| 
68
 | 
     1  | 
// Part 1 about Regular Expression Matching
  | 
| 
 | 
     2  | 
//==========================================
  | 
| 
4
 | 
     3  | 
  | 
| 
 | 
     4  | 
abstract class Rexp
  | 
| 
 | 
     5  | 
case object ZERO extends Rexp
  | 
| 
 | 
     6  | 
case object ONE extends Rexp
  | 
| 
 | 
     7  | 
case class CHAR(c: Char) extends Rexp
  | 
| 
68
 | 
     8  | 
case class ALT(r1: Rexp, r2: Rexp) extends Rexp   // alternative 
  | 
| 
 | 
     9  | 
case class SEQ(r1: Rexp, r2: Rexp) extends Rexp   // sequence
  | 
| 
 | 
    10  | 
case class STAR(r: Rexp) extends Rexp             // star
  | 
| 
 | 
    11  | 
  | 
| 
 | 
    12  | 
  | 
| 
 | 
    13  | 
// some convenience for typing in regular expressions
  | 
| 
 | 
    14  | 
  | 
| 
 | 
    15  | 
import scala.language.implicitConversions    
  | 
| 
 | 
    16  | 
import scala.language.reflectiveCalls 
  | 
| 
4
 | 
    17  | 
  | 
| 
68
 | 
    18  | 
def charlist2rexp(s: List[Char]): Rexp = s match {
 | 
| 
 | 
    19  | 
  case Nil => ONE
  | 
| 
 | 
    20  | 
  case c::Nil => CHAR(c)
  | 
| 
 | 
    21  | 
  case c::s => SEQ(CHAR(c), charlist2rexp(s))
  | 
| 
 | 
    22  | 
}
  | 
| 
 | 
    23  | 
implicit def string2rexp(s: String): Rexp = charlist2rexp(s.toList)
  | 
| 
 | 
    24  | 
  | 
| 
 | 
    25  | 
implicit def RexpOps (r: Rexp) = new {
 | 
| 
 | 
    26  | 
  def | (s: Rexp) = ALT(r, s)
  | 
| 
 | 
    27  | 
  def % = STAR(r)
  | 
| 
 | 
    28  | 
  def ~ (s: Rexp) = SEQ(r, s)
  | 
| 
4
 | 
    29  | 
}
  | 
| 
 | 
    30  | 
  | 
| 
68
 | 
    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 ~ (r: Rexp) = SEQ(s, r)
  | 
| 
 | 
    36  | 
  def ~ (r: String) = SEQ(s, r)
  | 
| 
4
 | 
    37  | 
}
  | 
| 
 | 
    38  | 
  | 
| 
68
 | 
    39  | 
// (1a) Complete the function nullable according to
  | 
| 
 | 
    40  | 
// the definition given in the coursework; this 
  | 
| 
 | 
    41  | 
// function checks whether a regular expression
  | 
| 
 | 
    42  | 
// can match the empty string
  | 
| 
 | 
    43  | 
  | 
| 
 | 
    44  | 
def nullable (r: Rexp) : Boolean = ...
  | 
| 
4
 | 
    45  | 
  | 
| 
 | 
    46  | 
  | 
| 
68
 | 
    47  | 
// (1b) Complete the function der according to
  | 
| 
 | 
    48  | 
// the definition given in the coursework; this
  | 
| 
 | 
    49  | 
// function calculates the derivative of a 
  | 
| 
 | 
    50  | 
// regular expression w.r.t. a character
  | 
| 
 | 
    51  | 
  | 
| 
 | 
    52  | 
def der (c: Char, r: Rexp) : Rexp = ...
  | 
| 
 | 
    53  | 
  | 
| 
 | 
    54  | 
// (1c) Complete the function der according to
  | 
| 
 | 
    55  | 
// the specification given in the coursework; this
  | 
| 
 | 
    56  | 
// function simplifies a regular expression;
  | 
| 
 | 
    57  | 
// however it does not simplify inside STAR-regular
  | 
| 
 | 
    58  | 
// expressions
  | 
| 
 | 
    59  | 
  | 
| 
 | 
    60  | 
def simp(r: Rexp) : Rexp = ... 
  | 
| 
 | 
    61  | 
  | 
| 
 | 
    62  | 
// (1d) Complete the two functions below; the first 
  | 
| 
 | 
    63  | 
// calculates the derivative w.r.t. a string; the second
  | 
| 
 | 
    64  | 
// is the regular expression matcher taking a regular
  | 
| 
 | 
    65  | 
// expression and a string and checks whether the
  | 
| 
 | 
    66  | 
// string matches the regular expression
  | 
| 
 | 
    67  | 
  | 
| 
 | 
    68  | 
def ders (s: List[Char], r: Rexp) : Rexp = ... 
  | 
| 
 | 
    69  | 
  | 
| 
 | 
    70  | 
def matcher(r: Rexp, s: String): Boolean = ...
  | 
| 
4
 | 
    71  | 
  | 
| 
 | 
    72  | 
  | 
| 
68
 | 
    73  | 
// (1e) Complete the function below: it searches (from the left to 
  | 
| 
 | 
    74  | 
// right) in string s1 all the non-empty substrings that match the 
  | 
| 
 | 
    75  | 
// regular expression -- these substrings are assumed to be
  | 
| 
 | 
    76  | 
// the longest substrings matched by the regular expression and
  | 
| 
 | 
    77  | 
// assumed to be non-overlapping. All these substrings in s1 are replaced
  | 
| 
 | 
    78  | 
// by s2.
  | 
| 
 | 
    79  | 
  | 
| 
 | 
    80  | 
def replace(r: Rexp, s1: String, s2: String): String = ...
  | 
| 
4
 | 
    81  | 
  | 
| 
68
 | 
    82  | 
  | 
| 
4
 | 
    83  | 
  | 
| 
68
 | 
    84  | 
// some testing data
  | 
| 
 | 
    85  | 
// the supposedly 'evil' regular expression (a*)* b
  | 
| 
 | 
    86  | 
/*
  | 
| 
 | 
    87  | 
val EVIL = SEQ(STAR(STAR(CHAR('a'))), CHAR('b'))
 | 
| 
 | 
    88  | 
println(matcher(EVIL, "a" * 1000 ++ "b"))
  | 
| 
 | 
    89  | 
println(matcher(EVIL, "a" * 1000))
  | 
| 
4
 | 
    90  | 
  | 
| 
 | 
    91  | 
  | 
| 
 | 
    92  | 
def time_needed[T](i: Int, code: => T) = {
 | 
| 
 | 
    93  | 
  val start = System.nanoTime()
  | 
| 
 | 
    94  | 
  for (j <- 1 to i) code
  | 
| 
 | 
    95  | 
  val end = System.nanoTime()
  | 
| 
 | 
    96  | 
  (end - start)/(i * 1.0e9)
  | 
| 
 | 
    97  | 
}
  | 
| 
 | 
    98  | 
  | 
| 
68
 | 
    99  | 
for (i <- 1 to 5000001 by 500000) {
 | 
| 
 | 
   100  | 
  println(i + " " + "%.5f".format(time_needed(2, matcher(EVIL, "a" * i))))
  | 
| 
4
 | 
   101  | 
}
  | 
| 
68
 | 
   102  | 
*/
  | 
| 
4
 | 
   103  | 
  | 
| 
 | 
   104  | 
  |