src/Pure/General/scan.scala
author wenzelm
Sun, 20 Dec 2009 17:47:59 +0100
changeset 34143 ded454429df3
parent 34140 31be1235d0fb
child 34144 bd7b3b91abab
permissions -rw-r--r--
added nested comments; tuned;
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
     1
/*  Title:      Pure/General/scan.scala
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
     2
    Author:     Makarius
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
     3
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
     4
Efficient scanning of keywords.
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
     5
*/
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
     6
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
     7
package isabelle
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
     8
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
     9
import scala.util.parsing.combinator.RegexParsers
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    10
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    11
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    12
object Scan
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    13
{
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    14
  /** Lexicon -- position tree **/
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    15
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    16
  object Lexicon
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    17
  {
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    18
    private case class Tree(val branches: Map[Char, (String, Tree)])
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    19
    private val empty_tree = Tree(Map())
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    20
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    21
    val empty: Lexicon = new Lexicon
31762
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    22
    def apply(elems: String*): Lexicon = empty ++ elems
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    23
  }
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    24
31764
e767fee21b22 moved string utilities to completion.scala;
wenzelm
parents: 31762
diff changeset
    25
  class Lexicon extends scala.collection.Set[String] with RegexParsers
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    26
  {
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    27
    /* representation */
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    28
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    29
    import Lexicon.Tree
31764
e767fee21b22 moved string utilities to completion.scala;
wenzelm
parents: 31762
diff changeset
    30
    protected val main_tree: Tree = Lexicon.empty_tree
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    31
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    32
31650
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    33
    /* auxiliary operations */
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    34
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    35
    private def content(tree: Tree, result: List[String]): List[String] =
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    36
      (result /: tree.branches.toList) ((res, entry) =>
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    37
        entry match { case (_, (s, tr)) =>
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    38
          if (s.isEmpty) content(tr, res) else content(tr, s :: res) })
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    39
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    40
    private def lookup(str: CharSequence): Option[(Boolean, Tree)] =
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    41
    {
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    42
      val len = str.length
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    43
      def look(tree: Tree, tip: Boolean, i: Int): Option[(Boolean, Tree)] =
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    44
      {
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    45
        if (i < len) {
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    46
          tree.branches.get(str.charAt(i)) match {
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    47
            case Some((s, tr)) => look(tr, !s.isEmpty, i + 1)
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    48
            case None => None
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    49
          }
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    50
        } else Some(tip, tree)
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    51
      }
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    52
      look(main_tree, false, 0)
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    53
    }
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    54
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    55
    def completions(str: CharSequence): List[String] =
31764
e767fee21b22 moved string utilities to completion.scala;
wenzelm
parents: 31762
diff changeset
    56
      lookup(str) match {
31650
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    57
        case Some((true, tree)) => content(tree, List(str.toString))
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    58
        case Some((false, tree)) => content(tree, Nil)
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    59
        case None => Nil
31764
e767fee21b22 moved string utilities to completion.scala;
wenzelm
parents: 31762
diff changeset
    60
      }
31650
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    61
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    62
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    63
    /* Set methods */
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    64
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    65
    override def stringPrefix = "Lexicon"
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    66
31753
a61475260e47 Lexicon: removed unused max_entry;
wenzelm
parents: 31701
diff changeset
    67
    override def isEmpty: Boolean = { main_tree.branches.isEmpty }
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
    68
31650
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    69
    def size: Int = content(main_tree, Nil).length
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    70
    def elements: Iterator[String] = content(main_tree, Nil).sort(_ <= _).elements
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    71
31762
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    72
    def contains(elem: String): Boolean =
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    73
      lookup(elem) match {
31650
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    74
        case Some((tip, _)) => tip
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    75
        case _ => false
cfaed41ee2c5 added completions;
wenzelm
parents: 31649
diff changeset
    76
      }
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    77
31762
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    78
    def + (elem: String): Lexicon =
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    79
      if (contains(elem)) this
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    80
      else {
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    81
        val len = elem.length
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    82
        def extend(tree: Tree, i: Int): Tree =
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    83
          if (i < len) {
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    84
            val c = elem.charAt(i)
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    85
            val end = (i + 1 == len)
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    86
            tree.branches.get(c) match {
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    87
              case Some((s, tr)) =>
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    88
                Tree(tree.branches +
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    89
                  (c -> (if (end) elem else s, extend(tr, i + 1))))
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    90
              case None =>
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    91
                Tree(tree.branches +
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    92
                  (c -> (if (end) elem else "", extend(Lexicon.empty_tree, i + 1))))
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    93
            }
34135
63dd95e3b393 indicate final state of keywords;
wenzelm
parents: 32450
diff changeset
    94
          }
63dd95e3b393 indicate final state of keywords;
wenzelm
parents: 32450
diff changeset
    95
          else tree
31762
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    96
        val old = this
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    97
        new Lexicon { override val main_tree = extend(old.main_tree, 0) }
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
    98
      }
31762
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
    99
31764
e767fee21b22 moved string utilities to completion.scala;
wenzelm
parents: 31762
diff changeset
   100
    def + (elem1: String, elem2: String, elems: String*): Lexicon =
31762
20745ab5b79a more precise implementation of trait methods -- oddly this seems to require copy/paste for +, ++;
wenzelm
parents: 31753
diff changeset
   101
      this + elem1 + elem2 ++ elems
31764
e767fee21b22 moved string utilities to completion.scala;
wenzelm
parents: 31762
diff changeset
   102
    def ++ (elems: Iterable[String]): Lexicon = (this /: elems) ((s, elem) => s + elem)
e767fee21b22 moved string utilities to completion.scala;
wenzelm
parents: 31762
diff changeset
   103
    def ++ (elems: Iterator[String]): Lexicon = (this /: elems) ((s, elem) => s + elem)
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   104
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   105
34137
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   106
    /** RegexParsers methods **/
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   107
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   108
    override val whiteSpace = "".r
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   109
34137
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   110
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   111
    /* keywords from lexicon */
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   112
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   113
    def keyword: Parser[String] = new Parser[String]
34135
63dd95e3b393 indicate final state of keywords;
wenzelm
parents: 32450
diff changeset
   114
    {
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   115
      def apply(in: Input) =
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   116
      {
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   117
        val source = in.source
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   118
        val offset = in.offset
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   119
        val len = source.length - offset
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   120
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   121
        def scan(tree: Tree, result: String, i: Int): String =
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   122
        {
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   123
          if (i < len) {
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   124
            tree.branches.get(source.charAt(offset + i)) match {
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   125
              case Some((s, tr)) => scan(tr, if (s.isEmpty) result else s, i + 1)
34135
63dd95e3b393 indicate final state of keywords;
wenzelm
parents: 32450
diff changeset
   126
              case None => result
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   127
            }
34135
63dd95e3b393 indicate final state of keywords;
wenzelm
parents: 32450
diff changeset
   128
          }
63dd95e3b393 indicate final state of keywords;
wenzelm
parents: 32450
diff changeset
   129
          else result
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   130
        }
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   131
        val result = scan(main_tree, "", 0)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   132
        if (result.isEmpty) Failure("keyword expected", in)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   133
        else Success(result, in.drop(result.length))
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
   134
      }
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   135
    }.named("keyword")
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
   136
34137
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   137
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   138
    /* symbols */
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   139
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   140
    def symbols(pred: String => Boolean, min_count: Int, max_count: Int): Parser[String] =
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   141
      new Parser[String]
34135
63dd95e3b393 indicate final state of keywords;
wenzelm
parents: 32450
diff changeset
   142
      {
34137
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   143
        def apply(in: Input) =
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   144
        {
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   145
          val start = in.offset
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   146
          val end = in.source.length
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   147
          val matcher = new Symbol.Matcher(in.source)
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   148
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   149
          var i = start
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   150
          var count = 0
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   151
          var finished = false
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   152
          while (!finished) {
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   153
            if (i < end && count < max_count) {
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   154
              val n = matcher(i, end)
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   155
              val sym = in.source.subSequence(i, i + n).toString
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   156
              if (pred(sym)) { i += n; count += 1 }
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   157
              else finished = true
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   158
            }
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   159
            else finished = true
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   160
          }
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   161
          if (count < min_count) Failure("bad input", in)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   162
          else Success(in.source.subSequence(start, i).toString, in.drop(i - start))
34137
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   163
        }
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   164
      }.named("symbols")
6cc9a0cbaf55 refined some Symbol operations/signatures;
wenzelm
parents: 34135
diff changeset
   165
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   166
    def one(pred: String => Boolean): Parser[String] = symbols(pred, 1, 1)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   167
    def many(pred: String => Boolean): Parser[String] = symbols(pred, 0, Integer.MAX_VALUE)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   168
    def many1(pred: String => Boolean): Parser[String] = symbols(pred, 1, Integer.MAX_VALUE)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   169
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   170
34143
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   171
    /* quoted strings */
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   172
34143
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   173
    private def quoted(quote: String): Parser[String] =
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   174
    {
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   175
      quote ~
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   176
        rep(many1(sym => sym != quote && sym != "\\" && Symbol.is_closed(sym)) |
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   177
          "\\" + quote | "\\\\" |
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   178
          (("""\\\d\d\d""".r) ^? { case x if x.substring(1, 4).toInt <= 255 => x })) ~
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   179
      quote ^^ { case x ~ ys ~ z => x + ys.mkString + z }
34143
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   180
    }.named("quoted")
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   181
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   182
    def quoted_content(quote: String, source: String): String =
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   183
    {
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   184
      require(parseAll(quoted(quote), source).successful)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   185
      source.substring(1, source.length - 1)  // FIXME proper escapes, recode utf8 (!?)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   186
    }
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   187
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   188
34143
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   189
    /* verbatim text */
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   190
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   191
    private def verbatim: Parser[String] =
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   192
    {
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   193
      "{*" ~ rep(many1(sym => sym != "*" && Symbol.is_closed(sym)) | """\*(?!\})""".r) ~ "*}" ^^
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   194
        { case x ~ ys ~ z => x + ys.mkString + z }
34143
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   195
    }.named("verbatim")
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   196
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   197
    def verbatim_content(source: String): String =
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   198
    {
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   199
      require(parseAll(verbatim, source).successful)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   200
      source.substring(2, source.length - 2)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   201
    }
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   202
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   203
34143
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   204
    /* nested comments */
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   205
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   206
    def comment: Parser[String] = new Parser[String]
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   207
    {
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   208
      val comment_text =
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   209
        rep(many1(sym => sym != "*" && sym != "(" && Symbol.is_closed(sym)) |
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   210
          """\*(?!\))|\((?!\*)""".r)
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   211
      val comment_open = "(*" ~ comment_text ^^ (_ => ())
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   212
      val comment_close = comment_text ~ "*)" ^^ (_ => ())
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   213
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   214
      def apply(in: Input) =
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   215
      {
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   216
        var rest = in
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   217
        def try_parse(p: Parser[Unit]): Boolean =
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   218
        {
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   219
          parse(p, rest) match {
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   220
            case Success(_, next) => { rest = next; true }
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   221
            case _ => false
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   222
          }
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   223
        }
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   224
        var depth = 0
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   225
        var finished = false
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   226
        while (!finished) {
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   227
          if (try_parse(comment_open)) depth += 1
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   228
          else if (depth > 0 && try_parse(comment_close)) depth -= 1
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   229
          else finished = true
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   230
        }
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   231
        if (in.offset < rest.offset && depth == 0)
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   232
          Success(in.source.subSequence(in.offset, rest.offset).toString, rest)
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   233
        else Failure("comment expected", in)
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   234
      }
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   235
    }.named("comment")
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   236
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   237
    def comment_content(source: String): String =
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   238
    {
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   239
      require(parseAll(comment, source).successful)
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   240
      source.substring(2, source.length - 2)
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   241
    }
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   242
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   243
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   244
    /* outer syntax tokens */
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   245
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   246
    def token(symbols: Symbol.Interpretation, is_command: String => Boolean):
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   247
      Parser[Outer_Lex.Token] =
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   248
    {
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   249
      import Outer_Lex._
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   250
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   251
      val id = one(symbols.is_letter) ~ many(symbols.is_letdig) ^^ { case x ~ y => x + y }
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   252
      val nat = many1(symbols.is_digit)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   253
      val id_nat = id ~ opt("." ~ nat) ^^ { case x ~ Some(y ~ z) => x + y + z case x ~ None => x }
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   254
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   255
      val ident = id ~ rep("." ~> id) ^^
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   256
        { case x ~ Nil => Ident(x)
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   257
          case x ~ ys => Long_Ident((x :: ys).mkString(".")) }
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   258
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   259
      val var_ = "?" ~ id_nat ^^ { case x ~ y => Var(x + y) }
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   260
      val type_ident = "'" ~ id ^^ { case x ~ y => Type_Ident(x + y) }
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   261
      val type_var = "?'" ~ id_nat ^^ { case x ~ y => Type_Var(x + y) }
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   262
      val nat_ = nat ^^ Nat
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   263
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   264
      val sym_ident =
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   265
        (many1(symbols.is_symbolic_char) |
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   266
          one(sym => symbols.is_symbolic(sym) & Symbol.is_closed(sym))) ^^ Sym_Ident
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   267
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   268
      val space = many1(symbols.is_blank) ^^ Space
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   269
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   270
      val string = quoted("\"") ^^ String_Token
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   271
      val alt_string = quoted("`") ^^ Alt_String_Token
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   272
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   273
      val bad_input = many1(sym => !(symbols.is_blank(sym))) ^^ Bad_Input
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   274
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   275
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   276
      /* tokens */
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   277
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   278
      // FIXME right-assoc !?
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   279
34143
ded454429df3 added nested comments;
wenzelm
parents: 34140
diff changeset
   280
      (string | alt_string | verbatim ^^ Verbatim | comment ^^ Comment | space) |
34140
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   281
      ((ident | var_ | type_ident | type_var | nat_ | sym_ident) |||
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   282
        keyword ^^ (x => if (is_command(x)) Command(x) else Keyword(x))) |
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   283
      bad_input
31be1235d0fb simplified result of keyword and symbols parser;
wenzelm
parents: 34137
diff changeset
   284
    }
31649
a11ea667d676 reorganized and abstracted version, via Set trait;
wenzelm
parents: 31648
diff changeset
   285
  }
31648
31b1f296515b Efficient scanning of literals.
wenzelm
parents:
diff changeset
   286
}