src/Pure/Isar/parse.scala
author wenzelm
Mon, 11 Jul 2016 18:18:24 +0200
changeset 63446 19162a9ef7e3
parent 62969 9f394a16c557
child 64471 c40c2975fb02
permissions -rw-r--r--
tunes signature;
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
43283
446e6621762d updated headers;
wenzelm
parents: 36956
diff changeset
     1
/*  Title:      Pure/Isar/parse.scala
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
     2
    Author:     Makarius
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
     3
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
     4
Generic parsers for Isabelle/Isar outer syntax.
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
     5
*/
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
     6
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
     7
package isabelle
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
     8
55618
995162143ef4 tuned imports;
wenzelm
parents: 51627
diff changeset
     9
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    10
import scala.util.parsing.combinator.Parsers
48599
5e64b7770f35 tuned signature;
wenzelm
parents: 48484
diff changeset
    11
import scala.annotation.tailrec
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    12
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    13
36948
d2cdad45fd14 renamed Outer_Parse to Parse (in Scala);
wenzelm
parents: 34311
diff changeset
    14
object Parse
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    15
{
34161
4c845a8f1357 consider proper input only;
wenzelm
parents: 34159
diff changeset
    16
  /* parsing tokens */
4c845a8f1357 consider proper input only;
wenzelm
parents: 34159
diff changeset
    17
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    18
  trait Parser extends Parsers
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    19
  {
36956
21be4832c362 renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
wenzelm
parents: 36948
diff changeset
    20
    type Elem = Token
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    21
48599
5e64b7770f35 tuned signature;
wenzelm
parents: 48484
diff changeset
    22
    def filter_proper: Boolean = true
34266
bfe8d6998734 added filter_proper parameter;
wenzelm
parents: 34168
diff changeset
    23
48599
5e64b7770f35 tuned signature;
wenzelm
parents: 48484
diff changeset
    24
    @tailrec private def proper(in: Input): Input =
5e64b7770f35 tuned signature;
wenzelm
parents: 48484
diff changeset
    25
      if (!filter_proper || in.atEnd || in.first.is_proper) in
34161
4c845a8f1357 consider proper input only;
wenzelm
parents: 34159
diff changeset
    26
      else proper(in.rest)
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    27
59692
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    28
    private def proper_position: Parser[Position.T] =
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    29
      new Parser[Position.T] {
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    30
        def apply(raw_input: Input) =
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    31
        {
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    32
          val in = proper(raw_input)
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    33
          val pos =
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    34
            in.pos match {
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    35
              case pos: Token.Pos => pos
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    36
              case _ => Token.Pos.none
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    37
            }
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    38
          Success(if (in.atEnd) pos.position() else pos.position(in.first), in)
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    39
        }
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    40
      }
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    41
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    42
    def position[A](parser: Parser[A]): Parser[(A, Position.T)] =
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    43
      proper_position ~ parser ^^ { case x ~ y => (y, x) }
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    44
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    45
    def token(s: String, pred: Elem => Boolean): Parser[Elem] =
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    46
      new Parser[Elem] {
56464
555f4be59be6 more precise token positions;
wenzelm
parents: 56209
diff changeset
    47
        def apply(raw_input: Input) =
555f4be59be6 more precise token positions;
wenzelm
parents: 56209
diff changeset
    48
        {
555f4be59be6 more precise token positions;
wenzelm
parents: 56209
diff changeset
    49
          val in = proper(raw_input)
555f4be59be6 more precise token positions;
wenzelm
parents: 56209
diff changeset
    50
          if (in.atEnd) Failure(s + " expected,\nbut end-of-input was found", in)
555f4be59be6 more precise token positions;
wenzelm
parents: 56209
diff changeset
    51
          else {
555f4be59be6 more precise token positions;
wenzelm
parents: 56209
diff changeset
    52
            val token = in.first
59692
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    53
            if (pred(token)) Success(token, proper(in.rest))
58861
5ff61774df11 command-line terminator ";" is no longer accepted;
wenzelm
parents: 56801
diff changeset
    54
            else Failure(s + " expected,\nbut " + token.kind + " was found:\n" + token.source, in)
56464
555f4be59be6 more precise token positions;
wenzelm
parents: 56209
diff changeset
    55
          }
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    56
        }
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    57
      }
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    58
34168
18843829c7f2 clarified atom parser: return content;
wenzelm
parents: 34161
diff changeset
    59
    def atom(s: String, pred: Elem => Boolean): Parser[String] =
59692
03aa1b63af10 position parser as in ML;
wenzelm
parents: 59671
diff changeset
    60
      token(s, pred) ^^ (_.content)
34168
18843829c7f2 clarified atom parser: return content;
wenzelm
parents: 34161
diff changeset
    61
63446
19162a9ef7e3 tunes signature;
wenzelm
parents: 62969
diff changeset
    62
    def command(name: String): Parser[String] = atom("command " + quote(name), _.is_command(name))
19162a9ef7e3 tunes signature;
wenzelm
parents: 62969
diff changeset
    63
    def $$$(name: String): Parser[String] = atom("keyword " + quote(name), _.is_keyword(name))
46943
ac1c41ea856d clarified syntax of prospective keywords;
wenzelm
parents: 44181
diff changeset
    64
    def string: Parser[String] = atom("string", _.is_string)
48349
a78e5d399599 support Session.Queue with ordering and dependencies;
wenzelm
parents: 46943
diff changeset
    65
    def nat: Parser[Int] = atom("natural number", _.is_nat) ^^ (s => Integer.parseInt(s))
62969
9f394a16c557 eliminated "xname" and variants;
wenzelm
parents: 62965
diff changeset
    66
    def name: Parser[String] = atom("name", _.is_name)
34168
18843829c7f2 clarified atom parser: return content;
wenzelm
parents: 34161
diff changeset
    67
    def text: Parser[String] = atom("text", _.is_text)
18843829c7f2 clarified atom parser: return content;
wenzelm
parents: 34161
diff changeset
    68
    def ML_source: Parser[String] = atom("ML source", _.is_text)
51627
589daaf48dba tuned signature -- agree with markup terminology;
wenzelm
parents: 48912
diff changeset
    69
    def document_source: Parser[String] = atom("document source", _.is_text)
59693
d96cb03caf9e tunes signature -- more uniform ML vs. Scala;
wenzelm
parents: 59692
diff changeset
    70
48484
70898d016538 more explicit checks during parsing;
wenzelm
parents: 48349
diff changeset
    71
    def path: Parser[String] =
62969
9f394a16c557 eliminated "xname" and variants;
wenzelm
parents: 62965
diff changeset
    72
      atom("file name/path specification", tok => tok.is_name && Path.is_wellformed(tok.content))
59693
d96cb03caf9e tunes signature -- more uniform ML vs. Scala;
wenzelm
parents: 59692
diff changeset
    73
48484
70898d016538 more explicit checks during parsing;
wenzelm
parents: 48349
diff changeset
    74
    def theory_name: Parser[String] =
56209
3c89e21d9be2 simplified (despite 70898d016538);
wenzelm
parents: 56208
diff changeset
    75
      atom("theory name", tok => tok.is_name && Path.is_wellformed(tok.content))
34168
18843829c7f2 clarified atom parser: return content;
wenzelm
parents: 34161
diff changeset
    76
18843829c7f2 clarified atom parser: return content;
wenzelm
parents: 34161
diff changeset
    77
    private def tag_name: Parser[String] =
18843829c7f2 clarified atom parser: return content;
wenzelm
parents: 34161
diff changeset
    78
      atom("tag name", tok =>
36956
21be4832c362 renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
wenzelm
parents: 36948
diff changeset
    79
          tok.kind == Token.Kind.IDENT ||
21be4832c362 renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
wenzelm
parents: 36948
diff changeset
    80
          tok.kind == Token.Kind.STRING)
34168
18843829c7f2 clarified atom parser: return content;
wenzelm
parents: 34161
diff changeset
    81
58908
58bedbc18915 tuned signature;
wenzelm
parents: 58861
diff changeset
    82
    def tags: Parser[List[String]] = rep($$$("%") ~> tag_name)
34161
4c845a8f1357 consider proper input only;
wenzelm
parents: 34159
diff changeset
    83
4c845a8f1357 consider proper input only;
wenzelm
parents: 34159
diff changeset
    84
4c845a8f1357 consider proper input only;
wenzelm
parents: 34159
diff changeset
    85
    /* wrappers */
4c845a8f1357 consider proper input only;
wenzelm
parents: 34159
diff changeset
    86
36956
21be4832c362 renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
wenzelm
parents: 36948
diff changeset
    87
    def parse[T](p: Parser[T], in: Token.Reader): ParseResult[T] = p(in)
48912
ffdb37019b2f improved errors of parser combinators;
wenzelm
parents: 48911
diff changeset
    88
48600
305ebcd9018a proper treatment of eof wrt. proper_input -- allow input of spaces/comments only;
wenzelm
parents: 48599
diff changeset
    89
    def parse_all[T](p: Parser[T], in: Token.Reader): ParseResult[T] =
48912
ffdb37019b2f improved errors of parser combinators;
wenzelm
parents: 48911
diff changeset
    90
    {
ffdb37019b2f improved errors of parser combinators;
wenzelm
parents: 48911
diff changeset
    91
      val result = parse(p, in)
ffdb37019b2f improved errors of parser combinators;
wenzelm
parents: 48911
diff changeset
    92
      val rest = proper(result.next)
ffdb37019b2f improved errors of parser combinators;
wenzelm
parents: 48911
diff changeset
    93
      if (result.successful && !rest.atEnd) Error("bad input", rest)
ffdb37019b2f improved errors of parser combinators;
wenzelm
parents: 48911
diff changeset
    94
      else result
ffdb37019b2f improved errors of parser combinators;
wenzelm
parents: 48911
diff changeset
    95
    }
34159
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    96
  }
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    97
}
903092d61519 Generic parsers for Isabelle/Isar outer syntax -- Scala version.
wenzelm
parents:
diff changeset
    98