diff -r e85600529ca5 -r 4794759139ea progs/html1.scala
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/progs/html1.scala	Sat Jun 15 09:23:18 2013 -0400
@@ -0,0 +1,75 @@
+
+
+//:load matcher.scala
+
+
+// some regular expressions
+val LETTER = RANGE("""ABCDEFGHIJKLMNOPQRSTUVXYZabcdefghijklmnopqrstuvwxyz.,!?-{[()]}':;%""")
+val DIGIT = RANGE("0123456789")
+val NONZERODIGIT = RANGE("123456789")
+
+val NAME = ALT(PLUS(LETTER), SEQS(PLUS(LETTER),"=", PLUS(LETTER)))
+val BTAG = SEQS("<", NAME, ">") 
+val ETAG = SEQS("</", PLUS(LETTER), ">")
+
+val WORD = PLUS(ALT(LETTER, DIGIT))
+val WHITESPACE = PLUS(RANGE(" \n"))
+
+// for classifying the strings that have been recognised
+abstract class Token
+case object T_WHITESPACE extends Token
+case class T_WORD(s: String) extends Token
+case class T_ETAG(s: String) extends Token
+case class T_BTAG(s: String) extends Token
+case class T_NT(s: String, rhs: List[Token]) extends Token
+
+def tokenizer(rs: List[Rule[Token]], s: String) : List[Token] = 
+  tokenize(rs, s.toList)
+
+
+
+// lexing rules for arithmetic expressions
+val lexing_rules: List[Rule[Token]]= 
+  List((BTAG, (s) => T_BTAG(s.mkString)),
+       (ETAG, (s) => T_ETAG(s.mkString)),
+       (WORD, (s) => T_WORD(s.mkString)),
+       (WHITESPACE, (s) => T_WHITESPACE))
+
+val ts = tokenize_file(lexing_rules, "test.html")
+
+
+val WIDTH = 60
+
+def is_tag(t: Token) = t match {
+  case T_BTAG(_) => true
+  case T_ETAG(_) => true
+  case _ => false
+}
+
+def interpret(ts: List[Token], c: Int, ctr: List[String]) : Unit= ts match {
+  case Nil => println(Console.RESET)
+  case T_WHITESPACE::rest => print(" "); interpret(rest, c + 1, ctr)
+  case T_WORD(s)::rest => {
+    val newc = c + s.length
+    val newstr = Console.RESET + ctr.reverse.mkString + s
+    if (newc < WIDTH) {
+      print(newstr);
+      interpret(rest, newc, ctr)
+    }
+    else {
+      print("\n" + newstr)
+      interpret(rest, s.length, ctr)
+    } 
+  }
+  case T_BTAG("<p>")::rest => print("\n"); interpret(rest, 0, ctr)
+  case T_ETAG("</p>")::rest => print("\n"); interpret(rest, 0, ctr)
+  case T_BTAG("<b>")::rest => interpret(rest, c, Console.BOLD :: ctr)
+  case T_BTAG("<A>")::rest => interpret(rest, c, Console.UNDERLINED :: ctr)
+  case T_BTAG("<cyan>")::rest => interpret(rest, c, Console.CYAN :: ctr)
+  case T_BTAG("<red>")::rest => interpret(rest, c, Console.RED :: ctr)
+  case T_BTAG("<blink>")::rest => interpret(rest, c, Console.BLINK :: ctr)
+  case T_ETAG(_)::rest => interpret(rest, c, ctr.tail)
+  case _::rest => interpret(rest, c, ctr)
+}
+ 
+interpret(ts, 0, Nil)