1 import scala.language.implicitConversions    

     2 import scala.language.reflectiveCalls 

     3 import scala.util._

     4 import scala.annotation.tailrec

     5 import scala.sys.process._

     6 

     7 def fromFile(name: String) : String = 

     8   io.Source.fromFile(name).mkString

     9 

    10 abstract class Rexp 

    11 case object NULL extends Rexp

    12 case object EMPTY extends Rexp

    13 case class CHAR(c: Char) extends Rexp

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

    15 case class RANGE(cs: List[Char]) extends Rexp 

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

    17 case class PLUS(r: Rexp) extends Rexp 

    18 case class STAR(r: Rexp) extends Rexp 

    19 case class NTIMES(r: Rexp, n: Int) extends Rexp 

    20 case class NUPTOM(r: Rexp, n: Int, m: Int) extends Rexp

    21 

    22 object RANGE {

    23   def apply(s: String) : RANGE = RANGE(s.toList)

    24 }

    25 def NMTIMES(r: Rexp, n: Int, m: Int) = {

    26   if (m < n) throw new IllegalArgumentException("the number m cannot be smaller than n.")

    27   else NUPTOM(r, n, m - n)

    28 }

    29 

    30 case class NOT(r: Rexp) extends Rexp 

    31 case class OPT(r: Rexp) extends Rexp 

    32 

    33 // some convenience for typing in regular expressions

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

    35   case Nil => EMPTY

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

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

    38 }

    39 implicit def string2rexp(s : String) : Rexp = charlist2rexp(s.toList)

    40 

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

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

    43   def % = STAR(r)

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

    45 }

    46 

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

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

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

    50   def % = STAR(s)

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

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

    53 }

    54 

    55 

    56 // nullable function: tests whether the regular 

    57 // expression can recognise the empty string

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

    59   case NULL => false

    60   case EMPTY => true

    61   case CHAR(_) => false

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

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

    64   case STAR(_) => true

    65   case PLUS(r) => nullable(r)

    66   case NTIMES(r, i) => if (i == 0) true else nullable(r)

    67   case NUPTOM(r, i, j) => if (i == 0) true else nullable(r)

    68   case RANGE(_) => false

    69   case NOT(r) => !(nullable(r))

    70   case OPT(_) => true

    71 }

    72 

    73 // derivative of a regular expression w.r.t. a character

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

    75   case NULL => NULL

    76   case EMPTY => NULL

    77   case CHAR(d) => if (c == d) EMPTY else NULL

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

    79   case SEQ(r1, r2) => 

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

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

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

    83   case PLUS(r) => SEQ(der(c, r), STAR(r))

    84   case NTIMES(r, i) => 

    85     if (i == 0) NULL else der(c, SEQ(r, NTIMES(r, i - 1)))

    86   case NUPTOM(r, i, j) =>

    87     if (i == 0 && j == 0) NULL else 

    88     if (i == 0) ALT(der(c, NTIMES(r, j)), der(c, NUPTOM(r, 0, j - 1)))

    89     else der(c, SEQ(r, NUPTOM(r, i - 1, j)))

    90   case RANGE(cs) => if (cs contains c) EMPTY else NULL

    91   case NOT(r) => NOT(der (c, r))

    92   case OPT(r) => der(c, r)

    93 }

    94 

    95 def zeroable (r: Rexp) : Boolean = r match {

    96   case NULL => true

    97   case EMPTY => false

    98   case CHAR(_) => false

    99   case ALT(r1, r2) => zeroable(r1) && zeroable(r2)

   100   case SEQ(r1, r2) => zeroable(r1) || zeroable(r2)

   101   case STAR(_) => false

   102   case PLUS(r) => zeroable(r)

   103   case NTIMES(r, i) => if (i == 0) false else zeroable(r)

   104   case NUPTOM(r, i, j) => if (i == 0) false else zeroable(r)

   105   case RANGE(_) => false

   106   case NOT(r) => !(zeroable(r))     // bug: incorrect definition for NOT

   107   case OPT(_) => false

   108 }

   109 

   110 // derivative w.r.t. a string (iterates der)

   111 def ders (s: List[Char], r: Rexp) : Rexp = s match {

   112   case Nil => r

   113   case c::s => ders(s, der(c, r))

   114 }

   115 

   116 

   117 // regular expressions for the While language

   118 val SYM = RANGE("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz_")

   119 val DIGIT = RANGE("0123456789")

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

   121 val NUM = PLUS(DIGIT)

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

   123 val SEMI: Rexp = ";"

   124 val COMMA: Rexp = ","

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

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

   127 val RPAREN: Rexp = ")"

   128 val LPAREN: Rexp = "("

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

   130 val ALL2 = ALL | "\n"

   131 //val COMMENT2 = ("/*" ~ NOT(ALL.% ~ "*/" ~ ALL.%) ~ "*/")

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

   133 

   134 

   135 // token for While language

   136 abstract class Token

   137 case object T_WHITESPACE extends Token

   138 case object T_SEMI extends Token

   139 case object T_COMMA extends Token

   140 case object T_LPAREN extends Token

   141 case object T_RPAREN extends Token

   142 case object T_COMMENT extends Token

   143 case class T_ID(s: String) extends Token

   144 case class T_OP(s: String) extends Token

   145 case class T_NUM(s: String) extends Token

   146 case class T_KWD(s: String) extends Token

   147 case class T_ERR(s: String) extends Token // special error token

   148 

   149 

   150 type TokenFun = String => Token

   151 type LexRules = List[(Rexp, TokenFun)]

   152 val While_lexing_rules: LexRules = 

   153   List((KEYWORD, (s) => T_KWD(s)),

   154        (ID, (s) => T_ID(s)),

   155        (COMMENT, (s) => T_COMMENT),

   156        (OP, (s) => T_OP(s)),

   157        (NUM, (s) => T_NUM(s)),

   158        (SEMI, (s) => T_SEMI),

   159        (COMMA, (s) => T_COMMA),

   160        (LPAREN, (s) => T_LPAREN),

   161        (RPAREN, (s) => T_RPAREN),

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

   163 

   164 

   165 // calculates derivatives until all of them are zeroable

   166 @tailrec

   167 def munch(s: List[Char], 

   168           pos: Int, 

   169           rs: LexRules, 

   170           last: Option[(Int, TokenFun)]): Option[(Int, TokenFun)] = {

   171   rs match {

   172   case Nil => last

   173   case rs if (s.length <= pos) => last

   174   case rs => {

   175     val ders = rs.map({case (r, tf) => (der(s(pos), r), tf)})

   176     val rs_nzero = ders.filterNot({case (r, _) => zeroable(r)})

   177     val rs_nulls = ders.filter({case (r, _) => nullable(r)})

   178     val new_last = if (rs_nulls != Nil) Some((pos, rs_nulls.head._2)) else last

   179     munch(s, 1 + pos, rs_nzero, new_last)

   180   }

   181 }}

   182 

   183 // iterates the munching function and returns a Token list

   184 def tokenize(s: String, rs: LexRules) : List[Token] = munch(s.toList, 0, rs, None) match {

   185   case None if (s == "") => Nil

   186   case None => List(T_ERR(s"Lexing error: $s"))

   187   case Some((n, tf)) => {

   188     val (head, tail) = s.splitAt(n + 1)

   189     tf(head)::tokenize(tail, rs)

   190   }

   191 }

   192 

   193 def tokenizer(s:String) : List[Token] = 

   194   tokenize(s, While_lexing_rules).filter {

   195     case T_ERR(s) => { println(s); sys.exit(-1) }

   196     case T_WHITESPACE => false

   197     case T_COMMENT => false

   198     case _ => true

   199   } 

   200 

   201 

   202 

   203 // Parser - Abstract syntax trees

   204 abstract class Exp

   205 abstract class BExp 

   206 abstract class Decl

   207 

   208 case class Def(name: String, args: List[String], body: Exp) extends Decl

   209 case class Main(e: Exp) extends Decl

   210 

   211 case class Call(name: String, args: List[Exp]) extends Exp

   212 case class If(a: BExp, e1: Exp, e2: Exp) extends Exp

   213 case class Write(e: Exp) extends Exp

   214 case class Var(s: String) extends Exp

   215 case class Num(i: Int) extends Exp

   216 case class Aop(o: String, a1: Exp, a2: Exp) extends Exp

   217 case class Sequ(e1: Exp, e2: Exp) extends Exp

   218 

   219 case class Bop(o: String, a1: Exp, a2: Exp) extends BExp

   220 

   221 // calculating the maximal needed stack size

   222 def max_stack_exp(e: Exp): Int = e match {

   223   case Call(_, args) => args.map(max_stack_exp).sum

   224   case If(a, e1, e2) => max_stack_bexp(a) + (List(max_stack_exp(e1), max_stack_exp(e2)).max)

   225   case Write(e) => max_stack_exp(e) + 1

   226   case Var(_) => 1

   227   case Num(_) => 1

   228   case Aop(_, a1, a2) => max_stack_exp(a1) + max_stack_exp(a2)

   229   case Sequ(e1, e2) => List(max_stack_exp(e1), max_stack_exp(e2)).max

   230 }

   231 def max_stack_bexp(e: BExp): Int = e match {

   232   case Bop(_, a1, a2) => max_stack_exp(a1) + max_stack_exp(a2)

   233 }

   234 

   235 

   236 

   237 // Parser combinators

   238 abstract class Parser[I <% Seq[_], T] {

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

   240 

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

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

   243 

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

   245     case List(t) => t

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

   247   }

   248 }

   249 

   250 class SeqParser[I <% Seq[_], T, S](p: => Parser[I, T], q: => Parser[I, S]) extends Parser[I, (T, S)] {

   251   def parse(sb: I) = 

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

   253          (head2, tail2) <- q.parse(tail1)) yield ((head1, head2), tail2)

   254 }

   255 

   256 class AltParser[I <% Seq[_], T](p: => Parser[I, T], q: => Parser[I, T]) extends Parser[I, T] {

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

   258 }

   259 

   260 class FunParser[I <% Seq[_], T, S](p: => Parser[I, T], f: T => S) extends Parser[I, S] {

   261   def parse(sb: I) = 

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

   263 }

   264 

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

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

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

   268     case _ => Set ()

   269   }

   270 }

   271 

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

   273 

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

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

   276     case T_NUM(s)::ts => Set((s.toInt, ts)) 

   277     case _ => Set ()

   278   }

   279 }

   280 

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

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

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

   284     case _ => Set ()

   285   }

   286 }

   287 

   288 

   289 implicit def ParserOps[I<% Seq[_], T](p: Parser[I, T]) = new {

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

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

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

   293 }

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

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

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

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

   298 }

   299 

   300 def ListParser[I <% Seq[_], T, S](p: => Parser[I, T], q: => Parser[I, S]): Parser[I, List[T]] = {

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

   302   (p ==> ((s) => List(s)))

   303 }

   304 

   305 

   306 // expressions

   307 lazy val Exp: Parser[List[Token], Exp] = 

   308   (T_KWD("if") ~ BExp ~ T_KWD("then") ~ Exp ~ T_KWD("else") ~ Exp) ==>

   309     { case (((((x, y), z), u), v), w) => If(y, u, w): Exp } ||

   310   (M ~ T_SEMI ~ Exp) ==> { case ((x, y), z) => Sequ(x, z): Exp } || M

   311 lazy val M: Parser[List[Token], Exp] =

   312   (T_KWD("write") ~ L) ==> { case (x, y) => Write(y): Exp } || L

   313 lazy val L: Parser[List[Token], Exp] = 

   314   (T ~ T_OP("+") ~ Exp) ==> { case ((x, y), z) => Aop("+", x, z): Exp } ||

   315   (T ~ T_OP("-") ~ Exp) ==> { case ((x, y), z) => Aop("-", x, z): Exp } || T  

   316 lazy val T: Parser[List[Token], Exp] = 

   317   (F ~ T_OP("*") ~ T) ==> { case ((x, y), z) => Aop("*", x, z): Exp } || 

   318   (F ~ T_OP("/") ~ T) ==> { case ((x, y), z) => Aop("/", x, z): Exp } || 

   319   (F ~ T_OP("%") ~ T) ==> { case ((x, y), z) => Aop("%", x, z): Exp } || F

   320 lazy val F: Parser[List[Token], Exp] = 

   321   (IdParser ~ T_LPAREN ~ ListParser(Exp, T_COMMA) ~ T_RPAREN) ==> 

   322     { case (((x, y), z), w) => Call(x, z): Exp } ||

   323   (T_LPAREN ~ Exp ~ T_RPAREN) ==> { case ((x, y), z) => y: Exp } || 

   324   IdParser ==> { case x => Var(x): Exp } || 

   325   NumParser ==> { case x => Num(x): Exp }

   326 

   327 // boolean expressions

   328 lazy val BExp: Parser[List[Token], BExp] = 

   329   (Exp ~ T_OP("==") ~ Exp) ==> { case ((x, y), z) => Bop("==", x, z): BExp } || 

   330   (Exp ~ T_OP("!=") ~ Exp) ==> { case ((x, y), z) => Bop("!=", x, z): BExp } || 

   331   (Exp ~ T_OP("<") ~ Exp) ==> { case ((x, y), z) => Bop("<", x, z): BExp } || 

   332   (Exp ~ T_OP(">") ~ Exp) ==> { case ((x, y), z) => Bop("<", z, x): BExp } || 

   333   (Exp ~ T_OP("<=") ~ Exp) ==> { case ((x, y), z) => Bop("<=", x, z): BExp } || 

   334   (Exp ~ T_OP("=>") ~ Exp) ==> { case ((x, y), z) => Bop("<=", z, x): BExp }  

   335 

   336 lazy val Defn: Parser[List[Token], Decl] =

   337    (T_KWD("def") ~ IdParser ~ T_LPAREN ~ ListParser(IdParser, T_COMMA) ~ T_RPAREN ~ T_OP("=") ~ Exp) ==>

   338      { case ((((((x, y), z), w), u), v), r) => Def(y, w, r): Decl }

   339 

   340 lazy val Prog: Parser[List[Token], List[Decl]] =

   341   (Defn ~ T_SEMI ~ Prog) ==> { case ((x, y), z) => x :: z : List[Decl] } ||

   342   (Exp ==> ((s) => List(Main(s)) : List[Decl]))

   343 

   344 // compiler - built-in functions 

   345 // copied from http://www.ceng.metu.edu.tr/courses/ceng444/link/jvm-cpm.html

   346 //

   347 

   348 val library = """

   349 .class public XXX.XXX

   350 .super java/lang/Object

   351 

   352 .method public <init>()V

   353         aload_0

   354         invokenonvirtual java/lang/Object/<init>()V

   355         return

   356 .end method

   357 

   358 .method public static write(I)V 

   359         .limit locals 5 

   360         .limit stack 5 

   361         iload 0 

   362         getstatic java/lang/System/out Ljava/io/PrintStream; 

   363         swap 

   364         invokevirtual java/io/PrintStream/println(I)V 

   365         return 

   366 .end method

   367 

   368 """

   369 

   370 // for generating new labels

   371 var counter = -1

   372 

   373 def Fresh(x: String) = {

   374   counter += 1

   375   x ++ "_" ++ counter.toString()

   376 }

   377 

   378 type Mem = Map[String, Int]

   379 type Instrs = List[String]

   380 

   381 def compile_expT(a: Exp, env : Mem, name: String) : Instrs = a match {

   382   case Num(i) => List("ldc " + i.toString + "\n")

   383   case Var(s) => List("iload " + env(s).toString + "\n")

   384   case Aop("+", a1, a2) => compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("iadd\n")

   385   case Aop("-", a1, a2) => compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("isub\n")

   386   case Aop("*", a1, a2) => compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("imul\n")

   387   case Aop("/", a1, a2) => compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("idiv\n")

   388   case Aop("%", a1, a2) => compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("irem\n")

   389   case If(b, a1, a2) => {

   390     val if_else = Fresh("If_else")

   391     val if_end = Fresh("If_end")

   392     compile_bexpT(b, env, if_else) ++

   393     compile_expT(a1, env, name) ++

   394     List("goto " + if_end + "\n") ++

   395     List("\n" + if_else + ":\n\n") ++

   396     compile_expT(a2, env, name) ++

   397     List("\n" + if_end + ":\n\n")

   398   }

   399   case Call(n, args) => if (name == n) { 

   400     val stores = args.zipWithIndex.map { case (x, y) => "istore " + y.toString + "\n" } 

   401     args.flatMap(a => compile_expT(a, env, "")) ++

   402     stores.reverse ++ 

   403     List ("goto " + n + "_Start\n") 

   404   } else {

   405     val is = "I" * args.length

   406     args.flatMap(a => compile_expT(a, env, "")) ++

   407     List ("invokestatic XXX/XXX/" + n + "(" + is + ")I\n")

   408   }

   409   case Sequ(a1, a2) => {

   410     compile_expT(a1, env, "") ++ List("pop\n") ++ compile_expT(a2, env, name)

   411   }

   412   case Write(a1) => {

   413     compile_expT(a1, env, "") ++

   414     List("dup\n",

   415          "invokestatic XXX/XXX/write(I)V\n")

   416   }

   417 }

   418 

   419 def compile_bexpT(b: BExp, env : Mem, jmp: String) : Instrs = b match {

   420   case Bop("==", a1, a2) => 

   421     compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("if_icmpne " + jmp + "\n")

   422   case Bop("!=", a1, a2) => 

   423     compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("if_icmpeq " + jmp + "\n")

   424   case Bop("<", a1, a2) => 

   425     compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("if_icmpge " + jmp + "\n")

   426   case Bop("<=", a1, a2) => 

   427     compile_expT(a1, env, "") ++ compile_expT(a2, env, "") ++ List("if_icmpgt " + jmp + "\n")

   428 }

   429 

   430 

   431 def compile_decl(d: Decl) : Instrs = d match {

   432   case Def(name, args, a) => { 

   433     val env = args.zipWithIndex.toMap

   434     val is = "I" * args.length

   435     List(".method public static " + name + "(" + is + ")I \n",

   436          ".limit locals " + args.length.toString + "\n",

   437          ".limit stack " + (1 + max_stack_exp(a)).toString + "\n",

   438          name + "_Start:\n") ++   

   439     compile_expT(a, env, name) ++

   440     List("ireturn\n",

   441          ".end method \n\n")

   442   }

   443   case Main(a) => {

   444     List(".method public static main([Ljava/lang/String;)V\n",

   445          ".limit locals 200\n",

   446          ".limit stack 200\n") ++

   447     compile_expT(a, Map(), "") ++

   448     List("invokestatic XXX/XXX/write(I)V\n",

   449          "return\n",

   450          ".end method\n")

   451   }

   452 }

   453 

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

   455   val tks = tokenizer(input)

   456   //println(Prog.parse(tks))

   457   val ast = Prog.parse_single(tks)

   458   val instructions = ast.flatMap(compile_decl).mkString

   459   (library + instructions).replaceAllLiterally("XXX", class_name)

   460 }

   461 

   462 

   463 def compile_file(file_name: String) = {

   464   val class_name = file_name.split('.')(0)

   465   val output = compile(class_name, fromFile(file_name))

   466   val fw = new java.io.FileWriter(class_name + ".j") 

   467   fw.write(output) 

   468   fw.close()

   469 }

   470 

   471 def time_needed[T](i: Int, code: => T) = {

   472   val start = System.nanoTime()

   473   for (j <- 1 to i) code

   474   val end = System.nanoTime()

   475   (end - start)/(i * 1.0e9)

   476 }

   477 

   478 def compile_run(file_name: String) : Unit = {

   479   val class_name = file_name.split('.')(0)

   480   compile_file(file_name)

   481   val test = ("java -jar jvm/jasmin-2.4/jasmin.jar " + class_name + ".j").!!

   482   println("Time: " + time_needed(2, ("java " + class_name + "/" + class_name).!))

   483 }

   484 

   485 

   486 //examples

   487 compile_run("defs.rec")

   488 //compile_run("fact.rec")