Converting PDF to HTML is inherently lossy — PDFs describe visual layout, not semantic structure. This tool does its best: detect heading candidates by font size, group nearby text into paragraphs, preserve hyperlinks, and reconstruct lists where indentation is consistent. The result is usable HTML but almost always needs a manual review pass.
What gets detected
Headings from font size relative to body text. Paragraphs from line-break gaps. Hyperlinks from PDF annotation objects. Lists from consistent indent+bullet patterns. Bold/italic from font metadata.
What you'll need to clean up
Multi-column layouts often merge columns incorrectly — a manual reflow is usually needed. Complex tables come out as flat divs. Images are extracted as separate files (referenced from the HTML). Footnotes may end up interleaved with body text.
Output modes
Clean HTML: minimal semantic markup, no inline styles. Preserved layout: absolute positioning to match the PDF exactly (heavy CSS but visually identical). Pick clean for publishing, preserved for archival.