> {@literal >} (same issue)
• {@code <p>} -> <code><p></code> (the <p> is not escaped, producing invalid XML)
• {@code &} -> <code>&amp;</code> (only & escaping works correctly)

Expected Behavior

After the fix, the generators should produce correct XML output:

• {@literal <} -> < in both meta XML and xdoc output
• {@literal >} -> > in both meta XML and xdoc output
• {@code <p>} -> <code>&lt;p&gt;</code> in both meta XML and xdoc output