| author | Christian Urban <urbanc@in.tum.de> | 
| Sun, 28 Jul 2019 16:15:03 +0100 | |
| changeset 625 | f8d3e2c7d8b2 | 
| parent 380 | 1e88390e81aa | 
| child 626 | c75c730bdf11 | 
| permissions | -rw-r--r-- | 
| 625 | 1 | // A Small Compiler for a Simple Functional Language | 
| 2 | // (includes a lexer and a parser) | |
| 3 | ||
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 4 | import scala.language.implicitConversions | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 5 | import scala.language.reflectiveCalls | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 6 | import scala.util._ | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 7 | import scala.annotation.tailrec | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 8 | import scala.sys.process._ | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 9 | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 10 | def fromFile(name: String) : String = | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 11 | io.Source.fromFile(name).mkString | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 12 | |
| 625 | 13 | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 14 | abstract class Rexp | 
| 625 | 15 | case object ZERO extends Rexp | 
| 16 | case object ONE extends Rexp | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 17 | case class CHAR(c: Char) extends Rexp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 18 | case class ALT(r1: Rexp, r2: Rexp) extends Rexp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 19 | case class SEQ(r1: Rexp, r2: Rexp) extends Rexp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 20 | case class STAR(r: Rexp) extends Rexp | 
| 625 | 21 | case class RECD(x: String, r: Rexp) extends Rexp | 
| 22 | ||
| 23 | abstract class Val | |
| 24 | case object Empty extends Val | |
| 25 | case class Chr(c: Char) extends Val | |
| 26 | case class Sequ(v1: Val, v2: Val) extends Val | |
| 27 | case class Left(v: Val) extends Val | |
| 28 | case class Right(v: Val) extends Val | |
| 29 | case class Stars(vs: List[Val]) extends Val | |
| 30 | case class Rec(x: String, v: Val) extends Val | |
| 31 | ||
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 32 | // some convenience for typing in regular expressions | 
| 625 | 33 | def charlist2rexp(s : List[Char]): Rexp = s match {
 | 
| 34 | case Nil => ONE | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 35 | case c::Nil => CHAR(c) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 36 | case c::s => SEQ(CHAR(c), charlist2rexp(s)) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 37 | } | 
| 625 | 38 | implicit def string2rexp(s : String) : Rexp = | 
| 39 | charlist2rexp(s.toList) | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 40 | |
| 625 | 41 | implicit def RexpOps(r: Rexp) = new {
 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 42 | def | (s: Rexp) = ALT(r, s) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 43 | def % = STAR(r) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 44 | def ~ (s: Rexp) = SEQ(r, s) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 45 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 46 | |
| 625 | 47 | implicit def stringOps(s: String) = new {
 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 48 | def | (r: Rexp) = ALT(s, r) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 49 | def | (r: String) = ALT(s, r) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 50 | def % = STAR(s) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 51 | def ~ (r: Rexp) = SEQ(s, r) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 52 | def ~ (r: String) = SEQ(s, r) | 
| 625 | 53 | def $ (r: Rexp) = RECD(s, r) | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 54 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 55 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 56 | def nullable (r: Rexp) : Boolean = r match {
 | 
| 625 | 57 | case ZERO => false | 
| 58 | case ONE => true | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 59 | case CHAR(_) => false | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 60 | case ALT(r1, r2) => nullable(r1) || nullable(r2) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 61 | case SEQ(r1, r2) => nullable(r1) && nullable(r2) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 62 | case STAR(_) => true | 
| 625 | 63 | case RECD(_, r1) => nullable(r1) | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 64 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 65 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 66 | def der (c: Char, r: Rexp) : Rexp = r match {
 | 
| 625 | 67 | case ZERO => ZERO | 
| 68 | case ONE => ZERO | |
| 69 | case CHAR(d) => if (c == d) ONE else ZERO | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 70 | case ALT(r1, r2) => ALT(der(c, r1), der(c, r2)) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 71 | case SEQ(r1, r2) => | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 72 | if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2)) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 73 | else SEQ(der(c, r1), r2) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 74 | case STAR(r) => SEQ(der(c, r), STAR(r)) | 
| 625 | 75 | case RECD(_, r1) => der(c, r1) | 
| 76 | } | |
| 77 | ||
| 78 | ||
| 79 | // extracts a string from value | |
| 80 | def flatten(v: Val) : String = v match {
 | |
| 81 | case Empty => "" | |
| 82 | case Chr(c) => c.toString | |
| 83 | case Left(v) => flatten(v) | |
| 84 | case Right(v) => flatten(v) | |
| 85 | case Sequ(v1, v2) => flatten(v1) + flatten(v2) | |
| 86 | case Stars(vs) => vs.map(flatten).mkString | |
| 87 | case Rec(_, v) => flatten(v) | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 88 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 89 | |
| 625 | 90 | // extracts an environment from a value; | 
| 91 | // used for tokenise a string | |
| 92 | def env(v: Val) : List[(String, String)] = v match {
 | |
| 93 | case Empty => Nil | |
| 94 | case Chr(c) => Nil | |
| 95 | case Left(v) => env(v) | |
| 96 | case Right(v) => env(v) | |
| 97 | case Sequ(v1, v2) => env(v1) ::: env(v2) | |
| 98 | case Stars(vs) => vs.flatMap(env) | |
| 99 | case Rec(x, v) => (x, flatten(v))::env(v) | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 100 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 101 | |
| 625 | 102 | // The Injection Part of the lexer | 
| 103 | ||
| 104 | // calculates a value for how a nullable regex | |
| 105 | // matches the empty string | |
| 106 | def mkeps(r: Rexp) : Val = r match {
 | |
| 107 | case ONE => Empty | |
| 108 | case ALT(r1, r2) => | |
| 109 | if (nullable(r1)) Left(mkeps(r1)) else Right(mkeps(r2)) | |
| 110 | case SEQ(r1, r2) => Sequ(mkeps(r1), mkeps(r2)) | |
| 111 | case STAR(r) => Stars(Nil) | |
| 112 | case RECD(x, r) => Rec(x, mkeps(r)) | |
| 113 | } | |
| 114 | ||
| 115 | // injects back a character into a value | |
| 116 | def inj(r: Rexp, c: Char, v: Val) : Val = (r, v) match {
 | |
| 117 | case (STAR(r), Sequ(v1, Stars(vs))) => Stars(inj(r, c, v1)::vs) | |
| 118 | case (SEQ(r1, r2), Sequ(v1, v2)) => Sequ(inj(r1, c, v1), v2) | |
| 119 | case (SEQ(r1, r2), Left(Sequ(v1, v2))) => Sequ(inj(r1, c, v1), v2) | |
| 120 | case (SEQ(r1, r2), Right(v2)) => Sequ(mkeps(r1), inj(r2, c, v2)) | |
| 121 | case (ALT(r1, r2), Left(v1)) => Left(inj(r1, c, v1)) | |
| 122 | case (ALT(r1, r2), Right(v2)) => Right(inj(r2, c, v2)) | |
| 123 | case (CHAR(d), Empty) => Chr(c) | |
| 124 | case (RECD(x, r1), _) => Rec(x, inj(r1, c, v)) | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 125 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 126 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 127 | |
| 625 | 128 | // some "rectification" functions for simplification | 
| 129 | def F_ID(v: Val): Val = v | |
| 130 | def F_RIGHT(f: Val => Val) = (v:Val) => Right(f(v)) | |
| 131 | def F_LEFT(f: Val => Val) = (v:Val) => Left(f(v)) | |
| 132 | def F_ALT(f1: Val => Val, f2: Val => Val) = (v:Val) => v match {
 | |
| 133 | case Right(v) => Right(f2(v)) | |
| 134 | case Left(v) => Left(f1(v)) | |
| 135 | } | |
| 136 | def F_SEQ(f1: Val => Val, f2: Val => Val) = (v:Val) => v match {
 | |
| 137 | case Sequ(v1, v2) => Sequ(f1(v1), f2(v2)) | |
| 138 | } | |
| 139 | def F_SEQ_Empty1(f1: Val => Val, f2: Val => Val) = | |
| 140 | (v:Val) => Sequ(f1(Empty), f2(v)) | |
| 141 | def F_SEQ_Empty2(f1: Val => Val, f2: Val => Val) = | |
| 142 | (v:Val) => Sequ(f1(v), f2(Empty)) | |
| 143 | def F_RECD(f: Val => Val) = (v:Val) => v match {
 | |
| 144 | case Rec(x, v) => Rec(x, f(v)) | |
| 145 | } | |
| 146 | def F_ERROR(v: Val): Val = throw new Exception("error")
 | |
| 147 | ||
| 148 | // simplification of regular expressions returns now also | |
| 149 | // an rectification function; no simplification under STAR | |
| 150 | def simp(r: Rexp): (Rexp, Val => Val) = r match {
 | |
| 151 |   case ALT(r1, r2) => {
 | |
| 152 | val (r1s, f1s) = simp(r1) | |
| 153 | val (r2s, f2s) = simp(r2) | |
| 154 |     (r1s, r2s) match {
 | |
| 155 | case (ZERO, _) => (r2s, F_RIGHT(f2s)) | |
| 156 | case (_, ZERO) => (r1s, F_LEFT(f1s)) | |
| 157 | case _ => if (r1s == r2s) (r1s, F_LEFT(f1s)) | |
| 158 | else (ALT (r1s, r2s), F_ALT(f1s, f2s)) | |
| 159 | } | |
| 160 | } | |
| 161 |   case SEQ(r1, r2) => {
 | |
| 162 | val (r1s, f1s) = simp(r1) | |
| 163 | val (r2s, f2s) = simp(r2) | |
| 164 |     (r1s, r2s) match {
 | |
| 165 | case (ZERO, _) => (ZERO, F_ERROR) | |
| 166 | case (_, ZERO) => (ZERO, F_ERROR) | |
| 167 | case (ONE, _) => (r2s, F_SEQ_Empty1(f1s, f2s)) | |
| 168 | case (_, ONE) => (r1s, F_SEQ_Empty2(f1s, f2s)) | |
| 169 | case _ => (SEQ(r1s,r2s), F_SEQ(f1s, f2s)) | |
| 170 | } | |
| 171 | } | |
| 172 |   case RECD(x, r1) => {
 | |
| 173 | val (r1s, f1s) = simp(r1) | |
| 174 | (RECD(x, r1s), F_RECD(f1s)) | |
| 175 | } | |
| 176 | case r => (r, F_ID) | |
| 177 | } | |
| 178 | ||
| 179 | // lexing functions including simplification | |
| 180 | def lex_simp(r: Rexp, s: List[Char]) : Val = s match {
 | |
| 181 |   case Nil => if (nullable(r)) mkeps(r) else { println ("Lexing Error") ; sys.exit(-1) } 
 | |
| 182 |   case c::cs => {
 | |
| 183 | val (r_simp, f_simp) = simp(der(c, r)) | |
| 184 | inj(r, c, f_simp(lex_simp(r_simp, cs))) | |
| 185 | } | |
| 186 | } | |
| 187 | ||
| 188 | def lexing_simp(r: Rexp, s: String) = env(lex_simp(r, s.toList)) | |
| 189 | ||
| 190 | ||
| 191 | // The Lexing Rules for the Fun Language | |
| 192 | ||
| 193 | def PLUS(r: Rexp) = r ~ r.% | |
| 194 | ||
| 195 | val SYM = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" | "j" | "k" | | |
| 196 | "l" | "m" | "n" | "o" | "p" | "q" | "r" | "s" | "t" | "u" | "v" | | |
| 197 | "w" | "x" | "y" | "z" | "T" | "_" | |
| 198 | val DIGIT = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 199 | val ID = SYM ~ (SYM | DIGIT).% | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 200 | val NUM = PLUS(DIGIT) | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 201 | val KEYWORD : Rexp = "if" | "then" | "else" | "write" | "def" | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 202 | val SEMI: Rexp = ";" | 
| 625 | 203 | val OP: Rexp = "=" | "==" | "-" | "+" | "*" | "!=" | "<" | ">" | "<=" | ">=" | "%" | "/" | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 204 | val WHITESPACE = PLUS(" " | "\n" | "\t")
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 205 | val RPAREN: Rexp = ")" | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 206 | val LPAREN: Rexp = "("
 | 
| 625 | 207 | val COMMA: Rexp = "," | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 208 | val ALL = SYM | DIGIT | OP | " " | ":" | ";" | "\"" | "=" | "," | "(" | ")"
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 209 | val ALL2 = ALL | "\n" | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 210 | val COMMENT = ("/*" ~ ALL2.% ~ "*/") | ("//" ~ ALL.% ~ "\n")
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 211 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 212 | |
| 625 | 213 | |
| 214 | val WHILE_REGS = (("k" $ KEYWORD) | 
 | |
| 215 |                   ("i" $ ID) | 
 | |
| 216 |                   ("o" $ OP) | 
 | |
| 217 |                   ("n" $ NUM) | 
 | |
| 218 |                   ("s" $ SEMI) | 
 | |
| 219 |                   ("c" $ COMMA) |
 | |
| 220 |                   ("pl" $ LPAREN) |
 | |
| 221 |                   ("pr" $ RPAREN) |
 | |
| 222 |                   ("w" $ (WHITESPACE | COMMENT))).%
 | |
| 223 | ||
| 224 | ||
| 225 | ||
| 226 | // The tokens for the While language | |
| 227 | ||
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 228 | abstract class Token | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 229 | case object T_SEMI extends Token | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 230 | case object T_COMMA extends Token | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 231 | case object T_LPAREN extends Token | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 232 | case object T_RPAREN extends Token | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 233 | case class T_ID(s: String) extends Token | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 234 | case class T_OP(s: String) extends Token | 
| 625 | 235 | case class T_NUM(n: Int) extends Token | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 236 | case class T_KWD(s: String) extends Token | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 237 | |
| 625 | 238 | val token : PartialFunction[(String, String), Token] = {
 | 
| 239 |   case ("k", s) => T_KWD(s)
 | |
| 240 |   case ("i", s) => T_ID(s)
 | |
| 241 |   case ("o", s) => T_OP(s)
 | |
| 242 |   case ("n", s) => T_NUM(s.toInt)
 | |
| 243 |   case ("s", _) => T_SEMI
 | |
| 244 |   case ("c", _) => T_COMMA
 | |
| 245 |   case ("pl", _) => T_LPAREN
 | |
| 246 |   case ("pr", _) => T_RPAREN
 | |
| 247 | } | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 248 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 249 | |
| 625 | 250 | def tokenise(s: String) : List[Token] = | 
| 251 | lexing_simp(WHILE_REGS, s).collect(token) | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 252 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 253 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 254 | // Parser - Abstract syntax trees | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 255 | abstract class Exp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 256 | abstract class BExp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 257 | abstract class Decl | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 258 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 259 | case class Def(name: String, args: List[String], body: Exp) extends Decl | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 260 | case class Main(e: Exp) extends Decl | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 261 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 262 | case class Call(name: String, args: List[Exp]) extends Exp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 263 | case class If(a: BExp, e1: Exp, e2: Exp) extends Exp | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 264 | case class Write(e: Exp) extends Exp | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 265 | case class Var(s: String) extends Exp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 266 | case class Num(i: Int) extends Exp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 267 | case class Aop(o: String, a1: Exp, a2: Exp) extends Exp | 
| 625 | 268 | case class Sequence(e1: Exp, e2: Exp) extends Exp | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 269 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 270 | case class Bop(o: String, a1: Exp, a2: Exp) extends BExp | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 271 | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 272 | // calculating the maximal needed stack size | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 273 | def max_stack_exp(e: Exp): Int = e match {
 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 274 | case Call(_, args) => args.map(max_stack_exp).sum | 
| 225 
ef573e8fc86e
bug found by Andres
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
223diff
changeset | 275 | case If(a, e1, e2) => max_stack_bexp(a) + (List(max_stack_exp(e1), max_stack_exp(e2)).max) | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 276 | case Write(e) => max_stack_exp(e) + 1 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 277 | case Var(_) => 1 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 278 | case Num(_) => 1 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 279 | case Aop(_, a1, a2) => max_stack_exp(a1) + max_stack_exp(a2) | 
| 625 | 280 | case Sequence(e1, e2) => List(max_stack_exp(e1), max_stack_exp(e2)).max | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 281 | } | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 282 | def max_stack_bexp(e: BExp): Int = e match {
 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 283 | case Bop(_, a1, a2) => max_stack_exp(a1) + max_stack_exp(a2) | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 284 | } | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 285 | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 286 | // Parser combinators | 
| 625 | 287 | abstract class Parser[I, T](implicit ev: I => Seq[_]) {
 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 288 | def parse(ts: I): Set[(T, I)] | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 289 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 290 | def parse_all(ts: I) : Set[T] = | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 291 | for ((head, tail) <- parse(ts); if (tail.isEmpty)) yield head | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 292 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 293 |   def parse_single(ts: I) : T = parse_all(ts).toList match {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 294 | case List(t) => t | 
| 625 | 295 |     case _ => { println ("Parse Error\n") ; sys.exit(-1) }
 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 296 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 297 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 298 | |
| 625 | 299 | class SeqParser[I, T, S](p: => Parser[I, T], | 
| 300 |                          q: => Parser[I, S])(implicit ev: I => Seq[_]) extends Parser[I, (T, S)] {
 | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 301 | def parse(sb: I) = | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 302 | for ((head1, tail1) <- p.parse(sb); | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 303 | (head2, tail2) <- q.parse(tail1)) yield ((head1, head2), tail2) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 304 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 305 | |
| 625 | 306 | class AltParser[I, T](p: => Parser[I, T], | 
| 307 |                       q: => Parser[I, T])(implicit ev: I => Seq[_]) extends Parser[I, T] {
 | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 308 | def parse(sb: I) = p.parse(sb) ++ q.parse(sb) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 309 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 310 | |
| 625 | 311 | class FunParser[I, T, S](p: => Parser[I, T], | 
| 312 |                          f: T => S)(implicit ev: I => Seq[_]) extends Parser[I, S] {
 | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 313 | def parse(sb: I) = | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 314 | for ((head, tail) <- p.parse(sb)) yield (f(head), tail) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 315 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 316 | |
| 625 | 317 | implicit def ParserOps[I, T](p: Parser[I, T])(implicit ev: I => Seq[_]) = new {
 | 
| 318 | def || (q : => Parser[I, T]) = new AltParser[I, T](p, q) | |
| 319 | def ==>[S] (f: => T => S) = new FunParser[I, T, S](p, f) | |
| 320 | def ~[S] (q : => Parser[I, S]) = new SeqParser[I, T, S](p, q) | |
| 321 | } | |
| 322 | ||
| 323 | def ListParser[I, T, S](p: => Parser[I, T], | |
| 324 |                         q: => Parser[I, S])(implicit ev: I => Seq[_]): Parser[I, List[T]] = {
 | |
| 325 |   (p ~ q ~ ListParser(p, q)) ==> { case ((x, y), z) => x :: z : List[T] } ||
 | |
| 326 | (p ==> ((s) => List(s))) | |
| 327 | } | |
| 328 | ||
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 329 | case class TokParser(tok: Token) extends Parser[List[Token], Token] {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 330 |   def parse(ts: List[Token]) = ts match {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 331 | case t::ts if (t == tok) => Set((t, ts)) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 332 | case _ => Set () | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 333 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 334 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 335 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 336 | implicit def token2tparser(t: Token) = TokParser(t) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 337 | |
| 625 | 338 | implicit def TokOps(t: Token) = new {
 | 
| 339 | def || (q : => Parser[List[Token], Token]) = new AltParser[List[Token], Token](t, q) | |
| 340 | def ==>[S] (f: => Token => S) = new FunParser[List[Token], Token, S](t, f) | |
| 341 | def ~[S](q : => Parser[List[Token], S]) = new SeqParser[List[Token], Token, S](t, q) | |
| 342 | } | |
| 343 | ||
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 344 | case object NumParser extends Parser[List[Token], Int] {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 345 |   def parse(ts: List[Token]) = ts match {
 | 
| 625 | 346 | case T_NUM(n)::ts => Set((n, ts)) | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 347 | case _ => Set () | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 348 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 349 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 350 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 351 | case object IdParser extends Parser[List[Token], String] {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 352 |   def parse(ts: List[Token]) = ts match {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 353 | case T_ID(s)::ts => Set((s, ts)) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 354 | case _ => Set () | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 355 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 356 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 357 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 358 | |
| 625 | 359 | // Grammar Rules | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 360 | |
| 625 | 361 | // arithmetic expressions | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 362 | lazy val Exp: Parser[List[Token], Exp] = | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 363 |   (T_KWD("if") ~ BExp ~ T_KWD("then") ~ Exp ~ T_KWD("else") ~ Exp) ==>
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 364 |     { case (((((x, y), z), u), v), w) => If(y, u, w): Exp } ||
 | 
| 625 | 365 |   (M ~ T_SEMI ~ Exp) ==> { case ((x, y), z) => Sequence(x, z): Exp } || M
 | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 366 | lazy val M: Parser[List[Token], Exp] = | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 367 |   (T_KWD("write") ~ L) ==> { case (x, y) => Write(y): Exp } || L
 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 368 | lazy val L: Parser[List[Token], Exp] = | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 369 |   (T ~ T_OP("+") ~ Exp) ==> { case ((x, y), z) => Aop("+", x, z): Exp } ||
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 370 |   (T ~ T_OP("-") ~ Exp) ==> { case ((x, y), z) => Aop("-", x, z): Exp } || T  
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 371 | lazy val T: Parser[List[Token], Exp] = | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 372 |   (F ~ T_OP("*") ~ T) ==> { case ((x, y), z) => Aop("*", x, z): Exp } || 
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 373 |   (F ~ T_OP("/") ~ T) ==> { case ((x, y), z) => Aop("/", x, z): Exp } || 
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 374 |   (F ~ T_OP("%") ~ T) ==> { case ((x, y), z) => Aop("%", x, z): Exp } || F
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 375 | lazy val F: Parser[List[Token], Exp] = | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 376 | (IdParser ~ T_LPAREN ~ ListParser(Exp, T_COMMA) ~ T_RPAREN) ==> | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 377 |     { case (((x, y), z), w) => Call(x, z): Exp } ||
 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 378 |   (T_LPAREN ~ Exp ~ T_RPAREN) ==> { case ((x, y), z) => y: Exp } || 
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 379 |   IdParser ==> { case x => Var(x): Exp } || 
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 380 |   NumParser ==> { case x => Num(x): Exp }
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 381 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 382 | // boolean expressions | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 383 | lazy val BExp: Parser[List[Token], BExp] = | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 384 |   (Exp ~ T_OP("==") ~ Exp) ==> { case ((x, y), z) => Bop("==", x, z): BExp } || 
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 385 |   (Exp ~ T_OP("!=") ~ Exp) ==> { case ((x, y), z) => Bop("!=", x, z): BExp } || 
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 386 |   (Exp ~ T_OP("<") ~ Exp) ==> { case ((x, y), z) => Bop("<", x, z): BExp } || 
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 387 |   (Exp ~ T_OP(">") ~ Exp) ==> { case ((x, y), z) => Bop("<", z, x): BExp } || 
 | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 388 |   (Exp ~ T_OP("<=") ~ Exp) ==> { case ((x, y), z) => Bop("<=", x, z): BExp } || 
 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 389 |   (Exp ~ T_OP("=>") ~ Exp) ==> { case ((x, y), z) => Bop("<=", z, x): BExp }  
 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 390 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 391 | lazy val Defn: Parser[List[Token], Decl] = | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 392 |    (T_KWD("def") ~ IdParser ~ T_LPAREN ~ ListParser(IdParser, T_COMMA) ~ T_RPAREN ~ T_OP("=") ~ Exp) ==>
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 393 |      { case ((((((x, y), z), w), u), v), r) => Def(y, w, r): Decl }
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 394 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 395 | lazy val Prog: Parser[List[Token], List[Decl]] = | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 396 |   (Defn ~ T_SEMI ~ Prog) ==> { case ((x, y), z) => x :: z : List[Decl] } ||
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 397 | (Exp ==> ((s) => List(Main(s)) : List[Decl])) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 398 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 399 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 400 | // compiler - built-in functions | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 401 | // copied from http://www.ceng.metu.edu.tr/courses/ceng444/link/jvm-cpm.html | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 402 | // | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 403 | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 404 | val library = """ | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 405 | .class public XXX.XXX | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 406 | .super java/lang/Object | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 407 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 408 | .method public <init>()V | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 409 | aload_0 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 410 | invokenonvirtual java/lang/Object/<init>()V | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 411 | return | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 412 | .end method | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 413 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 414 | .method public static write(I)V | 
| 380 
1e88390e81aa
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
323diff
changeset | 415 | .limit locals 1 | 
| 
1e88390e81aa
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
323diff
changeset | 416 | .limit stack 2 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 417 | getstatic java/lang/System/out Ljava/io/PrintStream; | 
| 380 
1e88390e81aa
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
323diff
changeset | 418 | iload 0 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 419 | invokevirtual java/io/PrintStream/println(I)V | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 420 | return | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 421 | .end method | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 422 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 423 | """ | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 424 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 425 | // for generating new labels | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 426 | var counter = -1 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 427 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 428 | def Fresh(x: String) = {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 429 | counter += 1 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 430 | x ++ "_" ++ counter.toString() | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 431 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 432 | |
| 625 | 433 | // convenient string interpolations | 
| 434 | // for instructions, labels and methods | |
| 435 | import scala.language.implicitConversions | |
| 436 | import scala.language.reflectiveCalls | |
| 437 | ||
| 438 | implicit def sring_inters(sc: StringContext) = new {
 | |
| 439 | def i(args: Any*): String = " " ++ sc.s(args:_*) ++ "\n" | |
| 440 | def l(args: Any*): String = sc.s(args:_*) ++ ":\n" | |
| 441 | def m(args: Any*): String = sc.s(args:_*) ++ "\n" | |
| 442 | } | |
| 443 | ||
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 444 | |
| 625 | 445 | type Env = Map[String, Int] | 
| 446 | ||
| 447 | // compile expressions | |
| 448 | def compile_exp(a: Exp, env : Env) : String = a match {
 | |
| 449 | case Num(i) => i"ldc $i" | |
| 450 |   case Var(s) => i"iload ${env(s)}"
 | |
| 451 |   case Aop("+", a1, a2) => compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"iadd"
 | |
| 452 |   case Aop("-", a1, a2) => compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"isub"
 | |
| 453 |   case Aop("*", a1, a2) => compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"imul"
 | |
| 454 |   case Aop("/", a1, a2) => compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"idiv"
 | |
| 455 |   case Aop("%", a1, a2) => compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"irem"
 | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 456 |   case If(b, a1, a2) => {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 457 |     val if_else = Fresh("If_else")
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 458 |     val if_end = Fresh("If_end")
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 459 | compile_bexp(b, env, if_else) ++ | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 460 | compile_exp(a1, env) ++ | 
| 625 | 461 | i"goto $if_end" ++ | 
| 462 | l"$if_else" ++ | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 463 | compile_exp(a2, env) ++ | 
| 625 | 464 | l"$if_end" | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 465 | } | 
| 625 | 466 |   case Call(name, args) => {
 | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 467 | val is = "I" * args.length | 
| 625 | 468 | args.map(a => compile_exp(a, env)).mkString ++ | 
| 469 | i"invokestatic XXX/XXX/$name($is)I" | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 470 | } | 
| 625 | 471 |   case Sequence(a1, a2) => {
 | 
| 472 | compile_exp(a1, env) ++ i"pop" ++ compile_exp(a2, env) | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 473 | } | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 474 |   case Write(a1) => {
 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 475 | compile_exp(a1, env) ++ | 
| 625 | 476 | i"dup" ++ | 
| 477 | i"invokestatic XXX/XXX/write(I)V" | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 478 | } | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 479 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 480 | |
| 625 | 481 | // compile boolean expressions | 
| 482 | def compile_bexp(b: BExp, env : Env, jmp: String) : String = b match {
 | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 483 |   case Bop("==", a1, a2) => 
 | 
| 625 | 484 | compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"if_icmpne $jmp" | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 485 |   case Bop("!=", a1, a2) => 
 | 
| 625 | 486 | compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"if_icmpeq $jmp" | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 487 |   case Bop("<", a1, a2) => 
 | 
| 625 | 488 | compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"if_icmpge $jmp" | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 489 |   case Bop("<=", a1, a2) => 
 | 
| 625 | 490 | compile_exp(a1, env) ++ compile_exp(a2, env) ++ i"if_icmpgt $jmp" | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 491 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 492 | |
| 625 | 493 | // compile function for declarations and main | 
| 494 | def compile_decl(d: Decl) : String = d match {
 | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 495 |   case Def(name, args, a) => { 
 | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 496 | val env = args.zipWithIndex.toMap | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 497 | val is = "I" * args.length | 
| 625 | 498 | m".method public static $name($is)I" ++ | 
| 499 |     m".limit locals ${args.length.toString}" ++
 | |
| 500 |     m".limit stack ${1 + max_stack_exp(a)}" ++
 | |
| 501 |     l"${name}_Start" ++   
 | |
| 223 
e4b29b57f6a3
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
221diff
changeset | 502 | compile_exp(a, env) ++ | 
| 625 | 503 | i"ireturn" ++ | 
| 504 | m".end method\n" | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 505 | } | 
| 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 506 |   case Main(a) => {
 | 
| 625 | 507 | m".method public static main([Ljava/lang/String;)V" ++ | 
| 508 | m".limit locals 200" ++ | |
| 509 | m".limit stack 200" ++ | |
| 223 
e4b29b57f6a3
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
221diff
changeset | 510 | compile_exp(a, Map()) ++ | 
| 625 | 511 | i"invokestatic XXX/XXX/write(I)V" ++ | 
| 512 | i"return" ++ | |
| 513 | m".end method\n" | |
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 514 | } | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 515 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 516 | |
| 625 | 517 | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 518 | def compile(class_name: String, input: String) : String = {
 | 
| 625 | 519 | val tks = tokenise(input) | 
| 323 
4ce07c4abdb4
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
311diff
changeset | 520 |   println(Prog.parse_single(tks).mkString("\n"))
 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 521 | val ast = Prog.parse_single(tks) | 
| 625 | 522 | val instructions = ast.map(compile_decl).mkString | 
| 221 
824ffbf66ab4
added fun tail
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
220diff
changeset | 523 |   (library + instructions).replaceAllLiterally("XXX", class_name)
 | 
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 524 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 525 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 526 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 527 | def compile_file(file_name: String) = {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 528 |   val class_name = file_name.split('.')(0)
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 529 | val output = compile(class_name, fromFile(file_name)) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 530 | val fw = new java.io.FileWriter(class_name + ".j") | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 531 | fw.write(output) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 532 | fw.close() | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 533 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 534 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 535 | def time_needed[T](i: Int, code: => T) = {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 536 | val start = System.nanoTime() | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 537 | for (j <- 1 to i) code | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 538 | val end = System.nanoTime() | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 539 | (end - start)/(i * 1.0e9) | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 540 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 541 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 542 | def compile_run(file_name: String) : Unit = {
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 543 |   val class_name = file_name.split('.')(0)
 | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 544 | compile_file(file_name) | 
| 625 | 545 |   val test = (s"java -jar jvm/jasmin-2.4/jasmin.jar ${class_name}.j").!!
 | 
| 546 |   println("Time: " + time_needed(2, (s"java ${class_name}/${class_name}").!))
 | |
| 220 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 547 | } | 
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 548 | |
| 
141041fc76b5
added
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: diff
changeset | 549 | |
| 625 | 550 | // some examples | 
| 323 
4ce07c4abdb4
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
311diff
changeset | 551 | compile_file("fact.rec")
 | 
| 625 | 552 | compile_run("defs.rec")
 | 
| 553 | compile_run("fact.rec")
 |