progs/re3.scala
author Christian Urban <christian dot urban at kcl dot ac dot uk>
Sat, 31 Oct 2015 11:16:52 +0000
changeset 363 0d6deecdb2eb
parent 341 ac1187b2e5c9
permissions -rw-r--r--
updated
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
     1
abstract class Rexp 
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     2
case object NULL extends Rexp
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     3
case object EMPTY extends Rexp
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     4
case class CHAR(c: Char) extends Rexp
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
     5
case class ALT(r1: Rexp, r2: Rexp) extends Rexp
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
     6
case class SEQ(r1: Rexp, r2: Rexp) extends Rexp 
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
     7
case class STAR(r: Rexp) extends Rexp 
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
     8
case class NTIMES(r: Rexp, n: Int) extends Rexp 
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
     9
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    10
def simp(r: Rexp): Rexp = r match {
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    11
  case ALT(r1, r2) => {
341
ac1187b2e5c9 updated program
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 261
diff changeset
    12
    (simp(r1), simp(r2)) match {
ac1187b2e5c9 updated program
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 261
diff changeset
    13
      case (NULL, r2s) => r2s
ac1187b2e5c9 updated program
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 261
diff changeset
    14
      case (r1s, NULL) => r1s
ac1187b2e5c9 updated program
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 261
diff changeset
    15
      case (r1s, r2s) => if (r1s == r2s) r1s else ALT(r1s, r2s)
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    16
    }
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    17
  }
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    18
  case SEQ(r1, r2) => {
341
ac1187b2e5c9 updated program
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 261
diff changeset
    19
    (simp(r1), simp(r2)) match {
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    20
      case (NULL, _) => NULL
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    21
      case (_, NULL) => NULL
341
ac1187b2e5c9 updated program
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 261
diff changeset
    22
      case (EMPTY, r2s) => r2s
ac1187b2e5c9 updated program
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 261
diff changeset
    23
      case (r1s, EMPTY) => r1s
ac1187b2e5c9 updated program
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 261
diff changeset
    24
      case (r1s, r2s) => SEQ(r1s, r2s)
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    25
    }
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    26
  }
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    27
  case NTIMES(r, n) => NTIMES(simp(r), n)    
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    28
  case r => r
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    29
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    30
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    31
def nullable (r: Rexp) : Boolean = r match {
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    32
  case NULL => false
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    33
  case EMPTY => true
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    34
  case CHAR(_) => false
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    35
  case ALT(r1, r2) => nullable(r1) || nullable(r2)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    36
  case SEQ(r1, r2) => nullable(r1) && nullable(r2)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    37
  case STAR(_) => true
59
Christian Urban <urbanc@in.tum.de>
parents: 49
diff changeset
    38
  case NTIMES(r, i) => if (i == 0) true else nullable(r)
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    39
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    40
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    41
def der (c: Char, r: Rexp) : Rexp = r match {
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    42
  case NULL => NULL
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    43
  case EMPTY => NULL
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    44
  case CHAR(d) => if (c == d) EMPTY else NULL
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    45
  case ALT(r1, r2) => ALT(der(c, r1), der(c, r2))
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    46
  case SEQ(r1, r2) => 
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    47
    if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2))
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    48
    else SEQ(der(c, r1), r2)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    49
  case STAR(r) => SEQ(der(c, r), STAR(r))
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    50
  case NTIMES(r, i) => 
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    51
    if (i == 0) NULL else SEQ(der(c, r), NTIMES(r, i - 1))
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    52
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    53
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    54
def ders (s: List[Char], r: Rexp) : Rexp = s match {
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    55
  case Nil => r
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    56
  case c::s => ders(s, simp(der(c, r)))
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    57
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    58
261
24531cfaa36a updated handouts
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 258
diff changeset
    59
def matches(r: Rexp, s: String) : Boolean = nullable(ders(s.toList, r))
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    60
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    61
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    62
//one or zero
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    63
def OPT(r: Rexp) = ALT(r, EMPTY)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    64
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    65
def EVIL(n: Int) = SEQ(NTIMES(OPT(CHAR('a')), n), NTIMES(CHAR('a'), n))
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    66
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    67
def time_needed[T](i: Int, code: => T) = {
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    68
  val start = System.nanoTime()
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    69
  for (j <- 1 to i) code
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    70
  val end = System.nanoTime()
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    71
  (end - start)/(i * 1.0e9)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    72
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    73
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    74
121
43c116860e47 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 119
diff changeset
    75
for (i <- 1 to 12001 by 500) {
261
24531cfaa36a updated handouts
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 258
diff changeset
    76
  println(i + " " + "%.5f".format(time_needed(1, matches(EVIL(i), "a" * i))))
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    77
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    78
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    79
363
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    80
// some convenience for typing in regular expressions
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    81
import scala.language.implicitConversions    
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    82
import scala.language.reflectiveCalls 
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    83
def charlist2rexp(s : List[Char]) : Rexp = s match {
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    84
  case Nil => EMPTY
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    85
  case c::Nil => CHAR(c)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    86
  case c::s => SEQ(CHAR(c), charlist2rexp(s))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    87
}
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    88
implicit def string2rexp(s : String) : Rexp = charlist2rexp(s.toList)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    89
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    90
implicit def RexpOps (r: Rexp) = new {
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    91
  def | (s: Rexp) = ALT(r, s)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    92
  def % = STAR(r)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    93
  def ~ (s: Rexp) = SEQ(r, s)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    94
}
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    95
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    96
implicit def stringOps (s: String) = new {
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    97
  def | (r: Rexp) = ALT(s, r)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    98
  def | (r: String) = ALT(s, r)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
    99
  def % = STAR(s)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   100
  def ~ (r: Rexp) = SEQ(s, r)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   101
  def ~ (r: String) = SEQ(s, r)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   102
}
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   103
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   104
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   105
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   106
def PLUS(r: Rexp) = SEQ(r, STAR(r))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   107
def RANGE(s: List[Char]) : Rexp = s match {
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   108
  case Nil => NULL
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   109
  case c::s => ALT(CHAR(c), RANGE(s))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   110
} 
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   111
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   112
println("EVILS:")
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   113
val EVIL1 = PLUS(PLUS("a" ~ "a" ~ "a"))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   114
val EVIL2 = PLUS(PLUS("aaaaaaaaaaaaaaaaaaa" ~ OPT("a")))  // 19 as ~ a?
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   115
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   116
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   117
//40 * aaa matches
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   118
//43 * aaa + aa does not
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   119
//45 * aaa + a
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   120
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   121
println("EVIL1:")
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   122
println(matches(EVIL1, "aaa" * 40))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   123
println(matches(EVIL1, "aaa" * 43 + "aa"))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   124
println(matches(EVIL1, "aaa" * 45 + "a"))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   125
println("EVIL2:")
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   126
println(matches(EVIL2, "aaa" * 40))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   127
println(matches(EVIL2, "aaa" * 43 + "aa"))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   128
println(matches(EVIL2, "aaa" * 45 + "a"))
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   129
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   130
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   131
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   132
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   133
println("EMAIL:")
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   134
val LOWERCASE = "abcdefghijklmnopqrstuvwxyz"
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   135
val DIGITS = "0123456789"
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   136
val EMAIL = PLUS(RANGE((LOWERCASE + DIGITS + "_" + "." + "-").toList)) ~ "@" ~ 
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   137
            PLUS(RANGE((LOWERCASE + DIGITS + "." + "-").toList)) ~ "." ~
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   138
            NMTIMES(RANGE((LOWERCASE + ".").toList), 2, 6)
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   139
0d6deecdb2eb updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 341
diff changeset
   140
val my_email = "christian.urban@kcl.ac.uk"