1 

     2 //:load matcher.scala

     3 

     4 // some regular expressions

     5 val SYM = RANGE("""ABCDEFGHIJKLMNOPQRSTUVXYZabcdefghijklmnopqrstuvwxyz_""")

     6 val DIGIT = RANGE("0123456789")

     7 val ID = SEQ(PLUS(SYM)) ~ STAR(ALT(SYM, DIGIT)) 

     8 

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

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

    11 

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

    13 

    14 // for classifying the strings that have been recognised

    15 abstract class Token

    16 case object T_WHITESPACE extends Token

    17 case class T_WORD(s: String) extends Token

    18 case class T_ETAG(s: String) extends Token

    19 case class T_BTAG(s: String) extends Token

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

    21 

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

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

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

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

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

    27 

    28 // the tokenizer

    29 val T = Tokenizer(lexing_rules)

    30 

    31 // width for printing

    32 val WIDTH = 60

    33 

    34 

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

    36   case Nil => println(Console.RESET)

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

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

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

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

    41       print(newstr);

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

    43     }

    44     else {

    45       print("\n" + newstr)

    46       interpret(rest, s.length, ctr)

    47     } 

    48   }

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

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

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

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

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

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

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

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

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

    58 }

    59  

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