1 |
|
2 abstract class Rexp |
|
3 |
|
4 case object NULL extends Rexp |
|
5 case object EMPTY extends Rexp |
|
6 case class CHAR(c: Char) extends Rexp |
|
7 case class ALT(r1: Rexp, r2: Rexp) extends Rexp |
|
8 case class SEQ(r1: Rexp, r2: Rexp) extends Rexp |
|
9 case class STAR(r: Rexp) extends Rexp |
|
10 case class NOT(r: Rexp) extends Rexp |
|
11 case class REP(r: Rexp, n: Int) extends Rexp |
|
12 |
|
13 // some convenience for typing in regular expressions |
|
14 def charlist2rexp(s : List[Char]) : Rexp = s match { |
|
15 case Nil => EMPTY |
|
16 case c::Nil => CHAR(c) |
|
17 case c::s => SEQ(CHAR(c), charlist2rexp(s)) |
|
18 } |
|
19 implicit def string2rexp(s : String) : Rexp = charlist2rexp(s.toList) |
|
20 |
|
21 implicit def RexpOps(r: Rexp) = new { |
|
22 def | (s: Rexp) = ALT(r, s) |
|
23 def % = STAR(r) |
|
24 def %(n: Int) = REP(r, n) |
|
25 def %%(n: Int) = SEQ(REP(r, n), STAR(r)) |
|
26 def ? = ALT(EMPTY, r) |
|
27 def unary_! = NOT(r) |
|
28 def ~ (s: Rexp) = SEQ(r, s) |
|
29 } |
|
30 |
|
31 implicit def stringOps(s: String) = new { |
|
32 def | (r: Rexp) = ALT(s, r) |
|
33 def | (r: String) = ALT(s, r) |
|
34 def % = STAR(s) |
|
35 def %(n: Int) = REP(s, n) |
|
36 def %%(n: Int) = SEQ(REP(s, n), STAR(s)) |
|
37 def ? = ALT(EMPTY, s) |
|
38 def unary_! = NOT(s) |
|
39 def ~ (r: Rexp) = SEQ(s, r) |
|
40 def ~ (r: String) = SEQ(s, r) |
|
41 } |
|
42 |
|
43 |
|
44 // nullable function: tests whether the regular |
|
45 // expression can recognise the empty string |
|
46 def nullable (r: Rexp) : Boolean = r match { |
|
47 case NULL => false |
|
48 case EMPTY => true |
|
49 case CHAR(_) => false |
|
50 case ALT(r1, r2) => nullable(r1) || nullable(r2) |
|
51 case SEQ(r1, r2) => nullable(r1) && nullable(r2) |
|
52 case STAR(_) => true |
|
53 case NOT(r) => !(nullable(r)) |
|
54 case REP(r, i) => if (i == 0) true else nullable(r) |
|
55 } |
|
56 |
|
57 // derivative of a regular expression w.r.t. a character |
|
58 def der (c: Char, r: Rexp) : Rexp = r match { |
|
59 case NULL => NULL |
|
60 case EMPTY => NULL |
|
61 case CHAR(d) => if (c == d) EMPTY else NULL |
|
62 case ALT(r1, r2) => ALT(der(c, r1), der(c, r2)) |
|
63 case SEQ(r1, r2) => |
|
64 if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2)) |
|
65 else SEQ(der(c, r1), r2) |
|
66 case STAR(r) => SEQ(der(c, r), STAR(r)) |
|
67 case NOT(r) => NOT(der (c, r)) |
|
68 case REP(r, i) => |
|
69 if (i == 0) NULL else SEQ(der(c, r), REP(r, i - 1)) |
|
70 } |
|
71 |
|
72 // derivative w.r.t. a string (iterates der) |
|
73 def ders (s: List[Char], r: Rexp) : Rexp = s match { |
|
74 case Nil => r |
|
75 case c::s => ders(s, der(c, r)) |
|
76 } |
|
77 |
|
78 // main matcher function |
|
79 def matcher(r: Rexp, s: String) : Boolean = nullable(ders(s.toList, r)) |
|
80 |
|
81 //examples |
|
82 val digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" |
|
83 val int = ("+" | "-").? ~ digit.%%(1) |
|
84 val real = ("+" | "-").? ~ digit.%%(1) ~ ("." ~ digit.%%(1)).? ~ (("e" | "E") ~ ("+" | "-").? ~ digit.%%(1)).? |
|
85 |
|
86 val ints = List("0", "-4534", "+049", "99") |
|
87 val reals = List("0.9", "-12.8", "+91.0", "9e12", "+9.21E-12", "-512E+01") |
|
88 val errs = List("", "-", "+", "+-1", "-+2", "2-") |
|
89 |
|
90 ints.map(s => matcher(int, s)) |
|
91 reals.map(s => matcher(int, s)) |
|
92 errs.map(s => matcher(int, s)) |
|
93 |
|
94 ints.map(s => matcher(real, s)) |
|
95 reals.map(s => matcher(real, s)) |
|
96 errs.map(s => matcher(real, s)) |
|
97 |
|
98 |
|
99 |
|
100 def RTEST(n: Int) = ("a".? %(n)) ~ ("a" %(n)) |
|
101 |
|
102 def time_needed[T](i: Int, code: => T) = { |
|
103 val start = System.nanoTime() |
|
104 for (j <- 1 to i) code |
|
105 val end = System.nanoTime() |
|
106 (end - start)/(i * 1.0e9) |
|
107 } |
|
108 |
|
109 for (i <- 1 to 29) { |
|
110 println(i + ": " + "%.5f".format(time_needed(1, matcher(RTEST(i), "a" * i)))) |
|
111 } |
|
112 |
|
113 |
|