need wisdom

I have been working with mnogosearch for two months.

I want to be able to search a repository of files with it.  I don't want to index any websites.

I have made reasonable progress, but I have found the requirements for indexing files are

quite different from those for indexing websites, and I would like some general guidance about 

how to alter the configuration to be more effective.

 

Sorry if the questions seem naïve.

 

One of the most important things when searching for a file is the kind of file it is.

Usually one knows this.  I would like to get the indexer to store information about the file

extension so it can be part of the query.  

 

It seems I should be able to harness existing mime types somehow.

Presently, I am using a battery of filters to convert everything to plain text with configurations

like the following:

 

Mime application/pdf          text/plain                 "/usr/local/filters/bin/pdftotext  -enc UTF-8 -q $1 -"

 

Of course, by the time mnogo sees it, it really is text, and I told it it was text, so I don't see how

the engine can help me look for .pdf files at that point.

 

So I am thinking I should be able to somehow add an attribute to the item that is the file extension

or else the real mime type, as just about every different type of file has a unique one.

 

How can I add this information such that someday I can build a search page where I can specify that 

I want to limit my search to .pdf files?

 

I think I can't just use the same filter and change the mime type above to application/pdf.  

 

What is the best approach to this?

 

Thanks

 

Stefan