Not a URL but arguably more useful:

    $ curl http://www.nickbostrom.com/extraterrestrial.pdf | pdftotext - -

Don't forget the two dashes, as it doesn't default to reading stdin, and doesn't default to writing to stdout when instructed to read from stdin.

Also, pdftotext is part of poppler in Arch and poppler-utils in Debian.

Embarrassingly enough, I'm a Python dev on Win7 - now you know why I'm so paranoid. :)

Thanks anyway, though.

The instructions work fine with cygwin on windows btw, pdftotext is in the poppler package.

Then, PDFMiner comes to mind. Pure python, so slower but no C/native dependencies.

Here we go: http://viewtext.org/article?url=http%3A%2F%2Fwww.nickbostrom...

I'm really paranoid about loading unfamiliar websites up (I've got a version of Chrome with js and plugins disabled that I use for random links), and pdfs are still a bit of a concern as an attack vector. It looks like this viewtext.org will come in handy.

</tangent>

EDIT: Reading comprehension problem -- you wanted clean HTML. I thought you meant you wanted an HTML5 viewer.

  - - -

This is nice:

https://docs.google.com/viewer

Example using this PDF:

https://docs.google.com/viewer?url=http://www.nickbostrom.co...

Clean, light, and doesn't require a GA login.

Usually the vacuum's functionality of scribd is used, but it seems to not work at this moment http://www.scribd.com/vacuum?url=http://www.nickbostrom.com/...

I paste the complete url into Google search, then hit "Quick View" on the first result.