src/Pure/Thy/thy_parse.ML
author paulson
Mon Dec 28 16:50:37 1998 +0100 (1998-12-28)
changeset 6043 3eecc7fbfad8
parent 6022 259e4f2114e1
child 6090 78c068b838ff
permissions -rw-r--r--
more efficient strip_quotes using "substring"
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@389
     8
infix 5 -- --$$ $$-- ^^;
wenzelm@389
     9
infix 3 >>;
wenzelm@389
    10
infix 0 ||;
wenzelm@389
    11
wenzelm@389
    12
signature THY_PARSE =
wenzelm@3977
    13
sig
wenzelm@389
    14
  type token
wenzelm@389
    15
  val !! : ('a -> 'b * 'c) -> 'a -> 'b * 'c
wenzelm@389
    16
  val >> : ('a -> 'b * 'c) * ('b -> 'd) -> 'a -> 'd * 'c
wenzelm@389
    17
  val || : ('a -> 'b) * ('a -> 'b) -> 'a -> 'b
wenzelm@389
    18
  val -- : ('a -> 'b * 'c) * ('c -> 'd * 'e) -> 'a -> ('b * 'd) * 'e
wenzelm@389
    19
  val ^^ : ('a -> string * 'b) * ('b -> string * 'c) -> 'a -> string * 'c
wenzelm@389
    20
  val $$ : string -> token list -> string * token list
wenzelm@389
    21
  val $$-- : string * (token list -> 'b * 'c) -> token list -> 'b * 'c
wenzelm@389
    22
  val --$$ : ('a -> 'b * token list) * string -> 'a -> 'b * token list
wenzelm@389
    23
  val ident: token list -> string * token list
wenzelm@389
    24
  val long_ident: token list -> string * token list
wenzelm@389
    25
  val long_id: token list -> string * token list
wenzelm@389
    26
  val type_var: token list -> string * token list
wenzelm@636
    27
  val type_args: token list -> string list * token list
wenzelm@389
    28
  val nat: token list -> string * token list
wenzelm@389
    29
  val string: token list -> string * token list
wenzelm@389
    30
  val verbatim: token list -> string * token list
wenzelm@389
    31
  val empty: 'a -> 'b list * 'a
wenzelm@389
    32
  val optional: ('a -> 'b * 'a) -> 'b -> 'a -> 'b * 'a
wenzelm@389
    33
  val repeat: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
wenzelm@389
    34
  val repeat1: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
wenzelm@389
    35
  val enum: string -> (token list -> 'a * token list)
wenzelm@389
    36
    -> token list -> 'a list * token list
wenzelm@389
    37
  val enum1: string -> (token list -> 'a * token list)
wenzelm@389
    38
    -> token list -> 'a list * token list
wenzelm@389
    39
  val list: (token list -> 'a * token list)
wenzelm@389
    40
    -> token list -> 'a list * token list
wenzelm@389
    41
  val list1: (token list -> 'a * token list)
wenzelm@389
    42
    -> token list -> 'a list * token list
wenzelm@389
    43
  val name: token list -> string * token list
wenzelm@389
    44
  val sort: token list -> string * token list
wenzelm@3977
    45
  val typ: token list -> string * token list
wenzelm@451
    46
  val opt_infix: token list -> string * token list
wenzelm@451
    47
  val opt_mixfix: token list -> string * token list
wenzelm@636
    48
  val opt_witness: token list -> string * token list
wenzelm@4099
    49
  val const_decls: token list -> string * token list
wenzelm@389
    50
  type syntax
wenzelm@389
    51
  val make_syntax: string list ->
wenzelm@389
    52
    (string * (token list -> (string * string) * token list)) list -> syntax
clasohm@476
    53
  val parse_thy: syntax -> string -> string -> string
wenzelm@389
    54
  val section: string -> string -> (token list -> string * token list)
wenzelm@389
    55
    -> (string * (token list -> (string * string) * token list))
wenzelm@389
    56
  val axm_section: string -> string
wenzelm@389
    57
    -> (token list -> (string * string list) * token list)
wenzelm@389
    58
    -> (string * (token list -> (string * string) * token list))
wenzelm@389
    59
  val pure_keywords: string list
wenzelm@389
    60
  val pure_sections:
wenzelm@389
    61
    (string * (token list -> (string * string) * token list)) list
lcp@570
    62
  (*items for building strings*)
wenzelm@710
    63
  val cat: string -> string -> string
wenzelm@656
    64
  val parens: string -> string
wenzelm@656
    65
  val brackets: string -> string
wenzelm@656
    66
  val mk_list: string list -> string
wenzelm@656
    67
  val mk_big_list: string list -> string
wenzelm@656
    68
  val mk_pair: string * string -> string
wenzelm@656
    69
  val mk_triple: string * string * string -> string
wenzelm@5058
    70
  val mk_triple1: (string * string) * string -> string
wenzelm@5058
    71
  val mk_triple2: string * (string * string) -> string
wenzelm@656
    72
  val strip_quotes: string -> string
wenzelm@3977
    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 list1 parse = enum1 "," parse;
wenzelm@3977
   157
fun list parse = enum "," parse;
wenzelm@3977
   158
wenzelm@389
   159
wenzelm@389
   160
wenzelm@389
   161
(** theory parsers **)
wenzelm@389
   162
wenzelm@389
   163
(* misc utilities *)
wenzelm@389
   164
wenzelm@389
   165
fun cat s1 s2 = s1 ^ " " ^ s2;
wenzelm@389
   166
wenzelm@558
   167
val parens = enclose "(" ")";
wenzelm@558
   168
val brackets = enclose "[" "]";
wenzelm@389
   169
wenzelm@389
   170
val mk_list = brackets o commas;
wenzelm@389
   171
val mk_big_list = brackets o space_implode ",\n ";
wenzelm@389
   172
wenzelm@558
   173
fun mk_pair (x, y) = parens (commas [x, y]);
wenzelm@558
   174
fun mk_triple (x, y, z) = parens (commas [x, y, z]);
wenzelm@389
   175
fun mk_triple1 ((x, y), z) = mk_triple (x, y, z);
wenzelm@389
   176
fun mk_triple2 (x, (y, z)) = mk_triple (x, y, z);
wenzelm@389
   177
paulson@2231
   178
fun split_decls l = flat (map (fn (xs, y) => map (rpair y) xs) l);
wenzelm@389
   179
paulson@6043
   180
(*Remove the leading and trailing chararacters.  Actually called to
paulson@6043
   181
  remove quotation marks.*)
paulson@6043
   182
fun strip_quotes s = String.substring (s, 1, size s - 2);
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@4965
   194
(* empty *)
wenzelm@4965
   195
wenzelm@4965
   196
fun empty_decl toks = (empty >> K "") toks;
wenzelm@4965
   197
wenzelm@4965
   198
wenzelm@389
   199
(* classes *)
wenzelm@389
   200
wenzelm@389
   201
val subclass = name -- optional ("<" $$-- !! name_list1) "[]";
wenzelm@389
   202
wenzelm@558
   203
val class_decls = repeat1 (subclass >> mk_pair) >> mk_big_list;
wenzelm@389
   204
wenzelm@389
   205
wenzelm@389
   206
(* arities *)
wenzelm@389
   207
wenzelm@389
   208
val sort =
wenzelm@389
   209
  name >> brackets ||
wenzelm@389
   210
  "{" $$-- name_list --$$ "}";
wenzelm@389
   211
wenzelm@389
   212
val sort_list1 = list1 sort >> mk_list;
wenzelm@389
   213
wenzelm@389
   214
wenzelm@389
   215
val arity = optional ("(" $$-- !! (sort_list1 --$$")")) "[]" -- sort;
wenzelm@389
   216
wenzelm@389
   217
val arity_decls = repeat1 (names1 --$$ "::" -- !! arity)
wenzelm@389
   218
  >> (mk_big_list o map mk_triple2 o split_decls);
wenzelm@389
   219
wenzelm@389
   220
wenzelm@389
   221
(* mixfix annotations *)
wenzelm@389
   222
wenzelm@2203
   223
val infxl =
wenzelm@2203
   224
  "infixl" $$-- !! (nat >> cat "Infixl" || string -- nat >> (cat "InfixlName" o mk_pair));
wenzelm@2203
   225
val infxr =
wenzelm@2203
   226
  "infixr" $$-- !! (nat >> cat "Infixr" || string -- nat >> (cat "InfixrName" o mk_pair));
wenzelm@389
   227
wenzelm@889
   228
val binder = "binder" $$--
wenzelm@2203
   229
  !! (string -- (("[" $$-- nat --$$ "]") -- nat || nat >> (fn n => (n, n))))
wenzelm@2203
   230
    >> (cat "Binder" o mk_triple2);
wenzelm@389
   231
wenzelm@389
   232
val opt_pris = optional ("[" $$-- !! (list nat --$$ "]")) [] >> mk_list;
wenzelm@389
   233
wenzelm@5687
   234
val mixfix = string -- !! (opt_pris -- optional nat "Syntax.max_pri")
wenzelm@389
   235
  >> (cat "Mixfix" o mk_triple2);
wenzelm@389
   236
wenzelm@558
   237
fun opt_syn fx = optional ("(" $$-- fx --$$ ")") "NoSyn";
wenzelm@389
   238
wenzelm@389
   239
val opt_infix = opt_syn (infxl || infxr);
wenzelm@389
   240
val opt_mixfix = opt_syn (mixfix || infxl || infxr || binder);
wenzelm@389
   241
wenzelm@389
   242
wenzelm@389
   243
(* types *)
wenzelm@389
   244
wenzelm@3977
   245
(* FIXME clean!! *)
wenzelm@3977
   246
clasohm@1705
   247
(*Parse an identifier, but only if it is not followed by "::", "=" or ",";
clasohm@1377
   248
  the exclusion of a postfix comma can be controlled to allow expressions
clasohm@1377
   249
  like "(id, id)" but disallow ones like "'a => id id,id :: ..."*)
clasohm@1377
   250
fun ident_no_colon _ [] = eof_err()
clasohm@1377
   251
  | ident_no_colon allow_comma ((Ident, s, n) :: (rest as (Keyword, s2, n2) ::
clasohm@1377
   252
                                toks)) =
clasohm@1705
   253
      if s2 = "::" orelse s2 = "=" orelse (not allow_comma andalso s2 = ",")
clasohm@1705
   254
      then syn_err (name_of_kind Ident) (quote s2) n2
clasohm@1377
   255
      else (s, rest)
clasohm@1377
   256
  | ident_no_colon _ ((Ident, s, n) :: toks) = (s, toks)
clasohm@1377
   257
  | ident_no_colon _ ((k, s, n) :: _) =
clasohm@1377
   258
      syn_err (name_of_kind Ident) (quote s) n;
clasohm@1321
   259
wenzelm@3110
   260
(*type used in types, consts and syntax sections*)
clasohm@1377
   261
fun const_type allow_comma toks =
wenzelm@3977
   262
  let
wenzelm@3977
   263
    val simple_type =
wenzelm@3977
   264
      (ident || kind TypeVar ^^ optional ($$ "::" ^^ ident) "") --
wenzelm@3977
   265
          repeat (ident_no_colon allow_comma)
wenzelm@3977
   266
          >> (fn (args, ts) => cat args (space_implode " " ts)) ||
wenzelm@3977
   267
        ("(" $$-- (list1 (const_type true)) --$$ ")" >> (parens o commas)) --
wenzelm@3977
   268
          repeat1 (ident_no_colon allow_comma)
wenzelm@3977
   269
          >> (fn (args, ts) => cat args (space_implode " " ts));
clasohm@1377
   270
clasohm@1377
   271
      val appl_param =
clasohm@1377
   272
        simple_type || "(" $$-- const_type true --$$ ")" >> parens || 
clasohm@1377
   273
        "[" $$-- (list1 (const_type true)) --$$ "]" --$$ "=>" --
clasohm@1377
   274
          const_type allow_comma >>
clasohm@1377
   275
          (fn (src, dest) => mk_list src ^ " => " ^ dest);
clasohm@1377
   276
  in ("[" $$-- (list1 (const_type true)) --$$ "]" --$$ "=>" --
clasohm@1377
   277
        const_type allow_comma >>
clasohm@1377
   278
        (fn (src, dest) => mk_list src ^ " => " ^ dest) ||
clasohm@1377
   279
      repeat1 (appl_param --$$ "=>") -- const_type allow_comma >>
clasohm@1377
   280
        (fn (src, dest) => space_implode " => " (src@[dest])) ||
clasohm@1377
   281
      simple_type ||
clasohm@1377
   282
      "(" $$-- const_type true --$$ ")" >> parens) toks
clasohm@1321
   283
  end;
clasohm@1321
   284
wenzelm@3977
   285
val typ = string || (const_type false >> quote);
wenzelm@3977
   286
wenzelm@3977
   287
clasohm@1383
   288
fun mk_old_type_decl ((ts, n), syn) =
clasohm@1383
   289
  map (fn t => (mk_triple (t, n, syn), false)) ts;
clasohm@1383
   290
clasohm@1383
   291
fun mk_type_decl (((xs, t), None), syn) =
clasohm@1383
   292
      [(mk_triple (t, string_of_int (length xs), syn), false)]
clasohm@1383
   293
  | mk_type_decl (((xs, t), Some rhs), syn) =
clasohm@1383
   294
      [(parens (commas [t, mk_list xs, rhs, syn]), true)];
clasohm@1383
   295
clasohm@1383
   296
fun mk_type_decls tys =
wenzelm@3764
   297
  "|> Theory.add_types\n" ^ mk_big_list (keyfilter tys false) ^ "\n\n\
wenzelm@3764
   298
  \|> Theory.add_tyabbrs\n" ^ mk_big_list (keyfilter tys true);
clasohm@1383
   299
clasohm@1383
   300
clasohm@1383
   301
val old_type_decl = names1 -- nat -- opt_infix >> mk_old_type_decl;
clasohm@1383
   302
clasohm@1383
   303
val type_args =
clasohm@1383
   304
  type_var >> (fn x => [x]) ||
clasohm@1383
   305
  "(" $$-- !! (list1 type_var --$$ ")") ||
clasohm@1383
   306
  empty >> K [];
clasohm@1383
   307
clasohm@1383
   308
val type_decl = type_args -- name --
wenzelm@3977
   309
  optional ("=" $$-- typ >> Some) None -- opt_infix >> mk_type_decl;
clasohm@1383
   310
wenzelm@2360
   311
val type_decls =
wenzelm@2360
   312
  repeat1 (old_type_decl || type_decl) >> (mk_type_decls o flat);
clasohm@1383
   313
clasohm@1383
   314
clasohm@1383
   315
(* consts *)
clasohm@1383
   316
wenzelm@2360
   317
val const_decls =
wenzelm@3977
   318
  repeat1 (names1 --$$ "::" -- !! (typ -- opt_mixfix))
wenzelm@2360
   319
  >> (mk_big_list o map mk_triple2 o split_decls);
wenzelm@389
   320
wenzelm@2360
   321
val opt_mode =
wenzelm@2360
   322
  optional
wenzelm@2360
   323
    ("(" $$-- !! (name -- optional ($$ "output" >> K "false") "true" --$$ ")"))
wenzelm@2360
   324
    ("\"\"", "true")
wenzelm@2360
   325
  >> mk_pair;
wenzelm@2360
   326
wenzelm@2360
   327
val syntax_decls = opt_mode -- const_decls >> (fn (mode, txt) => mode ^ "\n" ^ txt);
wenzelm@2203
   328
wenzelm@389
   329
wenzelm@389
   330
(* translations *)
wenzelm@389
   331
wenzelm@389
   332
val trans_pat =
wenzelm@389
   333
  optional ("(" $$-- !! (name --$$ ")")) "\"logic\"" -- string >> mk_pair;
wenzelm@389
   334
wenzelm@389
   335
val trans_arrow =
wenzelm@3528
   336
  $$ "=>" >> K "Syntax.ParseRule " ||
wenzelm@3528
   337
  $$ "<=" >> K "Syntax.PrintRule " ||
wenzelm@3528
   338
  $$ "==" >> K "Syntax.ParsePrintRule ";
wenzelm@389
   339
paulson@1810
   340
val trans_line =
wenzelm@2203
   341
  trans_pat -- !! (trans_arrow -- trans_pat)
wenzelm@2203
   342
    >> (fn (left, (arr, right)) => arr ^ mk_pair (left, right));
paulson@1810
   343
paulson@1810
   344
val trans_decls = repeat1 trans_line >> mk_big_list;
wenzelm@389
   345
wenzelm@389
   346
wenzelm@389
   347
(* ML translations *)
wenzelm@389
   348
wenzelm@4047
   349
val local_defs =
wenzelm@389
   350
  " val parse_ast_translation = [];\n\
wenzelm@389
   351
  \ val parse_translation = [];\n\
wenzelm@389
   352
  \ val print_translation = [];\n\
wenzelm@2385
   353
  \ val typed_print_translation = [];\n\
wenzelm@2694
   354
  \ val print_ast_translation = [];\n\
wenzelm@4852
   355
  \ val token_translation = [];";
wenzelm@389
   356
wenzelm@389
   357
val trfun_args =
wenzelm@389
   358
  "(parse_ast_translation, parse_translation, \
wenzelm@389
   359
  \print_translation, print_ast_translation)";
wenzelm@389
   360
wenzelm@389
   361
wenzelm@389
   362
(* axioms *)
wenzelm@389
   363
wenzelm@389
   364
val mk_axms = mk_big_list o map (mk_pair o apfst quote);
wenzelm@389
   365
wenzelm@389
   366
fun mk_axiom_decls axms = (mk_axms axms, map fst axms);
wenzelm@389
   367
wenzelm@389
   368
val axiom_decls = repeat1 (ident -- !! string) >> mk_axiom_decls;
wenzelm@389
   369
wenzelm@389
   370
wenzelm@3813
   371
(* oracle *)
wenzelm@3813
   372
wenzelm@3813
   373
val oracle_decl = (name --$$ "=" -- long_id) >> mk_pair;
wenzelm@3813
   374
wenzelm@3813
   375
clasohm@1555
   376
(* combined consts and axioms *)
clasohm@1555
   377
clasohm@1555
   378
fun mk_constaxiom_decls x =
clasohm@1555
   379
  let
clasohm@1555
   380
    val (axms_defs, axms_names) =
clasohm@1555
   381
      mk_axiom_decls (map (fn ((id, _), def) => (id ^ "_def", def)) x);
clasohm@1555
   382
  in ((mk_big_list o map mk_triple2 o map (apfst quote o fst)) x ^
wenzelm@4852
   383
       "\n\n|> (PureThy.add_defs o map Attribute.none)\n" ^ axms_defs, axms_names)
clasohm@1555
   384
  end;
clasohm@1555
   385
clasohm@1555
   386
val constaxiom_decls =
wenzelm@3977
   387
  repeat1 (ident --$$ "::" -- !! (typ -- opt_mixfix) -- !! string)
wenzelm@3977
   388
  >> mk_constaxiom_decls;
clasohm@1555
   389
clasohm@1555
   390
wenzelm@389
   391
(* axclass *)
wenzelm@389
   392
wenzelm@389
   393
fun mk_axclass_decl ((c, cs), axms) =
wenzelm@389
   394
  (mk_pair (c, cs) ^ "\n" ^ mk_axms axms,
wenzelm@389
   395
    (strip_quotes c ^ "I") :: map fst axms);
wenzelm@389
   396
wenzelm@389
   397
val axclass_decl = subclass -- repeat (ident -- !! string) >> mk_axclass_decl;
wenzelm@389
   398
wenzelm@389
   399
wenzelm@451
   400
(* instance *)
wenzelm@389
   401
wenzelm@425
   402
fun mk_witness (axths, opt_tac) =
wenzelm@389
   403
  mk_list (keyfilter axths false) ^ "\n" ^
wenzelm@389
   404
  mk_list (keyfilter axths true) ^ "\n" ^
wenzelm@389
   405
  opt_tac;
wenzelm@389
   406
wenzelm@389
   407
val axm_or_thm =
wenzelm@389
   408
  string >> rpair false ||
wenzelm@389
   409
  long_id >> rpair true;
wenzelm@389
   410
wenzelm@451
   411
wenzelm@425
   412
val opt_witness =
wenzelm@389
   413
  optional ("(" $$-- list1 axm_or_thm --$$ ")") [] --
wenzelm@558
   414
  optional (verbatim >> (parens o cat "Some" o parens)) "None"
wenzelm@425
   415
  >> mk_witness;
wenzelm@425
   416
wenzelm@425
   417
val instance_decl =
wenzelm@636
   418
  (name --$$ "<" -- name >> (pair "|> AxClass.add_inst_subclass" o mk_pair) ||
wenzelm@636
   419
    name --$$ "::" -- arity >> (pair "|> AxClass.add_inst_arity" o mk_triple2))
wenzelm@451
   420
  -- opt_witness
wenzelm@777
   421
  >> (fn ((x, y), z) => (cat_lines [x, y, z]));
wenzelm@389
   422
wenzelm@389
   423
wenzelm@5248
   424
(* locale *)
wenzelm@5248
   425
wenzelm@5248
   426
val locale_decl =
paulson@6022
   427
  (name --$$ "=") -- 
paulson@6022
   428
    (optional ((ident >> (fn x => parens ("Some" ^ quote x))) --$$ "+") ("None")) --
wenzelm@5248
   429
    ("fixes" $$--
paulson@6015
   430
      (repeat (name --$$ "::" -- !! (typ -- opt_mixfix)) 
paulson@6015
   431
       >> (mk_big_list o map mk_triple2))) --
paulson@6015
   432
    (optional 
paulson@6015
   433
     ("assumes" $$-- (repeat ((ident >> quote) -- !! string) 
paulson@6015
   434
		     >> (mk_list o map mk_pair)))
paulson@6015
   435
     "[]") --
paulson@6015
   436
    (optional 
paulson@6015
   437
     ("defines" $$-- (repeat ((ident >> quote) -- !! string) 
paulson@6015
   438
		      >> (mk_list o map mk_pair)))
paulson@6015
   439
     "[]")
paulson@6022
   440
  >> (fn ((((nm, ext), cs), asms), defs) => cat_lines [nm, ext, cs, asms, defs]);
wenzelm@5248
   441
wenzelm@5248
   442
wenzelm@389
   443
wenzelm@389
   444
(** theory syntax **)
wenzelm@389
   445
wenzelm@389
   446
type syntax =
wenzelm@4707
   447
  Scan.lexicon * (token list -> (string * string) * token list) Symtab.table;
wenzelm@389
   448
wenzelm@389
   449
fun make_syntax keywords sects =
wenzelm@4056
   450
  let
wenzelm@4056
   451
    val dups = duplicates (map fst sects);
wenzelm@4496
   452
    val sects' = gen_distinct eq_fst sects;
wenzelm@4707
   453
    val keys = map Symbol.explode (map fst sects' @ keywords);
wenzelm@4056
   454
  in
wenzelm@4056
   455
    if null dups then ()
wenzelm@4056
   456
    else warning ("Duplicate declaration of theory file sections:\n" ^ commas_quote dups);
wenzelm@4707
   457
    (Scan.make_lexicon keys, Symtab.make sects')
wenzelm@4056
   458
  end;
wenzelm@389
   459
wenzelm@389
   460
wenzelm@389
   461
(* header *)
wenzelm@389
   462
wenzelm@389
   463
fun mk_header (thy_name, bases) =
clasohm@586
   464
  (thy_name, "mk_base " ^ mk_list bases ^ " " ^ quote thy_name);
wenzelm@389
   465
wenzelm@389
   466
val base =
wenzelm@389
   467
  ident >> (cat "Thy" o quote) ||
wenzelm@389
   468
  string >> cat "File";
wenzelm@389
   469
wenzelm@389
   470
val header = ident --$$ "=" -- enum1 "+" base >> mk_header;
wenzelm@389
   471
wenzelm@389
   472
wenzelm@389
   473
(* extension *)
wenzelm@389
   474
wenzelm@3900
   475
fun mk_extension (txts, mltxt) =
wenzelm@389
   476
  let
wenzelm@389
   477
    val cat_sects = space_implode "\n\n" o filter_out (equal "");
wenzelm@389
   478
    val (extxts, postxts) = split_list txts;
wenzelm@389
   479
  in
wenzelm@3900
   480
    (cat_sects extxts, cat_sects postxts, mltxt)
wenzelm@389
   481
  end;
wenzelm@389
   482
wenzelm@389
   483
fun sect tab ((Keyword, s, n) :: toks) =
wenzelm@389
   484
      (case Symtab.lookup (tab, s) of
wenzelm@389
   485
        Some parse => !! parse toks
wenzelm@389
   486
      | None => syn_err "section" s n)
wenzelm@389
   487
  | sect _ ((_, s, n) :: _) = syn_err "section" s n
wenzelm@389
   488
  | sect _ [] = eof_err ();
wenzelm@389
   489
wenzelm@3900
   490
fun extension sectab = "+" $$-- !!
wenzelm@4707
   491
  (repeat (sect sectab) --$$ "end" -- optional verbatim "")
wenzelm@3900
   492
    >> mk_extension;
wenzelm@3813
   493
wenzelm@3900
   494
fun opt_extension sectab = optional (extension sectab) ("", "", "");
wenzelm@3875
   495
wenzelm@389
   496
wenzelm@389
   497
(* theory definition *)
wenzelm@389
   498
wenzelm@3900
   499
fun mk_structure tname ((thy_name, old_thys), (extxt, postxt, mltxt)) =
wenzelm@558
   500
  if thy_name <> tname then
wenzelm@558
   501
    error ("Filename \"" ^ tname ^ ".thy\" and theory name "
wenzelm@558
   502
      ^ quote thy_name ^ " are different")
wenzelm@558
   503
  else
wenzelm@3875
   504
    "val thy = " ^ old_thys ^ ";\n\n\
wenzelm@3875
   505
    \structure " ^ thy_name ^ " =\n\
wenzelm@3875
   506
    \struct\n\
wenzelm@3875
   507
    \\n\
wenzelm@4965
   508
    \local\n"
wenzelm@4047
   509
    ^ local_defs ^ "\n\
wenzelm@3875
   510
    \in\n\
wenzelm@3875
   511
    \\n"
wenzelm@3875
   512
    ^ mltxt ^ "\n\
wenzelm@3875
   513
    \\n\
wenzelm@3875
   514
    \val thy = thy\n\
wenzelm@4965
   515
    \|> PureThy.put_name " ^ quote thy_name ^ "\n\
wenzelm@4965
   516
    \|> PureThy.local_path\n\
wenzelm@3875
   517
    \|> Theory.add_trfuns\n"
wenzelm@3875
   518
    ^ trfun_args ^ "\n\
wenzelm@4047
   519
    \|> Theory.add_trfunsT typed_print_translation\n\
wenzelm@4047
   520
    \|> Theory.add_tokentrfuns token_translation\n\
wenzelm@3875
   521
    \\n"
wenzelm@3875
   522
    ^ extxt ^ "\n\
wenzelm@3875
   523
    \\n\
wenzelm@4965
   524
    \|> PureThy.end_theory\n\
wenzelm@3875
   525
    \\n\
wenzelm@4965
   526
    \val _ = store_theory thy;\n\
wenzelm@4056
   527
    \val _ = context thy;\n\
wenzelm@3875
   528
    \\n\
wenzelm@3875
   529
    \\n"
wenzelm@3875
   530
    ^ postxt ^ "\n\
wenzelm@3875
   531
    \\n\
wenzelm@3875
   532
    \end;\n\
wenzelm@3875
   533
    \end;\n\
wenzelm@3875
   534
    \\n\
wenzelm@3875
   535
    \open " ^ thy_name ^ ";\n\
wenzelm@3875
   536
    \\n";
wenzelm@389
   537
clasohm@476
   538
fun theory_defn sectab tname =
wenzelm@3875
   539
  header -- opt_extension sectab -- eof >> (mk_structure tname o #1);
wenzelm@389
   540
clasohm@476
   541
fun parse_thy (lex, sectab) tname str =
clasohm@476
   542
  #1 (!! (theory_defn sectab tname) (tokenize lex str));
wenzelm@389
   543
wenzelm@389
   544
wenzelm@389
   545
(* standard sections *)
wenzelm@389
   546
wenzelm@389
   547
fun mk_val ax = "val " ^ ax ^ " = get_axiom thy " ^ quote ax ^ ";";
wenzelm@777
   548
val mk_vals = cat_lines o map mk_val;
wenzelm@389
   549
wenzelm@777
   550
fun mk_axm_sect "" (txt, axs) = (txt, mk_vals axs)
wenzelm@777
   551
  | mk_axm_sect pretxt (txt, axs) = (pretxt ^ "\n" ^ txt, mk_vals axs);
wenzelm@389
   552
wenzelm@389
   553
fun axm_section name pretxt parse =
wenzelm@389
   554
  (name, parse >> mk_axm_sect pretxt);
wenzelm@389
   555
wenzelm@389
   556
fun section name pretxt parse =
wenzelm@389
   557
  axm_section name pretxt (parse >> rpair []);
wenzelm@389
   558
wenzelm@389
   559
wenzelm@389
   560
val pure_keywords =
wenzelm@4952
   561
 ["end", "ML", "mixfix", "infixr", "infixl", "binder", "output", "=",
wenzelm@4952
   562
  "+", ",", "<", "{", "}", "(", ")", "[", "]", "::", "==", "=>",
wenzelm@5248
   563
  "<=", "fixes", "assumes", "defines"];
wenzelm@389
   564
wenzelm@389
   565
val pure_sections =
wenzelm@3813
   566
 [section "classes" "|> Theory.add_classes" class_decls,
wenzelm@3764
   567
  section "default" "|> Theory.add_defsort" sort,
wenzelm@777
   568
  section "types" "" type_decls,
wenzelm@4852
   569
  section "nonterminals" "|> Theory.add_nonterminals" (repeat1 name >> mk_list),
wenzelm@3764
   570
  section "arities" "|> Theory.add_arities" arity_decls,
wenzelm@3764
   571
  section "consts" "|> Theory.add_consts" const_decls,
wenzelm@3764
   572
  section "syntax" "|> Theory.add_modesyntax" syntax_decls,
wenzelm@3764
   573
  section "translations" "|> Theory.add_trrules" trans_decls,
wenzelm@4852
   574
  axm_section "rules" "|> (PureThy.add_axioms o map Attribute.none)" axiom_decls,
wenzelm@4852
   575
  axm_section "defs" "|> (PureThy.add_defs o map Attribute.none)" axiom_decls,
wenzelm@3813
   576
  section "oracle" "|> Theory.add_oracle" oracle_decl,
wenzelm@3764
   577
  axm_section "constdefs" "|> Theory.add_consts" constaxiom_decls,
wenzelm@636
   578
  axm_section "axclass" "|> AxClass.add_axclass" axclass_decl,
wenzelm@3780
   579
  section "instance" "" instance_decl,
wenzelm@3900
   580
  section "path" "|> Theory.add_path" name,
wenzelm@4965
   581
  section "global" "|> PureThy.global_path" empty_decl,
wenzelm@4965
   582
  section "local" "|> PureThy.local_path" empty_decl,
wenzelm@5905
   583
  section "setup" "|> Library.apply" long_id,
wenzelm@5248
   584
  section "MLtext" "" verbatim,
wenzelm@5248
   585
  section "locale" "|> Locale.add_locale" locale_decl];
wenzelm@389
   586
wenzelm@3813
   587
wenzelm@389
   588
end;