| 725 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      1 | // (ASIDE!) A version which attempts to move whole strings, 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      2 | // not just characters, under derivatives whenever possible
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      3 | //    
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      4 | // call the test cases with X = {1,2}
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      5 | //
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      6 | //   amm re3.sc testX
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      7 | //
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      8 | // or 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |      9 | //
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     10 | //   amm re3.sc all
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     11 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     12 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     13 | abstract class Rexp
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     14 | case object ZERO extends Rexp
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     15 | case object ONE extends Rexp
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     16 | case class CHAR(c: Char) extends Rexp
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     17 | case class ALT(r1: Rexp, r2: Rexp) extends Rexp 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     18 | case class SEQ(r1: Rexp, r2: Rexp) extends Rexp 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     19 | case class STAR(r: Rexp) extends Rexp 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     20 | case class NTIMES(r: Rexp, n: Int) extends Rexp 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     21 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     22 | // the nullable function: tests whether the regular 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     23 | // expression can recognise the empty string
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     24 | def nullable (r: Rexp) : Boolean = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     25 |   case ZERO => false
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     26 |   case ONE => true
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     27 |   case CHAR(_) => false
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     28 |   case ALT(r1, r2) => nullable(r1) || nullable(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     29 |   case SEQ(r1, r2) => nullable(r1) && nullable(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     30 |   case STAR(_) => true
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     31 |   case NTIMES(r, i) => if (i == 0) true else nullable(r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     32 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     33 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     34 | // the derivative of a regular expression w.r.t. a character
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     35 | def der (c: Char, r: Rexp) : Rexp = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     36 |   case ZERO => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     37 |   case ONE => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     38 |   case CHAR(d) => if (c == d) ONE else ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     39 |   case ALT(r1, r2) => ALT(der(c, r1), der(c, r2))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     40 |   case SEQ(r1, r2) => 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     41 |     if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     42 |     else SEQ(der(c, r1), r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     43 |   case STAR(r1) => SEQ(der(c, r1), STAR(r1))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     44 |   case NTIMES(r1, i) => 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     45 |     if (i == 0) ZERO else SEQ(der(c, r1), NTIMES(r1, i - 1))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     46 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     47 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     48 | def simp(r: Rexp) : Rexp = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     49 |   case ALT(r1, r2) => (simp(r1), simp(r2)) match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     50 |     case (ZERO, r2s) => r2s
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     51 |     case (r1s, ZERO) => r1s
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     52 |     case (r1s, r2s) => if (r1s == r2s) r1s else ALT (r1s, r2s)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     53 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     54 |   case SEQ(r1, r2) =>  (simp(r1), simp(r2)) match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     55 |     case (ZERO, _) => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     56 |     case (_, ZERO) => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     57 |     case (ONE, r2s) => r2s
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     58 |     case (r1s, ONE) => r1s
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     59 |     case (r1s, r2s) => SEQ(r1s, r2s)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     60 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     61 |   case r => r
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     62 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     63 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     64 | // an example
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     65 | val r = SEQ(SEQ(CHAR('x'), CHAR('y')), CHAR('z'))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     66 | der('x', r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     67 | der('y', der('x', r))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     68 | der('z', der('y', der('x', r)))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     69 | simp(der('z', der('y', der('x', r))))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     70 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     71 | // *new*
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     72 | // the derivative w.r.t. a string (iterates der)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     73 | def ders2(s: List[Char], r: Rexp) : Rexp = (s, r) match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     74 |   case (Nil, r) => r
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     75 |   case (s, ZERO) => ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     76 |   case (s, ONE) => if (s == Nil) ONE else ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     77 |   case (s, CHAR(c)) => if (s == List(c)) ONE else 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     78 |                        if (s == Nil) CHAR(c) else ZERO
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     79 |   case (s, ALT(r1, r2)) => ALT(ders2(s, r1), ders2(s, r2))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     80 |   case (c::s, r) => ders2(s, simp(der(c, r)))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     81 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     82 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     83 | // the main matcher function
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     84 | def matcher(r: Rexp, s: String) : Boolean = 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     85 |   nullable(ders2(s.toList, r))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     86 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     87 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     88 | // one or zero
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     89 | def OPT(r: Rexp) = ALT(r, ONE)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     90 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     91 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     92 | // Test Cases
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     93 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     94 | def EVIL1(n: Int) = SEQ(NTIMES(OPT(CHAR('a')), n), NTIMES(CHAR('a'), n))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     95 | val EVIL2 = SEQ(STAR(STAR(CHAR('a'))), CHAR('b'))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     96 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     97 | // for measuring time
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     98 | def time_needed[T](i: Int, code: => T) = {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |     99 |   val start = System.nanoTime()
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    100 |   for (j <- 1 to i) code
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    101 |   val end = System.nanoTime()
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    102 |   (end - start) / (i * 1.0e9)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    103 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    104 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    105 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    106 | // test: (a?{n}) (a{n})
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    107 | @doc("Test (a?{n}) (a{n})")
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    108 | @main
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    109 | def test1() = {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    110 |   for (i <- 0 to 11000 by 1000) {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    111 |     println(f"$i: ${time_needed(2, matcher(EVIL1(i), "a" * i))}%.5f")
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    112 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    113 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    114 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    115 | // test: (a*)* b
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    116 | @doc("Test (a*)* b")
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    117 | @main
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    118 | def test2() = {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    119 |   for (i <- 0 to 7000000 by 500000) {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    120 |     println(f"$i: ${time_needed(2, matcher(EVIL2, "a" * i))}%.5f")
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    121 |   }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    122 | } 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    123 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    124 | @doc("All tests.")
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    125 | @main
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    126 | def all() = { test1(); test2() } 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    127 | 
 | 
| 
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 | // the size of a regular expressions - for testing purposes 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    131 | def size(r: Rexp) : Int = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    132 |   case ZERO => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    133 |   case ONE => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    134 |   case CHAR(_) => 1
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    135 |   case ALT(r1, r2) => 1 + size(r1) + size(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    136 |   case SEQ(r1, r2) => 1 + size(r1) + size(r2)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    137 |   case STAR(r) => 1 + size(r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    138 |   case NTIMES(r, _) => 1 + size(r)
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    139 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    140 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    141 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    142 | // string of a regular expressions - for testing purposes 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    143 | def string(r: Rexp) : String = r match {
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    144 |   case ZERO => "0"
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    145 |   case ONE => "1"
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    146 |   case CHAR(c) => c.toString 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    147 |   case ALT(r1, r2) => s"(${string(r1)} + ${string(r2)})"
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    148 |   case SEQ(CHAR(c), CHAR(d)) => s"${c}${d}"
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    149 |   case SEQ(r1, r2) => s"(${string(r1)} ~ ${string(r2)})"
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    150 |   case STAR(r) => s"(${string(r)})*"
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    151 |   case NTIMES(r, n) =>  s"(${string(r)}){${n}}"
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    152 | }
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    153 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    154 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    155 | // further testcases:
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    156 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    157 | // test: ("a" | "aa")*
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    158 | val EVIL3 = STAR(ALT(CHAR('a'), SEQ(CHAR('a'), CHAR('a'))))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    159 | 
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    160 | // test: ("" | "a" | "aa")*
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    161 | val EVIL4 = STAR(ALT(ONE, ALT(CHAR('a'), SEQ(CHAR('a'), CHAR('a')))))
 | 
| 
Christian Urban <christian.urban@kcl.ac.uk> parents: diff
changeset |    162 | 
 |