A parser is simple a class that implements

  #initialize(file_name, body, options)

and

  #scan

The initialize method takes a file name to be used, the body of the file, and an RDoc::Options object. The scan method is then called to return an appropriately parsed TopLevel code object.

The ParseFactory is used to redirect to the correct parser given a filename extension. This magic works because individual parsers have to register themselves with us as they are loaded in. The do this using the following incantation

   require "rdoc/parsers/parsefactory"

   module RDoc

     class XyzParser
       extend ParseFactory                  <<<<
       parse_files_matching /\.xyz$/        <<<<

       def initialize(file_name, body, options)
         ...
       end

       def scan
         ...
       end
     end
   end

Just to make life interesting, if we suspect a plain text file, we also look for a shebang line just in case it‘s a potential shell script