pdftohtml currently dumps document outlines in html mode but not in xml mode. The following patch adds generation of outlines in -xml mode. The generated .xml will have something like the following at the end if outlines were present (indentation added for readability):

<...>
</page>
<outline>
  <item page="5">Contents</item>
  <item page="9">Figures</item>
  <item page="13">Tables</item>
  <item page="23">Preface</item>
  <item page="25">1 Introduction</item>
  <outline>
	<item page="25">1.1 About This Book</item>
	<item page="28">1.2 Introduction to PDF 1.7 Features</item>
	<outline>
	  <item page="28">1.2.1 Presentation of 3D Artwork</item>
	  <item page="29">1.2.2 Interactive Features</item>
	  <outline>
		<item page="29">Interactive Features That Aid Technical Communication</item>
		<item page="29">Interactive Feature for Use in a Legal Setting</item>
	  </outline>
	  <item page="29">1.2.3 Accessibility Related Features</item>
	  <item page="30">1.2.4 Document Navigation Feature</item>
	  <item page="30">1.2.5 Security-Related Features</item>
	  <item page="31">1.2.6 General Features</item>
	  <item page="31">1.2.7 PDF Reference Changes</item>
<...>