#!/usr/bin/env ruby19
# Ruby Quiz #167
# http://splatbang.com/rubyquiz/quiz.rhtml?id=167_Statistician_I
# http://groups.google.com/group/comp.lang.ruby/browse_frm/thread/bea26955698bce12?hl=en#

require 'stringio'


class Reporter
    def initialize(rules)
        @rules = read_rules(rules)
    end

    # Syntax:
    # * 
    # * [optional group]
    # * \\escape
    def read_rules(rules_file)
        File.readlines(rules_file).map do |l0|
            state = :text
            level = 0
            l = l0.chomp.split(/([<>\[\]\\])/).map do |t|
                next if t.empty?
                case state
                when :escape
                    state = :text
                    Regexp.escape(t)
                when :group_name
                    raise "Syntax error 01: #{t}: #{l0}" if /\W/ =~ t
                    state = :group_name_end
                    t
                when :group_name_end
                    case t
                    when '>'
                        state = :text
                        '>.*?)'
                    else
                        raise "Syntax error 02: #{t}: #{l0}"
                    end
                when :text
                    case t
                    when '['
                        level += 1
                        '(?:'
                    when ']'
                        level -= 1
                        ')?'
                    when '<'
                        state = :group_name
                        '(?<'
                    when '\\'
                        state = :escape
                        nil
                    else
                        Regexp.escape(t)
                    end
                end
            end
            raise "Syntax error 00 #{level}: #{l0}" unless level == 0
            l.unshift('^')
            Regexp.new(l.join)
        end
    end

    def data_reader(io_data)
        Fiber.new do
            until io_data.eof?
                l = io_data.gets.chomp
                d = {:line => l}
                @rules.each_with_index do |r, i|
                    m = r.match(l)
                    if m
                        d[:match]  = m
                        d[:ruleno] = i
                        break
                    end
                end
                Fiber.yield d
            end
        end
    end

    def report(io_data, io_out)
        unmatched = [nil, 'Unmatched input:']
        reader = data_reader(io_data)
        while (d = reader.resume)
            if d[:match]
                io_out.puts %{Rule #{d[:ruleno]}: #{d[:match].captures.compact.join(', ')}}
            else
                unmatched << d[:line]
                io_out.puts '# No Match'
            end
        end
        io_out.puts unmatched.join("\n") if unmatched.size > 2
    end
end


if __FILE__ == $0
    rules, data = ARGV
    case rules
    when nil, '-h', '--help'
        puts "#$0 RULES.TXT [DATA.TXT]"
        exit 1
    end

    if data
        io_data = File.open(data)
        io_out  = StringIO.open
    else
        io_data = STDIN
        io_out  = STDOUT 
    end
    begin
        # IMHO the encoding handling in ruby19 is weird.
        io_data.set_encoding('ascii-8bit') if io_data.external_encoding == Encoding::US_ASCII
        Reporter.new(rules).report(io_data, io_out)
    ensure
        if data
            puts io_out.string
            io_out.close
            io_data.close
        end
    end
end