|
49
|
1 |
|
|
|
2 |
abstract class Rexp {
|
|
|
3 |
def simp : Rexp = this
|
|
|
4 |
}
|
|
|
5 |
|
|
|
6 |
case object NULL extends Rexp
|
|
|
7 |
case object EMPTY extends Rexp
|
|
|
8 |
case class CHAR(c: Char) extends Rexp
|
|
|
9 |
case class ALT(r1: Rexp, r2: Rexp) extends Rexp {
|
|
|
10 |
override def simp = (r1.simp, r2.simp) match {
|
|
|
11 |
case (NULL, r) => r
|
|
|
12 |
case (r, NULL) => r
|
|
|
13 |
case (r, EMPTY) => if (nullable(r)) r else ALT(r, EMPTY)
|
|
|
14 |
case (EMPTY, r) => if (nullable(r)) r else ALT(r, EMPTY)
|
|
|
15 |
case (r1, r2) => if (r1 == r2) r1 else ALT(r1, r2)
|
|
|
16 |
}
|
|
|
17 |
}
|
|
|
18 |
case class SEQ(r1: Rexp, r2: Rexp) extends Rexp {
|
|
|
19 |
override def simp = (r1.simp, r2.simp) match {
|
|
|
20 |
case (NULL, _) => NULL
|
|
|
21 |
case (_, NULL) => NULL
|
|
|
22 |
case (EMPTY, r) => r
|
|
|
23 |
case (r, EMPTY) => r
|
|
|
24 |
case (r1, r2) => SEQ(r1, r2)
|
|
|
25 |
}
|
|
|
26 |
}
|
|
|
27 |
case class STAR(r: Rexp) extends Rexp {
|
|
|
28 |
override def simp = r.simp match {
|
|
|
29 |
case NULL => EMPTY
|
|
|
30 |
case EMPTY => EMPTY
|
|
|
31 |
case r => STAR(r)
|
|
|
32 |
}
|
|
|
33 |
}
|
|
|
34 |
case class NTIMES(r: Rexp, n: Int) extends Rexp {
|
|
|
35 |
override def simp = r.simp match {
|
|
|
36 |
case NULL => NULL
|
|
|
37 |
case EMPTY => EMPTY
|
|
|
38 |
case r => NTIMES(r, n)
|
|
|
39 |
}
|
|
|
40 |
}
|
|
|
41 |
|
|
|
42 |
// some convenience for typing in regular expressions
|
|
|
43 |
def charlist2rexp(s : List[Char]) : Rexp = s match {
|
|
|
44 |
case Nil => EMPTY
|
|
|
45 |
case c::Nil => CHAR(c)
|
|
|
46 |
case c::s => SEQ(CHAR(c), charlist2rexp(s))
|
|
|
47 |
}
|
|
|
48 |
implicit def string2rexp(s : String) : Rexp = charlist2rexp(s.toList)
|
|
|
49 |
|
|
|
50 |
|
|
|
51 |
// nullable function: tests whether the regular
|
|
|
52 |
// expression can recognise the empty string
|
|
|
53 |
def nullable (r: Rexp) : Boolean = r match {
|
|
|
54 |
case NULL => false
|
|
|
55 |
case EMPTY => true
|
|
|
56 |
case CHAR(_) => false
|
|
|
57 |
case ALT(r1, r2) => nullable(r1) || nullable(r2)
|
|
|
58 |
case SEQ(r1, r2) => nullable(r1) && nullable(r2)
|
|
|
59 |
case STAR(_) => true
|
|
|
60 |
case NTIMES(r, i) => if (i == 0) false else nullable(r)
|
|
|
61 |
}
|
|
|
62 |
|
|
|
63 |
// derivative of a regular expression w.r.t. a character
|
|
|
64 |
def der (c: Char, r: Rexp) : Rexp = r match {
|
|
|
65 |
case NULL => NULL
|
|
|
66 |
case EMPTY => NULL
|
|
|
67 |
case CHAR(d) => if (c == d) EMPTY else NULL
|
|
|
68 |
case ALT(r1, r2) => ALT(der(c, r1), der(c, r2))
|
|
|
69 |
case SEQ(r1, r2) =>
|
|
|
70 |
if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2))
|
|
|
71 |
else SEQ(der(c, r1), r2)
|
|
|
72 |
case STAR(r) => SEQ(der(c, r), STAR(r))
|
|
|
73 |
case NTIMES(r, i) =>
|
|
|
74 |
if (i == 0) NULL else SEQ(der(c, r), NTIMES(r, i - 1))
|
|
|
75 |
}
|
|
|
76 |
|
|
|
77 |
// derivative w.r.t. a string (iterates der)
|
|
|
78 |
def ders (s: List[Char], r: Rexp) : Rexp = s match {
|
|
|
79 |
case Nil => r
|
|
|
80 |
case c::s => ders(s, der(c, r).simp)
|
|
|
81 |
}
|
|
|
82 |
|
|
|
83 |
// main matcher function
|
|
|
84 |
def matcher(r: Rexp, s: String) : Boolean = nullable(ders(s.toList, r))
|
|
|
85 |
|
|
|
86 |
|
|
|
87 |
|
|
|
88 |
//one or zero
|
|
|
89 |
def OPT(r: Rexp) = ALT(r, EMPTY)
|
|
|
90 |
|
|
|
91 |
//n-times
|
|
|
92 |
/*def NTIMES(r: Rexp, n: Int) : Rexp = n match {
|
|
|
93 |
case 0 => NULL
|
|
|
94 |
case 1 => r
|
|
|
95 |
case n => SEQ(r, NTIMES(r, n - 1))
|
|
|
96 |
}*/
|
|
|
97 |
|
|
|
98 |
def RTEST(n: Int) = SEQ(NTIMES(OPT("a"), n), NTIMES("a", n))
|
|
|
99 |
|
|
|
100 |
def time_needed[T](i: Int, code: => T) = {
|
|
|
101 |
val start = System.nanoTime()
|
|
|
102 |
for (j <- 1 to i) code
|
|
|
103 |
val end = System.nanoTime()
|
|
|
104 |
(end - start)/(i * 1.0e9)
|
|
|
105 |
}
|
|
|
106 |
|
|
|
107 |
|
|
|
108 |
for (i <- 1 to 10001 by 500) {
|
|
|
109 |
println(i + " " + time_needed(1, matcher(RTEST(i), "a" * i)))
|
|
|
110 |
}
|
|
|
111 |
|
|
|
112 |
|