Docx IO list whitespace can drop list item text

Problem

pnpm check failed in packages/docx-io/src/lib/html-to-docx/html-to-docx.spec.ts for nested lists and indented lists.

The generated word/document.xml kept the list numbering paragraphs but dropped visible item text like Item 1 and Indented item.

Root Cause

buildList treated formatting whitespace around <li> nodes as real content.

When a <ul> or <ol> started with indentation text nodes, the reducer created a blank paragraph first, then appended the real <li> node into that paragraph instead of treating it as a list item. That left Word XML with numbering and whitespace runs, but no actual list text.

Solution

Ignore whitespace-only text nodes while walking list children, and do not append <li> nodes into an already-open paragraph accumulator.

Relevant fix:

const isBlankTextNode =
  isVText(childVNode) &&
  (childVNode as VTextType).text.trim() === '';

if (isBlankTextNode) {
  return accumulator;
}

// ...
} else if (
  accumulator.length > 0 &&
  isVNode(accumulator.at(-1)!.node) &&
  ((accumulator.at(-1)!.node as VNodeType).tagName || '').toLowerCase() ===
    'p' &&
  (childNode.tagName || '').toLowerCase() !== 'li'
) {

Prevention

• Treat indentation whitespace in parsed HTML as layout noise, not list content.
• For list walkers, keep <li> as a structural boundary. Do not merge it into a text paragraph.
• When a DOCX list test loses text but keeps numbering, inspect the intermediate list reducer before blaming ZIP or XML generation.