| author | nipkow | 
| Thu, 03 Feb 2022 10:33:55 +0100 | |
| changeset 75061 | 57df04e4f018 | 
| parent 74887 | 56247fdb8bbb | 
| child 75393 | 87ebf5a50283 | 
| permissions | -rw-r--r-- | 
| 43283 | 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 | 9  | 
|
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
10  | 
import scala.util.parsing.combinator.Parsers  | 
| 48599 | 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 | 14  | 
object Parse  | 
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
15  | 
{
 | 
| 34161 | 16  | 
/* parsing tokens */  | 
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 | 22  | 
def filter_proper: Boolean = true  | 
| 34266 | 23  | 
|
| 48599 | 24  | 
@tailrec private def proper(in: Input): Input =  | 
25  | 
if (!filter_proper || in.atEnd || in.first.is_proper) in  | 
|
| 34161 | 26  | 
else proper(in.rest)  | 
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
27  | 
|
| 59692 | 28  | 
private def proper_position: Parser[Position.T] =  | 
29  | 
      new Parser[Position.T] {
 | 
|
30  | 
def apply(raw_input: Input) =  | 
|
31  | 
        {
 | 
|
32  | 
val in = proper(raw_input)  | 
|
33  | 
val pos =  | 
|
34  | 
            in.pos match {
 | 
|
35  | 
case pos: Token.Pos => pos  | 
|
36  | 
case _ => Token.Pos.none  | 
|
37  | 
}  | 
|
38  | 
Success(if (in.atEnd) pos.position() else pos.position(in.first), in)  | 
|
39  | 
}  | 
|
40  | 
}  | 
|
41  | 
||
42  | 
def position[A](parser: Parser[A]): Parser[(A, Position.T)] =  | 
|
43  | 
      proper_position ~ parser ^^ { case x ~ y => (y, x) }
 | 
|
44  | 
||
45  | 
def token(s: String, pred: Elem => Boolean): Parser[Elem] =  | 
|
46  | 
      new Parser[Elem] {
 | 
|
| 56464 | 47  | 
def apply(raw_input: Input) =  | 
48  | 
        {
 | 
|
49  | 
val in = proper(raw_input)  | 
|
50  | 
if (in.atEnd) Failure(s + " expected,\nbut end-of-input was found", in)  | 
|
51  | 
          else {
 | 
|
52  | 
val token = in.first  | 
|
| 59692 | 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 | 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 | 59  | 
def atom(s: String, pred: Elem => Boolean): Parser[String] =  | 
| 59692 | 60  | 
token(s, pred) ^^ (_.content)  | 
| 34168 | 61  | 
|
| 63446 | 62  | 
    def command(name: String): Parser[String] = atom("command " + quote(name), _.is_command(name))
 | 
63  | 
    def $$$(name: String): Parser[String] = atom("keyword " + quote(name), _.is_keyword(name))
 | 
|
| 46943 | 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 | 66  | 
    def name: Parser[String] = atom("name", _.is_name)
 | 
| 
64471
 
c40c2975fb02
more uniform path syntax, as in ML (see 5a7c919a4ada);
 
wenzelm 
parents: 
63446 
diff
changeset
 | 
67  | 
    def embedded: Parser[String] = atom("embedded content", _.is_embedded)
 | 
| 74887 | 68  | 
    def text: Parser[String] = atom("text", _.is_embedded)
 | 
69  | 
    def ML_source: Parser[String] = atom("ML source", _.is_embedded)
 | 
|
70  | 
    def document_source: Parser[String] = atom("document source", _.is_embedded)
 | 
|
| 59693 | 71  | 
|
| 70668 | 72  | 
def opt_keyword(s: String): Parser[Boolean] =  | 
73  | 
      ($$$("(") ~! $$$(s) ~ $$$(")")) ^^ { case _ => true } | success(false)
 | 
|
74  | 
||
| 48484 | 75  | 
def path: Parser[String] =  | 
| 
64471
 
c40c2975fb02
more uniform path syntax, as in ML (see 5a7c919a4ada);
 
wenzelm 
parents: 
63446 
diff
changeset
 | 
76  | 
      atom("file name/path specification", tok => tok.is_embedded && Path.is_wellformed(tok.content))
 | 
| 59693 | 77  | 
|
| 66914 | 78  | 
    def session_name: Parser[String] = atom("session name", _.is_system_name)
 | 
79  | 
    def theory_name: Parser[String] = atom("theory name", _.is_system_name)
 | 
|
| 34168 | 80  | 
|
81  | 
private def tag_name: Parser[String] =  | 
|
82  | 
      atom("tag name", tok =>
 | 
|
| 
36956
 
21be4832c362
renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
 
wenzelm 
parents: 
36948 
diff
changeset
 | 
83  | 
tok.kind == Token.Kind.IDENT ||  | 
| 
 
21be4832c362
renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
 
wenzelm 
parents: 
36948 
diff
changeset
 | 
84  | 
tok.kind == Token.Kind.STRING)  | 
| 34168 | 85  | 
|
| 69887 | 86  | 
    def tag: Parser[String] = $$$("%") ~> tag_name
 | 
87  | 
def tags: Parser[List[String]] = rep(tag)  | 
|
88  | 
||
| 
69891
 
def3ec9cdb7e
document markers are formal comments, and may thus occur anywhere in the command-span;
 
wenzelm 
parents: 
69887 
diff
changeset
 | 
89  | 
    def marker: Parser[String] = token("marker", _.is_marker) ^^ (_.content)
 | 
| 69887 | 90  | 
|
| 
69891
 
def3ec9cdb7e
document markers are formal comments, and may thus occur anywhere in the command-span;
 
wenzelm 
parents: 
69887 
diff
changeset
 | 
91  | 
    def annotation: Parser[Unit] = rep(tag | marker) ^^ { case _ => () }
 | 
| 34161 | 92  | 
|
93  | 
||
94  | 
/* wrappers */  | 
|
95  | 
||
| 
36956
 
21be4832c362
renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
 
wenzelm 
parents: 
36948 
diff
changeset
 | 
96  | 
def parse[T](p: Parser[T], in: Token.Reader): ParseResult[T] = p(in)  | 
| 48912 | 97  | 
|
| 
48600
 
305ebcd9018a
proper treatment of eof wrt. proper_input -- allow input of spaces/comments only;
 
wenzelm 
parents: 
48599 
diff
changeset
 | 
98  | 
def parse_all[T](p: Parser[T], in: Token.Reader): ParseResult[T] =  | 
| 48912 | 99  | 
    {
 | 
100  | 
val result = parse(p, in)  | 
|
101  | 
val rest = proper(result.next)  | 
|
102  | 
      if (result.successful && !rest.atEnd) Error("bad input", rest)
 | 
|
103  | 
else result  | 
|
104  | 
}  | 
|
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
105  | 
}  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
106  | 
}  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
107  |