Convert PDF to HTML

One file, no dependencies, fixed layout

What comes out is a single .html file with nothing beside it. Every picture is inlined as a data URI, the stylesheet is in the head, and no font is fetched from anywhere — so the page opens identically on a machine with no internet connection, which is the only kind of HTML export worth having. The cost is arithmetic: encoding binary image data as text makes it about a third larger, so a 10 MB PDF of scans lands near 14 MB of HTML, all of which a browser must parse before it shows anything.

Layout is fixed, not responsive, and this is a choice rather than an omission. Each page is a box of the PDF's exact dimensions in points, and every line of text is an absolutely positioned span at its original coordinates, with its size and colour carried across. Nothing reflows, ever — on a phone you get the whole page scaled down, not a column of readable text. A print stylesheet is included, so printing the HTML breaks cleanly at the original page boundaries.

Because no fonts travel with the file, text is set in whatever Helvetica or Arial the reader's machine has. The size and position are right and the letterforms are not, so a line can run slightly wide of the artwork behind it. There is one neat trick worth knowing about: when a page's words are part of the drawn artwork rather than real text, the text layer is still written out, but transparently — so the page stays selectable and searchable without the words appearing twice.

Links inside the PDF arrive as plain text rather than anchors, headings arrive as positioned spans rather than heading elements, and there is no reading order to speak of. The output is an accurate picture of a document that happens to be made of HTML — it is not structured, semantic markup, and it should not be used where a screen reader or a search engine has to make sense of it.

When you want something else

pdf2htmlEX is the tool that does this properly. It extracts and embeds the PDF's own fonts as web fonts, so the text is not merely positioned correctly but set in the right typeface at the right width, and it reconstructs vector artwork as vectors instead of flattening the page to a picture. The output is dramatically more faithful and dramatically more complicated. Poppler's pdftohtml -s -c in.pdf out.html is the quick middle ground.

If what you actually want is a web page rather than a picture of a page — something that reflows on a phone, that Google can read, that a screen reader can navigate — no PDF-to-HTML converter will give it to you, because the structure it needs was thrown away when the PDF was made. Convert to Word in flowing mode and export from there, or go back to whatever the PDF was printed from.