| author | Christian Urban <christian.urban@kcl.ac.uk> | 
| Mon, 21 Sep 2020 10:44:48 +0100 | |
| changeset 759 | 636ad159b4eb | 
| parent 742 | 155426396b5f | 
| permissions | -rw-r--r-- | 
| 623 | 1 | // A simple matcher for basic regular expressions | 
| 477 | 2 | |
| 54 | 3 | abstract class Rexp | 
| 471 | 4 | case object ZERO extends Rexp // matches nothing | 
| 623 | 5 | case object ONE extends Rexp // matches an empty string | 
| 471 | 6 | case class CHAR(c: Char) extends Rexp // matches a character c | 
| 7 | case class ALT(r1: Rexp, r2: Rexp) extends Rexp // alternative | |
| 8 | case class SEQ(r1: Rexp, r2: Rexp) extends Rexp // sequence | |
| 9 | case class STAR(r: Rexp) extends Rexp // star | |
| 49 | 10 | |
| 471 | 11 | // nullable function: tests whether a regular | 
| 49 | 12 | // expression can recognise the empty string | 
| 586 | 13 | def nullable(r: Rexp) : Boolean = r match {
 | 
| 424 
1129024b26d5
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
422diff
changeset | 14 | case ZERO => false | 
| 
1129024b26d5
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
422diff
changeset | 15 | case ONE => true | 
| 49 | 16 | case CHAR(_) => false | 
| 17 | case ALT(r1, r2) => nullable(r1) || nullable(r2) | |
| 18 | case SEQ(r1, r2) => nullable(r1) && nullable(r2) | |
| 19 | case STAR(_) => true | |
| 20 | } | |
| 21 | ||
| 623 | 22 | // the derivative of a regular expression w.r.t. a character | 
| 49 | 23 | def der (c: Char, r: Rexp) : Rexp = r match {
 | 
| 424 
1129024b26d5
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
422diff
changeset | 24 | case ZERO => ZERO | 
| 
1129024b26d5
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
422diff
changeset | 25 | case ONE => ZERO | 
| 
1129024b26d5
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
422diff
changeset | 26 | case CHAR(d) => if (c == d) ONE else ZERO | 
| 49 | 27 | case ALT(r1, r2) => ALT(der(c, r1), der(c, r2)) | 
| 28 | case SEQ(r1, r2) => | |
| 29 | if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2)) | |
| 30 | else SEQ(der(c, r1), r2) | |
| 440 
e14cd32ad497
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
434diff
changeset | 31 | case STAR(r1) => SEQ(der(c, r1), STAR(r1)) | 
| 49 | 32 | } | 
| 33 | ||
| 623 | 34 | // the derivative w.r.t. a string (iterates der) | 
| 49 | 35 | def ders (s: List[Char], r: Rexp) : Rexp = s match {
 | 
| 36 | case Nil => r | |
| 54 | 37 | case c::s => ders(s, der(c, r)) | 
| 49 | 38 | } | 
| 39 | ||
| 623 | 40 | // the main matcher function | 
| 631 | 41 | def matcher(r: Rexp, s: String) : Boolean = | 
| 586 | 42 | nullable(ders(s.toList, r)) | 
| 49 | 43 | |
| 477 | 44 | |
| 623 | 45 | // examples from the homework | 
| 477 | 46 | |
| 343 
539b2e88f5b9
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
261diff
changeset | 47 | val r = STAR(ALT(SEQ(CHAR('a'), CHAR('b')), CHAR('b')))
 | 
| 
539b2e88f5b9
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
261diff
changeset | 48 | der('a', r)
 | 
| 
539b2e88f5b9
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
261diff
changeset | 49 | der('b', r)
 | 
| 
539b2e88f5b9
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
261diff
changeset | 50 | der('c', r)
 | 
| 49 | 51 | |
| 550 | 52 | val r2 = SEQ(SEQ(CHAR('x'), CHAR('y')), CHAR('z'))
 | 
| 53 | der('x', r2)
 | |
| 54 | der('y', der('x', r2))
 | |
| 55 | der('z', der('y', der('x', r2)))
 | |
| 546 | 56 | |
| 57 | ||
| 623 | 58 | // the optional regular expression (one or zero times) | 
| 424 
1129024b26d5
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
422diff
changeset | 59 | def OPT(r: Rexp) = ALT(r, ONE) | 
| 49 | 60 | |
| 623 | 61 | // the n-times regular expression (explicitly expanded) | 
| 49 | 62 | def NTIMES(r: Rexp, n: Int) : Rexp = n match {
 | 
| 424 
1129024b26d5
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
422diff
changeset | 63 | case 0 => ONE | 
| 49 | 64 | case 1 => r | 
| 65 | case n => SEQ(r, NTIMES(r, n - 1)) | |
| 66 | } | |
| 67 | ||
| 477 | 68 | |
| 69 | // Test Cases | |
| 70 | ||
| 258 
1e4da6d2490c
updated programs
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
119diff
changeset | 71 | // the evil regular expression  a?{n} a{n}
 | 
| 415 | 72 | def EVIL1(n: Int) = SEQ(NTIMES(OPT(CHAR('a')), n), NTIMES(CHAR('a'), n))
 | 
| 422 | 73 | |
| 74 | // the evil regular expression (a*)*b | |
| 440 
e14cd32ad497
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
434diff
changeset | 75 | val EVIL2 = SEQ(STAR(STAR(CHAR('a'))), CHAR('b'))
 | 
| 49 | 76 | |
| 623 | 77 | // for measuring time | 
| 469 
226f294987f2
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
467diff
changeset | 78 | def time_needed[T](i: Int, code: => T) = {
 | 
| 49 | 79 | val start = System.nanoTime() | 
| 469 
226f294987f2
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
467diff
changeset | 80 | for (j <- 1 to i) code | 
| 49 | 81 | val end = System.nanoTime() | 
| 631 | 82 | (end - start) / (i * 1.0e9) | 
| 49 | 83 | } | 
| 84 | ||
| 480 | 85 | |
| 623 | 86 | // test: (a?{n}) (a{n})
 | 
| 556 | 87 | println("Test (a?{n}) (a{n})")
 | 
| 433 
c08290ee4f1f
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
424diff
changeset | 88 | |
| 623 | 89 | for (i <- 0 to 20 by 2) {
 | 
| 631 | 90 |   println(f"$i: ${time_needed(2, matcher(EVIL1(i), "a" * i))}%.5f")
 | 
| 49 | 91 | } | 
| 363 
0d6deecdb2eb
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
343diff
changeset | 92 | |
| 623 | 93 | // test: (a*)* b | 
| 556 | 94 | println("Test (a*)* b")
 | 
| 95 | ||
| 623 | 96 | for (i <- 0 to 20 by 2) {
 | 
| 631 | 97 |   println(f"$i: ${time_needed(2, matcher(EVIL2, "a" * i))}%.5f")
 | 
| 415 | 98 | } | 
| 363 
0d6deecdb2eb
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
343diff
changeset | 99 | |
| 477 | 100 | |
| 623 | 101 | // the size of a regular expressions - for testing purposes | 
| 477 | 102 | def size(r: Rexp) : Int = r match {
 | 
| 103 | case ZERO => 1 | |
| 104 | case ONE => 1 | |
| 105 | case CHAR(_) => 1 | |
| 106 | case ALT(r1, r2) => 1 + size(r1) + size(r2) | |
| 107 | case SEQ(r1, r2) => 1 + size(r1) + size(r2) | |
| 108 | case STAR(r) => 1 + size(r) | |
| 109 | } | |
| 110 | ||
| 111 | // the expicit expansion in EVIL1(n) increases | |
| 112 | // drastically its size | |
| 113 | ||
| 114 | size(EVIL1(1)) // 5 | |
| 115 | size(EVIL1(3)) // 17 | |
| 116 | size(EVIL1(5)) // 29 | |
| 117 | size(EVIL1(7)) // 41 | |
| 623 | 118 | size(EVIL1(20)) // 119 | 
| 479 | 119 | |
| 120 | // given a regular expression and building successive | |
| 121 | // derivatives might result into bigger and bigger | |
| 122 | // regular expressions...here is an example for this: | |
| 123 | ||
| 563 | 124 | // (a+b)* o a o b o (a+b)* | 
| 479 | 125 | val BIG_aux = STAR(ALT(CHAR('a'), CHAR('b')))
 | 
| 126 | val BIG = SEQ(BIG_aux, SEQ(CHAR('a'),SEQ(CHAR('b'), BIG_aux)))
 | |
| 127 | ||
| 128 | size(ders("".toList, BIG))              // 13
 | |
| 129 | size(ders("ab".toList, BIG))            // 51
 | |
| 130 | size(ders("abab".toList, BIG))          // 112
 | |
| 131 | size(ders("ababab".toList, BIG))        // 191
 | |
| 132 | size(ders("abababab".toList, BIG))      // 288
 | |
| 133 | size(ders("ababababab".toList, BIG))    // 403
 | |
| 134 | size(ders("abababababab".toList, BIG))  // 536
 | |
| 135 | ||
| 136 | ||
| 137 | size(ders(("ab" * 200).toList, BIG))    // 366808
 | |
| 138 | ||
| 623 | 139 | for (i <- 0 to 200 by 10) {
 | 
| 631 | 140 |   println(f"$i: ${time_needed(2, matcher(BIG, "ab" * i))}%.5f")
 | 
| 556 | 141 | } | 
| 647 | 142 | |
| 143 | ||
| 144 | ||
| 145 | ||
| 146 | ////////////////////////////////////// | |
| 147 | def concat(A: Set[String], B: Set[String]) : Set[String] = | |
| 148 | for (s1 <- A; s2 <- B) yield s1 ++ s2 | |
| 149 | ||
| 150 | ||
| 151 | val A = Set("foo", "bar")
 | |
| 152 | val B = Set("a", "b")
 | |
| 153 | ||
| 154 |