src/Pure/Thy/thy_parse.ML
author wenzelm
Mon Oct 20 17:08:18 1997 +0200 (1997-10-20)
changeset 3958 78a8e9a1c2f9
parent 3931 c3c287d3f502
child 3977 9b3cbfd6a936
permissions -rw-r--r--
make SML/NJ happy;
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@3900
     9
val global_names = ref true;
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 =
paulson@1512
    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@451
    49
  val opt_infix: token list -> string * token list
wenzelm@451
    50
  val opt_mixfix: token list -> string * token list
wenzelm@636
    51
  val opt_witness: token list -> string * token list
wenzelm@389
    52
  type syntax
wenzelm@389
    53
  val make_syntax: string list ->
wenzelm@389
    54
    (string * (token list -> (string * string) * token list)) list -> syntax
clasohm@476
    55
  val parse_thy: syntax -> string -> string -> string
wenzelm@389
    56
  val section: string -> string -> (token list -> string * token list)
wenzelm@389
    57
    -> (string * (token list -> (string * string) * token list))
wenzelm@389
    58
  val axm_section: string -> string
wenzelm@389
    59
    -> (token list -> (string * string list) * token list)
wenzelm@389
    60
    -> (string * (token list -> (string * string) * token list))
wenzelm@389
    61
  val pure_keywords: string list
wenzelm@389
    62
  val pure_sections:
wenzelm@389
    63
    (string * (token list -> (string * string) * token list)) list
lcp@570
    64
  (*items for building strings*)
wenzelm@710
    65
  val cat: string -> string -> string
wenzelm@656
    66
  val parens: string -> string
wenzelm@656
    67
  val brackets: string -> string
wenzelm@656
    68
  val mk_list: string list -> string
wenzelm@656
    69
  val mk_big_list: string list -> string
wenzelm@656
    70
  val mk_pair: string * string -> string
wenzelm@656
    71
  val mk_triple: string * string * string -> string
wenzelm@656
    72
  val strip_quotes: string -> string
paulson@1512
    73
  end;
wenzelm@389
    74
paulson@1512
    75
paulson@1512
    76
structure ThyParse : THY_PARSE=
wenzelm@389
    77
struct
wenzelm@389
    78
wenzelm@389
    79
open ThyScan;
wenzelm@389
    80
wenzelm@389
    81
wenzelm@389
    82
(** parser toolbox **)
wenzelm@389
    83
wenzelm@389
    84
type token = token_kind * string * int;
wenzelm@389
    85
wenzelm@389
    86
wenzelm@389
    87
(* errors *)
wenzelm@389
    88
wenzelm@389
    89
exception SYNTAX_ERROR of string * string * int;
wenzelm@389
    90
wenzelm@389
    91
fun syn_err s1 s2 n = raise SYNTAX_ERROR (s1, s2, n);
wenzelm@389
    92
wenzelm@389
    93
fun eof_err () = error "Unexpected end-of-file";
wenzelm@389
    94
lcp@570
    95
(*Similar to Prolog's cut: reports any syntax error instead of backtracking
lcp@570
    96
  through a superior || *)
wenzelm@389
    97
fun !! parse toks = parse toks
wenzelm@389
    98
  handle SYNTAX_ERROR (s1, s2, n) => error ("Syntax error on line " ^
wenzelm@389
    99
    string_of_int n ^ ": " ^ s1 ^ " expected and " ^ s2 ^ " was found");
wenzelm@389
   100
wenzelm@389
   101
wenzelm@389
   102
(* parser combinators *)
wenzelm@389
   103
wenzelm@389
   104
fun (parse >> f) toks = apfst f (parse toks);
wenzelm@389
   105
wenzelm@389
   106
fun (parse1 || parse2) toks =
wenzelm@389
   107
  parse1 toks handle SYNTAX_ERROR _ => parse2 toks;
wenzelm@389
   108
wenzelm@389
   109
fun (parse1 -- parse2) toks =
wenzelm@389
   110
  let
wenzelm@389
   111
    val (x, toks') = parse1 toks;
wenzelm@389
   112
    val (y, toks'') = parse2 toks';
wenzelm@389
   113
  in
wenzelm@389
   114
    ((x, y), toks'')
wenzelm@389
   115
  end;
wenzelm@389
   116
wenzelm@389
   117
fun (parse1 ^^ parse2) = parse1 -- parse2 >> op ^;
wenzelm@389
   118
wenzelm@389
   119
wenzelm@389
   120
(* generic parsers *)
wenzelm@389
   121
wenzelm@389
   122
fun $$ a ((k, b, n) :: toks) =
wenzelm@389
   123
      if k = Keyword andalso a = b then (a, toks)
wenzelm@389
   124
      else syn_err (quote a) (quote b) n
wenzelm@389
   125
  | $$ _ [] = eof_err ();
wenzelm@389
   126
wenzelm@389
   127
fun (a $$-- parse) = $$ a -- parse >> #2;
wenzelm@389
   128
wenzelm@389
   129
fun (parse --$$ a) = parse -- $$ a >> #1;
wenzelm@389
   130
wenzelm@389
   131
wenzelm@389
   132
fun kind k1 ((k2, s, n) :: toks) =
wenzelm@389
   133
      if k1 = k2 then (s, toks)
wenzelm@389
   134
      else syn_err (name_of_kind k1) (quote s) n
wenzelm@389
   135
  | kind _ [] = eof_err ();
wenzelm@389
   136
wenzelm@389
   137
val ident = kind Ident;
wenzelm@389
   138
val long_ident = kind LongIdent;
wenzelm@389
   139
val long_id = ident || long_ident;
wenzelm@389
   140
val type_var = kind TypeVar >> quote;
wenzelm@389
   141
val nat = kind Nat;
wenzelm@389
   142
val string = kind String;
wenzelm@389
   143
val verbatim = kind Verbatim;
wenzelm@389
   144
val eof = kind EOF;
wenzelm@389
   145
wenzelm@389
   146
fun empty toks = ([], toks);
wenzelm@389
   147
wenzelm@389
   148
fun optional parse def = parse || empty >> K def;
wenzelm@389
   149
wenzelm@389
   150
fun repeat parse toks = (parse -- repeat parse >> op :: || empty) toks;
wenzelm@389
   151
fun repeat1 parse = parse -- repeat parse >> op ::;
wenzelm@389
   152
wenzelm@389
   153
fun enum1 sep parse = parse -- repeat (sep $$-- parse) >> op ::;
wenzelm@389
   154
fun enum sep parse = enum1 sep parse || empty;
wenzelm@389
   155
paulson@2231
   156
fun list  parse = enum  "," parse;
paulson@2231
   157
fun list1 parse = enum1 "," parse;
wenzelm@389
   158
wenzelm@389
   159
wenzelm@389
   160
(** theory parsers **)
wenzelm@389
   161
wenzelm@389
   162
(* misc utilities *)
wenzelm@389
   163
wenzelm@389
   164
fun cat s1 s2 = s1 ^ " " ^ s2;
wenzelm@389
   165
wenzelm@558
   166
val parens = enclose "(" ")";
wenzelm@558
   167
val brackets = enclose "[" "]";
wenzelm@389
   168
wenzelm@389
   169
val mk_list = brackets o commas;
wenzelm@389
   170
val mk_big_list = brackets o space_implode ",\n ";
wenzelm@389
   171
wenzelm@558
   172
fun mk_pair (x, y) = parens (commas [x, y]);
wenzelm@558
   173
fun mk_triple (x, y, z) = parens (commas [x, y, z]);
wenzelm@389
   174
fun mk_triple1 ((x, y), z) = mk_triple (x, y, z);
wenzelm@389
   175
fun mk_triple2 (x, (y, z)) = mk_triple (x, y, z);
wenzelm@389
   176
paulson@2231
   177
fun split_decls l = flat (map (fn (xs, y) => map (rpair y) xs) l);
wenzelm@389
   178
wenzelm@389
   179
fun strip_quotes str =
wenzelm@389
   180
  implode (tl (take (size str - 1, explode str)));
wenzelm@389
   181
wenzelm@389
   182
wenzelm@389
   183
(* names *)
wenzelm@389
   184
wenzelm@389
   185
val name = ident >> quote || string;
wenzelm@389
   186
val names = list name;
wenzelm@389
   187
val names1 = list1 name;
wenzelm@389
   188
val name_list = names >> mk_list;
wenzelm@389
   189
val name_list1 = names1 >> mk_list;
wenzelm@389
   190
wenzelm@389
   191
wenzelm@389
   192
(* classes *)
wenzelm@389
   193
wenzelm@389
   194
val subclass = name -- optional ("<" $$-- !! name_list1) "[]";
wenzelm@389
   195
wenzelm@558
   196
val class_decls = repeat1 (subclass >> mk_pair) >> mk_big_list;
wenzelm@389
   197
wenzelm@389
   198
wenzelm@389
   199
(* arities *)
wenzelm@389
   200
wenzelm@389
   201
val sort =
wenzelm@389
   202
  name >> brackets ||
wenzelm@389
   203
  "{" $$-- name_list --$$ "}";
wenzelm@389
   204
wenzelm@389
   205
val sort_list1 = list1 sort >> mk_list;
wenzelm@389
   206
wenzelm@389
   207
wenzelm@389
   208
val arity = optional ("(" $$-- !! (sort_list1 --$$")")) "[]" -- sort;
wenzelm@389
   209
wenzelm@389
   210
val arity_decls = repeat1 (names1 --$$ "::" -- !! arity)
wenzelm@389
   211
  >> (mk_big_list o map mk_triple2 o split_decls);
wenzelm@389
   212
wenzelm@389
   213
wenzelm@389
   214
(* mixfix annotations *)
wenzelm@389
   215
wenzelm@2203
   216
val infxl =
wenzelm@2203
   217
  "infixl" $$-- !! (nat >> cat "Infixl" || string -- nat >> (cat "InfixlName" o mk_pair));
wenzelm@2203
   218
val infxr =
wenzelm@2203
   219
  "infixr" $$-- !! (nat >> cat "Infixr" || string -- nat >> (cat "InfixrName" o mk_pair));
wenzelm@389
   220
wenzelm@889
   221
val binder = "binder" $$--
wenzelm@2203
   222
  !! (string -- (("[" $$-- nat --$$ "]") -- nat || nat >> (fn n => (n, n))))
wenzelm@2203
   223
    >> (cat "Binder" o mk_triple2);
wenzelm@389
   224
wenzelm@389
   225
val opt_pris = optional ("[" $$-- !! (list nat --$$ "]")) [] >> mk_list;
wenzelm@389
   226
wenzelm@389
   227
val mixfix = string -- !! (opt_pris -- optional nat "max_pri")
wenzelm@389
   228
  >> (cat "Mixfix" o mk_triple2);
wenzelm@389
   229
wenzelm@558
   230
fun opt_syn fx = optional ("(" $$-- fx --$$ ")") "NoSyn";
wenzelm@389
   231
wenzelm@389
   232
val opt_infix = opt_syn (infxl || infxr);
wenzelm@389
   233
val opt_mixfix = opt_syn (mixfix || infxl || infxr || binder);
wenzelm@389
   234
wenzelm@389
   235
wenzelm@389
   236
(* types *)
wenzelm@389
   237
clasohm@1705
   238
(*Parse an identifier, but only if it is not followed by "::", "=" or ",";
clasohm@1377
   239
  the exclusion of a postfix comma can be controlled to allow expressions
clasohm@1377
   240
  like "(id, id)" but disallow ones like "'a => id id,id :: ..."*)
clasohm@1377
   241
fun ident_no_colon _ [] = eof_err()
clasohm@1377
   242
  | ident_no_colon allow_comma ((Ident, s, n) :: (rest as (Keyword, s2, n2) ::
clasohm@1377
   243
                                toks)) =
clasohm@1705
   244
      if s2 = "::" orelse s2 = "=" orelse (not allow_comma andalso s2 = ",")
clasohm@1705
   245
      then syn_err (name_of_kind Ident) (quote s2) n2
clasohm@1377
   246
      else (s, rest)
clasohm@1377
   247
  | ident_no_colon _ ((Ident, s, n) :: toks) = (s, toks)
clasohm@1377
   248
  | ident_no_colon _ ((k, s, n) :: _) =
clasohm@1377
   249
      syn_err (name_of_kind Ident) (quote s) n;
clasohm@1321
   250
wenzelm@3110
   251
(*type used in types, consts and syntax sections*)
clasohm@1377
   252
fun const_type allow_comma toks =
clasohm@1377
   253
  let val simple_type =
clasohm@1377
   254
        (ident ||
clasohm@1377
   255
         kind TypeVar -- optional ("::" $$-- ident >> cat "::") "" >>
clasohm@1377
   256
           (fn (tv, cl) => cat tv cl)) --
clasohm@1377
   257
           repeat (ident_no_colon allow_comma) >>
clasohm@1377
   258
           (fn (args, ts) => cat args (space_implode " " ts)) ||
clasohm@1377
   259
         ("(" $$-- (list1 (const_type true)) --$$ ")" >> (parens o commas)) --
clasohm@1377
   260
           repeat1 (ident_no_colon allow_comma) >>
clasohm@1377
   261
           (fn (args, ts) => cat args (space_implode " " ts));
clasohm@1377
   262
clasohm@1377
   263
      val appl_param =
clasohm@1377
   264
        simple_type || "(" $$-- const_type true --$$ ")" >> parens || 
clasohm@1377
   265
        "[" $$-- (list1 (const_type true)) --$$ "]" --$$ "=>" --
clasohm@1377
   266
          const_type allow_comma >>
clasohm@1377
   267
          (fn (src, dest) => mk_list src ^ " => " ^ dest);
clasohm@1377
   268
  in ("[" $$-- (list1 (const_type true)) --$$ "]" --$$ "=>" --
clasohm@1377
   269
        const_type allow_comma >>
clasohm@1377
   270
        (fn (src, dest) => mk_list src ^ " => " ^ dest) ||
clasohm@1377
   271
      repeat1 (appl_param --$$ "=>") -- const_type allow_comma >>
clasohm@1377
   272
        (fn (src, dest) => space_implode " => " (src@[dest])) ||
clasohm@1377
   273
      simple_type ||
clasohm@1377
   274
      "(" $$-- const_type true --$$ ")" >> parens) toks
clasohm@1321
   275
  end;
clasohm@1321
   276
clasohm@1383
   277
fun mk_old_type_decl ((ts, n), syn) =
clasohm@1383
   278
  map (fn t => (mk_triple (t, n, syn), false)) ts;
clasohm@1383
   279
clasohm@1383
   280
fun mk_type_decl (((xs, t), None), syn) =
clasohm@1383
   281
      [(mk_triple (t, string_of_int (length xs), syn), false)]
clasohm@1383
   282
  | mk_type_decl (((xs, t), Some rhs), syn) =
clasohm@1383
   283
      [(parens (commas [t, mk_list xs, rhs, syn]), true)];
clasohm@1383
   284
clasohm@1383
   285
fun mk_type_decls tys =
wenzelm@3764
   286
  "|> Theory.add_types\n" ^ mk_big_list (keyfilter tys false) ^ "\n\n\
wenzelm@3764
   287
  \|> Theory.add_tyabbrs\n" ^ mk_big_list (keyfilter tys true);
clasohm@1383
   288
clasohm@1383
   289
clasohm@1383
   290
val old_type_decl = names1 -- nat -- opt_infix >> mk_old_type_decl;
clasohm@1383
   291
clasohm@1383
   292
val type_args =
clasohm@1383
   293
  type_var >> (fn x => [x]) ||
clasohm@1383
   294
  "(" $$-- !! (list1 type_var --$$ ")") ||
clasohm@1383
   295
  empty >> K [];
clasohm@1383
   296
clasohm@1383
   297
val type_decl = type_args -- name --
clasohm@1383
   298
  optional ("=" $$-- (string || (const_type false >> quote)) >> Some) None
clasohm@1383
   299
  -- opt_infix >> mk_type_decl;
clasohm@1383
   300
wenzelm@2360
   301
val type_decls =
wenzelm@2360
   302
  repeat1 (old_type_decl || type_decl) >> (mk_type_decls o flat);
clasohm@1383
   303
clasohm@1383
   304
clasohm@1383
   305
(* consts *)
clasohm@1383
   306
wenzelm@2360
   307
val const_decls =
wenzelm@2360
   308
  repeat1
wenzelm@2360
   309
    (names1 --$$ "::" -- !! ((string || const_type false >> quote) -- opt_mixfix))
wenzelm@2360
   310
  >> (mk_big_list o map mk_triple2 o split_decls);
wenzelm@389
   311
wenzelm@2360
   312
val opt_mode =
wenzelm@2360
   313
  optional
wenzelm@2360
   314
    ("(" $$-- !! (name -- optional ($$ "output" >> K "false") "true" --$$ ")"))
wenzelm@2360
   315
    ("\"\"", "true")
wenzelm@2360
   316
  >> mk_pair;
wenzelm@2360
   317
wenzelm@2360
   318
val syntax_decls = opt_mode -- const_decls >> (fn (mode, txt) => mode ^ "\n" ^ txt);
wenzelm@2203
   319
wenzelm@389
   320
wenzelm@389
   321
(* translations *)
wenzelm@389
   322
wenzelm@389
   323
val trans_pat =
wenzelm@389
   324
  optional ("(" $$-- !! (name --$$ ")")) "\"logic\"" -- string >> mk_pair;
wenzelm@389
   325
wenzelm@389
   326
val trans_arrow =
wenzelm@3528
   327
  $$ "=>" >> K "Syntax.ParseRule " ||
wenzelm@3528
   328
  $$ "<=" >> K "Syntax.PrintRule " ||
wenzelm@3528
   329
  $$ "==" >> K "Syntax.ParsePrintRule ";
wenzelm@389
   330
paulson@1810
   331
val trans_line =
wenzelm@2203
   332
  trans_pat -- !! (trans_arrow -- trans_pat)
wenzelm@2203
   333
    >> (fn (left, (arr, right)) => arr ^ mk_pair (left, right));
paulson@1810
   334
paulson@1810
   335
val trans_decls = repeat1 trans_line >> mk_big_list;
wenzelm@389
   336
wenzelm@389
   337
wenzelm@389
   338
(* ML translations *)
wenzelm@389
   339
wenzelm@389
   340
val trfun_defs =
wenzelm@389
   341
  " val parse_ast_translation = [];\n\
wenzelm@389
   342
  \ val parse_translation = [];\n\
wenzelm@389
   343
  \ val print_translation = [];\n\
wenzelm@2385
   344
  \ val typed_print_translation = [];\n\
wenzelm@2694
   345
  \ val print_ast_translation = [];\n\
wenzelm@2694
   346
  \ val token_translation = [];";
wenzelm@389
   347
wenzelm@389
   348
val trfun_args =
wenzelm@389
   349
  "(parse_ast_translation, parse_translation, \
wenzelm@389
   350
  \print_translation, print_ast_translation)";
wenzelm@389
   351
wenzelm@389
   352
wenzelm@389
   353
(* axioms *)
wenzelm@389
   354
wenzelm@389
   355
val mk_axms = mk_big_list o map (mk_pair o apfst quote);
wenzelm@389
   356
wenzelm@389
   357
fun mk_axiom_decls axms = (mk_axms axms, map fst axms);
wenzelm@389
   358
wenzelm@389
   359
val axiom_decls = repeat1 (ident -- !! string) >> mk_axiom_decls;
wenzelm@389
   360
wenzelm@389
   361
wenzelm@3813
   362
(* oracle *)
wenzelm@3813
   363
wenzelm@3813
   364
val oracle_decl = (name --$$ "=" -- long_id) >> mk_pair;
wenzelm@3813
   365
wenzelm@3813
   366
clasohm@1555
   367
(* combined consts and axioms *)
clasohm@1555
   368
clasohm@1555
   369
fun mk_constaxiom_decls x =
clasohm@1555
   370
  let
clasohm@1555
   371
    val (axms_defs, axms_names) =
clasohm@1555
   372
      mk_axiom_decls (map (fn ((id, _), def) => (id ^ "_def", def)) x);
clasohm@1555
   373
  in ((mk_big_list o map mk_triple2 o map (apfst quote o fst)) x ^
wenzelm@3764
   374
       "\n\n|> Theory.add_defs\n" ^ axms_defs, axms_names)
clasohm@1555
   375
  end;
clasohm@1555
   376
clasohm@1555
   377
val constaxiom_decls =
clasohm@1555
   378
  repeat1 (ident --$$ "::" -- !!
clasohm@1555
   379
           ((string || const_type false >> quote) -- opt_mixfix) -- !!
clasohm@1555
   380
           string) >> mk_constaxiom_decls;
clasohm@1555
   381
clasohm@1555
   382
wenzelm@389
   383
(* axclass *)
wenzelm@389
   384
wenzelm@389
   385
fun mk_axclass_decl ((c, cs), axms) =
wenzelm@389
   386
  (mk_pair (c, cs) ^ "\n" ^ mk_axms axms,
wenzelm@389
   387
    (strip_quotes c ^ "I") :: map fst axms);
wenzelm@389
   388
wenzelm@389
   389
val axclass_decl = subclass -- repeat (ident -- !! string) >> mk_axclass_decl;
wenzelm@389
   390
wenzelm@389
   391
wenzelm@451
   392
(* instance *)
wenzelm@389
   393
wenzelm@425
   394
fun mk_witness (axths, opt_tac) =
wenzelm@389
   395
  mk_list (keyfilter axths false) ^ "\n" ^
wenzelm@389
   396
  mk_list (keyfilter axths true) ^ "\n" ^
wenzelm@389
   397
  opt_tac;
wenzelm@389
   398
wenzelm@389
   399
val axm_or_thm =
wenzelm@389
   400
  string >> rpair false ||
wenzelm@389
   401
  long_id >> rpair true;
wenzelm@389
   402
wenzelm@451
   403
wenzelm@425
   404
val opt_witness =
wenzelm@389
   405
  optional ("(" $$-- list1 axm_or_thm --$$ ")") [] --
wenzelm@558
   406
  optional (verbatim >> (parens o cat "Some" o parens)) "None"
wenzelm@425
   407
  >> mk_witness;
wenzelm@425
   408
wenzelm@425
   409
val instance_decl =
wenzelm@636
   410
  (name --$$ "<" -- name >> (pair "|> AxClass.add_inst_subclass" o mk_pair) ||
wenzelm@636
   411
    name --$$ "::" -- arity >> (pair "|> AxClass.add_inst_arity" o mk_triple2))
wenzelm@451
   412
  -- opt_witness
wenzelm@777
   413
  >> (fn ((x, y), z) => (cat_lines [x, y, z]));
wenzelm@389
   414
wenzelm@389
   415
wenzelm@3931
   416
(* local, global path *)
wenzelm@3931
   417
wenzelm@3958
   418
fun empty_decl toks = (empty >> K "") toks;
wenzelm@3900
   419
wenzelm@3931
   420
val global_path =
wenzelm@3931
   421
  "|> (fn thy => if ! global_names then thy else Theory.add_path \"/\" thy)";
wenzelm@3931
   422
wenzelm@3931
   423
val local_path =
wenzelm@3931
   424
  global_path ^ "\n\
wenzelm@3931
   425
  \|> (fn thy => if ! global_names then thy\
wenzelm@3931
   426
  \ else Theory.add_path thy_name thy)";
wenzelm@3900
   427
wenzelm@3900
   428
wenzelm@389
   429
wenzelm@389
   430
(** theory syntax **)
wenzelm@389
   431
wenzelm@389
   432
type syntax =
wenzelm@389
   433
  lexicon * (token list -> (string * string) * token list) Symtab.table;
wenzelm@389
   434
wenzelm@389
   435
fun make_syntax keywords sects =
wenzelm@451
   436
  (make_lexicon (map fst sects @ keywords),
wenzelm@451
   437
    Symtab.make sects handle Symtab.DUPS dups =>
wenzelm@451
   438
      error ("Duplicate sections in theory file syntax: " ^ commas_quote dups));
wenzelm@389
   439
wenzelm@389
   440
wenzelm@389
   441
(* header *)
wenzelm@389
   442
wenzelm@389
   443
fun mk_header (thy_name, bases) =
clasohm@586
   444
  (thy_name, "mk_base " ^ mk_list bases ^ " " ^ quote thy_name);
wenzelm@389
   445
wenzelm@389
   446
val base =
wenzelm@389
   447
  ident >> (cat "Thy" o quote) ||
wenzelm@389
   448
  string >> cat "File";
wenzelm@389
   449
wenzelm@389
   450
val header = ident --$$ "=" -- enum1 "+" base >> mk_header;
wenzelm@389
   451
wenzelm@389
   452
wenzelm@389
   453
(* extension *)
wenzelm@389
   454
wenzelm@3900
   455
fun mk_extension (txts, mltxt) =
wenzelm@389
   456
  let
wenzelm@389
   457
    val cat_sects = space_implode "\n\n" o filter_out (equal "");
wenzelm@389
   458
    val (extxts, postxts) = split_list txts;
wenzelm@389
   459
  in
wenzelm@3900
   460
    (cat_sects extxts, cat_sects postxts, mltxt)
wenzelm@389
   461
  end;
wenzelm@389
   462
wenzelm@389
   463
fun sect tab ((Keyword, s, n) :: toks) =
wenzelm@389
   464
      (case Symtab.lookup (tab, s) of
wenzelm@389
   465
        Some parse => !! parse toks
wenzelm@389
   466
      | None => syn_err "section" s n)
wenzelm@389
   467
  | sect _ ((_, s, n) :: _) = syn_err "section" s n
wenzelm@389
   468
  | sect _ [] = eof_err ();
wenzelm@389
   469
wenzelm@3900
   470
fun extension sectab = "+" $$-- !!
wenzelm@3900
   471
  (repeat (sect sectab) --$$ "end" -- optional ("ML" $$-- verbatim) "")
wenzelm@3900
   472
    >> mk_extension;
wenzelm@3813
   473
wenzelm@3900
   474
fun opt_extension sectab = optional (extension sectab) ("", "", "");
wenzelm@3875
   475
wenzelm@389
   476
wenzelm@389
   477
(* theory definition *)
wenzelm@389
   478
wenzelm@3900
   479
fun mk_structure tname ((thy_name, old_thys), (extxt, postxt, mltxt)) =
wenzelm@558
   480
  if thy_name <> tname then
wenzelm@558
   481
    error ("Filename \"" ^ tname ^ ".thy\" and theory name "
wenzelm@558
   482
      ^ quote thy_name ^ " are different")
wenzelm@558
   483
  else
wenzelm@3875
   484
    "val thy = " ^ old_thys ^ ";\n\n\
wenzelm@3875
   485
    \structure " ^ thy_name ^ " =\n\
wenzelm@3875
   486
    \struct\n\
wenzelm@3875
   487
    \\n\
wenzelm@3931
   488
    \local\n\
wenzelm@3931
   489
    \ val thy_name = \"" ^ tname ^ "\";\n"
wenzelm@3875
   490
    ^ trfun_defs ^ "\n\
wenzelm@3875
   491
    \in\n\
wenzelm@3875
   492
    \\n"
wenzelm@3875
   493
    ^ mltxt ^ "\n\
wenzelm@3875
   494
    \\n\
wenzelm@3875
   495
    \val thy = thy\n\
wenzelm@3931
   496
    \\n"
wenzelm@3931
   497
    ^ local_path ^
wenzelm@3875
   498
    "\n\
wenzelm@3875
   499
    \|> Theory.add_trfuns\n"
wenzelm@3875
   500
    ^ trfun_args ^ "\n\
wenzelm@3875
   501
    \|> Theory.add_trfunsT typed_print_translation \n\
wenzelm@3875
   502
    \|> Theory.add_tokentrfuns token_translation \n\
wenzelm@3875
   503
    \\n"
wenzelm@3875
   504
    ^ extxt ^ "\n\
wenzelm@3875
   505
    \\n\
wenzelm@3875
   506
    \|> Theory.add_name " ^ quote thy_name ^ ";\n\
wenzelm@3875
   507
    \\n\
wenzelm@3875
   508
    \val _ = store_theory (thy, " ^ quote thy_name ^ ");\n\
wenzelm@3875
   509
    \\n\
wenzelm@3875
   510
    \\n"
wenzelm@3875
   511
    ^ postxt ^ "\n\
wenzelm@3875
   512
    \\n\
wenzelm@3875
   513
    \end;\n\
wenzelm@3875
   514
    \end;\n\
wenzelm@3875
   515
    \\n\
wenzelm@3875
   516
    \open " ^ thy_name ^ ";\n\
wenzelm@3875
   517
    \\n";
wenzelm@389
   518
clasohm@476
   519
fun theory_defn sectab tname =
wenzelm@3875
   520
  header -- opt_extension sectab -- eof >> (mk_structure tname o #1);
wenzelm@389
   521
clasohm@476
   522
fun parse_thy (lex, sectab) tname str =
clasohm@476
   523
  #1 (!! (theory_defn sectab tname) (tokenize lex str));
wenzelm@389
   524
wenzelm@389
   525
wenzelm@389
   526
(* standard sections *)
wenzelm@389
   527
wenzelm@389
   528
fun mk_val ax = "val " ^ ax ^ " = get_axiom thy " ^ quote ax ^ ";";
wenzelm@777
   529
val mk_vals = cat_lines o map mk_val;
wenzelm@389
   530
wenzelm@777
   531
fun mk_axm_sect "" (txt, axs) = (txt, mk_vals axs)
wenzelm@777
   532
  | mk_axm_sect pretxt (txt, axs) = (pretxt ^ "\n" ^ txt, mk_vals axs);
wenzelm@389
   533
wenzelm@389
   534
fun axm_section name pretxt parse =
wenzelm@389
   535
  (name, parse >> mk_axm_sect pretxt);
wenzelm@389
   536
wenzelm@389
   537
fun section name pretxt parse =
wenzelm@389
   538
  axm_section name pretxt (parse >> rpair []);
wenzelm@389
   539
wenzelm@389
   540
wenzelm@389
   541
val pure_keywords =
wenzelm@3900
   542
 ["end", "ML", "mixfix", "infixr", "infixl", "binder", "global",
wenzelm@3931
   543
  "local", "output", "=", "+", ",", "<", "{", "}", "(", ")", "[", "]",
wenzelm@3931
   544
  "::", "==", "=>", "<="];
wenzelm@389
   545
wenzelm@389
   546
val pure_sections =
wenzelm@3813
   547
 [section "classes" "|> Theory.add_classes" class_decls,
wenzelm@3764
   548
  section "default" "|> Theory.add_defsort" sort,
wenzelm@777
   549
  section "types" "" type_decls,
wenzelm@3764
   550
  section "arities" "|> Theory.add_arities" arity_decls,
wenzelm@3764
   551
  section "consts" "|> Theory.add_consts" const_decls,
wenzelm@3764
   552
  section "syntax" "|> Theory.add_modesyntax" syntax_decls,
wenzelm@3764
   553
  section "translations" "|> Theory.add_trrules" trans_decls,
wenzelm@3764
   554
  axm_section "rules" "|> Theory.add_axioms" axiom_decls,
wenzelm@3764
   555
  axm_section "defs" "|> Theory.add_defs" axiom_decls,
wenzelm@3813
   556
  section "oracle" "|> Theory.add_oracle" oracle_decl,
wenzelm@3764
   557
  axm_section "constdefs" "|> Theory.add_consts" constaxiom_decls,
wenzelm@636
   558
  axm_section "axclass" "|> AxClass.add_axclass" axclass_decl,
wenzelm@3780
   559
  section "instance" "" instance_decl,
wenzelm@3900
   560
  section "path" "|> Theory.add_path" name,
wenzelm@3931
   561
  section "global" global_path empty_decl,
wenzelm@3931
   562
  section "local" local_path empty_decl];
wenzelm@389
   563
wenzelm@3813
   564
wenzelm@389
   565
end;