src/Pure/Isar/outer_parse.ML
author wenzelm
Mon Nov 05 21:00:45 2001 +0100 (2001-11-05)
changeset 12056 5b5ed7eec3a8
parent 12047 e151e66da2d6
child 12064 34c270893ecb
permissions -rw-r--r--
locale_element: optional typ;
wenzelm@5826
     1
(*  Title:      Pure/Isar/outer_parse.ML
wenzelm@5826
     2
    ID:         $Id$
wenzelm@5826
     3
    Author:     Markus Wenzel, TU Muenchen
wenzelm@8807
     4
    License:    GPL (GNU GENERAL PUBLIC LICENSE)
wenzelm@5826
     5
wenzelm@5826
     6
Generic parsers for Isabelle/Isar outer syntax.
wenzelm@5826
     7
*)
wenzelm@5826
     8
wenzelm@5826
     9
signature OUTER_PARSE =
wenzelm@5826
    10
sig
wenzelm@5826
    11
  type token
wenzelm@5826
    12
  val group: string -> (token list -> 'a) -> token list -> 'a
wenzelm@5826
    13
  val !!! : (token list -> 'a) -> token list -> 'a
wenzelm@8581
    14
  val !!!! : (token list -> 'a) -> token list -> 'a
wenzelm@12047
    15
  val triple1: ('a * 'b) * 'c -> 'a * 'b * 'c
wenzelm@12047
    16
  val triple2: 'a * ('b * 'c) -> 'a * 'b * 'c
wenzelm@12047
    17
  val triple_swap: ('a * 'b) * 'c -> ('a * 'c) * 'b
wenzelm@5826
    18
  val $$$ : string -> token list -> string * token list
wenzelm@9131
    19
  val semicolon: token list -> string * token list
wenzelm@11792
    20
  val underscore: token list -> string * token list
wenzelm@5826
    21
  val position: (token list -> 'a * 'b) -> token list -> ('a * Position.T) * 'b
wenzelm@7026
    22
  val command: token list -> string * token list
wenzelm@5826
    23
  val keyword: token list -> string * token list
wenzelm@5826
    24
  val short_ident: token list -> string * token list
wenzelm@5826
    25
  val long_ident: token list -> string * token list
wenzelm@5826
    26
  val sym_ident: token list -> string * token list
wenzelm@5826
    27
  val term_var: token list -> string * token list
wenzelm@5826
    28
  val type_ident: token list -> string * token list
wenzelm@5826
    29
  val type_var: token list -> string * token list
wenzelm@5826
    30
  val number: token list -> string * token list
wenzelm@5826
    31
  val string: token list -> string * token list
wenzelm@5826
    32
  val verbatim: token list -> string * token list
wenzelm@6860
    33
  val sync: token list -> string * token list
wenzelm@5826
    34
  val eof: token list -> string * token list
wenzelm@5826
    35
  val not_eof: token list -> token * token list
wenzelm@7930
    36
  val opt_unit: token list -> unit * token list
wenzelm@5826
    37
  val nat: token list -> int * token list
wenzelm@5826
    38
  val enum: string -> (token list -> 'a * token list) -> token list -> 'a list * token list
wenzelm@5826
    39
  val enum1: string -> (token list -> 'a * token list) -> token list -> 'a list * token list
wenzelm@5826
    40
  val list: (token list -> 'a * token list) -> token list -> 'a list * token list
wenzelm@5826
    41
  val list1: (token list -> 'a * token list) -> token list -> 'a list * token list
wenzelm@6013
    42
  val and_list: (token list -> 'a * token list) -> token list -> 'a list * token list
wenzelm@6013
    43
  val and_list1: (token list -> 'a * token list) -> token list -> 'a list * token list
wenzelm@5826
    44
  val name: token list -> bstring * token list
wenzelm@5826
    45
  val xname: token list -> xstring * token list
wenzelm@5826
    46
  val text: token list -> string * token list
wenzelm@6553
    47
  val comment: token list -> Comment.text * token list
wenzelm@6553
    48
  val marg_comment: token list -> Comment.text * token list
wenzelm@6553
    49
  val interest: token list -> Comment.interest * token list
wenzelm@8897
    50
  val sort: token list -> string * token list
wenzelm@8897
    51
  val arity: token list -> (string list * string) * token list
wenzelm@8897
    52
  val simple_arity: token list -> (string list * xclass) * token list
wenzelm@5826
    53
  val type_args: token list -> string list * token list
wenzelm@5826
    54
  val typ: token list -> string * token list
wenzelm@5826
    55
  val opt_infix: token list -> Syntax.mixfix * token list
wenzelm@5826
    56
  val opt_mixfix: token list -> Syntax.mixfix * token list
wenzelm@9037
    57
  val opt_infix': token list -> Syntax.mixfix * token list
wenzelm@9037
    58
  val opt_mixfix': token list -> Syntax.mixfix * token list
wenzelm@5826
    59
  val const: token list -> (bstring * string * Syntax.mixfix) * token list
wenzelm@5826
    60
  val term: token list -> string * token list
wenzelm@5826
    61
  val prop: token list -> string * token list
wenzelm@6935
    62
  val propp: token list -> (string * (string list * string list)) * token list
wenzelm@6949
    63
  val termp: token list -> (string * string list) * token list
wenzelm@9131
    64
  val arguments: token list -> Args.T list * token list
wenzelm@5826
    65
  val attribs: token list -> Args.src list * token list
wenzelm@5826
    66
  val opt_attribs: token list -> Args.src list * token list
wenzelm@5917
    67
  val thm_name: string -> token list -> (bstring * Args.src list) * token list
wenzelm@5917
    68
  val opt_thm_name: string -> token list -> (bstring * Args.src list) * token list
wenzelm@6372
    69
  val spec_name: token list -> ((bstring * string) * Args.src list) * token list
wenzelm@6372
    70
  val spec_opt_name: token list -> ((bstring * string) * Args.src list) * token list
wenzelm@5917
    71
  val xthm: token list -> (xstring * Args.src list) * token list
wenzelm@5917
    72
  val xthms1: token list -> (xstring * Args.src list) list * token list
wenzelm@12047
    73
  val locale_element: token list -> Args.src Locale.element * token list
wenzelm@5826
    74
  val method: token list -> Method.text * token list
wenzelm@5826
    75
end;
wenzelm@5826
    76
wenzelm@5826
    77
structure OuterParse: OUTER_PARSE =
wenzelm@5826
    78
struct
wenzelm@5826
    79
wenzelm@9131
    80
structure T = OuterLex;
wenzelm@9131
    81
type token = T.token;
wenzelm@5826
    82
wenzelm@5826
    83
wenzelm@5826
    84
(** error handling **)
wenzelm@5826
    85
wenzelm@5826
    86
(* group atomic parsers (no cuts!) *)
wenzelm@5826
    87
wenzelm@5826
    88
fun fail_with s = Scan.fail_with
wenzelm@5826
    89
  (fn [] => s ^ " expected (past end-of-file!)"
wenzelm@9155
    90
    | (tok :: _) => s ^ " expected,\nbut " ^ T.name_of tok ^ T.pos_of tok ^ " was found");
wenzelm@5826
    91
wenzelm@5826
    92
fun group s scan = scan || fail_with s;
wenzelm@5826
    93
wenzelm@5826
    94
wenzelm@5877
    95
(* cut *)
wenzelm@5826
    96
wenzelm@8581
    97
fun cut kind scan =
wenzelm@5826
    98
  let
wenzelm@5826
    99
    fun get_pos [] = " (past end-of-file!)"
wenzelm@9131
   100
      | get_pos (tok :: _) = T.pos_of tok;
wenzelm@5826
   101
wenzelm@8581
   102
    fun err (toks, None) = kind ^ get_pos toks
wenzelm@8581
   103
      | err (toks, Some msg) = kind ^ get_pos toks ^ ": " ^ msg;
wenzelm@5826
   104
  in Scan.!! err scan end;
wenzelm@5826
   105
wenzelm@8586
   106
fun !!! scan = cut "Outer syntax error" scan;
wenzelm@8586
   107
fun !!!! scan = cut "Corrupted outer syntax in presentation" scan;
wenzelm@8581
   108
wenzelm@5826
   109
wenzelm@5826
   110
wenzelm@5826
   111
(** basic parsers **)
wenzelm@5826
   112
wenzelm@5826
   113
(* utils *)
wenzelm@5826
   114
wenzelm@5826
   115
fun triple1 ((x, y), z) = (x, y, z);
wenzelm@5826
   116
fun triple2 (x, (y, z)) = (x, y, z);
wenzelm@6430
   117
fun triple_swap ((x, y), z) = ((x, z), y);
wenzelm@5826
   118
wenzelm@5826
   119
wenzelm@5826
   120
(* tokens *)
wenzelm@5826
   121
wenzelm@5826
   122
fun position scan =
wenzelm@9131
   123
  (Scan.ahead (Scan.one T.not_eof) >> T.position_of) -- scan >> Library.swap;
wenzelm@5826
   124
wenzelm@5826
   125
fun kind k =
wenzelm@9131
   126
  group (T.str_of_kind k) (Scan.one (T.is_kind k) >> T.val_of);
wenzelm@5826
   127
wenzelm@9131
   128
val command = kind T.Command;
wenzelm@9131
   129
val keyword = kind T.Keyword;
wenzelm@9131
   130
val short_ident = kind T.Ident;
wenzelm@9131
   131
val long_ident = kind T.LongIdent;
wenzelm@9131
   132
val sym_ident = kind T.SymIdent;
wenzelm@9131
   133
val term_var = kind T.Var;
wenzelm@9131
   134
val type_ident = kind T.TypeIdent;
wenzelm@9131
   135
val type_var = kind T.TypeVar;
wenzelm@9131
   136
val number = kind T.Nat;
wenzelm@9131
   137
val string = kind T.String;
wenzelm@9131
   138
val verbatim = kind T.Verbatim;
wenzelm@9131
   139
val sync = kind T.Sync;
wenzelm@9131
   140
val eof = kind T.EOF;
wenzelm@5826
   141
wenzelm@5826
   142
fun $$$ x =
wenzelm@9131
   143
  group (T.str_of_kind T.Keyword ^ " " ^ quote x)
wenzelm@9131
   144
    (Scan.one (T.keyword_with (equal x)) >> T.val_of);
wenzelm@9131
   145
wenzelm@9131
   146
val semicolon = $$$ ";";
wenzelm@5826
   147
wenzelm@11792
   148
val underscore = sym_ident :-- (fn "_" => Scan.succeed () | _ => Scan.fail) >> #1;
wenzelm@11792
   149
wenzelm@5877
   150
val nat = number >> (fst o Term.read_int o Symbol.explode);
wenzelm@5826
   151
wenzelm@9131
   152
val not_eof = Scan.one T.not_eof;
wenzelm@5826
   153
wenzelm@7930
   154
val opt_unit = Scan.optional ($$$ "(" -- $$$ ")" >> (K ())) ();
wenzelm@7930
   155
wenzelm@5826
   156
wenzelm@5826
   157
(* enumerations *)
wenzelm@5826
   158
wenzelm@6003
   159
fun enum1 sep scan = scan -- Scan.repeat ($$$ sep |-- !!! scan) >> op ::;
wenzelm@5826
   160
fun enum sep scan = enum1 sep scan || Scan.succeed [];
wenzelm@5826
   161
wenzelm@5826
   162
fun list1 scan = enum1 "," scan;
wenzelm@5826
   163
fun list scan = enum "," scan;
wenzelm@5826
   164
wenzelm@6013
   165
fun and_list1 scan = enum1 "and" scan;
wenzelm@6013
   166
fun and_list scan = enum "and" scan;
wenzelm@6013
   167
wenzelm@5826
   168
wenzelm@5960
   169
(* names and text *)
wenzelm@5826
   170
wenzelm@8146
   171
val name = group "name declaration" (short_ident || sym_ident || string || number);
wenzelm@8146
   172
val xname = group "name reference" (short_ident || long_ident || sym_ident || string || number);
wenzelm@8146
   173
val text = group "text" (short_ident || long_ident || sym_ident || string || number || verbatim);
wenzelm@6553
   174
wenzelm@6553
   175
wenzelm@6553
   176
(* formal comments *)
wenzelm@6553
   177
wenzelm@8077
   178
val comment = text >> (Comment.plain o Library.single);
wenzelm@8077
   179
val marg_comment = Scan.repeat ($$$ "--" |-- text) >> Comment.plain;
wenzelm@7171
   180
wenzelm@7171
   181
val interest = Scan.optional ($$$ "%%" >> K Comment.no_interest ||
wenzelm@7171
   182
  $$$ "%" |-- Scan.optional nat 1 >> Comment.interest) Comment.default_interest;
wenzelm@5826
   183
wenzelm@5826
   184
wenzelm@6372
   185
(* sorts and arities *)
wenzelm@5826
   186
wenzelm@8897
   187
val sort = group "sort" xname;
wenzelm@5826
   188
wenzelm@6372
   189
fun gen_arity cod =
wenzelm@7352
   190
  Scan.optional ($$$ "(" |-- !!! (list1 sort --| $$$ ")")) [] -- cod;
wenzelm@6372
   191
wenzelm@6372
   192
val arity = gen_arity sort;
wenzelm@8897
   193
val simple_arity = gen_arity xname;
wenzelm@5826
   194
wenzelm@5826
   195
wenzelm@5826
   196
(* types *)
wenzelm@5826
   197
wenzelm@8146
   198
val typ = group "type"
wenzelm@8146
   199
  (short_ident || long_ident || sym_ident || type_ident || type_var || string || number);
wenzelm@5826
   200
wenzelm@5826
   201
val type_args =
wenzelm@5826
   202
  type_ident >> single ||
wenzelm@5826
   203
  $$$ "(" |-- !!! (list1 type_ident --| $$$ ")") ||
wenzelm@5826
   204
  Scan.succeed [];
wenzelm@5826
   205
wenzelm@5826
   206
wenzelm@5826
   207
(* mixfix annotations *)
wenzelm@5826
   208
wenzelm@11651
   209
val infx = $$$ "infix" |-- !!! (nat >> Syntax.Infix || string -- nat >> Syntax.InfixName);
wenzelm@5826
   210
val infxl = $$$ "infixl" |-- !!! (nat >> Syntax.Infixl || string -- nat >> Syntax.InfixlName);
wenzelm@5826
   211
val infxr = $$$ "infixr" |-- !!! (nat >> Syntax.Infixr || string -- nat >> Syntax.InfixrName);
wenzelm@5826
   212
wenzelm@5826
   213
val binder =
wenzelm@5826
   214
  $$$ "binder" |--
wenzelm@5826
   215
    !!! (string -- ($$$ "[" |-- nat --| $$$ "]" -- nat || nat >> (fn n => (n, n))))
wenzelm@5826
   216
  >> (Syntax.Binder o triple2);
wenzelm@5826
   217
wenzelm@5826
   218
wenzelm@5826
   219
val opt_pris = Scan.optional ($$$ "[" |-- !!! (list nat --| $$$ "]")) [];
wenzelm@5826
   220
wenzelm@5826
   221
val mixfix =
wenzelm@8648
   222
  (string -- !!! (opt_pris -- Scan.optional nat Syntax.max_pri))
wenzelm@8648
   223
  >> (Syntax.Mixfix o triple2);
wenzelm@5826
   224
wenzelm@9037
   225
fun opt_fix guard fix =
wenzelm@9037
   226
  Scan.optional ($$$ "(" |-- guard (fix --| $$$ ")")) Syntax.NoSyn;
wenzelm@5826
   227
wenzelm@11651
   228
val opt_infix = opt_fix !!! (infxl || infxr || infx);
wenzelm@11651
   229
val opt_mixfix = opt_fix !!! (mixfix || binder || infxl || infxr || infx);
wenzelm@9037
   230
wenzelm@11651
   231
val opt_infix' = opt_fix I (infxl || infxr || infx);
wenzelm@11651
   232
val opt_mixfix' = opt_fix I (mixfix || binder || infxl || infxr || infx);
wenzelm@5826
   233
wenzelm@5826
   234
wenzelm@5826
   235
(* consts *)
wenzelm@5826
   236
wenzelm@5826
   237
val const =
wenzelm@5826
   238
  name -- ($$$ "::" |-- !!! (typ -- opt_mixfix)) >> triple2;
wenzelm@5826
   239
wenzelm@5826
   240
wenzelm@5826
   241
(* terms *)
wenzelm@5826
   242
wenzelm@7477
   243
val trm = short_ident || long_ident || sym_ident || term_var || number || string;
wenzelm@5826
   244
wenzelm@5826
   245
val term = group "term" trm;
wenzelm@5826
   246
val prop = group "proposition" trm;
wenzelm@5826
   247
wenzelm@5826
   248
wenzelm@6949
   249
(* patterns *)
wenzelm@6935
   250
wenzelm@6949
   251
val is_terms = Scan.repeat1 ($$$ "is" |-- term);
wenzelm@6935
   252
val is_props = Scan.repeat1 ($$$ "is" |-- prop);
wenzelm@6935
   253
val concl_props = $$$ "concl" |-- !!! is_props;
wenzelm@7418
   254
val any_props = concl_props >> pair [] || is_props -- Scan.optional concl_props [];
wenzelm@6935
   255
wenzelm@6935
   256
val propp = prop -- Scan.optional ($$$ "(" |-- !!! (any_props --| $$$ ")")) ([], []);
wenzelm@12047
   257
val propp' = prop -- Scan.optional ($$$ "(" |-- !!! (is_props --| $$$ ")")) [];
wenzelm@6949
   258
val termp = term -- Scan.optional ($$$ "(" |-- !!! (is_terms --| $$$ ")")) [];
wenzelm@6935
   259
wenzelm@6935
   260
wenzelm@5826
   261
(* arguments *)
wenzelm@5826
   262
wenzelm@9131
   263
fun keyword_symid is_symid = Scan.one (T.keyword_with is_symid) >> T.val_of;
wenzelm@9131
   264
val keyword_sid = keyword_symid T.is_sid;
wenzelm@5826
   265
wenzelm@6983
   266
fun atom_arg is_symid blk =
wenzelm@5826
   267
  group "argument"
wenzelm@7477
   268
    (position (short_ident || long_ident || sym_ident || term_var ||
wenzelm@5877
   269
        type_ident || type_var || number) >> Args.ident ||
wenzelm@6983
   270
      position (keyword_symid is_symid) >> Args.keyword ||
wenzelm@8350
   271
      position (string || verbatim) >> Args.string ||
wenzelm@5877
   272
      position (if blk then $$$ "," else Scan.fail) >> Args.keyword);
wenzelm@5826
   273
wenzelm@5877
   274
fun paren_args l r scan = position ($$$ l) -- !!! (scan true -- position ($$$ r))
wenzelm@5826
   275
  >> (fn (x, (ys, z)) => Args.keyword x :: ys @ [Args.keyword z]);
wenzelm@5826
   276
wenzelm@6983
   277
fun args is_symid blk x = Scan.optional (args1 is_symid blk) [] x
wenzelm@6983
   278
and args1 is_symid blk x =
wenzelm@5826
   279
  ((Scan.repeat1
wenzelm@6983
   280
    (Scan.repeat1 (atom_arg is_symid blk) ||
wenzelm@6983
   281
      paren_args "(" ")" (args is_symid) ||
wenzelm@6983
   282
      paren_args "[" "]" (args is_symid))) >> flat) x;
wenzelm@5826
   283
wenzelm@9131
   284
val arguments = args T.is_sid false;
wenzelm@9131
   285
wenzelm@5826
   286
wenzelm@6372
   287
(* theorem specifications *)
wenzelm@5826
   288
wenzelm@9131
   289
val attrib = position ((keyword_sid || xname) -- !!! arguments) >> Args.src;
wenzelm@5826
   290
val attribs = $$$ "[" |-- !!! (list attrib --| $$$ "]");
wenzelm@5826
   291
val opt_attribs = Scan.optional attribs [];
wenzelm@5826
   292
wenzelm@6398
   293
fun thm_name s = name -- opt_attribs --| $$$ s;
wenzelm@6511
   294
fun opt_thm_name s =
wenzelm@6511
   295
  Scan.optional ((name -- opt_attribs || (attribs >> pair "")) --| $$$ s) ("", []);;
wenzelm@5917
   296
wenzelm@6372
   297
val spec_name = thm_name ":" -- prop >> (fn ((x, y), z) => ((x, z), y));
wenzelm@6372
   298
val spec_opt_name = opt_thm_name ":" -- prop >> (fn ((x, y), z) => ((x, z), y));
wenzelm@6372
   299
wenzelm@5917
   300
val xthm = xname -- opt_attribs;
wenzelm@6372
   301
val xthms1 = Scan.repeat1 xthm;
wenzelm@5826
   302
wenzelm@5826
   303
wenzelm@12047
   304
(* locale elements *)
wenzelm@12047
   305
wenzelm@12047
   306
val loc_mixfix = $$$ "(" -- $$$ "structure" -- !!! ($$$ ")") >> K None || opt_mixfix >> Some;
wenzelm@12047
   307
wenzelm@12047
   308
val locale_element =
wenzelm@12056
   309
  $$$ "fixes" |-- !!! (and_list1 (name -- Scan.option ($$$ "::" |-- typ) -- loc_mixfix >> triple1))
wenzelm@12047
   310
    >> Locale.Fixes ||
wenzelm@12047
   311
  $$$ "assumes" |-- !!! (and_list1 (opt_thm_name ":" -- Scan.repeat1 propp))
wenzelm@12047
   312
    >> Locale.Assumes ||
wenzelm@12047
   313
  $$$ "defines" |-- !!! (and_list1 (opt_thm_name ":" -- propp'))
wenzelm@12047
   314
    >> Locale.Defines ||
wenzelm@12047
   315
  $$$ "notes" |-- !!! (and_list1 (opt_thm_name "=" -- xthms1)) >> Locale.Notes ||
wenzelm@12047
   316
  $$$ "uses" |-- (!!! ($$$ "FIXME" >> K ())) >> Locale.Uses;
wenzelm@12047
   317
wenzelm@12047
   318
wenzelm@5826
   319
(* proof methods *)
wenzelm@5826
   320
wenzelm@6983
   321
fun is_symid_meth s =
wenzelm@9131
   322
  s <> "|" andalso s <> "?" andalso s <> "+" andalso T.is_sid s;
wenzelm@6983
   323
wenzelm@5826
   324
fun meth4 x =
wenzelm@5877
   325
 (position (xname >> rpair []) >> (Method.Source o Args.src) ||
wenzelm@6983
   326
  $$$ "(" |-- !!! (meth0 --| $$$ ")")) x
wenzelm@5826
   327
and meth3 x =
wenzelm@5826
   328
 (meth4 --| $$$ "?" >> Method.Try ||
wenzelm@5826
   329
  meth4 --| $$$ "+" >> Method.Repeat1 ||
wenzelm@5940
   330
  meth4) x
wenzelm@5940
   331
and meth2 x =
wenzelm@6983
   332
 (position (xname -- args1 is_symid_meth false) >> (Method.Source o Args.src) ||
wenzelm@5826
   333
  meth3) x
wenzelm@5826
   334
and meth1 x = (enum1 "," meth2 >> (fn [m] => m | ms => Method.Then ms)) x
wenzelm@5826
   335
and meth0 x = (enum1 "|" meth1 >> (fn [m] => m | ms => Method.Orelse ms)) x;
wenzelm@5826
   336
wenzelm@6558
   337
val method = meth3;
wenzelm@5826
   338
wenzelm@5826
   339
wenzelm@5826
   340
end;