PDF to HTML

No. pdf.js extracts the text in a Web Worker on your device, and pandoc compiled to WebAssembly converts it to HTML in that same worker. No request goes out, no copy is kept, and the whole thing works offline after your first visit to the site.

No, you get body markup. Pandoc is called with only a source and a target format, with no standalone option, so there is no doctype, no head, no title and no stylesheet. Paste the result into a page of your own, or into a CMS field that expects HTML fragments.

No, and that is deliberate, because the file came from someone else. Two layers stop it. The extractor backslash-escapes every character Markdown gives meaning to, and the reader is then configured with raw_html and raw_tex both switched off. A page drawing the literal text of a script tag can only ever come out as inert escaped characters.

Because the smart extension is switched off on purpose. With it on, pandoc rewrites the document's own text: two hyphens become an en dash and straight quotes become curly ones. That silently changes what the page said, so this tool leaves the characters exactly as they were drawn.

It means pdf.js found no text objects to work from. Scanned documents are the usual cause, since they contain page images rather than characters. Run PDF OCR to add a text layer first, then convert the OCR output here. The conversion stops at that point rather than writing you an empty HTML file.

It is your PDF name with .html in place of .pdf, and yes: the HTML is shown on the page as a preview of exactly the same bytes that get written to the file. Nothing differs between what you read and what you download. Your original PDF is not modified.