intprog_hackery/pdf-parser.rb

require 'pdf-reader'
require_relative 'Method_desc'

def parse(exam_text)
  reader = PDF::Reader.new(exam_text)

  text = reader.pages[0].text.split "\n"

  content = text.slice(3, 15).map {|x| x.split "  "}.map {|x| x.filter {|y| y.length > 1 and y.strip != "*"} }.filter {|x| x.length > 0}


  prim_class = content[0][0]
  if content[0].length == 1
    container = content[1][0].strip
  else
    container = content[0][1].strip
  end


  # Fuck mig
  container_params = content[2].length > 1 ? [content[2][0].strip] : [content[3][0].strip]


  container_methods = content.slice(2, content.length-1).filter { |x|
    if x[0].include? "("
      x[0]
    end
  }.map {|x| x[0].strip}

  class_fields = content.slice(3, content.length-1).filter { |x|
    (x.length > 1) and (not x[1].include? "(")
  }.map {|x| x[1].strip}

  if content[4].length == 1
    tmp = content[4][0].strip
    class_fields = [tmp] + class_fields
  end

  #print(content.slice(0, content.length-1))

  regex = /\A(\S+) (\w+)\((.*)\)\Z/

  container_methods.map! do |x|
    matches = x.match(regex)
    Method_desc.new(*matches.captures)
  end


  return prim_class, container, class_fields, container_params, container_methods
end

puts parse("ferry.pdf")