Two questions:
1. Is there any way to force libprs500 to generate metadata from the filename first (or exclusively) even if the file has metadata within it?
2. This is more of a feature request as I doubt there currently is a way to do this:
Some books in my collection I have saved in the form of <author> - <series> - <series_index> - <title> and those without series are <author> - <title>. Perhaps you could add to the metadata from filename section something that allows for two filename schemes (tries to match the first and then the second)

1. No
2. Yeah that's a feature request. Open a ticket.

I actually had the same issue as you. I did the following:

Use a program called PDFInfo to remove the metadata (you may need to use a scripting tool like AutoIT to automate it) then I use the following regular expression:

(?P<authors>[^-]+)\s*-+\s*((?P<series>[^-\d]+)\s*(?P<series_index>\d)*\s*-+\s*)*(?P<title>.+)

Edit: (?P<authors>[^-]+)\s*-+\s*((?P<series>[^-]+)\s*-+\s*)*(?P<title>.+) works without the error of the series_index not being there all the time

It has support for the Authors name, followed by the - separator then an optional Series name & optional Series_Index, followed by an optional separator (tied to the Series option) and finally the Title.

I am trying to work out why sometimes it fails when there is no series_index, but I suspect that once you define a ?P<id> it MUST be used and not be make optional using a *. I may force it picking up a space (\s), unfortunately not all filenames contain a space next to the separator (-) and I much prefer using an optional space between any of the fields (\s*)

If this doesn't end up working, I may end up using Pyton/Perl/VBScript to split the file name up and use PDFInfo to populate the Metadata fields instead, unfortunately there is no MetaData field for Series or Series_Index (at least using PDFInfo) so I may need to find another tool that allows me to add fields and hope Calibre can map these fields to Series & Series_Index.

P.S. I realise you may not be using PDF's, but thats all that applicable for me, but the regular expression may still end up working with a bit of tweaking.