| 644 |      1 | // A Small Compiler for a Simple Functional Language
 | 
|  |      2 | // (includes a lexer and a parser)
 | 
|  |      3 | 
 | 
|  |      4 | object Fun_Parser {
 | 
|  |      5 | 
 | 
|  |      6 | import scala.language.implicitConversions    
 | 
|  |      7 | import scala.language.reflectiveCalls
 | 
|  |      8 | import java.io._
 | 
|  |      9 | 
 | 
|  |     10 | abstract class Token extends Serializable 
 | 
|  |     11 | case object T_SEMI extends Token
 | 
|  |     12 | case object T_COMMA extends Token
 | 
|  |     13 | case object T_LPAREN extends Token
 | 
|  |     14 | case object T_RPAREN extends Token
 | 
|  |     15 | case class T_ID(s: String) extends Token
 | 
|  |     16 | case class T_OP(s: String) extends Token
 | 
|  |     17 | case class T_NUM(n: Int) extends Token
 | 
|  |     18 | case class T_KWD(s: String) extends Token
 | 
|  |     19 | 
 | 
|  |     20 | 
 | 
|  |     21 | // Parser combinators
 | 
|  |     22 | //    type parameter I needs to be of Seq-type
 | 
|  |     23 | //
 | 
|  |     24 | abstract class Parser[I, T](implicit ev: I => Seq[_]) {
 | 
|  |     25 |   def parse(ts: I): Set[(T, I)]
 | 
|  |     26 | 
 | 
|  |     27 |   def parse_all(ts: I) : Set[T] =
 | 
|  |     28 |     for ((head, tail) <- parse(ts); if (tail.isEmpty)) yield head
 | 
|  |     29 | 
 | 
|  |     30 |   def parse_single(ts: I) : T = parse_all(ts).toList match {
 | 
|  |     31 |     case List(t) => t
 | 
|  |     32 |     case _ => { println ("Parse Error\n") ; sys.exit(-1) }
 | 
|  |     33 |   }
 | 
|  |     34 | }
 | 
|  |     35 | 
 | 
|  |     36 | // convenience for writing grammar rules
 | 
|  |     37 | case class ~[+A, +B](_1: A, _2: B)
 | 
|  |     38 | 
 | 
|  |     39 | class SeqParser[I, T, S](p: => Parser[I, T], 
 | 
|  |     40 |                          q: => Parser[I, S])(implicit ev: I => Seq[_]) extends Parser[I, ~[T, S]] {
 | 
|  |     41 |   def parse(sb: I) = 
 | 
|  |     42 |     for ((head1, tail1) <- p.parse(sb); 
 | 
|  |     43 |          (head2, tail2) <- q.parse(tail1)) yield (new ~(head1, head2), tail2)
 | 
|  |     44 | }
 | 
|  |     45 | 
 | 
|  |     46 | class AltParser[I, T](p: => Parser[I, T], 
 | 
|  |     47 |                       q: => Parser[I, T])(implicit ev: I => Seq[_]) extends Parser[I, T] {
 | 
|  |     48 |   def parse(sb: I) = p.parse(sb) ++ q.parse(sb)   
 | 
|  |     49 | }
 | 
|  |     50 | 
 | 
|  |     51 | class FunParser[I, T, S](p: => Parser[I, T], 
 | 
|  |     52 |                          f: T => S)(implicit ev: I => Seq[_]) extends Parser[I, S] {
 | 
|  |     53 |   def parse(sb: I) = 
 | 
|  |     54 |     for ((head, tail) <- p.parse(sb)) yield (f(head), tail)
 | 
|  |     55 | }
 | 
|  |     56 | 
 | 
|  |     57 | // convenient combinators
 | 
|  |     58 | implicit def ParserOps[I, T](p: Parser[I, T])(implicit ev: I => Seq[_]) = new {
 | 
|  |     59 |   def || (q : => Parser[I, T]) = new AltParser[I, T](p, q)
 | 
|  |     60 |   def ==>[S] (f: => T => S) = new FunParser[I, T, S](p, f)
 | 
|  |     61 |   def ~[S] (q : => Parser[I, S]) = new SeqParser[I, T, S](p, q)
 | 
|  |     62 | }
 | 
|  |     63 | 
 | 
|  |     64 | def ListParser[I, T, S](p: => Parser[I, T], 
 | 
|  |     65 |                         q: => Parser[I, S])(implicit ev: I => Seq[_]): Parser[I, List[T]] = {
 | 
|  |     66 |   (p ~ q ~ ListParser(p, q)) ==> { case x ~ _ ~ z => x :: z : List[T] } ||
 | 
|  |     67 |   (p ==> ((s) => List(s)))
 | 
|  |     68 | }
 | 
|  |     69 | 
 | 
|  |     70 | case class TokParser(tok: Token) extends Parser[List[Token], Token] {
 | 
|  |     71 |   def parse(ts: List[Token]) = ts match {
 | 
|  |     72 |     case t::ts if (t == tok) => Set((t, ts)) 
 | 
|  |     73 |     case _ => Set ()
 | 
|  |     74 |   }
 | 
|  |     75 | }
 | 
|  |     76 | 
 | 
|  |     77 | implicit def token2tparser(t: Token) = TokParser(t)
 | 
|  |     78 | 
 | 
|  |     79 | implicit def TokOps(t: Token) = new {
 | 
|  |     80 |   def || (q : => Parser[List[Token], Token]) = new AltParser[List[Token], Token](t, q)
 | 
|  |     81 |   def ==>[S] (f: => Token => S) = new FunParser[List[Token], Token, S](t, f)
 | 
|  |     82 |   def ~[S](q : => Parser[List[Token], S]) = new SeqParser[List[Token], Token, S](t, q)
 | 
|  |     83 | }
 | 
|  |     84 | 
 | 
|  |     85 | case object NumParser extends Parser[List[Token], Int] {
 | 
|  |     86 |   def parse(ts: List[Token]) = ts match {
 | 
|  |     87 |     case T_NUM(n)::ts => Set((n, ts)) 
 | 
|  |     88 |     case _ => Set ()
 | 
|  |     89 |   }
 | 
|  |     90 | }
 | 
|  |     91 | 
 | 
|  |     92 | case object IdParser extends Parser[List[Token], String] {
 | 
|  |     93 |   def parse(ts: List[Token]) = ts match {
 | 
|  |     94 |     case T_ID(s)::ts => Set((s, ts)) 
 | 
|  |     95 |     case _ => Set ()
 | 
|  |     96 |   }
 | 
|  |     97 | }
 | 
|  |     98 | 
 | 
|  |     99 | 
 | 
|  |    100 | 
 | 
|  |    101 | // Abstract syntax trees for the Fun language
 | 
|  |    102 | abstract class Exp extends Serializable 
 | 
|  |    103 | abstract class BExp extends Serializable 
 | 
|  |    104 | abstract class Decl extends Serializable 
 | 
|  |    105 | 
 | 
|  |    106 | case class Def(name: String, args: List[String], body: Exp) extends Decl
 | 
|  |    107 | case class Main(e: Exp) extends Decl
 | 
|  |    108 | 
 | 
|  |    109 | case class Call(name: String, args: List[Exp]) extends Exp
 | 
|  |    110 | case class If(a: BExp, e1: Exp, e2: Exp) extends Exp
 | 
|  |    111 | case class Write(e: Exp) extends Exp
 | 
|  |    112 | case class Var(s: String) extends Exp
 | 
|  |    113 | case class Num(i: Int) extends Exp
 | 
|  |    114 | case class Aop(o: String, a1: Exp, a2: Exp) extends Exp
 | 
|  |    115 | case class Sequence(e1: Exp, e2: Exp) extends Exp
 | 
|  |    116 | case class Bop(o: String, a1: Exp, a2: Exp) extends BExp
 | 
|  |    117 | 
 | 
|  |    118 | 
 | 
|  |    119 | 
 | 
|  |    120 | // Grammar Rules for the Fun language
 | 
|  |    121 | 
 | 
|  |    122 | // arithmetic expressions
 | 
|  |    123 | lazy val Exp: Parser[List[Token], Exp] = 
 | 
|  |    124 |   (T_KWD("if") ~ BExp ~ T_KWD("then") ~ Exp ~ T_KWD("else") ~ Exp) ==>
 | 
|  |    125 |     { case _ ~ x ~ _ ~ y ~ _ ~ z => If(x, y, z): Exp } ||
 | 
|  |    126 |   (M ~ T_SEMI ~ Exp) ==> { case x ~ _ ~ y => Sequence(x, y): Exp } || M
 | 
|  |    127 | lazy val M: Parser[List[Token], Exp] =
 | 
|  |    128 |   (T_KWD("write") ~ L) ==> { case _ ~ y => Write(y): Exp } || L
 | 
|  |    129 | lazy val L: Parser[List[Token], Exp] = 
 | 
|  |    130 |   (T ~ T_OP("+") ~ Exp) ==> { case x ~ _ ~ z => Aop("+", x, z): Exp } ||
 | 
|  |    131 |   (T ~ T_OP("-") ~ Exp) ==> { case x ~ _ ~ z => Aop("-", x, z): Exp } || T  
 | 
|  |    132 | lazy val T: Parser[List[Token], Exp] = 
 | 
|  |    133 |   (F ~ T_OP("*") ~ T) ==> { case x ~ _ ~ z => Aop("*", x, z): Exp } || 
 | 
|  |    134 |   (F ~ T_OP("/") ~ T) ==> { case x ~ _ ~ z => Aop("/", x, z): Exp } || 
 | 
|  |    135 |   (F ~ T_OP("%") ~ T) ==> { case x ~ _ ~ z => Aop("%", x, z): Exp } || F
 | 
|  |    136 | lazy val F: Parser[List[Token], Exp] = 
 | 
|  |    137 |   (IdParser ~ T_LPAREN ~ ListParser(Exp, T_COMMA) ~ T_RPAREN) ==> 
 | 
|  |    138 |     { case x ~ _ ~ z ~ _ => Call(x, z): Exp } ||
 | 
|  |    139 |   (T_LPAREN ~ Exp ~ T_RPAREN) ==> { case _ ~ y ~ _ => y: Exp } || 
 | 
|  |    140 |   IdParser ==> { case x => Var(x): Exp } || 
 | 
|  |    141 |   NumParser ==> { case x => Num(x): Exp }
 | 
|  |    142 | 
 | 
|  |    143 | // boolean expressions
 | 
|  |    144 | lazy val BExp: Parser[List[Token], BExp] = 
 | 
|  |    145 |   (Exp ~ T_OP("==") ~ Exp) ==> { case x ~ _ ~ z => Bop("==", x, z): BExp } || 
 | 
|  |    146 |   (Exp ~ T_OP("!=") ~ Exp) ==> { case x ~ _ ~ z => Bop("!=", x, z): BExp } || 
 | 
|  |    147 |   (Exp ~ T_OP("<") ~ Exp) ==> { case x ~ _ ~ z => Bop("<", x, z): BExp } || 
 | 
|  |    148 |   (Exp ~ T_OP(">") ~ Exp) ==> { case x ~ _ ~ z => Bop("<", z, x): BExp } || 
 | 
|  |    149 |   (Exp ~ T_OP("<=") ~ Exp) ==> { case x ~ _ ~ z => Bop("<=", x, z): BExp } || 
 | 
|  |    150 |   (Exp ~ T_OP("=>") ~ Exp) ==> { case x ~ _ ~ z => Bop("<=", z, x): BExp }  
 | 
|  |    151 | 
 | 
|  |    152 | lazy val Defn: Parser[List[Token], Decl] =
 | 
|  |    153 |    (T_KWD("def") ~ IdParser ~ T_LPAREN ~ ListParser(IdParser, T_COMMA) ~ T_RPAREN ~ T_OP("=") ~ Exp) ==>
 | 
|  |    154 |      { case _ ~ y ~ _ ~ w ~ _ ~ _ ~ r => Def(y, w, r): Decl }
 | 
|  |    155 | 
 | 
|  |    156 | lazy val Prog: Parser[List[Token], List[Decl]] =
 | 
|  |    157 |   (Defn ~ T_SEMI ~ Prog) ==> { case x ~ _ ~ z => x :: z : List[Decl] } ||
 | 
|  |    158 |   (Exp ==> ((s) => List(Main(s)) : List[Decl]))
 | 
|  |    159 | 
 | 
|  |    160 | 
 | 
|  |    161 | 
 | 
|  |    162 | // Reading tokens and Writing parse trees
 | 
|  |    163 | 
 | 
|  |    164 | def serialise[T](fname: String, data: T) = {
 | 
|  |    165 |   val out = new ObjectOutputStream(new FileOutputStream(fname))
 | 
|  |    166 |   out.writeObject(data)
 | 
|  |    167 |   out.close
 | 
|  |    168 | }
 | 
|  |    169 | 
 | 
|  |    170 | def deserialise[T](fname: String) : T = {
 | 
|  |    171 |   val in = new ObjectInputStream(new FileInputStream(fname))
 | 
|  |    172 |   val data = in.readObject.asInstanceOf[T]
 | 
|  |    173 |   in.close
 | 
|  |    174 |   data
 | 
|  |    175 | }
 | 
|  |    176 | 
 | 
|  |    177 | 
 | 
|  |    178 | def main(args: Array[String]) = {
 | 
|  |    179 |   val fname = args(0)
 | 
|  |    180 |   val pname = fname.stripSuffix(".tks") ++ ".prs"
 | 
|  |    181 |   val tks = deserialise[List[Token]](fname)
 | 
|  |    182 |   serialise(pname, Prog.parse_single(tks))
 | 
|  |    183 | 
 | 
|  |    184 |   // testing whether read-back is working
 | 
|  |    185 |   //val ptree = deserialise[List[Decl]](pname)
 | 
|  |    186 |   //println(s"Reading back from ${pname}:\n${ptree.mkString("\n")}")  
 | 
|  |    187 | }
 | 
|  |    188 | 
 | 
|  |    189 | } |