progs/re2.scala
author Christian Urban <christian dot urban at kcl dot ac dot uk>
Sun, 02 Oct 2016 15:07:21 +0100
changeset 434 8664ff87cd77
parent 424 1129024b26d5
child 440 e14cd32ad497
permissions -rw-r--r--
updated
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
422
5deefcc8cffa updated programs
Christian Urban <urbanc@in.tum.de>
parents: 415
diff changeset
     1
// version with explicit n-times regular expression
5deefcc8cffa updated programs
Christian Urban <urbanc@in.tum.de>
parents: 415
diff changeset
     2
120
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
     3
abstract class Rexp 
424
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
     4
case object ZERO extends Rexp
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
     5
case object ONE extends Rexp
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     6
case class CHAR(c: Char) extends Rexp
120
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
     7
case class ALT(r1: Rexp, r2: Rexp) extends Rexp 
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
     8
case class SEQ(r1: Rexp, r2: Rexp) extends Rexp 
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
     9
case class STAR(r: Rexp) extends Rexp 
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    10
case class NTIMES(r: Rexp, n: Int) extends Rexp //explicit constructor
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    11
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    12
def nullable (r: Rexp) : Boolean = r match {
424
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
    13
  case ZERO => false
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
    14
  case ONE => true
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    15
  case CHAR(_) => false
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    16
  case ALT(r1, r2) => nullable(r1) || nullable(r2)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    17
  case SEQ(r1, r2) => nullable(r1) && nullable(r2)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    18
  case STAR(_) => true
120
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
    19
  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
    20
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    21
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    22
def der (c: Char, r: Rexp) : Rexp = r match {
424
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
    23
  case ZERO => ZERO
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
    24
  case ONE => ZERO
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
    25
  case CHAR(d) => if (c == d) ONE else ZERO
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    26
  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
    27
  case SEQ(r1, r2) => 
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    28
    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
    29
    else SEQ(der(c, r1), r2)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    30
  case STAR(r) => SEQ(der(c, r), STAR(r))
120
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
    31
  case NTIMES(r, i) => 
424
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
    32
    if (i == 0) ZERO else SEQ(der(c, r), NTIMES(r, i - 1))
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    33
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    34
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    35
def ders (s: List[Char], r: Rexp) : Rexp = s match {
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    36
  case Nil => r
120
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
    37
  case c::s => ders(s, der(c, r))
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    38
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    39
261
24531cfaa36a updated handouts
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 258
diff changeset
    40
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
    41
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    42
258
1e4da6d2490c updated programs
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 121
diff changeset
    43
//optional: one or zero times
424
1129024b26d5 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 422
diff changeset
    44
def OPT(r: Rexp) = ALT(r, ONE)
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    45
422
5deefcc8cffa updated programs
Christian Urban <urbanc@in.tum.de>
parents: 415
diff changeset
    46
//evil regular expressions
414
065ca01b62ae updated
Christian Urban <urbanc@in.tum.de>
parents: 343
diff changeset
    47
def EVIL1(n: Int) = SEQ(NTIMES(OPT(CHAR('a')), n), NTIMES(CHAR('a'), n))
065ca01b62ae updated
Christian Urban <urbanc@in.tum.de>
parents: 343
diff changeset
    48
val EVIL2 = SEQ(STAR(CHAR('a')), CHAR('b'))
49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    49
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    50
def time_needed[T](i: Int, code: => T) = {
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    51
  val start = System.nanoTime()
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    52
  for (j <- 1 to i) code
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    53
  val end = System.nanoTime()
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    54
  (end - start)/(i * 1.0e9)
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    55
}
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    56
d2c6852ca8da added programs and slides
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    57
434
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    58
//test: (a?{n}) (a{n})
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    59
for (i <- 1 to 1101 by 100) {
415
4ae59fd3b174 updated
Christian Urban <urbanc@in.tum.de>
parents: 414
diff changeset
    60
  println(i + " " + "%.5f".format(time_needed(2, matches(EVIL1(i), "a" * i))))
120
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
    61
}
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
    62
434
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    63
for (i <- 1 to 1101 by 100) {
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    64
  println(i + " " + "%.5f".format(time_needed(2, matches(EVIL1(i), "a" * i))))
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    65
}
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    66
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    67
//test: (a*)* b
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    68
for (i <- 1 to 4501 by 500) {
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    69
  println(i + " " + "%.5f".format(time_needed(2, matches(EVIL2, "a" * i))))
8664ff87cd77 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 424
diff changeset
    70
}
120
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
    71
415
4ae59fd3b174 updated
Christian Urban <urbanc@in.tum.de>
parents: 414
diff changeset
    72
for (i <- 1 to 4501 by 500) {
4ae59fd3b174 updated
Christian Urban <urbanc@in.tum.de>
parents: 414
diff changeset
    73
  println(i + " " + "%.5f".format(time_needed(2, matches(EVIL2, "a" * i))))
414
065ca01b62ae updated
Christian Urban <urbanc@in.tum.de>
parents: 343
diff changeset
    74
}
065ca01b62ae updated
Christian Urban <urbanc@in.tum.de>
parents: 343
diff changeset
    75
065ca01b62ae updated
Christian Urban <urbanc@in.tum.de>
parents: 343
diff changeset
    76