Turning notes and web pages into a PDF you can file

Everything that becomes a PDF has to be laid out by something

Markdown has structure and no pages. HTML has structure and a design that depends on a stylesheet. A spreadsheet has a grid and no idea where a page ends. A PDF has pages and nothing else, so every conversion into one is a typesetting job: decide where lines break, how much bigger a heading is, how far a list indents, and when the next line will not fit.

The interesting question is which input the layout is done from. There are two answers and they suit different jobs.

Your browser is better at appearance

If what you want is a page that looks the way it looks on screen, use Print, then Save as PDF. Nothing else comes close: the browser has the stylesheet, the webfonts, the images and a complete layout engine, and it is already rendering exactly what you are asking to keep.

Converters that claim to match this are usually shipping a headless browser on a server, which means sending it the page.

A converter is better at everything else

Browser printing needs the page open, one at a time, at whatever width the window happened to be. That fails the moment you have a folder of files, or a file rather than a page, or a need to do the same thing next week without remembering how.

That is the job Markdown to PDF and HTML to PDF do. They take the structure and ignore the appearance: headings stay headings, lists stay lists, paragraphs wrap to a measured column and break across pages when they run out of room.

Two consequences worth being clear about, because they are limits and not oversights.

No CSS is applied. A converter that fetched the stylesheets and images a page references would be making requests to third parties on your behalf, which is the thing a local tool exists not to do.

And no script runs, so a page whose words are drawn into the document by JavaScript has nothing in the file to convert. Saving the page from your browser after it has loaded puts the text where a converter can see it.

Raw HTML written inside a Markdown file is printed as the text you wrote rather than acted on. A Markdown file is not a place anyone means to write markup that does something, and printing it is the reading that cannot surprise anybody.

Tables and pictures have their own routes

A grid needs a different layout from prose, so CSV to PDF and Excel to PDF go through a table renderer rather than the prose one, and through the same table renderer as each other, so a CSV and the spreadsheet made from it come out the same shape.

The decision that renderer has to make is what happens when a table is wider than the page: shrink the type until nobody can read it, truncate every cell, or drop the columns that will not fit and say how many. The last is the only one where you can see what happened.

Images to PDF is for the scans, receipts and photographed documents that arrive as a pile of JPEGs and need to be one file before anybody will accept them.

Then fit more onto less paper

Pages per sheet arranges two, four, six, nine or sixteen pages onto each sheet. It is a printing tool and a reviewing one: four slides to a page is a handout, two pages side by side is how a long document reads on a wide screen, and either cuts a print job substantially.

It rearranges rather than removing. The text stays selectable, because the pages are placed as pages rather than rasterised into pictures of themselves.

If the result then needs to fit an upload limit, compressing is a separate step and belongs after this one.

The fonts are the honest caveat

The prose converters set type in the standard PDF fonts, which every reader has and which cover Latin text and not much else. Characters they cannot draw are replaced, and the result tells you how many were replaced rather than letting them disappear between the file you had and the file you got.

If your document is in a script those fonts do not cover, that count will be large and you will know immediately. That is the intended behaviour: a conversion that quietly dropped a third of the characters would be worse in every way than one that says so.

None of it is uploaded

Every tool named here runs in your browser, which is why none of them can fetch a URL for you and why none of them needs to. The notes, drafts and exports people convert are frequently the ones they would least like to hand to a server.