| 645 |      1 | // A parser for the Fun language
 | 
|  |      2 | //================================
 | 
|  |      3 | //
 | 
|  |      4 | // call with 
 | 
|  |      5 | //
 | 
|  |      6 | //     scala fun_parser.scala fact.tks
 | 
|  |      7 | //
 | 
|  |      8 | //     scala fun_parser.scala defs.tks
 | 
|  |      9 | //
 | 
|  |     10 | // this will generate a .prs file that can be deserialised back
 | 
|  |     11 | // into a list of declarations
 | 
| 644 |     12 | 
 | 
|  |     13 | object Fun_Parser {
 | 
|  |     14 | 
 | 
|  |     15 | import scala.language.implicitConversions    
 | 
|  |     16 | import scala.language.reflectiveCalls
 | 
| 645 |     17 | import scala.util._ 
 | 
| 644 |     18 | import java.io._
 | 
|  |     19 | 
 | 
|  |     20 | abstract class Token extends Serializable 
 | 
|  |     21 | case object T_SEMI extends Token
 | 
|  |     22 | case object T_COMMA extends Token
 | 
|  |     23 | case object T_LPAREN extends Token
 | 
|  |     24 | case object T_RPAREN extends Token
 | 
|  |     25 | case class T_ID(s: String) extends Token
 | 
|  |     26 | case class T_OP(s: String) extends Token
 | 
|  |     27 | case class T_NUM(n: Int) extends Token
 | 
|  |     28 | case class T_KWD(s: String) extends Token
 | 
|  |     29 | 
 | 
|  |     30 | 
 | 
|  |     31 | // Parser combinators
 | 
|  |     32 | //    type parameter I needs to be of Seq-type
 | 
|  |     33 | //
 | 
|  |     34 | abstract class Parser[I, T](implicit ev: I => Seq[_]) {
 | 
|  |     35 |   def parse(ts: I): Set[(T, I)]
 | 
|  |     36 | 
 | 
|  |     37 |   def parse_all(ts: I) : Set[T] =
 | 
|  |     38 |     for ((head, tail) <- parse(ts); if (tail.isEmpty)) yield head
 | 
|  |     39 | 
 | 
|  |     40 |   def parse_single(ts: I) : T = parse_all(ts).toList match {
 | 
|  |     41 |     case List(t) => t
 | 
|  |     42 |     case _ => { println ("Parse Error\n") ; sys.exit(-1) }
 | 
|  |     43 |   }
 | 
|  |     44 | }
 | 
|  |     45 | 
 | 
|  |     46 | // convenience for writing grammar rules
 | 
|  |     47 | case class ~[+A, +B](_1: A, _2: B)
 | 
|  |     48 | 
 | 
|  |     49 | class SeqParser[I, T, S](p: => Parser[I, T], 
 | 
|  |     50 |                          q: => Parser[I, S])(implicit ev: I => Seq[_]) extends Parser[I, ~[T, S]] {
 | 
|  |     51 |   def parse(sb: I) = 
 | 
|  |     52 |     for ((head1, tail1) <- p.parse(sb); 
 | 
|  |     53 |          (head2, tail2) <- q.parse(tail1)) yield (new ~(head1, head2), tail2)
 | 
|  |     54 | }
 | 
|  |     55 | 
 | 
|  |     56 | class AltParser[I, T](p: => Parser[I, T], 
 | 
|  |     57 |                       q: => Parser[I, T])(implicit ev: I => Seq[_]) extends Parser[I, T] {
 | 
|  |     58 |   def parse(sb: I) = p.parse(sb) ++ q.parse(sb)   
 | 
|  |     59 | }
 | 
|  |     60 | 
 | 
|  |     61 | class FunParser[I, T, S](p: => Parser[I, T], 
 | 
|  |     62 |                          f: T => S)(implicit ev: I => Seq[_]) extends Parser[I, S] {
 | 
|  |     63 |   def parse(sb: I) = 
 | 
|  |     64 |     for ((head, tail) <- p.parse(sb)) yield (f(head), tail)
 | 
|  |     65 | }
 | 
|  |     66 | 
 | 
|  |     67 | // convenient combinators
 | 
|  |     68 | implicit def ParserOps[I, T](p: Parser[I, T])(implicit ev: I => Seq[_]) = new {
 | 
|  |     69 |   def || (q : => Parser[I, T]) = new AltParser[I, T](p, q)
 | 
|  |     70 |   def ==>[S] (f: => T => S) = new FunParser[I, T, S](p, f)
 | 
|  |     71 |   def ~[S] (q : => Parser[I, S]) = new SeqParser[I, T, S](p, q)
 | 
|  |     72 | }
 | 
|  |     73 | 
 | 
|  |     74 | def ListParser[I, T, S](p: => Parser[I, T], 
 | 
|  |     75 |                         q: => Parser[I, S])(implicit ev: I => Seq[_]): Parser[I, List[T]] = {
 | 
|  |     76 |   (p ~ q ~ ListParser(p, q)) ==> { case x ~ _ ~ z => x :: z : List[T] } ||
 | 
|  |     77 |   (p ==> ((s) => List(s)))
 | 
|  |     78 | }
 | 
|  |     79 | 
 | 
|  |     80 | case class TokParser(tok: Token) extends Parser[List[Token], Token] {
 | 
|  |     81 |   def parse(ts: List[Token]) = ts match {
 | 
|  |     82 |     case t::ts if (t == tok) => Set((t, ts)) 
 | 
|  |     83 |     case _ => Set ()
 | 
|  |     84 |   }
 | 
|  |     85 | }
 | 
|  |     86 | 
 | 
|  |     87 | implicit def token2tparser(t: Token) = TokParser(t)
 | 
|  |     88 | 
 | 
|  |     89 | implicit def TokOps(t: Token) = new {
 | 
|  |     90 |   def || (q : => Parser[List[Token], Token]) = new AltParser[List[Token], Token](t, q)
 | 
|  |     91 |   def ==>[S] (f: => Token => S) = new FunParser[List[Token], Token, S](t, f)
 | 
|  |     92 |   def ~[S](q : => Parser[List[Token], S]) = new SeqParser[List[Token], Token, S](t, q)
 | 
|  |     93 | }
 | 
|  |     94 | 
 | 
|  |     95 | case object NumParser extends Parser[List[Token], Int] {
 | 
|  |     96 |   def parse(ts: List[Token]) = ts match {
 | 
|  |     97 |     case T_NUM(n)::ts => Set((n, ts)) 
 | 
|  |     98 |     case _ => Set ()
 | 
|  |     99 |   }
 | 
|  |    100 | }
 | 
|  |    101 | 
 | 
|  |    102 | case object IdParser extends Parser[List[Token], String] {
 | 
|  |    103 |   def parse(ts: List[Token]) = ts match {
 | 
|  |    104 |     case T_ID(s)::ts => Set((s, ts)) 
 | 
|  |    105 |     case _ => Set ()
 | 
|  |    106 |   }
 | 
|  |    107 | }
 | 
|  |    108 | 
 | 
|  |    109 | 
 | 
|  |    110 | 
 | 
|  |    111 | // Abstract syntax trees for the Fun language
 | 
|  |    112 | abstract class Exp extends Serializable 
 | 
|  |    113 | abstract class BExp extends Serializable 
 | 
|  |    114 | abstract class Decl extends Serializable 
 | 
|  |    115 | 
 | 
|  |    116 | case class Def(name: String, args: List[String], body: Exp) extends Decl
 | 
|  |    117 | case class Main(e: Exp) extends Decl
 | 
|  |    118 | 
 | 
|  |    119 | case class Call(name: String, args: List[Exp]) extends Exp
 | 
|  |    120 | case class If(a: BExp, e1: Exp, e2: Exp) extends Exp
 | 
|  |    121 | case class Write(e: Exp) extends Exp
 | 
|  |    122 | case class Var(s: String) extends Exp
 | 
|  |    123 | case class Num(i: Int) extends Exp
 | 
|  |    124 | case class Aop(o: String, a1: Exp, a2: Exp) extends Exp
 | 
|  |    125 | case class Sequence(e1: Exp, e2: Exp) extends Exp
 | 
|  |    126 | case class Bop(o: String, a1: Exp, a2: Exp) extends BExp
 | 
|  |    127 | 
 | 
|  |    128 | 
 | 
|  |    129 | 
 | 
|  |    130 | // Grammar Rules for the Fun language
 | 
|  |    131 | 
 | 
|  |    132 | // arithmetic expressions
 | 
|  |    133 | lazy val Exp: Parser[List[Token], Exp] = 
 | 
|  |    134 |   (T_KWD("if") ~ BExp ~ T_KWD("then") ~ Exp ~ T_KWD("else") ~ Exp) ==>
 | 
|  |    135 |     { case _ ~ x ~ _ ~ y ~ _ ~ z => If(x, y, z): Exp } ||
 | 
|  |    136 |   (M ~ T_SEMI ~ Exp) ==> { case x ~ _ ~ y => Sequence(x, y): Exp } || M
 | 
|  |    137 | lazy val M: Parser[List[Token], Exp] =
 | 
|  |    138 |   (T_KWD("write") ~ L) ==> { case _ ~ y => Write(y): Exp } || L
 | 
|  |    139 | lazy val L: Parser[List[Token], Exp] = 
 | 
|  |    140 |   (T ~ T_OP("+") ~ Exp) ==> { case x ~ _ ~ z => Aop("+", x, z): Exp } ||
 | 
|  |    141 |   (T ~ T_OP("-") ~ Exp) ==> { case x ~ _ ~ z => Aop("-", x, z): Exp } || T  
 | 
|  |    142 | lazy val T: Parser[List[Token], Exp] = 
 | 
|  |    143 |   (F ~ T_OP("*") ~ T) ==> { case x ~ _ ~ z => Aop("*", x, z): Exp } || 
 | 
|  |    144 |   (F ~ T_OP("/") ~ T) ==> { case x ~ _ ~ z => Aop("/", x, z): Exp } || 
 | 
|  |    145 |   (F ~ T_OP("%") ~ T) ==> { case x ~ _ ~ z => Aop("%", x, z): Exp } || F
 | 
|  |    146 | lazy val F: Parser[List[Token], Exp] = 
 | 
|  |    147 |   (IdParser ~ T_LPAREN ~ ListParser(Exp, T_COMMA) ~ T_RPAREN) ==> 
 | 
|  |    148 |     { case x ~ _ ~ z ~ _ => Call(x, z): Exp } ||
 | 
|  |    149 |   (T_LPAREN ~ Exp ~ T_RPAREN) ==> { case _ ~ y ~ _ => y: Exp } || 
 | 
|  |    150 |   IdParser ==> { case x => Var(x): Exp } || 
 | 
|  |    151 |   NumParser ==> { case x => Num(x): Exp }
 | 
|  |    152 | 
 | 
|  |    153 | // boolean expressions
 | 
|  |    154 | lazy val BExp: Parser[List[Token], BExp] = 
 | 
|  |    155 |   (Exp ~ T_OP("==") ~ Exp) ==> { case x ~ _ ~ z => Bop("==", x, z): BExp } || 
 | 
|  |    156 |   (Exp ~ T_OP("!=") ~ Exp) ==> { case x ~ _ ~ z => Bop("!=", x, z): BExp } || 
 | 
|  |    157 |   (Exp ~ T_OP("<") ~ Exp) ==> { case x ~ _ ~ z => Bop("<", x, z): BExp } || 
 | 
|  |    158 |   (Exp ~ T_OP(">") ~ Exp) ==> { case x ~ _ ~ z => Bop("<", z, x): BExp } || 
 | 
|  |    159 |   (Exp ~ T_OP("<=") ~ Exp) ==> { case x ~ _ ~ z => Bop("<=", x, z): BExp } || 
 | 
|  |    160 |   (Exp ~ T_OP("=>") ~ Exp) ==> { case x ~ _ ~ z => Bop("<=", z, x): BExp }  
 | 
|  |    161 | 
 | 
|  |    162 | lazy val Defn: Parser[List[Token], Decl] =
 | 
|  |    163 |    (T_KWD("def") ~ IdParser ~ T_LPAREN ~ ListParser(IdParser, T_COMMA) ~ T_RPAREN ~ T_OP("=") ~ Exp) ==>
 | 
|  |    164 |      { case _ ~ y ~ _ ~ w ~ _ ~ _ ~ r => Def(y, w, r): Decl }
 | 
|  |    165 | 
 | 
|  |    166 | lazy val Prog: Parser[List[Token], List[Decl]] =
 | 
|  |    167 |   (Defn ~ T_SEMI ~ Prog) ==> { case x ~ _ ~ z => x :: z : List[Decl] } ||
 | 
|  |    168 |   (Exp ==> ((s) => List(Main(s)) : List[Decl]))
 | 
|  |    169 | 
 | 
|  |    170 | 
 | 
|  |    171 | 
 | 
|  |    172 | // Reading tokens and Writing parse trees
 | 
|  |    173 | 
 | 
|  |    174 | def serialise[T](fname: String, data: T) = {
 | 
| 645 |    175 |   import scala.util.Using
 | 
|  |    176 |   Using(new ObjectOutputStream(new FileOutputStream(fname))) {
 | 
|  |    177 |     out => out.writeObject(data)
 | 
|  |    178 |   }
 | 
| 644 |    179 | }
 | 
|  |    180 | 
 | 
| 645 |    181 | def deserialise[T](fname: String) : Try[T] = {
 | 
|  |    182 |   import scala.util.Using
 | 
|  |    183 |   Using(new ObjectInputStream(new FileInputStream(fname))) {
 | 
|  |    184 |     in => in.readObject.asInstanceOf[T]
 | 
|  |    185 |   }
 | 
| 644 |    186 | }
 | 
|  |    187 | 
 | 
|  |    188 | 
 | 
| 645 |    189 | def main(args: Array[String]) : Unit= {
 | 
| 644 |    190 |   val fname = args(0)
 | 
|  |    191 |   val pname = fname.stripSuffix(".tks") ++ ".prs"
 | 
| 645 |    192 |   val tks = deserialise[List[Token]](fname).getOrElse(Nil)
 | 
| 644 |    193 |   serialise(pname, Prog.parse_single(tks))
 | 
|  |    194 | 
 | 
|  |    195 |   // testing whether read-back is working
 | 
| 645 |    196 |   //val ptree = deserialise[List[Decl]](pname).get
 | 
| 644 |    197 |   //println(s"Reading back from ${pname}:\n${ptree.mkString("\n")}")  
 | 
|  |    198 | }
 | 
|  |    199 | 
 | 
| 645 |    200 | }
 |