Fine grained events

  def parse(input, handler):
   for line in input:
     ...
     if line.startswith("AC  "):
       words = line.replace(";", " ").split()
       for word in words:
         handler.element("accession", word)
   ...
   return handler.get_record()

Even this isn't quite good enough.
  • modifies data
  • don't know position information
  • still have to write all that recognition code by hand
 
.