2 // (includes a lexer and a parser)

     2 // (includes an external lexer and parser)

     3 

     3 

     4 import scala.language.implicitConversions    

     4 import java.io._

     5 import scala.language.reflectiveCalls 

     5 

     6 

     6 object Compiler {

     7 abstract class Rexp 

     7 

     8 case object ZERO extends Rexp

     8 // Abstract syntax trees for the Fun language

     9 case object ONE extends Rexp

     9 abstract class Exp extends Serializable 

    10 case class CHAR(c: Char) extends Rexp

    10 abstract class BExp extends Serializable 

    11 case class ALT(r1: Rexp, r2: Rexp) extends Rexp 

    11 abstract class Decl extends Serializable 

    12 case class SEQ(r1: Rexp, r2: Rexp) extends Rexp 

    13 case class STAR(r: Rexp) extends Rexp 

    14 case class RECD(x: String, r: Rexp) extends Rexp

    15   

    16 abstract class Val

    17 case object Empty extends Val

    18 case class Chr(c: Char) extends Val

    19 case class Sequ(v1: Val, v2: Val) extends Val

    20 case class Left(v: Val) extends Val

    21 case class Right(v: Val) extends Val

    22 case class Stars(vs: List[Val]) extends Val

    23 case class Rec(x: String, v: Val) extends Val

    24    

    25 // some convenience for typing in regular expressions

    26 def charlist2rexp(s : List[Char]): Rexp = s match {

    27   case Nil => ONE

    28   case c::Nil => CHAR(c)

    29   case c::s => SEQ(CHAR(c), charlist2rexp(s))

    30 }

    31 implicit def string2rexp(s : String) : Rexp = 

    32   charlist2rexp(s.toList)

    33 

    34 implicit def RexpOps(r: Rexp) = new {

    35   def | (s: Rexp) = ALT(r, s)

    36   def % = STAR(r)

    37   def ~ (s: Rexp) = SEQ(r, s)

    38 }

    39 

    40 implicit def stringOps(s: String) = new {

    41   def | (r: Rexp) = ALT(s, r)

    42   def | (r: String) = ALT(s, r)

    43   def % = STAR(s)

    44   def ~ (r: Rexp) = SEQ(s, r)

    45   def ~ (r: String) = SEQ(s, r)

    46   def $ (r: Rexp) = RECD(s, r)

    47 }

    48 

    49 def nullable (r: Rexp) : Boolean = r match {

    50   case ZERO => false

    51   case ONE => true

    52   case CHAR(_) => false

    53   case ALT(r1, r2) => nullable(r1) || nullable(r2)

    54   case SEQ(r1, r2) => nullable(r1) && nullable(r2)

    55   case STAR(_) => true

    56   case RECD(_, r1) => nullable(r1)

    57 }

    58 

    59 def der (c: Char, r: Rexp) : Rexp = r match {

    60   case ZERO => ZERO

    61   case ONE => ZERO

    62   case CHAR(d) => if (c == d) ONE else ZERO

    63   case ALT(r1, r2) => ALT(der(c, r1), der(c, r2))

    64   case SEQ(r1, r2) => 

    65     if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2))

    66     else SEQ(der(c, r1), r2)

    67   case STAR(r) => SEQ(der(c, r), STAR(r))

    68   case RECD(_, r1) => der(c, r1)

    69 }

    70 

    71 

    72 // extracts a string from value

    73 def flatten(v: Val) : String = v match {

    74   case Empty => ""

    75   case Chr(c) => c.toString

    76   case Left(v) => flatten(v)

    77   case Right(v) => flatten(v)

    78   case Sequ(v1, v2) => flatten(v1) + flatten(v2)

    79   case Stars(vs) => vs.map(flatten).mkString

    80   case Rec(_, v) => flatten(v)

    81 }

    82 

    83 // extracts an environment from a value;

    84 // used for tokenise a string

    85 def env(v: Val) : List[(String, String)] = v match {

    86   case Empty => Nil

    87   case Chr(c) => Nil

    88   case Left(v) => env(v)

    89   case Right(v) => env(v)

    90   case Sequ(v1, v2) => env(v1) ::: env(v2)

    91   case Stars(vs) => vs.flatMap(env)

    92   case Rec(x, v) => (x, flatten(v))::env(v)

    93 }

    94 

    95 // The Injection Part of the lexer

    96 

    97 def mkeps(r: Rexp) : Val = r match {

    98   case ONE => Empty

    99   case ALT(r1, r2) => 

   100     if (nullable(r1)) Left(mkeps(r1)) else Right(mkeps(r2))

   101   case SEQ(r1, r2) => Sequ(mkeps(r1), mkeps(r2))

   102   case STAR(r) => Stars(Nil)

   103   case RECD(x, r) => Rec(x, mkeps(r))

   104 }

   105 

   106 def inj(r: Rexp, c: Char, v: Val) : Val = (r, v) match {

   107   case (STAR(r), Sequ(v1, Stars(vs))) => Stars(inj(r, c, v1)::vs)

   108   case (SEQ(r1, r2), Sequ(v1, v2)) => Sequ(inj(r1, c, v1), v2)

   109   case (SEQ(r1, r2), Left(Sequ(v1, v2))) => Sequ(inj(r1, c, v1), v2)

   110   case (SEQ(r1, r2), Right(v2)) => Sequ(mkeps(r1), inj(r2, c, v2))

   111   case (ALT(r1, r2), Left(v1)) => Left(inj(r1, c, v1))

   112   case (ALT(r1, r2), Right(v2)) => Right(inj(r2, c, v2))

   113   case (CHAR(d), Empty) => Chr(c) 

   114   case (RECD(x, r1), _) => Rec(x, inj(r1, c, v))

   115   case _ => { println ("Injection error") ; sys.exit(-1) } 

   116 }

   117 

   118 // some "rectification" functions for simplification

   119 def F_ID(v: Val): Val = v

   120 def F_RIGHT(f: Val => Val) = (v:Val) => Right(f(v))

   121 def F_LEFT(f: Val => Val) = (v:Val) => Left(f(v))

   122 def F_ALT(f1: Val => Val, f2: Val => Val) = (v:Val) => v match {

   123   case Right(v) => Right(f2(v))

   124   case Left(v) => Left(f1(v))

   125 }

   126 def F_SEQ(f1: Val => Val, f2: Val => Val) = (v:Val) => v match {

   127   case Sequ(v1, v2) => Sequ(f1(v1), f2(v2))

   128 }

   129 def F_SEQ_Empty1(f1: Val => Val, f2: Val => Val) = 

   130   (v:Val) => Sequ(f1(Empty), f2(v))

   131 def F_SEQ_Empty2(f1: Val => Val, f2: Val => Val) = 

   132   (v:Val) => Sequ(f1(v), f2(Empty))

   133 def F_RECD(f: Val => Val) = (v:Val) => v match {

   134   case Rec(x, v) => Rec(x, f(v))

   135 }

   136 def F_ERROR(v: Val): Val = throw new Exception("error")

   137 

   138 def simp(r: Rexp): (Rexp, Val => Val) = r match {

   139   case ALT(r1, r2) => {

   140     val (r1s, f1s) = simp(r1)

   141     val (r2s, f2s) = simp(r2)

   142     (r1s, r2s) match {

   143       case (ZERO, _) => (r2s, F_RIGHT(f2s))

   144       case (_, ZERO) => (r1s, F_LEFT(f1s))

   145       case _ => if (r1s == r2s) (r1s, F_LEFT(f1s))

   146                 else (ALT (r1s, r2s), F_ALT(f1s, f2s)) 

   147     }

   148   }

   149   case SEQ(r1, r2) => {

   150     val (r1s, f1s) = simp(r1)

   151     val (r2s, f2s) = simp(r2)

   152     (r1s, r2s) match {

   153       case (ZERO, _) => (ZERO, F_ERROR)

   154       case (_, ZERO) => (ZERO, F_ERROR)

   155       case (ONE, _) => (r2s, F_SEQ_Empty1(f1s, f2s))

   156       case (_, ONE) => (r1s, F_SEQ_Empty2(f1s, f2s))

   157       case _ => (SEQ(r1s,r2s), F_SEQ(f1s, f2s))

   158     }

   159   }

   160   case RECD(x, r1) => {

   161     val (r1s, f1s) = simp(r1)

   162     (RECD(x, r1s), F_RECD(f1s))

   163   }

   164   case r => (r, F_ID)

   165 }

   166 

   167 // lexing functions including simplification

   168 def lex_simp(r: Rexp, s: List[Char]) : Val = s match {

   169   case Nil => if (nullable(r)) mkeps(r) else { println ("Lexing Error") ; sys.exit(-1) } 

   170   case c::cs => {

   171     val (r_simp, f_simp) = simp(der(c, r))

   172     inj(r, c, f_simp(lex_simp(r_simp, cs)))

   173   }

   174 }

   175 

   176 def lexing_simp(r: Rexp, s: String) = env(lex_simp(r, s.toList))

   177 

   178 

   179 // The Lexing Rules for the Fun Language

   180 

   181 def PLUS(r: Rexp) = r ~ r.%

   182 

   183 val SYM = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" | "j" | "k" | 

   184           "l" | "m" | "n" | "o" | "p" | "q" | "r" | "s" | "t" | "u" | "v" | 

   185           "w" | "x" | "y" | "z" | "T" | "_"

   186 val DIGIT = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9"

   187 val ID = SYM ~ (SYM | DIGIT).% 

   188 val NUM = PLUS(DIGIT)

   189 val KEYWORD : Rexp = "if" | "then" | "else" | "write" | "def"

   190 val SEMI: Rexp = ";"

   191 val OP: Rexp = "=" | "==" | "-" | "+" | "*" | "!=" | "<" | ">" | "<=" | ">=" | "%" | "/"

   192 val WHITESPACE = PLUS(" " | "\n" | "\t")

   193 val RPAREN: Rexp = ")"

   194 val LPAREN: Rexp = "("

   195 val COMMA: Rexp = ","

   196 val ALL = SYM | DIGIT | OP | " " | ":" | ";" | "\"" | "=" | "," | "(" | ")"

   197 val ALL2 = ALL | "\n"

   198 val COMMENT = ("/*" ~ ALL2.% ~ "*/") | ("//" ~ ALL.% ~ "\n")

   199 

   200 

   201 val WHILE_REGS = (("k" $ KEYWORD) | 

   202                   ("i" $ ID) | 

   203                   ("o" $ OP) | 

   204                   ("n" $ NUM) | 

   205                   ("s" $ SEMI) | 

   206                   ("c" $ COMMA) |

   207                   ("pl" $ LPAREN) |

   208                   ("pr" $ RPAREN) |

   209                   ("w" $ (WHITESPACE | COMMENT))).%

   210 

   211 

   212 

   213 // The tokens for the Fun language

   214 

   215 abstract class Token

   216 case object T_SEMI extends Token

   217 case object T_COMMA extends Token

   218 case object T_LPAREN extends Token

   219 case object T_RPAREN extends Token

   220 case class T_ID(s: String) extends Token

   221 case class T_OP(s: String) extends Token

   222 case class T_NUM(n: Int) extends Token

   223 case class T_KWD(s: String) extends Token

   224 

   225 val token : PartialFunction[(String, String), Token] = {

   226   case ("k", s) => T_KWD(s)

   227   case ("i", s) => T_ID(s)

   228   case ("o", s) => T_OP(s)

   229   case ("n", s) => T_NUM(s.toInt)

   230   case ("s", _) => T_SEMI

   231   case ("c", _) => T_COMMA

   232   case ("pl", _) => T_LPAREN

   233   case ("pr", _) => T_RPAREN

   234 }

   235 

   236 

   237 def tokenise(s: String) : List[Token] = 

   238   lexing_simp(WHILE_REGS, s).collect(token)

   239 

   240 

   241 

   242 // Parser combinators

   243 abstract class Parser[I, T](implicit ev: I => Seq[_]) {

   244   def parse(ts: I): Set[(T, I)]

   245 

   246   def parse_all(ts: I) : Set[T] =

   247     for ((head, tail) <- parse(ts); if (tail.isEmpty)) yield head

   248 

   249   def parse_single(ts: I) : T = parse_all(ts).toList match {

   250     case List(t) => t

   251     case _ => { println ("Parse Error\n") ; sys.exit(-1) }

   252   }

   253 }

   254 

   255 case class ~[+A, +B](_1: A, _2: B)

   256 

   257 class SeqParser[I, T, S](p: => Parser[I, T], 

   258                          q: => Parser[I, S])(implicit ev: I => Seq[_]) extends Parser[I, ~[T, S]] {

   259   def parse(sb: I) = 

   260     for ((head1, tail1) <- p.parse(sb); 

   261          (head2, tail2) <- q.parse(tail1)) yield (new ~(head1, head2), tail2)

   262 }

   263 

   264 class AltParser[I, T](p: => Parser[I, T], 

   265                       q: => Parser[I, T])(implicit ev: I => Seq[_]) extends Parser[I, T] {

   266   def parse(sb: I) = p.parse(sb) ++ q.parse(sb)   

   267 }

   268 

   269 class FunParser[I, T, S](p: => Parser[I, T], 

   270                          f: T => S)(implicit ev: I => Seq[_]) extends Parser[I, S] {

   271   def parse(sb: I) = 

   272     for ((head, tail) <- p.parse(sb)) yield (f(head), tail)

   273 }

   274 

   275 implicit def ParserOps[I, T](p: Parser[I, T])(implicit ev: I => Seq[_]) = new {

   276   def || (q : => Parser[I, T]) = new AltParser[I, T](p, q)

   277   def ==>[S] (f: => T => S) = new FunParser[I, T, S](p, f)

   278   def ~[S] (q : => Parser[I, S]) = new SeqParser[I, T, S](p, q)

   279 }

   280 

   281 def ListParser[I, T, S](p: => Parser[I, T], 

   282                         q: => Parser[I, S])(implicit ev: I => Seq[_]): Parser[I, List[T]] = {

   283   (p ~ q ~ ListParser(p, q)) ==> { case x ~ _ ~ z => x :: z : List[T] } ||

   284   (p ==> ((s) => List(s)))

   285 }

   286 

   287 case class TokParser(tok: Token) extends Parser[List[Token], Token] {

   288   def parse(ts: List[Token]) = ts match {

   289     case t::ts if (t == tok) => Set((t, ts)) 

   290     case _ => Set ()

   291   }

   292 }

   293 

   294 implicit def token2tparser(t: Token) = TokParser(t)

   295 

   296 implicit def TokOps(t: Token) = new {

   297   def || (q : => Parser[List[Token], Token]) = new AltParser[List[Token], Token](t, q)

   298   def ==>[S] (f: => Token => S) = new FunParser[List[Token], Token, S](t, f)

   299   def ~[S](q : => Parser[List[Token], S]) = new SeqParser[List[Token], Token, S](t, q)

   300 }

   301 

   302 case object NumParser extends Parser[List[Token], Int] {

   303   def parse(ts: List[Token]) = ts match {

   304     case T_NUM(n)::ts => Set((n, ts)) 

   305     case _ => Set ()

   306   }

   307 }

   308 

   309 case object IdParser extends Parser[List[Token], String] {

   310   def parse(ts: List[Token]) = ts match {

   311     case T_ID(s)::ts => Set((s, ts)) 

   312     case _ => Set ()

   313   }

   314 }

   315 

   316 

   317 

   318 // Abstract syntax trees for Fun

   319 abstract class Exp

   320 abstract class BExp 

   321 abstract class Decl

   519 def compile(class_name: String, input: String) : String = {

   168 def deserialise[T](fname: String) : T = {

   520   val tks = tokenise(input)

   169   val in = new ObjectInputStream(new FileInputStream(fname))

   521   val ast = Prog.parse_single(tks)

   170   val data = in.readObject.asInstanceOf[T]

   171   in.close

   172   data

   173 }

   174 

   175 

   176 def compile(class_name: String) : String = {

   177   val ast = deserialise[List[Decl]](class_name ++ ".prs") 

   526 def compile_file(class_name: String) = {

   182 def compile_to_file(class_name: String) = {

   527   val input = io.Source.fromFile(s"${class_name}.fun").mkString

   183   val output = compile(class_name)

   528   val output = compile(class_name, input)

   535   compile_file(class_name)

   190   compile_to_file(class_name)

   544 compile_run("fact")

   199 //compile_run("fact")

   200 

   201 def main(args: Array[String]) = 

   202    compile_run(args(0))

   203 

   204 

   205 }