1 

     2 //:load matcher.scala

     3 

     4 // some regular expressions

     5 val SYM = RANGE("""ABCDEFGHIJKLMNOPQRSTUVXYZabcdefghijklmnopqrstuvwxyz.,!?-{[()]}':;%0123456789""")

     6 val WORD = PLUS(SYM)

     7 

     8 val BTAG = SEQS("<", WORD, ">") 

     9 val ETAG = SEQS("</", WORD, ">")

    10 

    11 val WHITESPACE = PLUS(RANGE(" \n"))

    12 

    13 // for classifying the strings that have been recognised

    14 abstract class Token

    15 case object T_WHITESPACE extends Token

    16 case class T_WORD(s: String) extends Token

    17 case class T_ETAG(s: String) extends Token

    18 case class T_BTAG(s: String) extends Token

    19 case class T_NT(s: String, rhs: List[Token]) extends Token

    20 

    21 val lexing_rules: List[Rule[Token]] = 

    22   List((BTAG, (s) => T_BTAG(s.mkString)),

    23        (ETAG, (s) => T_ETAG(s.mkString)),

    24        (WORD, (s) => T_WORD(s.mkString)),

    25        (WHITESPACE, (s) => T_WHITESPACE))

    26 

    27 // the tokenizer

    28 val T = Tokenizer(lexing_rules)

    29 

    30 // width for printing

    31 val WIDTH = 60

    32 

    33 

    34 def interpret(ts: List[Token], c: Int, ctr: List[String]) : Unit= ts match {

    35   case Nil => println(Console.RESET)

    36   case T_WHITESPACE::rest => print(Console.RESET + " "); interpret(rest, c + 1, ctr)

    37   case T_WORD(s)::rest => {

    38     val newstr = Console.RESET + ctr.reverse.mkString + s

    39     if (c + s.length < WIDTH) {

    40       print(newstr);

    41       interpret(rest, c + s.length, ctr)

    42     }

    43     else {

    44       print("\n" + newstr)

    45       interpret(rest, s.length, ctr)

    46     } 

    47   }

    48   case T_BTAG("<p>")::rest => print("\n"); interpret(rest, 0, ctr)

    49   case T_ETAG("</p>")::rest => print("\n"); interpret(rest, 0, ctr)

    50   case T_BTAG("<b>")::rest => interpret(rest, c, Console.BOLD :: ctr)

    51   case T_BTAG("<a>")::rest => interpret(rest, c, Console.UNDERLINED :: ctr)

    52   case T_BTAG("<cyan>")::rest => interpret(rest, c, Console.CYAN :: ctr)

    53   case T_BTAG("<red>")::rest => interpret(rest, c, Console.RED :: ctr)

    54   case T_BTAG("<blink>")::rest => interpret(rest, c, Console.BLINK :: ctr)

    55   case T_ETAG(_)::rest => interpret(rest, c, ctr.tail)

    56   case _::rest => interpret(rest, c, ctr)

    57 }

    58  

    59 interpret(T.fromFile("test.html"), 0, Nil)