| author | wenzelm | 
| Thu, 20 Feb 2014 14:36:17 +0100 | |
| changeset 55618 | 995162143ef4 | 
| parent 51627 | 589daaf48dba | 
| child 55879 | ac979f750c1a | 
| 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  | 
|
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
28  | 
def token(s: String, pred: Elem => Boolean): Parser[Elem] = new Parser[Elem]  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
29  | 
    {
 | 
| 34161 | 30  | 
def apply(raw_input: Input) =  | 
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
31  | 
      {
 | 
| 34161 | 32  | 
val in = proper(raw_input)  | 
| 
48911
 
5debc3e4fa81
tuned messages: end-of-input rarely means physical end-of-file from the past;
 
wenzelm 
parents: 
48718 
diff
changeset
 | 
33  | 
if (in.atEnd) Failure(s + " expected,\nbut end-of-input was found", in)  | 
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
34  | 
        else {
 | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
35  | 
val token = in.first  | 
| 34161 | 36  | 
if (pred(token)) Success(token, proper(in.rest))  | 
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
37  | 
else  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
38  | 
            token.text match {
 | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
39  | 
case (txt, "") =>  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
40  | 
Failure(s + " expected,\nbut " + txt + " was found", in)  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
41  | 
case (txt1, txt2) =>  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
42  | 
Failure(s + " expected,\nbut " + txt1 + " was found:\n" + txt2, in)  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
43  | 
}  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
44  | 
}  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
45  | 
}  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
46  | 
}  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
47  | 
|
| 34168 | 48  | 
def atom(s: String, pred: Elem => Boolean): Parser[String] =  | 
49  | 
token(s, pred) ^^ (_.content)  | 
|
50  | 
||
| 48718 | 51  | 
def command(name: String): Parser[String] =  | 
52  | 
      atom("command " + quote(name), tok => tok.is_command && tok.source == name)
 | 
|
53  | 
||
| 34168 | 54  | 
def keyword(name: String): Parser[String] =  | 
| 48718 | 55  | 
      atom("keyword " + quote(name), tok => tok.is_keyword && tok.source == name)
 | 
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
56  | 
|
| 46943 | 57  | 
    def string: Parser[String] = atom("string", _.is_string)
 | 
| 
48349
 
a78e5d399599
support Session.Queue with ordering and dependencies;
 
wenzelm 
parents: 
46943 
diff
changeset
 | 
58  | 
    def nat: Parser[Int] = atom("natural number", _.is_nat) ^^ (s => Integer.parseInt(s))
 | 
| 34168 | 59  | 
    def name: Parser[String] = atom("name declaration", _.is_name)
 | 
60  | 
    def xname: Parser[String] = atom("name reference", _.is_xname)
 | 
|
61  | 
    def text: Parser[String] = atom("text", _.is_text)
 | 
|
62  | 
    def ML_source: Parser[String] = atom("ML source", _.is_text)
 | 
|
| 
51627
 
589daaf48dba
tuned signature -- agree with markup terminology;
 
wenzelm 
parents: 
48912 
diff
changeset
 | 
63  | 
    def document_source: Parser[String] = atom("document source", _.is_text)
 | 
| 48484 | 64  | 
def path: Parser[String] =  | 
65  | 
      atom("file name/path specification", tok => tok.is_name && Path.is_ok(tok.content))
 | 
|
66  | 
def theory_name: Parser[String] =  | 
|
67  | 
      atom("theory name", tok => tok.is_name && Thy_Load.is_ok(tok.content))
 | 
|
| 34168 | 68  | 
|
69  | 
private def tag_name: Parser[String] =  | 
|
70  | 
      atom("tag name", tok =>
 | 
|
| 
36956
 
21be4832c362
renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
 
wenzelm 
parents: 
36948 
diff
changeset
 | 
71  | 
tok.kind == Token.Kind.IDENT ||  | 
| 
 
21be4832c362
renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
 
wenzelm 
parents: 
36948 
diff
changeset
 | 
72  | 
tok.kind == Token.Kind.STRING)  | 
| 34168 | 73  | 
|
74  | 
    def tags: Parser[List[String]] = rep(keyword("%") ~> tag_name)
 | 
|
| 34161 | 75  | 
|
76  | 
||
77  | 
/* wrappers */  | 
|
78  | 
||
| 
36956
 
21be4832c362
renamed class Outer_Lex to Token and Token_Kind to Token.Kind;
 
wenzelm 
parents: 
36948 
diff
changeset
 | 
79  | 
def parse[T](p: Parser[T], in: Token.Reader): ParseResult[T] = p(in)  | 
| 48912 | 80  | 
|
| 
48600
 
305ebcd9018a
proper treatment of eof wrt. proper_input -- allow input of spaces/comments only;
 
wenzelm 
parents: 
48599 
diff
changeset
 | 
81  | 
def parse_all[T](p: Parser[T], in: Token.Reader): ParseResult[T] =  | 
| 48912 | 82  | 
    {
 | 
83  | 
val result = parse(p, in)  | 
|
84  | 
val rest = proper(result.next)  | 
|
85  | 
      if (result.successful && !rest.atEnd) Error("bad input", rest)
 | 
|
86  | 
else result  | 
|
87  | 
}  | 
|
| 
34159
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
88  | 
}  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
89  | 
}  | 
| 
 
903092d61519
Generic parsers for Isabelle/Isar outer syntax -- Scala version.
 
wenzelm 
parents:  
diff
changeset
 | 
90  |