src/Pure/Thy/thy_parse.ML
author wenzelm
Thu Oct 30 17:00:34 1997 +0100 (1997-10-30)
changeset 4047 67b5552b1067
parent 4020 f88775cc8d17
child 4056 abb0f4742ed7
permissions -rw-r--r--
added thy_data;
wenzelm@389
     1
(*  Title:      Pure/Thy/thy_parse.ML
wenzelm@389
     2
    ID:         $Id$
wenzelm@389
     3
    Author:     Markus Wenzel, TU Muenchen
wenzelm@389
     4
wenzelm@389
     5
The parser for theory files.
wenzelm@389
     6
*)
wenzelm@389
     7
wenzelm@3900
     8
(* FIXME tmp *)
wenzelm@4011
     9
val global_names = ref false;
wenzelm@3900
    10
wenzelm@3900
    11
wenzelm@389
    12
infix 5 -- --$$ $$-- ^^;
wenzelm@389
    13
infix 3 >>;
wenzelm@389
    14
infix 0 ||;
wenzelm@389
    15
wenzelm@389
    16
signature THY_PARSE =
wenzelm@3977
    17
sig
wenzelm@389
    18
  type token
wenzelm@389
    19
  val !! : ('a -> 'b * 'c) -> 'a -> 'b * 'c
wenzelm@389
    20
  val >> : ('a -> 'b * 'c) * ('b -> 'd) -> 'a -> 'd * 'c
wenzelm@389
    21
  val || : ('a -> 'b) * ('a -> 'b) -> 'a -> 'b
wenzelm@389
    22
  val -- : ('a -> 'b * 'c) * ('c -> 'd * 'e) -> 'a -> ('b * 'd) * 'e
wenzelm@389
    23
  val ^^ : ('a -> string * 'b) * ('b -> string * 'c) -> 'a -> string * 'c
wenzelm@389
    24
  val $$ : string -> token list -> string * token list
wenzelm@389
    25
  val $$-- : string * (token list -> 'b * 'c) -> token list -> 'b * 'c
wenzelm@389
    26
  val --$$ : ('a -> 'b * token list) * string -> 'a -> 'b * token list
wenzelm@389
    27
  val ident: token list -> string * token list
wenzelm@389
    28
  val long_ident: token list -> string * token list
wenzelm@389
    29
  val long_id: token list -> string * token list
wenzelm@389
    30
  val type_var: token list -> string * token list
wenzelm@636
    31
  val type_args: token list -> string list * token list
wenzelm@389
    32
  val nat: token list -> string * token list
wenzelm@389
    33
  val string: token list -> string * token list
wenzelm@389
    34
  val verbatim: token list -> string * token list
wenzelm@389
    35
  val empty: 'a -> 'b list * 'a
wenzelm@389
    36
  val optional: ('a -> 'b * 'a) -> 'b -> 'a -> 'b * 'a
wenzelm@389
    37
  val repeat: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
wenzelm@389
    38
  val repeat1: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
wenzelm@389
    39
  val enum: string -> (token list -> 'a * token list)
wenzelm@389
    40
    -> token list -> 'a list * token list
wenzelm@389
    41
  val enum1: string -> (token list -> 'a * token list)
wenzelm@389
    42
    -> token list -> 'a list * token list
wenzelm@389
    43
  val list: (token list -> 'a * token list)
wenzelm@389
    44
    -> token list -> 'a list * token list
wenzelm@389
    45
  val list1: (token list -> 'a * token list)
wenzelm@389
    46
    -> token list -> 'a list * token list
wenzelm@389
    47
  val name: token list -> string * token list
wenzelm@389
    48
  val sort: token list -> string * token list
wenzelm@3977
    49
  val typ: token list -> string * token list
wenzelm@451
    50
  val opt_infix: token list -> string * token list
wenzelm@451
    51
  val opt_mixfix: token list -> string * token list
wenzelm@636
    52
  val opt_witness: token list -> string * token list
wenzelm@389
    53
  type syntax
wenzelm@389
    54
  val make_syntax: string list ->
wenzelm@389
    55
    (string * (token list -> (string * string) * token list)) list -> syntax
clasohm@476
    56
  val parse_thy: syntax -> string -> string -> string
wenzelm@389
    57
  val section: string -> string -> (token list -> string * token list)
wenzelm@389
    58
    -> (string * (token list -> (string * string) * token list))
wenzelm@389
    59
  val axm_section: string -> string
wenzelm@389
    60
    -> (token list -> (string * string list) * token list)
wenzelm@389
    61
    -> (string * (token list -> (string * string) * token list))
wenzelm@389
    62
  val pure_keywords: string list
wenzelm@389
    63
  val pure_sections:
wenzelm@389
    64
    (string * (token list -> (string * string) * token list)) list
lcp@570
    65
  (*items for building strings*)
wenzelm@710
    66
  val cat: string -> string -> string
wenzelm@656
    67
  val parens: string -> string
wenzelm@656
    68
  val brackets: string -> string
wenzelm@656
    69
  val mk_list: string list -> string
wenzelm@656
    70
  val mk_big_list: string list -> string
wenzelm@656
    71
  val mk_pair: string * string -> string
wenzelm@656
    72
  val mk_triple: string * string * string -> string
wenzelm@656
    73
  val strip_quotes: string -> string
wenzelm@3977
    74
end;
wenzelm@389
    75
paulson@1512
    76
paulson@1512
    77
structure ThyParse : THY_PARSE=
wenzelm@389
    78
struct
wenzelm@389
    79
wenzelm@389
    80
open ThyScan;
wenzelm@389
    81
wenzelm@389
    82
wenzelm@389
    83
(** parser toolbox **)
wenzelm@389
    84
wenzelm@389
    85
type token = token_kind * string * int;
wenzelm@389
    86
wenzelm@389
    87
wenzelm@389
    88
(* errors *)
wenzelm@389
    89
wenzelm@389
    90
exception SYNTAX_ERROR of string * string * int;
wenzelm@389
    91
wenzelm@389
    92
fun syn_err s1 s2 n = raise SYNTAX_ERROR (s1, s2, n);
wenzelm@389
    93
wenzelm@389
    94
fun eof_err () = error "Unexpected end-of-file";
wenzelm@389
    95
lcp@570
    96
(*Similar to Prolog's cut: reports any syntax error instead of backtracking
lcp@570
    97
  through a superior || *)
wenzelm@389
    98
fun !! parse toks = parse toks
wenzelm@389
    99
  handle SYNTAX_ERROR (s1, s2, n) => error ("Syntax error on line " ^
wenzelm@389
   100
    string_of_int n ^ ": " ^ s1 ^ " expected and " ^ s2 ^ " was found");
wenzelm@389
   101
wenzelm@389
   102
wenzelm@389
   103
(* parser combinators *)
wenzelm@389
   104
wenzelm@389
   105
fun (parse >> f) toks = apfst f (parse toks);
wenzelm@389
   106
wenzelm@389
   107
fun (parse1 || parse2) toks =
wenzelm@389
   108
  parse1 toks handle SYNTAX_ERROR _ => parse2 toks;
wenzelm@389
   109
wenzelm@389
   110
fun (parse1 -- parse2) toks =
wenzelm@389
   111
  let
wenzelm@389
   112
    val (x, toks') = parse1 toks;
wenzelm@389
   113
    val (y, toks'') = parse2 toks';
wenzelm@389
   114
  in
wenzelm@389
   115
    ((x, y), toks'')
wenzelm@389
   116
  end;
wenzelm@389
   117
wenzelm@389
   118
fun (parse1 ^^ parse2) = parse1 -- parse2 >> op ^;
wenzelm@389
   119
wenzelm@389
   120
wenzelm@389
   121
(* generic parsers *)
wenzelm@389
   122
wenzelm@389
   123
fun $$ a ((k, b, n) :: toks) =
wenzelm@389
   124
      if k = Keyword andalso a = b then (a, toks)
wenzelm@389
   125
      else syn_err (quote a) (quote b) n
wenzelm@389
   126
  | $$ _ [] = eof_err ();
wenzelm@389
   127
wenzelm@389
   128
fun (a $$-- parse) = $$ a -- parse >> #2;
wenzelm@389
   129
wenzelm@389
   130
fun (parse --$$ a) = parse -- $$ a >> #1;
wenzelm@389
   131
wenzelm@389
   132
wenzelm@389
   133
fun kind k1 ((k2, s, n) :: toks) =
wenzelm@389
   134
      if k1 = k2 then (s, toks)
wenzelm@389
   135
      else syn_err (name_of_kind k1) (quote s) n
wenzelm@389
   136
  | kind _ [] = eof_err ();
wenzelm@389
   137
wenzelm@389
   138
val ident = kind Ident;
wenzelm@389
   139
val long_ident = kind LongIdent;
wenzelm@389
   140
val long_id = ident || long_ident;
wenzelm@389
   141
val type_var = kind TypeVar >> quote;
wenzelm@389
   142
val nat = kind Nat;
wenzelm@389
   143
val string = kind String;
wenzelm@389
   144
val verbatim = kind Verbatim;
wenzelm@389
   145
val eof = kind EOF;
wenzelm@389
   146
wenzelm@389
   147
fun empty toks = ([], toks);
wenzelm@389
   148
wenzelm@389
   149
fun optional parse def = parse || empty >> K def;
wenzelm@389
   150
wenzelm@389
   151
fun repeat parse toks = (parse -- repeat parse >> op :: || empty) toks;
wenzelm@389
   152
fun repeat1 parse = parse -- repeat parse >> op ::;
wenzelm@389
   153
wenzelm@389
   154
fun enum1 sep parse = parse -- repeat (sep $$-- parse) >> op ::;
wenzelm@389
   155
fun enum sep parse = enum1 sep parse || empty;
wenzelm@389
   156
paulson@2231
   157
fun list1 parse = enum1 "," parse;
wenzelm@3977
   158
fun list parse = enum "," parse;
wenzelm@3977
   159
wenzelm@389
   160
wenzelm@389
   161
wenzelm@389
   162
(** theory parsers **)
wenzelm@389
   163
wenzelm@389
   164
(* misc utilities *)
wenzelm@389
   165
wenzelm@389
   166
fun cat s1 s2 = s1 ^ " " ^ s2;
wenzelm@389
   167
wenzelm@558
   168
val parens = enclose "(" ")";
wenzelm@558
   169
val brackets = enclose "[" "]";
wenzelm@389
   170
wenzelm@389
   171
val mk_list = brackets o commas;
wenzelm@389
   172
val mk_big_list = brackets o space_implode ",\n ";
wenzelm@389
   173
wenzelm@558
   174
fun mk_pair (x, y) = parens (commas [x, y]);
wenzelm@558
   175
fun mk_triple (x, y, z) = parens (commas [x, y, z]);
wenzelm@389
   176
fun mk_triple1 ((x, y), z) = mk_triple (x, y, z);
wenzelm@389
   177
fun mk_triple2 (x, (y, z)) = mk_triple (x, y, z);
wenzelm@389
   178
paulson@2231
   179
fun split_decls l = flat (map (fn (xs, y) => map (rpair y) xs) l);
wenzelm@389
   180
wenzelm@389
   181
fun strip_quotes str =
wenzelm@389
   182
  implode (tl (take (size str - 1, explode str)));
wenzelm@389
   183
wenzelm@389
   184
wenzelm@389
   185
(* names *)
wenzelm@389
   186
wenzelm@389
   187
val name = ident >> quote || string;
wenzelm@389
   188
val names = list name;
wenzelm@389
   189
val names1 = list1 name;
wenzelm@389
   190
val name_list = names >> mk_list;
wenzelm@389
   191
val name_list1 = names1 >> mk_list;
wenzelm@389
   192
wenzelm@389
   193
wenzelm@389
   194
(* classes *)
wenzelm@389
   195
wenzelm@389
   196
val subclass = name -- optional ("<" $$-- !! name_list1) "[]";
wenzelm@389
   197
wenzelm@558
   198
val class_decls = repeat1 (subclass >> mk_pair) >> mk_big_list;
wenzelm@389
   199
wenzelm@389
   200
wenzelm@389
   201
(* arities *)
wenzelm@389
   202
wenzelm@389
   203
val sort =
wenzelm@389
   204
  name >> brackets ||
wenzelm@389
   205
  "{" $$-- name_list --$$ "}";
wenzelm@389
   206
wenzelm@389
   207
val sort_list1 = list1 sort >> mk_list;
wenzelm@389
   208
wenzelm@389
   209
wenzelm@389
   210
val arity = optional ("(" $$-- !! (sort_list1 --$$")")) "[]" -- sort;
wenzelm@389
   211
wenzelm@389
   212
val arity_decls = repeat1 (names1 --$$ "::" -- !! arity)
wenzelm@389
   213
  >> (mk_big_list o map mk_triple2 o split_decls);
wenzelm@389
   214
wenzelm@389
   215
wenzelm@389
   216
(* mixfix annotations *)
wenzelm@389
   217
wenzelm@2203
   218
val infxl =
wenzelm@2203
   219
  "infixl" $$-- !! (nat >> cat "Infixl" || string -- nat >> (cat "InfixlName" o mk_pair));
wenzelm@2203
   220
val infxr =
wenzelm@2203
   221
  "infixr" $$-- !! (nat >> cat "Infixr" || string -- nat >> (cat "InfixrName" o mk_pair));
wenzelm@389
   222
wenzelm@889
   223
val binder = "binder" $$--
wenzelm@2203
   224
  !! (string -- (("[" $$-- nat --$$ "]") -- nat || nat >> (fn n => (n, n))))
wenzelm@2203
   225
    >> (cat "Binder" o mk_triple2);
wenzelm@389
   226
wenzelm@389
   227
val opt_pris = optional ("[" $$-- !! (list nat --$$ "]")) [] >> mk_list;
wenzelm@389
   228
wenzelm@389
   229
val mixfix = string -- !! (opt_pris -- optional nat "max_pri")
wenzelm@389
   230
  >> (cat "Mixfix" o mk_triple2);
wenzelm@389
   231
wenzelm@558
   232
fun opt_syn fx = optional ("(" $$-- fx --$$ ")") "NoSyn";
wenzelm@389
   233
wenzelm@389
   234
val opt_infix = opt_syn (infxl || infxr);
wenzelm@389
   235
val opt_mixfix = opt_syn (mixfix || infxl || infxr || binder);
wenzelm@389
   236
wenzelm@389
   237
wenzelm@389
   238
(* types *)
wenzelm@389
   239
wenzelm@3977
   240
(* FIXME clean!! *)
wenzelm@3977
   241
clasohm@1705
   242
(*Parse an identifier, but only if it is not followed by "::", "=" or ",";
clasohm@1377
   243
  the exclusion of a postfix comma can be controlled to allow expressions
clasohm@1377
   244
  like "(id, id)" but disallow ones like "'a => id id,id :: ..."*)
clasohm@1377
   245
fun ident_no_colon _ [] = eof_err()
clasohm@1377
   246
  | ident_no_colon allow_comma ((Ident, s, n) :: (rest as (Keyword, s2, n2) ::
clasohm@1377
   247
                                toks)) =
clasohm@1705
   248
      if s2 = "::" orelse s2 = "=" orelse (not allow_comma andalso s2 = ",")
clasohm@1705
   249
      then syn_err (name_of_kind Ident) (quote s2) n2
clasohm@1377
   250
      else (s, rest)
clasohm@1377
   251
  | ident_no_colon _ ((Ident, s, n) :: toks) = (s, toks)
clasohm@1377
   252
  | ident_no_colon _ ((k, s, n) :: _) =
clasohm@1377
   253
      syn_err (name_of_kind Ident) (quote s) n;
clasohm@1321
   254
wenzelm@3110
   255
(*type used in types, consts and syntax sections*)
clasohm@1377
   256
fun const_type allow_comma toks =
wenzelm@3977
   257
  let
wenzelm@3977
   258
    val simple_type =
wenzelm@3977
   259
      (ident || kind TypeVar ^^ optional ($$ "::" ^^ ident) "") --
wenzelm@3977
   260
          repeat (ident_no_colon allow_comma)
wenzelm@3977
   261
          >> (fn (args, ts) => cat args (space_implode " " ts)) ||
wenzelm@3977
   262
        ("(" $$-- (list1 (const_type true)) --$$ ")" >> (parens o commas)) --
wenzelm@3977
   263
          repeat1 (ident_no_colon allow_comma)
wenzelm@3977
   264
          >> (fn (args, ts) => cat args (space_implode " " ts));
clasohm@1377
   265
clasohm@1377
   266
      val appl_param =
clasohm@1377
   267
        simple_type || "(" $$-- const_type true --$$ ")" >> parens || 
clasohm@1377
   268
        "[" $$-- (list1 (const_type true)) --$$ "]" --$$ "=>" --
clasohm@1377
   269
          const_type allow_comma >>
clasohm@1377
   270
          (fn (src, dest) => mk_list src ^ " => " ^ dest);
clasohm@1377
   271
  in ("[" $$-- (list1 (const_type true)) --$$ "]" --$$ "=>" --
clasohm@1377
   272
        const_type allow_comma >>
clasohm@1377
   273
        (fn (src, dest) => mk_list src ^ " => " ^ dest) ||
clasohm@1377
   274
      repeat1 (appl_param --$$ "=>") -- const_type allow_comma >>
clasohm@1377
   275
        (fn (src, dest) => space_implode " => " (src@[dest])) ||
clasohm@1377
   276
      simple_type ||
clasohm@1377
   277
      "(" $$-- const_type true --$$ ")" >> parens) toks
clasohm@1321
   278
  end;
clasohm@1321
   279
wenzelm@3977
   280
val typ = string || (const_type false >> quote);
wenzelm@3977
   281
wenzelm@3977
   282
clasohm@1383
   283
fun mk_old_type_decl ((ts, n), syn) =
clasohm@1383
   284
  map (fn t => (mk_triple (t, n, syn), false)) ts;
clasohm@1383
   285
clasohm@1383
   286
fun mk_type_decl (((xs, t), None), syn) =
clasohm@1383
   287
      [(mk_triple (t, string_of_int (length xs), syn), false)]
clasohm@1383
   288
  | mk_type_decl (((xs, t), Some rhs), syn) =
clasohm@1383
   289
      [(parens (commas [t, mk_list xs, rhs, syn]), true)];
clasohm@1383
   290
clasohm@1383
   291
fun mk_type_decls tys =
wenzelm@3764
   292
  "|> Theory.add_types\n" ^ mk_big_list (keyfilter tys false) ^ "\n\n\
wenzelm@3764
   293
  \|> Theory.add_tyabbrs\n" ^ mk_big_list (keyfilter tys true);
clasohm@1383
   294
clasohm@1383
   295
clasohm@1383
   296
val old_type_decl = names1 -- nat -- opt_infix >> mk_old_type_decl;
clasohm@1383
   297
clasohm@1383
   298
val type_args =
clasohm@1383
   299
  type_var >> (fn x => [x]) ||
clasohm@1383
   300
  "(" $$-- !! (list1 type_var --$$ ")") ||
clasohm@1383
   301
  empty >> K [];
clasohm@1383
   302
clasohm@1383
   303
val type_decl = type_args -- name --
wenzelm@3977
   304
  optional ("=" $$-- typ >> Some) None -- opt_infix >> mk_type_decl;
clasohm@1383
   305
wenzelm@2360
   306
val type_decls =
wenzelm@2360
   307
  repeat1 (old_type_decl || type_decl) >> (mk_type_decls o flat);
clasohm@1383
   308
clasohm@1383
   309
clasohm@1383
   310
(* consts *)
clasohm@1383
   311
wenzelm@2360
   312
val const_decls =
wenzelm@3977
   313
  repeat1 (names1 --$$ "::" -- !! (typ -- opt_mixfix))
wenzelm@2360
   314
  >> (mk_big_list o map mk_triple2 o split_decls);
wenzelm@389
   315
wenzelm@2360
   316
val opt_mode =
wenzelm@2360
   317
  optional
wenzelm@2360
   318
    ("(" $$-- !! (name -- optional ($$ "output" >> K "false") "true" --$$ ")"))
wenzelm@2360
   319
    ("\"\"", "true")
wenzelm@2360
   320
  >> mk_pair;
wenzelm@2360
   321
wenzelm@2360
   322
val syntax_decls = opt_mode -- const_decls >> (fn (mode, txt) => mode ^ "\n" ^ txt);
wenzelm@2203
   323
wenzelm@389
   324
wenzelm@389
   325
(* translations *)
wenzelm@389
   326
wenzelm@389
   327
val trans_pat =
wenzelm@389
   328
  optional ("(" $$-- !! (name --$$ ")")) "\"logic\"" -- string >> mk_pair;
wenzelm@389
   329
wenzelm@389
   330
val trans_arrow =
wenzelm@3528
   331
  $$ "=>" >> K "Syntax.ParseRule " ||
wenzelm@3528
   332
  $$ "<=" >> K "Syntax.PrintRule " ||
wenzelm@3528
   333
  $$ "==" >> K "Syntax.ParsePrintRule ";
wenzelm@389
   334
paulson@1810
   335
val trans_line =
wenzelm@2203
   336
  trans_pat -- !! (trans_arrow -- trans_pat)
wenzelm@2203
   337
    >> (fn (left, (arr, right)) => arr ^ mk_pair (left, right));
paulson@1810
   338
paulson@1810
   339
val trans_decls = repeat1 trans_line >> mk_big_list;
wenzelm@389
   340
wenzelm@389
   341
wenzelm@389
   342
(* ML translations *)
wenzelm@389
   343
wenzelm@4047
   344
val local_defs =
wenzelm@389
   345
  " val parse_ast_translation = [];\n\
wenzelm@389
   346
  \ val parse_translation = [];\n\
wenzelm@389
   347
  \ val print_translation = [];\n\
wenzelm@2385
   348
  \ val typed_print_translation = [];\n\
wenzelm@2694
   349
  \ val print_ast_translation = [];\n\
wenzelm@4047
   350
  \ val token_translation = [];\n\
wenzelm@4047
   351
  \ val thy_data = []";
wenzelm@389
   352
wenzelm@389
   353
val trfun_args =
wenzelm@389
   354
  "(parse_ast_translation, parse_translation, \
wenzelm@389
   355
  \print_translation, print_ast_translation)";
wenzelm@389
   356
wenzelm@389
   357
wenzelm@389
   358
(* axioms *)
wenzelm@389
   359
wenzelm@389
   360
val mk_axms = mk_big_list o map (mk_pair o apfst quote);
wenzelm@389
   361
wenzelm@389
   362
fun mk_axiom_decls axms = (mk_axms axms, map fst axms);
wenzelm@389
   363
wenzelm@389
   364
val axiom_decls = repeat1 (ident -- !! string) >> mk_axiom_decls;
wenzelm@389
   365
wenzelm@389
   366
wenzelm@3813
   367
(* oracle *)
wenzelm@3813
   368
wenzelm@3813
   369
val oracle_decl = (name --$$ "=" -- long_id) >> mk_pair;
wenzelm@3813
   370
wenzelm@3813
   371
clasohm@1555
   372
(* combined consts and axioms *)
clasohm@1555
   373
clasohm@1555
   374
fun mk_constaxiom_decls x =
clasohm@1555
   375
  let
clasohm@1555
   376
    val (axms_defs, axms_names) =
clasohm@1555
   377
      mk_axiom_decls (map (fn ((id, _), def) => (id ^ "_def", def)) x);
clasohm@1555
   378
  in ((mk_big_list o map mk_triple2 o map (apfst quote o fst)) x ^
wenzelm@4020
   379
       "\n\n|> PureThy.add_store_defs\n" ^ axms_defs, axms_names)
clasohm@1555
   380
  end;
clasohm@1555
   381
clasohm@1555
   382
val constaxiom_decls =
wenzelm@3977
   383
  repeat1 (ident --$$ "::" -- !! (typ -- opt_mixfix) -- !! string)
wenzelm@3977
   384
  >> mk_constaxiom_decls;
clasohm@1555
   385
clasohm@1555
   386
wenzelm@389
   387
(* axclass *)
wenzelm@389
   388
wenzelm@389
   389
fun mk_axclass_decl ((c, cs), axms) =
wenzelm@389
   390
  (mk_pair (c, cs) ^ "\n" ^ mk_axms axms,
wenzelm@389
   391
    (strip_quotes c ^ "I") :: map fst axms);
wenzelm@389
   392
wenzelm@389
   393
val axclass_decl = subclass -- repeat (ident -- !! string) >> mk_axclass_decl;
wenzelm@389
   394
wenzelm@389
   395
wenzelm@451
   396
(* instance *)
wenzelm@389
   397
wenzelm@425
   398
fun mk_witness (axths, opt_tac) =
wenzelm@389
   399
  mk_list (keyfilter axths false) ^ "\n" ^
wenzelm@389
   400
  mk_list (keyfilter axths true) ^ "\n" ^
wenzelm@389
   401
  opt_tac;
wenzelm@389
   402
wenzelm@389
   403
val axm_or_thm =
wenzelm@389
   404
  string >> rpair false ||
wenzelm@389
   405
  long_id >> rpair true;
wenzelm@389
   406
wenzelm@451
   407
wenzelm@425
   408
val opt_witness =
wenzelm@389
   409
  optional ("(" $$-- list1 axm_or_thm --$$ ")") [] --
wenzelm@558
   410
  optional (verbatim >> (parens o cat "Some" o parens)) "None"
wenzelm@425
   411
  >> mk_witness;
wenzelm@425
   412
wenzelm@425
   413
val instance_decl =
wenzelm@636
   414
  (name --$$ "<" -- name >> (pair "|> AxClass.add_inst_subclass" o mk_pair) ||
wenzelm@636
   415
    name --$$ "::" -- arity >> (pair "|> AxClass.add_inst_arity" o mk_triple2))
wenzelm@451
   416
  -- opt_witness
wenzelm@777
   417
  >> (fn ((x, y), z) => (cat_lines [x, y, z]));
wenzelm@389
   418
wenzelm@389
   419
wenzelm@3931
   420
(* local, global path *)
wenzelm@3931
   421
wenzelm@3958
   422
fun empty_decl toks = (empty >> K "") toks;
wenzelm@3900
   423
wenzelm@3931
   424
val global_path =
wenzelm@3931
   425
  "|> (fn thy => if ! global_names then thy else Theory.add_path \"/\" thy)";
wenzelm@3931
   426
wenzelm@3931
   427
val local_path =
wenzelm@3931
   428
  global_path ^ "\n\
wenzelm@3931
   429
  \|> (fn thy => if ! global_names then thy\
wenzelm@3931
   430
  \ else Theory.add_path thy_name thy)";
wenzelm@3900
   431
wenzelm@3900
   432
wenzelm@389
   433
wenzelm@389
   434
(** theory syntax **)
wenzelm@389
   435
wenzelm@389
   436
type syntax =
wenzelm@389
   437
  lexicon * (token list -> (string * string) * token list) Symtab.table;
wenzelm@389
   438
wenzelm@389
   439
fun make_syntax keywords sects =
wenzelm@451
   440
  (make_lexicon (map fst sects @ keywords),
wenzelm@451
   441
    Symtab.make sects handle Symtab.DUPS dups =>
wenzelm@451
   442
      error ("Duplicate sections in theory file syntax: " ^ commas_quote dups));
wenzelm@389
   443
wenzelm@389
   444
wenzelm@389
   445
(* header *)
wenzelm@389
   446
wenzelm@389
   447
fun mk_header (thy_name, bases) =
clasohm@586
   448
  (thy_name, "mk_base " ^ mk_list bases ^ " " ^ quote thy_name);
wenzelm@389
   449
wenzelm@389
   450
val base =
wenzelm@389
   451
  ident >> (cat "Thy" o quote) ||
wenzelm@389
   452
  string >> cat "File";
wenzelm@389
   453
wenzelm@389
   454
val header = ident --$$ "=" -- enum1 "+" base >> mk_header;
wenzelm@389
   455
wenzelm@389
   456
wenzelm@389
   457
(* extension *)
wenzelm@389
   458
wenzelm@3900
   459
fun mk_extension (txts, mltxt) =
wenzelm@389
   460
  let
wenzelm@389
   461
    val cat_sects = space_implode "\n\n" o filter_out (equal "");
wenzelm@389
   462
    val (extxts, postxts) = split_list txts;
wenzelm@389
   463
  in
wenzelm@3900
   464
    (cat_sects extxts, cat_sects postxts, mltxt)
wenzelm@389
   465
  end;
wenzelm@389
   466
wenzelm@389
   467
fun sect tab ((Keyword, s, n) :: toks) =
wenzelm@389
   468
      (case Symtab.lookup (tab, s) of
wenzelm@389
   469
        Some parse => !! parse toks
wenzelm@389
   470
      | None => syn_err "section" s n)
wenzelm@389
   471
  | sect _ ((_, s, n) :: _) = syn_err "section" s n
wenzelm@389
   472
  | sect _ [] = eof_err ();
wenzelm@389
   473
wenzelm@3900
   474
fun extension sectab = "+" $$-- !!
wenzelm@3900
   475
  (repeat (sect sectab) --$$ "end" -- optional ("ML" $$-- verbatim) "")
wenzelm@3900
   476
    >> mk_extension;
wenzelm@3813
   477
wenzelm@3900
   478
fun opt_extension sectab = optional (extension sectab) ("", "", "");
wenzelm@3875
   479
wenzelm@389
   480
wenzelm@389
   481
(* theory definition *)
wenzelm@389
   482
wenzelm@3900
   483
fun mk_structure tname ((thy_name, old_thys), (extxt, postxt, mltxt)) =
wenzelm@558
   484
  if thy_name <> tname then
wenzelm@558
   485
    error ("Filename \"" ^ tname ^ ".thy\" and theory name "
wenzelm@558
   486
      ^ quote thy_name ^ " are different")
wenzelm@558
   487
  else
wenzelm@3875
   488
    "val thy = " ^ old_thys ^ ";\n\n\
wenzelm@3875
   489
    \structure " ^ thy_name ^ " =\n\
wenzelm@3875
   490
    \struct\n\
wenzelm@3875
   491
    \\n\
wenzelm@3931
   492
    \local\n\
wenzelm@3931
   493
    \ val thy_name = \"" ^ tname ^ "\";\n"
wenzelm@4047
   494
    ^ local_defs ^ "\n\
wenzelm@3875
   495
    \in\n\
wenzelm@3875
   496
    \\n"
wenzelm@3875
   497
    ^ mltxt ^ "\n\
wenzelm@3875
   498
    \\n\
wenzelm@3875
   499
    \val thy = thy\n\
wenzelm@3931
   500
    \\n"
wenzelm@3931
   501
    ^ local_path ^
wenzelm@3875
   502
    "\n\
wenzelm@3875
   503
    \|> Theory.add_trfuns\n"
wenzelm@3875
   504
    ^ trfun_args ^ "\n\
wenzelm@4047
   505
    \|> Theory.add_trfunsT typed_print_translation\n\
wenzelm@4047
   506
    \|> Theory.add_tokentrfuns token_translation\n\
wenzelm@4047
   507
    \|> Theory.init_data thy_data\n\
wenzelm@3875
   508
    \\n"
wenzelm@3875
   509
    ^ extxt ^ "\n\
wenzelm@3875
   510
    \\n\
wenzelm@3875
   511
    \|> Theory.add_name " ^ quote thy_name ^ ";\n\
wenzelm@3875
   512
    \\n\
wenzelm@3875
   513
    \val _ = store_theory (thy, " ^ quote thy_name ^ ");\n\
wenzelm@3875
   514
    \\n\
wenzelm@3875
   515
    \\n"
wenzelm@3875
   516
    ^ postxt ^ "\n\
wenzelm@3875
   517
    \\n\
wenzelm@3875
   518
    \end;\n\
wenzelm@3875
   519
    \end;\n\
wenzelm@3875
   520
    \\n\
wenzelm@3875
   521
    \open " ^ thy_name ^ ";\n\
wenzelm@3875
   522
    \\n";
wenzelm@389
   523
clasohm@476
   524
fun theory_defn sectab tname =
wenzelm@3875
   525
  header -- opt_extension sectab -- eof >> (mk_structure tname o #1);
wenzelm@389
   526
clasohm@476
   527
fun parse_thy (lex, sectab) tname str =
clasohm@476
   528
  #1 (!! (theory_defn sectab tname) (tokenize lex str));
wenzelm@389
   529
wenzelm@389
   530
wenzelm@389
   531
(* standard sections *)
wenzelm@389
   532
wenzelm@389
   533
fun mk_val ax = "val " ^ ax ^ " = get_axiom thy " ^ quote ax ^ ";";
wenzelm@777
   534
val mk_vals = cat_lines o map mk_val;
wenzelm@389
   535
wenzelm@777
   536
fun mk_axm_sect "" (txt, axs) = (txt, mk_vals axs)
wenzelm@777
   537
  | mk_axm_sect pretxt (txt, axs) = (pretxt ^ "\n" ^ txt, mk_vals axs);
wenzelm@389
   538
wenzelm@389
   539
fun axm_section name pretxt parse =
wenzelm@389
   540
  (name, parse >> mk_axm_sect pretxt);
wenzelm@389
   541
wenzelm@389
   542
fun section name pretxt parse =
wenzelm@389
   543
  axm_section name pretxt (parse >> rpair []);
wenzelm@389
   544
wenzelm@389
   545
wenzelm@389
   546
val pure_keywords =
wenzelm@3900
   547
 ["end", "ML", "mixfix", "infixr", "infixl", "binder", "global",
wenzelm@3931
   548
  "local", "output", "=", "+", ",", "<", "{", "}", "(", ")", "[", "]",
wenzelm@3931
   549
  "::", "==", "=>", "<="];
wenzelm@389
   550
wenzelm@389
   551
val pure_sections =
wenzelm@3813
   552
 [section "classes" "|> Theory.add_classes" class_decls,
wenzelm@3764
   553
  section "default" "|> Theory.add_defsort" sort,
wenzelm@777
   554
  section "types" "" type_decls,
wenzelm@3764
   555
  section "arities" "|> Theory.add_arities" arity_decls,
wenzelm@3764
   556
  section "consts" "|> Theory.add_consts" const_decls,
wenzelm@3764
   557
  section "syntax" "|> Theory.add_modesyntax" syntax_decls,
wenzelm@3764
   558
  section "translations" "|> Theory.add_trrules" trans_decls,
wenzelm@4020
   559
  axm_section "rules" "|> PureThy.add_store_axioms" axiom_decls,
wenzelm@4020
   560
  axm_section "defs" "|> PureThy.add_store_defs" axiom_decls,
wenzelm@3813
   561
  section "oracle" "|> Theory.add_oracle" oracle_decl,
wenzelm@3764
   562
  axm_section "constdefs" "|> Theory.add_consts" constaxiom_decls,
wenzelm@636
   563
  axm_section "axclass" "|> AxClass.add_axclass" axclass_decl,
wenzelm@3780
   564
  section "instance" "" instance_decl,
wenzelm@3900
   565
  section "path" "|> Theory.add_path" name,
wenzelm@3931
   566
  section "global" global_path empty_decl,
wenzelm@3931
   567
  section "local" local_path empty_decl];
wenzelm@389
   568
wenzelm@3813
   569
wenzelm@389
   570
end;