| 929 |      1 | // A Version of the simple matcher with an explicit 
 | 
| 919 |      2 | // n-times regular expression
 | 
|  |      3 | //
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      4 | // this keeps the size of the regular expression in the
 | 
| 919 |      5 | // EVIL1 testcase small
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      6 | //
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      7 | // call the test cases with X = {1,2}
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      8 | //
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      9 | //   amm re2.sc testX
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     10 | //
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     11 | // or 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     12 | //
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     13 | //   amm re2.sc all
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     14 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     15 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     16 | abstract class Rexp 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     17 | case object ZERO extends Rexp
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     18 | case object ONE extends Rexp
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     19 | case class CHAR(c: Char) extends Rexp
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     20 | case class ALT(r1: Rexp, r2: Rexp) extends Rexp 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     21 | case class SEQ(r1: Rexp, r2: Rexp) extends Rexp 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     22 | case class STAR(r: Rexp) extends Rexp 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     23 | case class NTIMES(r: Rexp, n: Int) extends Rexp   //explicit constructor for n-times
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     24 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     25 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     26 | def nullable (r: Rexp) : Boolean = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     27 |   case ZERO => false
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     28 |   case ONE => true
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     29 |   case CHAR(_) => false
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     30 |   case ALT(r1, r2) => nullable(r1) || nullable(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     31 |   case SEQ(r1, r2) => nullable(r1) && nullable(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     32 |   case STAR(_) => true
 | 
| 769 |     33 |   case NTIMES(r, n) => if (n == 0) true else nullable(r)
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     34 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     35 | 
 | 
| 825 |     36 | def der(c: Char, r: Rexp) : Rexp = r match {
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     37 |   case ZERO => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     38 |   case ONE => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     39 |   case CHAR(d) => if (c == d) ONE else ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     40 |   case ALT(r1, r2) => ALT(der(c, r1), der(c, r2))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     41 |   case SEQ(r1, r2) => 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     42 |     if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     43 |     else SEQ(der(c, r1), r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     44 |   case STAR(r1) => SEQ(der(c, r1), STAR(r1))
 | 
| 769 |     45 |   case NTIMES(r, n) => 
 | 
|  |     46 |     if (n == 0) ZERO else SEQ(der(c, r), NTIMES(r, n - 1))
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     47 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     48 | 
 | 
| 825 |     49 | def ders(s: List[Char], r: Rexp) : Rexp = s match {
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     50 |   case Nil => r
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     51 |   case c::s => ders(s, der(c, r))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     52 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     53 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     54 | def matcher(r: Rexp, s: String) : Boolean = 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     55 |   nullable(ders(s.toList, r))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     56 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     57 | 
 | 
| 919 |     58 | // Test Cases
 | 
|  |     59 | 
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     60 | // the optional regular expression: one or zero times
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     61 | // this regular expression is still defined in terms of ALT
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     62 | def OPT(r: Rexp) = ALT(r, ONE)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     63 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     64 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     65 | // evil regular expressions
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     66 | def EVIL1(n: Int) = SEQ(NTIMES(OPT(CHAR('a')), n), NTIMES(CHAR('a'), n))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     67 | val EVIL2 = SEQ(STAR(STAR(CHAR('a'))), CHAR('b'))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     68 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     69 | def time_needed[T](i: Int, code: => T) = {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     70 |   val start = System.nanoTime()
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     71 |   for (j <- 1 to i) code
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     72 |   val end = System.nanoTime()
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     73 |   (end - start) / (i * 1.0e9)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     74 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     75 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     76 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     77 | 
 | 
| 826 |     78 | @arg(doc = "Test (a?{n}) (a{n})")
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     79 | @main
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     80 | def test1() = {
 | 
| 742 |     81 |   println("Test (a?{n}) (a{n})")
 | 
|  |     82 | 
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     83 |   for (i <- 0 to 1000 by 100) {
 | 
| 769 |     84 |     println(f"$i: ${time_needed(1, matcher(EVIL1(i), "a" * i))}%.5f")
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     85 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     86 | }  
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     87 | 
 | 
| 742 |     88 | 
 | 
| 826 |     89 | @arg(doc = "Test (a*)* b")
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     90 | @main
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     91 | def test2() = {
 | 
| 742 |     92 |   println("Test (a*)* b")
 | 
|  |     93 | 
 | 
| 929 |     94 |   for (i <- 0 to 20 by 2) {
 | 
| 769 |     95 |     println(f"$i: ${time_needed(1, matcher(EVIL2, "a" * i))}%.5f")
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     96 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     97 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     98 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     99 | // the size of a regular expressions - for testing purposes 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    100 | def size(r: Rexp) : Int = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    101 |   case ZERO => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    102 |   case ONE => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    103 |   case CHAR(_) => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    104 |   case ALT(r1, r2) => 1 + size(r1) + size(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    105 |   case SEQ(r1, r2) => 1 + size(r1) + size(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    106 |   case STAR(r) => 1 + size(r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    107 |   case NTIMES(r, _) => 1 + size(r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    108 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    109 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    110 | // EVIL1(n) has now a constant size, no matter
 | 
| 929 |    111 | // what n is
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    112 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    113 | size(EVIL1(1))  // 7
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    114 | size(EVIL1(3))  // 7
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    115 | size(EVIL1(5))  // 7
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    116 | size(EVIL1(7))  // 7
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    117 | size(EVIL1(20)) // 7
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    118 | 
 | 
| 929 |    119 | // also the derivative only grows much more 
 | 
|  |    120 | // moderately and actually maxes out at size 211
 | 
|  |    121 | // (for n = 5)
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    122 | size(ders("".toList, EVIL1(5)))       // 7
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    123 | size(ders("a".toList, EVIL1(5)))      // 16
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    124 | size(ders("aa".toList, EVIL1(5)))     // 35
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    125 | size(ders("aaa".toList, EVIL1(5)))    // 59
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    126 | size(ders("aaaa".toList, EVIL1(5)))   // 88
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    127 | size(ders("aaaaa".toList, EVIL1(5)))  // 122
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    128 | size(ders("aaaaaa".toList, EVIL1(5))) // 151
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    129 | 
 | 
| 929 |    130 | size(ders(("a" * 20).toList, EVIL1(5)))   // 211
 | 
|  |    131 | size(ders(("a" * 200).toList, EVIL1(5)))  // 211
 | 
|  |    132 | size(ders(("a" * 2000).toList, EVIL1(5))) // 211
 | 
| 769 |    133 | 
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    134 | // but the size of the derivatives can still grow 
 | 
| 769 |    135 | // quite dramatically in case of EVIL2:  (a*)* b
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    136 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    137 | size(ders("".toList, EVIL2))       // 5
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    138 | size(ders("a".toList, EVIL2))      // 12
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    139 | size(ders("aa".toList, EVIL2))     // 28
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    140 | size(ders("aaa".toList, EVIL2))    // 58
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    141 | size(ders("aaaa".toList, EVIL2))   // 116
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    142 | size(ders("aaaaa".toList, EVIL2))  // 230
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    143 | size(ders("aaaaaa".toList, EVIL2)) // 456
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    144 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    145 | size(ders(("a" * 20).toList, EVIL2)) // 7340068
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    146 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    147 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    148 | 
 | 
| 826 |    149 | @arg(doc = "All tests.")
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    150 | @main
 | 
| 825 |    151 | def all() = { test1(); test2() } 
 | 
| 826 |    152 | 
 |