src/Pure/library.scala
author wenzelm
Sun, 27 Dec 2009 23:09:16 +0100
changeset 34191 b6960fc09ef3
parent 34141 297b2149077d
child 34196 c352f679dcca
permissions -rw-r--r--
factored-out Library.decode_permissive_utf8;
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
34136
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     1
/*  Title:      Pure/library.scala
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     2
    Author:     Makarius
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     3
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     4
Basic library.
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     5
*/
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     6
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     7
package isabelle
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     8
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
     9
import java.lang.System
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    10
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    11
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    12
object Library
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    13
{
34141
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    14
  /* reverse CharSequence */
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    15
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    16
  class Reverse(text: CharSequence, start: Int, end: Int) extends CharSequence
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    17
  {
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    18
    require(0 <= start && start <= end && end <= text.length)
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    19
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    20
    def this(text: CharSequence) = this(text, 0, text.length)
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    21
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    22
    def length: Int = end - start
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    23
    def charAt(i: Int): Char = text.charAt(end - i - 1)
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    24
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    25
    def subSequence(i: Int, j: Int): CharSequence =
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    26
      if (0 <= i && i <= j && j <= length) new Reverse(text, end - j, end - i)
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    27
      else throw new IndexOutOfBoundsException
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    28
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    29
    override def toString: String =
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    30
    {
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    31
      val buf = new StringBuilder(length)
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    32
      for (i <- 0 until length)
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    33
        buf.append(charAt(i))
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    34
      buf.toString
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    35
    }
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    36
  }
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    37
297b2149077d simiplified result of keyword parser (again);
wenzelm
parents: 34136
diff changeset
    38
34191
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    39
  /* permissive UTF-8 decoding */
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    40
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    41
  // see also http://en.wikipedia.org/wiki/UTF-8#Description
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    42
  def decode_permissive_utf8(text: CharSequence): String =
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    43
  {
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    44
    val buf = new java.lang.StringBuilder(text.length)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    45
    var code = -1
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    46
    var rest = 0
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    47
    def flush()
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    48
    {
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    49
      if (code != -1) {
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    50
        if (rest == 0 && Character.isValidCodePoint(code))
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    51
          buf.appendCodePoint(code)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    52
        else buf.append('\uFFFD')
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    53
        code = -1
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    54
        rest = 0
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    55
      }
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    56
    }
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    57
    def init(x: Int, n: Int)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    58
    {
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    59
      flush()
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    60
      code = x
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    61
      rest = n
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    62
    }
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    63
    def push(x: Int)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    64
    {
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    65
      if (rest <= 0) init(x, -1)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    66
      else {
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    67
        code <<= 6
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    68
        code += x
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    69
        rest -= 1
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    70
      }
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    71
    }
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    72
    for (i <- 0 until text.length) {
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    73
      val c = text.charAt(i)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    74
      if (c < 128) { flush(); buf.append(c) }
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    75
      else if ((c & 0xC0) == 0x80) push(c & 0x3F)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    76
      else if ((c & 0xE0) == 0xC0) init(c & 0x1F, 1)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    77
      else if ((c & 0xF0) == 0xE0) init(c & 0x0F, 2)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    78
      else if ((c & 0xF8) == 0xF0) init(c & 0x07, 3)
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    79
    }
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    80
    flush()
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    81
    buf.toString
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    82
  }
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    83
b6960fc09ef3 factored-out Library.decode_permissive_utf8;
wenzelm
parents: 34141
diff changeset
    84
34136
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    85
  /* timing */
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    86
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    87
  def timeit[A](e: => A) =
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    88
  {
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    89
    val start = System.currentTimeMillis()
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    90
    val result = Exn.capture(e)
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    91
    val stop = System.currentTimeMillis()
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    92
    System.err.println((stop - start) + "ms elapsed time")
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    93
    Exn.release(result)
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    94
  }
3dcb46ae6185 added basic library -- Scala version;
wenzelm
parents:
diff changeset
    95
}