| 153 |      1 | // Part 1 about Regular Expression Matching
 | 
|  |      2 | //==========================================
 | 
|  |      3 | 
 | 
| 249 |      4 | //object CW9a {
 | 
|  |      5 | 
 | 
| 221 |      6 | // Regular Expressions
 | 
| 153 |      7 | abstract class Rexp
 | 
|  |      8 | case object ZERO extends Rexp
 | 
|  |      9 | case object ONE extends Rexp
 | 
|  |     10 | case class CHAR(c: Char) extends Rexp
 | 
| 249 |     11 | case class ALT(r1: Rexp, r2: Rexp) extends Rexp 
 | 
|  |     12 | case class SEQ(r1: Rexp, r2: Rexp) extends Rexp 
 | 
|  |     13 | case class STAR(r: Rexp) extends Rexp 
 | 
| 153 |     14 | 
 | 
| 249 |     15 | // some convenience for typing in regular expressions
 | 
| 153 |     16 | 
 | 
| 229 |     17 | import scala.language.implicitConversions    
 | 
|  |     18 | import scala.language.reflectiveCalls 
 | 
|  |     19 | 
 | 
| 249 |     20 | 
 | 
| 153 |     21 | def charlist2rexp(s: List[Char]): Rexp = s match {
 | 
|  |     22 |   case Nil => ONE
 | 
|  |     23 |   case c::Nil => CHAR(c)
 | 
|  |     24 |   case c::s => SEQ(CHAR(c), charlist2rexp(s))
 | 
|  |     25 | }
 | 
|  |     26 | implicit def string2rexp(s: String): Rexp = charlist2rexp(s.toList)
 | 
|  |     27 | 
 | 
|  |     28 | implicit def RexpOps (r: Rexp) = new {
 | 
|  |     29 |   def | (s: Rexp) = ALT(r, s)
 | 
|  |     30 |   def % = STAR(r)
 | 
|  |     31 |   def ~ (s: Rexp) = SEQ(r, s)
 | 
|  |     32 | }
 | 
|  |     33 | 
 | 
|  |     34 | implicit def stringOps (s: String) = new {
 | 
|  |     35 |   def | (r: Rexp) = ALT(s, r)
 | 
|  |     36 |   def | (r: String) = ALT(s, r)
 | 
|  |     37 |   def % = STAR(s)
 | 
|  |     38 |   def ~ (r: Rexp) = SEQ(s, r)
 | 
|  |     39 |   def ~ (r: String) = SEQ(s, r)
 | 
|  |     40 | }
 | 
|  |     41 | 
 | 
| 221 |     42 | // (1) Complete the function nullable according to
 | 
| 229 |     43 | // the definition given in the coursework; this 
 | 
| 153 |     44 | // function checks whether a regular expression
 | 
| 221 |     45 | // can match the empty string and Returns a boolean
 | 
|  |     46 | // accordingly.
 | 
| 153 |     47 | 
 | 
| 229 |     48 | def nullable (r: Rexp) : Boolean = r match {
 | 
| 249 |     49 |   case ZERO => false
 | 
|  |     50 |   case ONE => true
 | 
|  |     51 |   case CHAR(_) => false
 | 
|  |     52 |   case ALT(r1, r2) => nullable(r1) || nullable(r2)
 | 
|  |     53 |   case SEQ(r1, r2) => nullable(r1) && nullable(r2)
 | 
|  |     54 |   case STAR(_) => true
 | 
| 153 |     55 | }
 | 
|  |     56 | 
 | 
| 221 |     57 | // (2) Complete the function der according to
 | 
| 153 |     58 | // the definition given in the coursework; this
 | 
| 229 |     59 | // function calculates the derivative of a 
 | 
| 221 |     60 | // regular expression w.r.t. a character.
 | 
| 153 |     61 | 
 | 
| 229 |     62 | def der (c: Char, r: Rexp) : Rexp = r match {
 | 
| 249 |     63 |   case ZERO => ZERO
 | 
|  |     64 |   case ONE => ZERO
 | 
|  |     65 |   case CHAR(d) => if (c == d) ONE else ZERO
 | 
|  |     66 |   case ALT(r1, r2) => ALT(der(c, r1), der(c, r2))
 | 
|  |     67 |   case SEQ(r1, r2) => 
 | 
|  |     68 |     if (nullable(r1)) ALT(SEQ(der(c, r1), r2), der(c, r2))
 | 
|  |     69 |     else SEQ(der(c, r1), r2)
 | 
|  |     70 |   case STAR(r1) => SEQ(der(c, r1), STAR(r1))
 | 
| 153 |     71 | }
 | 
|  |     72 | 
 | 
| 221 |     73 | // (3) Complete the simp function according to
 | 
| 153 |     74 | // the specification given in the coursework; this
 | 
| 221 |     75 | // function simplifies a regular expression from
 | 
| 229 |     76 | // the inside out, like you would simplify arithmetic 
 | 
|  |     77 | // expressions; however it does not simplify inside 
 | 
| 221 |     78 | // STAR-regular expressions.
 | 
| 153 |     79 | 
 | 
| 229 |     80 | def simp(r: Rexp) : Rexp = r match {
 | 
| 249 |     81 |   case ALT(r1, r2) => (simp(r1), simp(r2)) match {
 | 
|  |     82 |     case (ZERO, r2s) => r2s
 | 
|  |     83 |     case (r1s, ZERO) => r1s
 | 
|  |     84 |     case (r1s, r2s) => if (r1s == r2s) r1s else ALT (r1s, r2s)
 | 
|  |     85 |   }
 | 
|  |     86 |   case SEQ(r1, r2) =>  (simp(r1), simp(r2)) match {
 | 
|  |     87 |     case (ZERO, _) => ZERO
 | 
|  |     88 |     case (_, ZERO) => ZERO
 | 
|  |     89 |     case (ONE, r2s) => r2s
 | 
|  |     90 |     case (r1s, ONE) => r1s
 | 
|  |     91 |     case (r1s, r2s) => SEQ(r1s, r2s)
 | 
|  |     92 |   }
 | 
|  |     93 |   case r => r
 | 
| 153 |     94 | }
 | 
|  |     95 | 
 | 
| 221 |     96 | 
 | 
| 229 |     97 | // (4) Complete the two functions below; the first 
 | 
| 153 |     98 | // calculates the derivative w.r.t. a string; the second
 | 
|  |     99 | // is the regular expression matcher taking a regular
 | 
|  |    100 | // expression and a string and checks whether the
 | 
| 249 |    101 | // string matches the regular expression.
 | 
| 153 |    102 | 
 | 
| 229 |    103 | def ders (s: List[Char], r: Rexp) : Rexp = s match {
 | 
| 249 |    104 |   case Nil => r
 | 
|  |    105 |   case c::s => ders(s, simp(der(c, r)))
 | 
| 153 |    106 | }
 | 
|  |    107 | 
 | 
| 249 |    108 | // main matcher function
 | 
|  |    109 | def matcher(r: Rexp, s: String) = nullable(ders(s.toList, r))
 | 
| 153 |    110 | 
 | 
| 221 |    111 | // (5) Complete the size function for regular
 | 
| 229 |    112 | // expressions according to the specification 
 | 
| 153 |    113 | // given in the coursework.
 | 
|  |    114 | 
 | 
| 249 |    115 | 
 | 
| 229 |    116 | def size(r: Rexp): Int = r match {
 | 
| 249 |    117 |   case ZERO => 1
 | 
|  |    118 |   case ONE => 1
 | 
|  |    119 |   case CHAR(_) => 1
 | 
|  |    120 |   case ALT(r1, r2) => 1 + size(r1) + size (r2)
 | 
|  |    121 |   case SEQ(r1, r2) => 1 + size(r1) + size (r2)
 | 
|  |    122 |   case STAR(r1) => 1 + size(r1)
 | 
| 153 |    123 | }
 | 
|  |    124 | 
 | 
| 228 |    125 | 
 | 
| 249 |    126 | 
 | 
| 236 |    127 | // some testing data
 | 
| 249 |    128 | /*
 | 
|  |    129 | matcher(("a" ~ "b") ~ "c", "abc")  // => true
 | 
|  |    130 | matcher(("a" ~ "b") ~ "c", "ab")   // => false
 | 
| 229 |    131 | 
 | 
|  |    132 | // the supposedly 'evil' regular expression (a*)* b
 | 
|  |    133 | val EVIL = SEQ(STAR(STAR(CHAR('a'))), CHAR('b'))
 | 
|  |    134 | 
 | 
| 249 |    135 | matcher(EVIL, "a" * 1000 ++ "b")   // => true
 | 
|  |    136 | matcher(EVIL, "a" * 1000)          // => false
 | 
| 153 |    137 | 
 | 
|  |    138 | // size without simplifications
 | 
| 249 |    139 | size(der('a', der('a', EVIL)))             // => 28
 | 
|  |    140 | size(der('a', der('a', der('a', EVIL))))   // => 58
 | 
| 153 |    141 | 
 | 
|  |    142 | // size with simplification
 | 
| 249 |    143 | size(simp(der('a', der('a', EVIL))))           // => 8
 | 
|  |    144 | size(simp(der('a', der('a', der('a', EVIL))))) // => 8
 | 
| 228 |    145 | 
 | 
| 229 |    146 | // Python needs around 30 seconds for matching 28 a's with EVIL. 
 | 
| 221 |    147 | // Java 9 and later increase this to an "astonishing" 40000 a's in
 | 
| 249 |    148 | // around 30 seconds.
 | 
| 153 |    149 | //
 | 
| 249 |    150 | // Lets see how long it takes to match strings with 
 | 
|  |    151 | // 5 Million a's...it should be in the range of a 
 | 
|  |    152 | // couple of seconds.
 | 
| 153 |    153 | 
 | 
|  |    154 | def time_needed[T](i: Int, code: => T) = {
 | 
|  |    155 |   val start = System.nanoTime()
 | 
|  |    156 |   for (j <- 1 to i) code
 | 
|  |    157 |   val end = System.nanoTime()
 | 
|  |    158 |   (end - start)/(i * 1.0e9)
 | 
|  |    159 | }
 | 
|  |    160 | 
 | 
| 249 |    161 | for (i <- 0 to 5000000 by 500000) {
 | 
|  |    162 |   println(i + " " + "%.5f".format(time_needed(2, matcher(EVIL, "a" * i))))
 | 
|  |    163 | }
 | 
| 221 |    164 | 
 | 
| 229 |    165 | // another "power" test case 
 | 
| 249 |    166 | simp(Iterator.iterate(ONE:Rexp)(r => SEQ(r, ONE | ONE)).drop(100).next) == ONE
 | 
| 221 |    167 | 
 | 
|  |    168 | // the Iterator produces the rexp
 | 
|  |    169 | //
 | 
|  |    170 | //      SEQ(SEQ(SEQ(..., ONE | ONE) , ONE | ONE), ONE | ONE)
 | 
|  |    171 | //
 | 
| 249 |    172 | //    where SEQ is nested 100 times.
 | 
|  |    173 |  
 | 
|  |    174 | */
 | 
| 228 |    175 | 
 | 
| 249 |    176 | //}
 |