Mercurial Mercurial > hg > afl-material / comparison
summary | shortlog | changelog | graph | tags | bookmarks | branches | files | changeset | file | latest | revisions | annotate | diff | comparison | raw | help
Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
solutions/cw3/lexer.sc
changeset 919 53f08d873e09
parent 905 15973df32613
child 920 7af2eea19646
equal deleted inserted replaced
918:53e7da9f372a 919:53f08d873e09 
    35 }
 
    35 }
 
    36 
    36 
    37 implicit def string2rexp(s : String) : Rexp = 
    37 implicit def string2rexp(s : String) : Rexp = 
    38   charlist2rexp(s.toList)
 
    38   charlist2rexp(s.toList)
 
    39 
    39 
    40 implicit def RexpOps(r: Rexp) = new {
 
    40 extension (r: Rexp) {
 
    41   def | (s: Rexp) = ALT(r, s)
 
    41   def | (s: Rexp) = ALT(r, s)
 
    42   def % = STAR(r)
 
    42   def % = STAR(r)
 
    43   def ~ (s: Rexp) = SEQ(r, s)
 
    43   def ~ (s: Rexp) = SEQ(r, s)
 
    44 }
 
    44 }
 
    45 
    45 
    46 implicit def stringOps(s: String) = new {
 
    46 extension (s: String) {
 
    47   def | (r: Rexp) = ALT(s, r)
 
    47   def | (r: Rexp) = ALT(s, r)
 
    48   def | (r: String) = ALT(s, r)
 
    48   def | (r: String) = ALT(s, r)
 
    49   def % = STAR(s)
 
    49   def % = STAR(s)
 
    50   def ~ (r: Rexp) = SEQ(s, r)
 
    50   def ~ (r: Rexp) = SEQ(s, r)
 
    51   def ~ (r: String) = SEQ(s, r)
 
    51   def ~ (r: String) = SEQ(s, r)
 
    80   case STAR(r) => SEQ(der(c, r), STAR(r))
 
    80   case STAR(r) => SEQ(der(c, r), STAR(r))
 
    81 
    81 
    82   case RECD(_, r1) => der(c, r1)
 
    82   case RECD(_, r1) => der(c, r1)
 
    83   case RANGE(s) => if (s.contains(c)) ONE else ZERO 
    83   case RANGE(s) => if (s.contains(c)) ONE else ZERO 
    84   case PLUS(r1) => SEQ(der(c, r1), STAR(r1))
 
    84   case PLUS(r1) => SEQ(der(c, r1), STAR(r1))
 
    85   case OPTIONAL(r1) => ALT(der(c, r1), ZERO)
 
    85   case OPTIONAL(r1) => der(c, r1)
 
    86   case NTIMES(r, i) => 
    86   case NTIMES(r, i) => 
    87     if (i == 0) ZERO else SEQ(der(c, r), NTIMES(r, i - 1))
 
    87     if (i == 0) ZERO else SEQ(der(c, r), NTIMES(r, i - 1))
 
    88 }
 
    88 }
 
    89 
    89 
    90 // Flatten
 
    90 // Flatten
 
   117   case SEQ(r1, r2) => Sequ(mkeps(r1), mkeps(r2))
 
   117   case SEQ(r1, r2) => Sequ(mkeps(r1), mkeps(r2))
 
   118   case STAR(r) => Stars(Nil)
 
   118   case STAR(r) => Stars(Nil)
 
   119   case RECD(x, r) => Rec(x, mkeps(r))
 
   119   case RECD(x, r) => Rec(x, mkeps(r))
 
   120 
   120 
   121   case PLUS(r) => Stars(List(mkeps(r)))   // the first copy must match the empty string
 
   121   case PLUS(r) => Stars(List(mkeps(r)))   // the first copy must match the empty string
 
   122   case OPTIONAL(r) => Right(Empty)
 
   122   case OPTIONAL(r) => if (nullable(r)) Stars(List(mkeps(r))) else Stars(Nil)
 
   123   case NTIMES(r, i) => Stars(List.fill(i)(mkeps(r)))
 
   123   case NTIMES(r, i) => Stars(List.fill(i)(mkeps(r)))
 
   124 }
 
   124 }
 
   125 
   125 
   126 // Inj
 
   126 // Inj
 
   127 def inj(r: Rexp, c: Char, v: Val) : Val = (r, v) match {
 
   127 def inj(r: Rexp, c: Char, v: Val) : Val = (r, v) match {
 
   134   case (CHAR(d), Empty) => Chr(c) 
   134   case (CHAR(d), Empty) => Chr(c) 
   135   case (RECD(x, r1), _) => Rec(x, inj(r1, c, v))
 
   135   case (RECD(x, r1), _) => Rec(x, inj(r1, c, v))
 
   136 
   136 
   137   case (RANGE(_), Empty) => Chr(c)
 
   137   case (RANGE(_), Empty) => Chr(c)
 
   138   case (PLUS(r), Sequ(v1, Stars(vs))) => Stars(inj(r, c, v1)::vs)
 
   138   case (PLUS(r), Sequ(v1, Stars(vs))) => Stars(inj(r, c, v1)::vs)
 
   139   case (OPTIONAL(r), Left(v1)) => Left(inj(r, c, v1))
 
   139   case (OPTIONAL(r), Left(v1)) => Stars(List(inj(r, c, v1)))
 
   140   case (NTIMES(r, n), Sequ(v1, Stars(vs))) => Stars(inj(r, c, v1)::vs)
 
   140   case (NTIMES(r, n), Sequ(v1, Stars(vs))) => Stars(inj(r, c, v1)::vs)
 
   141 }
 
   141 }
 
   142 
   142 
   143 // Rectification functions
 
   143 // Rectification functions
 
   144 def F_ID(v: Val): Val = v
 
   144 def F_ID(v: Val): Val = v
 
   199 def lexing_simp(r: Rexp, s: String) = env(lex_simp(r, s.toList))
 
   199 def lexing_simp(r: Rexp, s: String) = env(lex_simp(r, s.toList))
 
   200 
   200 
   201 // Language specific code
 
   201 // Language specific code
 
   202 val KEYWORD : Rexp = "while" | "if" | "then" | "else" | "do" | "for" | "to" | "true" | "false" | "read" | "write" | "skip" 
   202 val KEYWORD : Rexp = "while" | "if" | "then" | "else" | "do" | "for" | "to" | "true" | "false" | "read" | "write" | "skip" 
   203 val OP : Rexp = "+" | "-" | "*" | "%" | "/" | "==" | "!=" | ">" | "<" | ">=" | "<=" | ":=" | "&&" | "||"
 
   203 val OP : Rexp = "+" | "-" | "*" | "%" | "/" | "==" | "!=" | ">" | "<" | ">=" | "<=" | ":=" | "&&" | "||"
 
   204 val LET: Rexp = RANGE(('A' to 'Z').toSet ++ ('a' to 'z'))
 
   204 val LET: Rexp = RANGE(('A' to 'Z').toSet ++ ('a' to 'z').toSet)
 
   205 val SYM : Rexp = (LET | RANGE(Set('.', '_', '>', '<', '=', ';', ',', ':')))
 
   205 val SYM : Rexp = RANGE(Set('.', '_', '>', '<', '=', ';', ',', ':', ')', '('))
 
   206 val PARENS : Rexp = "(" | "{" | ")" | "}"
 
   206 val PARENS : Rexp = "(" | "{" | ")" | "}"
 
   207 val SEMI : Rexp = ";"
 
   207 val SEMI : Rexp = ";"
 
   208 val WHITESPACE : Rexp = PLUS(" ") | "\n" | "\t" | "\r"
 
   208 val WHITESPACE : Rexp = PLUS(" ") | "\n" | "\t" | "\r"
 
   209 val DIGIT : Rexp = RANGE(('0' to '9').toSet)
 
   209 val DIGIT : Rexp = RANGE(('0' to '9').toSet)
 
   210 val DIGIT1 : Rexp = RANGE(('1' to '9').toSet)
 
   210 val DIGIT1 : Rexp = RANGE(('1' to '9').toSet)
 
   211 val STRING : Rexp = "\"" ~ (SYM | " " | "\\n" | DIGIT).% ~ "\""
 
   211 val STRING : Rexp = "\"" ~ (LET | SYM | DIGIT | " " | "\\n").% ~ "\""
 
   212 val ID : Rexp = LET ~ (LET | "_" | DIGIT).%
 
   212 val ID : Rexp = LET ~ (LET | "_" | DIGIT).%
 
   213 val NUM : Rexp = "0" | (DIGIT1 ~ DIGIT.%)
 
   213 val NUM : Rexp = "0" | (DIGIT1 ~ DIGIT.%)
 
   214 val EOL : Rexp = "\n" | "\r\n"
 
   214 val EOL : Rexp = "\n" | "\r\n"
 
   215 val COMMENT : Rexp = "//" ~ (SYM | PARENS | " " | DIGIT).% ~ EOL 
   215 val COMMENT : Rexp = "//" ~ (LET | SYM | PARENS | " " | DIGIT).% ~ EOL 
   216 
   216 
   217 val WHILE_REGS = (("k" $ KEYWORD) | 
   217 val WHILE_REGS = (("k" $ KEYWORD) | 
   218                   ("o" $ OP) | 
   218                   ("o" $ OP) | 
   219                   ("str" $ STRING) |
 
   219                   ("str" $ STRING) |
 
   220                   ("p" $ PARENS) |
 
   220                   ("p" $ PARENS) |
 
   222                   ("w" $ WHITESPACE) | 
   222                   ("w" $ WHITESPACE) | 
   223                   ("i" $ ID) | 
   223                   ("i" $ ID) | 
   224                   ("n" $ NUM) |
 
   224                   ("n" $ NUM) |
 
   225 		  ("c" $ COMMENT)).%
 
   225 		  ("c" $ COMMENT)).%
 
   226 
   226 
   227 // Token
 
   227 
       
   228 
       
   229 def esc(raw: String): String = {
 
       
   230   import scala.reflect.runtime.universe._
 
       
   231   Literal(Constant(raw)).toString
 
       
   232 }
 
       
   233 
       
   234 def escape(tks: List[(String, String)]) =
 
       
   235   tks.map{ case (s1, s2) => (s1, esc(s2))}
 
       
   236 
       
   237 
       
   238 // Tokens
 
   228 abstract class Token extends Serializable 
   239 abstract class Token extends Serializable 
   229 case class T_KEYWORD(s: String) extends Token
 
   240 case class T_KEYWORD(s: String) extends Token
 
   230 case class T_OP(s: String) extends Token
 
   241 case class T_OP(s: String) extends Token
 
   231 case class T_STRING(s: String) extends Token
 
   242 case class T_STRING(s: String) extends Token
 
   232 case class T_PAREN(s: String) extends Token
 
   243 case class T_PAREN(s: String) extends Token
 
   245 }
 
   256 }
 
   246 
   257 
   247 // Tokenise
 
   258 // Tokenise
 
   248 def tokenise(s: String) : List[Token] = 
   259 def tokenise(s: String) : List[Token] = 
   249   lexing_simp(WHILE_REGS, s).collect(token)
 
   260   lexing_simp(WHILE_REGS, s).collect(token)
 
   250 
       
   251 
       
   252 val fact = """
 
       
   253 write "Input n please";
 
       
   254 read n;
 
       
   255 write "The factors of n are";
 
       
   256 f := 2;
 
       
   257 while n != 1 do {
 
       
   258     while (n / f) * f == n do {
 
       
   259         write f;
 
       
   260         n := n / f
 
       
   261     };
 
       
   262     f := f + 1
 
       
   263 }
 
       
   264 """
 
       
   265 
       
   266 //println(tokenise(fact))
 
       
   267
afl-material