ruby/reader.rb

   1 require_relative "types"
   2
   3 class Reader
   4     def initialize(tokens)
   5         @position = 0
   6         @tokens = tokens
   7     end
   8     def peek
   9         return @tokens[@position]
  10     end
  11     def next
  12         @position += 1
  13         return @tokens[@position-1]
  14     end
  15 end
  16
  17
  18 def tokenize(str)
  19     re = /[\s,]*(~@|[\[\]{}()'`~^@]|"(?:\\.|[^\\"])*"?|;.*|[^\s\[\]{}('"`,;)]*)/
  20     return str.scan(re).map{|m| m[0]}.select{ |t|
  21         t != "" && t[0..0] != ";"
  22     }
  23 end
  24
  25 def parse_str(t) # trim and unescape
  26     return t[1..-2].gsub(/\\./, {"\\\\" => "\\", "\\n" => "\n", "\\\"" => '"'})
  27 end
  28
  29 def read_atom(rdr)
  30     token = rdr.next
  31     return case token
  32         when /^-?[0-9]+$/ then       token.to_i # integer
  33         when /^-?[0-9][0-9.]*$/ then token.to_f # float
  34         when /^".*"$/ then           parse_str(token) # string
  35         when /^".*$/ then            raise "expected '\"', got EOF"
  36         when /^:/ then               "\u029e" + token[1..-1] # keyword
  37         when "nil" then              nil
  38         when "true" then             true
  39         when "false" then            false
  40         else                         token.to_sym # symbol
  41     end
  42 end
  43
  44 def read_list(rdr, klass, start="(", last =")")
  45     ast = klass.new
  46     token = rdr.next()
  47     if token != start
  48         raise "expected '" + start + "'"
  49     end
  50     while (token = rdr.peek) != last
  51         if not token
  52             raise "expected '" + last + "', got EOF"
  53         end
  54         ast.push(read_form(rdr))
  55     end
  56     rdr.next
  57     return ast
  58 end
  59
  60 def read_form(rdr)
  61     return case rdr.peek
  62         when ";" then  nil
  63         when "'" then  rdr.next; List.new [:quote, read_form(rdr)]
  64         when "`" then  rdr.next; List.new [:quasiquote, read_form(rdr)]
  65         when "~" then  rdr.next; List.new [:unquote, read_form(rdr)]
  66         when "~@" then rdr.next; List.new [:"splice-unquote", read_form(rdr)]
  67         when "^" then  rdr.next; meta = read_form(rdr);
  68                        List.new [:"with-meta", read_form(rdr), meta]
  69         when "@" then  rdr.next; List.new [:deref, read_form(rdr)]
  70
  71         when "(" then  read_list(rdr, List, "(", ")")
  72         when ")" then  raise "unexpected ')'"
  73         when "[" then  read_list(rdr, Vector, "[", "]")
  74         when "]" then  raise "unexpected ']'"
  75         when "{" then  Hash[read_list(rdr, List, "{", "}").each_slice(2).to_a]
  76         when "}" then  raise "unexpected '}'"
  77         else           read_atom(rdr)
  78     end
  79 end
  80
  81 def read_str(str)
  82     tokens = tokenize(str)
  83     return nil if tokens.size == 0
  84     return read_form(Reader.new(tokens))
  85 end
  86