src/Pure/General/symbol.scala
author wenzelm
Thu, 17 Dec 2009 19:30:12 +0100
changeset 34104 2b9cdf23c188
parent 34010 6e5eafb373b3
child 34143 d8d9df8407f6
permissions -rw-r--r--
tuned signature;
wenzelm@27901
     1
/*  Title:      Pure/General/symbol.scala
wenzelm@27901
     2
    Author:     Makarius
wenzelm@27901
     3
wenzelm@27924
     4
Detecting and recoding Isabelle symbols.
wenzelm@27901
     5
*/
wenzelm@27901
     6
wenzelm@27901
     7
package isabelle
wenzelm@27901
     8
wenzelm@27918
     9
import scala.io.Source
wenzelm@31939
    10
import scala.collection.{jcl, mutable}
wenzelm@31537
    11
import scala.util.matching.Regex
wenzelm@27901
    12
wenzelm@27901
    13
wenzelm@31537
    14
object Symbol
wenzelm@31537
    15
{
wenzelm@34007
    16
  /* Symbol regexps */
wenzelm@27901
    17
wenzelm@31537
    18
  private val plain = new Regex("""(?xs)
wenzelm@31537
    19
    [^\\ \ud800-\udfff] | [\ud800-\udbff][\udc00-\udfff] """)
wenzelm@27901
    20
wenzelm@31537
    21
  private val symbol = new Regex("""(?xs)
wenzelm@31548
    22
      \\ < (?:
wenzelm@27924
    23
      \^? [A-Za-z][A-Za-z0-9_']* |
wenzelm@27924
    24
      \^raw: [\x20-\x7e\u0100-\uffff && [^.>]]* ) >""")
wenzelm@27923
    25
wenzelm@31537
    26
  private val bad_symbol = new Regex("(?xs) (?!" + symbol + ")" +
wenzelm@31548
    27
    """ \\ < (?: (?! \s | [\"`\\] | \(\* | \*\) | \{\* | \*\} ) . )*""")
wenzelm@27923
    28
wenzelm@27939
    29
  // total pattern
wenzelm@31537
    30
  val regex = new Regex(plain + "|" + symbol + "|" + bad_symbol + "| .")
wenzelm@27918
    31
wenzelm@31537
    32
  // prefix of another symbol
wenzelm@34007
    33
  def is_open(s: CharSequence): Boolean =
wenzelm@31537
    34
  {
wenzelm@31537
    35
    val len = s.length
wenzelm@34007
    36
    len == 1 && Character.isHighSurrogate(s.charAt(0)) ||
wenzelm@31537
    37
    s == "\\" ||
wenzelm@31537
    38
    s == "\\<" ||
wenzelm@34007
    39
    len > 2 && s.charAt(len - 1) != '>'
wenzelm@31537
    40
  }
wenzelm@27901
    41
wenzelm@27901
    42
wenzelm@34007
    43
  /* elements */
wenzelm@31939
    44
wenzelm@34007
    45
  private def could_open(c: Char): Boolean =
wenzelm@34007
    46
    c == '\\' || Character.isHighSurrogate(c)
wenzelm@34007
    47
wenzelm@34010
    48
  def elements(text: CharSequence) = new Iterator[String] {
wenzelm@34007
    49
    private val matcher = regex.pattern.matcher(text)
wenzelm@34007
    50
    private var i = 0
wenzelm@34007
    51
    def hasNext = i < text.length
wenzelm@34007
    52
    def next = {
wenzelm@34007
    53
      val len =
wenzelm@34007
    54
        if (could_open(text.charAt(i))) {
wenzelm@34007
    55
          matcher.region(i, text.length).lookingAt
wenzelm@34007
    56
          matcher.group.length
wenzelm@34007
    57
        }
wenzelm@34007
    58
        else 1
wenzelm@34007
    59
      val s = text.subSequence(i, i + len)
wenzelm@34007
    60
      i += len
wenzelm@34010
    61
      s.toString
wenzelm@34007
    62
    }
wenzelm@34007
    63
  }
wenzelm@34007
    64
wenzelm@34007
    65
wenzelm@34007
    66
  /* decoding offsets */
wenzelm@34007
    67
wenzelm@34007
    68
  class Index(text: CharSequence)
wenzelm@31939
    69
  {
wenzelm@31939
    70
    case class Entry(chr: Int, sym: Int)
wenzelm@31939
    71
    val index: Array[Entry] =
wenzelm@31939
    72
    {
wenzelm@31939
    73
      val matcher = regex.pattern.matcher(text)
wenzelm@31939
    74
      val buf = new mutable.ArrayBuffer[Entry]
wenzelm@31939
    75
      var chr = 0
wenzelm@31939
    76
      var sym = 0
wenzelm@34007
    77
      while (chr < text.length) {
wenzelm@31939
    78
        val len =
wenzelm@34007
    79
          if (could_open(text.charAt(chr))) {
wenzelm@34007
    80
            matcher.region(chr, text.length).lookingAt
wenzelm@31939
    81
            matcher.group.length
wenzelm@34007
    82
          }
wenzelm@34007
    83
          else 1
wenzelm@31939
    84
        chr += len
wenzelm@31939
    85
        sym += 1
wenzelm@31939
    86
        if (len > 1) buf += Entry(chr, sym)
wenzelm@31939
    87
      }
wenzelm@31939
    88
      buf.toArray
wenzelm@31939
    89
    }
wenzelm@31939
    90
    def decode(sym: Int): Int =
wenzelm@31939
    91
    {
wenzelm@31939
    92
      val end = index.length
wenzelm@31939
    93
      def bisect(a: Int, b: Int): Int =
wenzelm@31939
    94
      {
wenzelm@31939
    95
        if (a < b) {
wenzelm@31939
    96
          val c = (a + b) / 2
wenzelm@31939
    97
          if (sym < index(c).sym) bisect(a, c)
wenzelm@31939
    98
          else if (c + 1 == end || sym < index(c + 1).sym) c
wenzelm@31939
    99
          else bisect(c + 1, b)
wenzelm@31939
   100
        }
wenzelm@31939
   101
        else -1
wenzelm@31939
   102
      }
wenzelm@31939
   103
      val i = bisect(0, end)
wenzelm@31939
   104
      if (i < 0) sym
wenzelm@31939
   105
      else index(i).chr + sym - index(i).sym
wenzelm@31939
   106
    }
wenzelm@31939
   107
  }
wenzelm@31939
   108
wenzelm@31939
   109
wenzelm@34007
   110
  /* recoding text */
wenzelm@27924
   111
wenzelm@31537
   112
  private class Recoder(list: List[(String, String)])
wenzelm@31537
   113
  {
wenzelm@31537
   114
    private val (min, max) =
wenzelm@31537
   115
    {
wenzelm@27937
   116
      var min = '\uffff'
wenzelm@27937
   117
      var max = '\u0000'
wenzelm@27937
   118
      for ((x, _) <- list) {
wenzelm@27937
   119
        val c = x(0)
wenzelm@27937
   120
        if (c < min) min = c
wenzelm@27937
   121
        if (c > max) max = c
wenzelm@27937
   122
      }
wenzelm@27937
   123
      (min, max)
wenzelm@27937
   124
    }
wenzelm@31537
   125
    private val table =
wenzelm@31537
   126
    {
wenzelm@31537
   127
      val table = new jcl.HashMap[String, String]   // reasonably efficient?
wenzelm@27937
   128
      for ((x, y) <- list) table + (x -> y)
wenzelm@27927
   129
      table
wenzelm@27927
   130
    }
wenzelm@31537
   131
    def recode(text: String): String =
wenzelm@31537
   132
    {
wenzelm@27937
   133
      val len = text.length
wenzelm@31537
   134
      val matcher = regex.pattern.matcher(text)
wenzelm@27937
   135
      val result = new StringBuilder(len)
wenzelm@27937
   136
      var i = 0
wenzelm@27937
   137
      while (i < len) {
wenzelm@27937
   138
        val c = text(i)
wenzelm@27937
   139
        if (min <= c && c <= max) {
wenzelm@31939
   140
          matcher.region(i, len).lookingAt
wenzelm@27938
   141
          val x = matcher.group
wenzelm@31537
   142
          result.append(table.get(x) getOrElse x)
wenzelm@27937
   143
          i = matcher.end
wenzelm@27937
   144
        }
wenzelm@27937
   145
        else { result.append(c); i += 1 }
wenzelm@27937
   146
      }
wenzelm@27937
   147
      result.toString
wenzelm@27924
   148
    }
wenzelm@27924
   149
  }
wenzelm@27918
   150
wenzelm@27918
   151
wenzelm@27937
   152
wenzelm@27937
   153
  /** Symbol interpretation **/
wenzelm@27937
   154
wenzelm@29569
   155
  class Interpretation(symbol_decls: Iterator[String])
wenzelm@29569
   156
  {
wenzelm@31537
   157
    /* read symbols */
wenzelm@31537
   158
wenzelm@31537
   159
    private val empty = new Regex("""(?xs) ^\s* (?: \#.* )? $ """)
wenzelm@31537
   160
    private val key = new Regex("""(?xs) (.+): """)
wenzelm@31537
   161
wenzelm@31537
   162
    private def read_decl(decl: String): (String, Map[String, String]) =
wenzelm@31537
   163
    {
wenzelm@31537
   164
      def err() = error("Bad symbol declaration: " + decl)
wenzelm@31537
   165
wenzelm@31537
   166
      def read_props(props: List[String]): Map[String, String] =
wenzelm@31537
   167
      {
wenzelm@31537
   168
        props match {
wenzelm@31537
   169
          case Nil => Map()
wenzelm@31537
   170
          case _ :: Nil => err()
wenzelm@31537
   171
          case key(x) :: y :: rest => read_props(rest) + (x -> y)
wenzelm@31537
   172
          case _ => err()
wenzelm@31537
   173
        }
wenzelm@31537
   174
      }
wenzelm@31537
   175
      decl.split("\\s+").toList match {
wenzelm@31537
   176
        case Nil => err()
wenzelm@31537
   177
        case sym :: props => (sym, read_props(props))
wenzelm@31537
   178
      }
wenzelm@31537
   179
    }
wenzelm@31537
   180
wenzelm@31537
   181
    private val symbols: List[(String, Map[String, String])] =
wenzelm@31537
   182
      for (decl <- symbol_decls.toList if !empty.pattern.matcher(decl).matches)
wenzelm@31537
   183
        yield read_decl(decl)
wenzelm@31537
   184
wenzelm@31537
   185
wenzelm@31651
   186
    /* misc properties */
wenzelm@31651
   187
wenzelm@31651
   188
    val names: Map[String, String] = {
wenzelm@31651
   189
      val name = new Regex("""\\<([A-Za-z][A-Za-z0-9_']*)>""")
wenzelm@31651
   190
      Map((for ((sym @ name(a), _) <- symbols) yield (sym -> a)): _*)
wenzelm@31651
   191
    }
wenzelm@31651
   192
wenzelm@31651
   193
    val abbrevs: Map[String, String] = Map((
wenzelm@31651
   194
      for ((sym, props) <- symbols if props.isDefinedAt("abbrev"))
wenzelm@31651
   195
        yield (sym -> props("abbrev"))): _*)
wenzelm@31651
   196
wenzelm@31651
   197
wenzelm@31537
   198
    /* main recoder methods */
wenzelm@31537
   199
wenzelm@31537
   200
    private val (decoder, encoder) =
wenzelm@31537
   201
    {
wenzelm@31537
   202
      val mapping =
wenzelm@31537
   203
        for {
wenzelm@31537
   204
          (sym, props) <- symbols
wenzelm@31537
   205
          val code =
wenzelm@31537
   206
            try { Integer.decode(props("code")).intValue }
wenzelm@31537
   207
            catch {
wenzelm@31537
   208
              case _: NoSuchElementException => error("Missing code for symbol " + sym)
wenzelm@31537
   209
              case _: NumberFormatException => error("Bad code for symbol " + sym)
wenzelm@31537
   210
            }
wenzelm@31537
   211
          val ch = new String(Character.toChars(code))
wenzelm@31537
   212
        } yield (sym, ch)
wenzelm@31548
   213
      (new Recoder(mapping),
wenzelm@31551
   214
       new Recoder(mapping map { case (x, y) => (y, x) }))
wenzelm@31537
   215
    }
wenzelm@27918
   216
wenzelm@34104
   217
    def decode(text: String): String = decoder.recode(text)
wenzelm@34104
   218
    def encode(text: String): String = encoder.recode(text)
wenzelm@27918
   219
  }
wenzelm@27901
   220
}