| 928 |      1 | // A version of the simple matcher with simplification 
 | 
| 919 |      2 | // of derivatives
 | 
|  |      3 | //
 | 
| 928 |      4 | // this keeps the regular expressions (often) small, which
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      5 | // is good for the run-time
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      6 | // 
 | 
| 928 |      7 | // call the test cases with X = {1,2,3,4}
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      8 | //
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      9 | //   amm re3.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 re3.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 
 | 
| 
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 | // the nullable function: tests whether the regular 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     27 | // expression can recognise the empty string
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     28 | def nullable (r: Rexp) : Boolean = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     29 |   case ZERO => false
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     30 |   case ONE => true
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     31 |   case CHAR(_) => false
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     32 |   case ALT(r1, r2) => nullable(r1) || nullable(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     33 |   case SEQ(r1, r2) => nullable(r1) && nullable(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     34 |   case STAR(_) => true
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     35 |   case NTIMES(r, i) => if (i == 0) true else nullable(r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     36 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     37 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     38 | // the derivative of a regular expression w.r.t. a character
 | 
| 825 |     39 | def der(c: Char, r: Rexp) : Rexp = r match {
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     40 |   case ZERO => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     41 |   case ONE => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     42 |   case CHAR(d) => if (c == d) ONE else ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     43 |   case ALT(r1, r2) => ALT(der(c, r1), der(c, r2))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     44 |   case SEQ(r1, r2) => 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     45 |     if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     46 |     else SEQ(der(c, r1), r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     47 |   case STAR(r1) => SEQ(der(c, r1), STAR(r1))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     48 |   case NTIMES(r, i) => 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     49 |     if (i == 0) ZERO else SEQ(der(c, r), NTIMES(r, i - 1))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     50 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     51 | 
 | 
| 919 |     52 | // simplification
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     53 | def simp(r: Rexp) : Rexp = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     54 |   case ALT(r1, r2) => (simp(r1), simp(r2)) match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     55 |     case (ZERO, r2s) => r2s
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     56 |     case (r1s, ZERO) => r1s
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     57 |     case (r1s, r2s) => if (r1s == r2s) r1s else ALT (r1s, r2s)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     58 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     59 |   case SEQ(r1, r2) =>  (simp(r1), simp(r2)) match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     60 |     case (ZERO, _) => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     61 |     case (_, ZERO) => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     62 |     case (ONE, r2s) => r2s
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     63 |     case (r1s, ONE) => r1s
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     64 |     case (r1s, r2s) => SEQ(r1s, r2s)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     65 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     66 |   case r => r
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     67 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     68 | 
 | 
| 928 |     69 | // the derivative w.r.t. a string (iterates der and simp)
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     70 | def ders(s: List[Char], r: Rexp) : Rexp = s match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     71 |   case Nil => r
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     72 |   case c::s => ders(s, simp(der(c, r)))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     73 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     74 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     75 | // the main matcher function
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     76 | def matcher(r: Rexp, s: String) : Boolean = 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     77 |   nullable(ders(s.toList, r))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     78 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     79 | 
 | 
| 919 |     80 | // Test Cases
 | 
|  |     81 | //============
 | 
|  |     82 | 
 | 
| 928 |     83 | // optional is still just defined
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     84 | def OPT(r: Rexp) = ALT(r, ONE)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     85 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     86 | // evil regular expressions: (a?){n} a{n}  and (a*)* b
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     87 | def EVIL1(n: Int) = SEQ(NTIMES(OPT(CHAR('a')), n), NTIMES(CHAR('a'), n))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     88 | val EVIL2 = SEQ(STAR(STAR(CHAR('a'))), CHAR('b'))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     89 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     90 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     91 | def time_needed[T](i: Int, code: => T) = {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     92 |   val start = System.nanoTime()
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     93 |   for (j <- 1 to i) code
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     94 |   val end = System.nanoTime()
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     95 |   (end - start)/(i * 1.0e9)
 | 
| 
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 | 
 | 
| 826 |     99 | @arg(doc = "Test (a?{n}) (a{n})")
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    100 | @main
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    101 | def test1() = {
 | 
| 742 |    102 |   println("Test (a?{n}) (a{n})")
 | 
|  |    103 | 
 | 
| 769 |    104 |   for (i <- 0 to 9000 by 1000) {
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    105 |     println(f"$i: ${time_needed(3, matcher(EVIL1(i), "a" * i))}%.5f")
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    106 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    107 | }  
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    108 | 
 | 
| 826 |    109 | @arg(doc = "Test (a*)* b")
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    110 | @main
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    111 | def test2() = {
 | 
| 742 |    112 |   println("Test (a*)* b")
 | 
|  |    113 | 
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    114 |   for (i <- 0 to 6000000 by 500000) {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    115 |     println(f"$i: ${time_needed(3, matcher(EVIL2, "a" * i))}%.5f")
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    116 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    117 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    118 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    119 | // size of a regular expressions - for testing purposes 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    120 | def size(r: Rexp) : Int = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    121 |   case ZERO => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    122 |   case ONE => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    123 |   case CHAR(_) => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    124 |   case ALT(r1, r2) => 1 + size(r1) + size(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    125 |   case SEQ(r1, r2) => 1 + size(r1) + size(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    126 |   case STAR(r) => 1 + size(r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    127 |   case NTIMES(r, _) => 1 + size(r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    128 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    129 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    130 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    131 | // now the size of the derivatives grows 
 | 
| 928 |    132 | // much, much slower and actually maxes out
 | 
|  |    133 | // at size 8
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    134 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    135 | size(ders("".toList, EVIL2))      // 5
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    136 | size(ders("a".toList, EVIL2))     // 8
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    137 | size(ders("aa".toList, EVIL2))    // 8
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    138 | size(ders("aaa".toList, EVIL2))   // 8
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    139 | size(ders("aaaa".toList, EVIL2))  // 8
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    140 | size(ders("aaaaa".toList, EVIL2)) // 8
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    141 | 
 | 
| 928 |    142 | size(ders(("a" * 20000).toList, EVIL2)) // 8
 | 
|  |    143 | 
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    144 | 
 | 
| 826 |    145 | @arg(doc = "All tests.")
 | 
| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    146 | @main
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    147 | def all() = { test1(); test2() } 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    148 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    149 | 
 | 
| 782 |    150 | 
 | 
|  |    151 | 
 | 
|  |    152 | 
 | 
| 826 |    153 | // PS:
 | 
|  |    154 | //
 | 
|  |    155 | // If you want to dig deeper into the topic, you can have 
 | 
|  |    156 | // a look at these examples which still explode. They
 | 
|  |    157 | // need an even more aggressive simplification.
 | 
|  |    158 | 
 | 
| 782 |    159 | // test: (a + aa)*
 | 
|  |    160 | val EVIL3 = STAR(ALT(CHAR('a'), SEQ(CHAR('a'), CHAR('a'))))
 | 
|  |    161 | 
 | 
|  |    162 | 
 | 
| 826 |    163 | @arg(doc = "Test EVIL3")
 | 
| 782 |    164 | @main
 | 
|  |    165 | def test3() = {
 | 
|  |    166 |   println("Test (a + aa)*")
 | 
|  |    167 | 
 | 
| 826 |    168 |   for (i <- 0 to 30 by 5) {
 | 
| 782 |    169 |     println(f"$i: ${time_needed(1, matcher(EVIL3, "a" * i))}%.5f")
 | 
|  |    170 |   }
 | 
|  |    171 | }
 | 
|  |    172 | 
 | 
| 826 |    173 | 
 | 
|  |    174 | // test: (1 + a + aa)*
 | 
|  |    175 | val EVIL4 = STAR(ALT(ONE, ALT(CHAR('a'), SEQ(CHAR('a'), CHAR('a')))))
 | 
|  |    176 | 
 | 
|  |    177 | @arg(doc = "Test EVIL4")
 | 
| 782 |    178 | @main
 | 
|  |    179 | def test4() = {
 | 
|  |    180 |   println("Test (1 + a + aa)*")
 | 
|  |    181 | 
 | 
| 826 |    182 |   for (i <- 0 to 30 by 5) {
 | 
| 782 |    183 |     println(f"$i: ${time_needed(1, matcher(EVIL4, "a" * i))}%.5f")
 | 
|  |    184 |   }
 | 
| 825 |    185 | }
 | 
| 826 |    186 | 
 | 
|  |    187 | @arg(doc = "Tests that show not all is hunky-dory, but a solution leads too far afield.")
 | 
|  |    188 | @main
 | 
|  |    189 | def fail() = { test3(); test4() } 
 | 
|  |    190 | 
 | 
|  |    191 | 
 |