src/Pure/Isar/outer_syntax.ML
author aspinall
Wed Aug 18 16:04:39 2004 +0200 (2004-08-18 ago)
changeset 15144 85929e1b307d
parent 14981 e73f8140af78
child 15156 daa9f645a26e
permissions -rw-r--r--
Remove isar_readstring. Split read into scanner and parser.
wenzelm@5829
     1
(*  Title:      Pure/Isar/outer_syntax.ML
wenzelm@5829
     2
    ID:         $Id$
wenzelm@5829
     3
    Author:     Markus Wenzel, TU Muenchen
wenzelm@5829
     4
wenzelm@5829
     5
The global Isabelle/Isar outer syntax.
wenzelm@5829
     6
*)
wenzelm@5829
     7
wenzelm@5829
     8
signature BASIC_OUTER_SYNTAX =
wenzelm@5829
     9
sig
wenzelm@5829
    10
  val main: unit -> unit
wenzelm@5883
    11
  val loop: unit -> unit
wenzelm@6860
    12
  val sync_main: unit -> unit
wenzelm@6860
    13
  val sync_loop: unit -> unit
wenzelm@5829
    14
end;
wenzelm@5829
    15
wenzelm@5829
    16
signature OUTER_SYNTAX =
wenzelm@5829
    17
sig
wenzelm@5829
    18
  include BASIC_OUTER_SYNTAX
wenzelm@6722
    19
  structure Keyword:
wenzelm@6722
    20
    sig
wenzelm@6722
    21
      val control: string
wenzelm@6722
    22
      val diag: string
wenzelm@6722
    23
      val thy_begin: string
wenzelm@7104
    24
      val thy_switch: string
wenzelm@6722
    25
      val thy_end: string
wenzelm@6722
    26
      val thy_heading: string
wenzelm@6722
    27
      val thy_decl: string
wenzelm@9588
    28
      val thy_script: string
wenzelm@6722
    29
      val thy_goal: string
wenzelm@6722
    30
      val qed: string
wenzelm@6733
    31
      val qed_block: string
wenzelm@8209
    32
      val qed_global: string
wenzelm@9552
    33
      val prf_heading: string
wenzelm@6722
    34
      val prf_goal: string
wenzelm@6722
    35
      val prf_block: string
wenzelm@9056
    36
      val prf_open: string
wenzelm@9056
    37
      val prf_close: string
wenzelm@6722
    38
      val prf_chain: string
wenzelm@6722
    39
      val prf_decl: string
wenzelm@6868
    40
      val prf_asm: string
wenzelm@7676
    41
      val prf_asm_goal: string
wenzelm@6722
    42
      val prf_script: string
wenzelm@6722
    43
      val kinds: string list
wenzelm@6722
    44
    end
wenzelm@5829
    45
  type token
wenzelm@5829
    46
  type parser
wenzelm@6722
    47
  val command: string -> string -> string ->
wenzelm@6373
    48
    (token list -> (Toplevel.transition -> Toplevel.transition) * token list) -> parser
wenzelm@9132
    49
  val markup_command: IsarOutput.markup -> string -> string -> string ->
wenzelm@7789
    50
    (token list -> (Toplevel.transition -> Toplevel.transition) * token list) -> parser
wenzelm@6722
    51
  val improper_command: string -> string -> string ->
wenzelm@6373
    52
    (token list -> (Toplevel.transition -> Toplevel.transition) * token list) -> parser
wenzelm@14687
    53
  val is_keyword: string -> bool
wenzelm@7026
    54
  val dest_keywords: unit -> string list
wenzelm@7026
    55
  val dest_parsers: unit -> (string * string * string * bool) list
wenzelm@5883
    56
  val print_outer_syntax: unit -> unit
wenzelm@9223
    57
  val print_commands: Toplevel.transition -> Toplevel.transition
wenzelm@5829
    58
  val add_keywords: string list -> unit
wenzelm@5829
    59
  val add_parsers: parser list -> unit
wenzelm@12943
    60
  val check_text: string * Position.T -> bool -> Toplevel.state -> unit
wenzelm@7940
    61
  val deps_thy: string -> bool -> Path.T -> string list * Path.T list
wenzelm@6199
    62
  val load_thy: string -> bool -> bool -> Path.T -> unit
nipkow@14091
    63
  val isar: bool -> bool -> unit Toplevel.isar
aspinall@15144
    64
  val scan: string -> OuterLex.token list
aspinall@15144
    65
  val read: OuterLex.token list -> 
aspinall@15144
    66
		(string * OuterLex.token list * Toplevel.transition) list
wenzelm@5829
    67
end;
wenzelm@5829
    68
wenzelm@5829
    69
structure OuterSyntax: OUTER_SYNTAX =
wenzelm@5829
    70
struct
wenzelm@5829
    71
wenzelm@7750
    72
structure T = OuterLex;
wenzelm@6860
    73
structure P = OuterParse;
wenzelm@6860
    74
wenzelm@5829
    75
wenzelm@5829
    76
(** outer syntax **)
wenzelm@5829
    77
wenzelm@6722
    78
(* command keyword classification *)
wenzelm@6722
    79
wenzelm@6722
    80
structure Keyword =
wenzelm@6722
    81
struct
wenzelm@6722
    82
  val control = "control";
wenzelm@6722
    83
  val diag = "diag";
wenzelm@6722
    84
  val thy_begin = "theory-begin";
wenzelm@7104
    85
  val thy_switch = "theory-switch";
wenzelm@6722
    86
  val thy_end = "theory-end";
wenzelm@6722
    87
  val thy_heading = "theory-heading";
wenzelm@6722
    88
  val thy_decl = "theory-decl";
wenzelm@9588
    89
  val thy_script = "theory-script";
wenzelm@6722
    90
  val thy_goal = "theory-goal";
wenzelm@6722
    91
  val qed = "qed";
wenzelm@6733
    92
  val qed_block = "qed-block";
wenzelm@8209
    93
  val qed_global = "qed-global";
wenzelm@9552
    94
  val prf_heading = "proof-heading";
wenzelm@6722
    95
  val prf_goal = "proof-goal";
wenzelm@6722
    96
  val prf_block = "proof-block";
wenzelm@9056
    97
  val prf_open = "proof-open";
wenzelm@9056
    98
  val prf_close = "proof-close";
wenzelm@6722
    99
  val prf_chain = "proof-chain";
wenzelm@6722
   100
  val prf_decl = "proof-decl";
wenzelm@6868
   101
  val prf_asm = "proof-asm";
wenzelm@7676
   102
  val prf_asm_goal = "proof-asm-goal";
wenzelm@6722
   103
  val prf_script = "proof-script";
wenzelm@6722
   104
wenzelm@9588
   105
  val kinds = [control, diag, thy_begin, thy_switch, thy_end, thy_heading, thy_decl, thy_script,
wenzelm@9588
   106
    thy_goal, qed, qed_block, qed_global, prf_heading, prf_goal, prf_block, prf_open, prf_close,
wenzelm@9588
   107
    prf_chain, prf_decl, prf_asm, prf_asm_goal, prf_script];
wenzelm@6722
   108
end;
wenzelm@6722
   109
wenzelm@6722
   110
wenzelm@5829
   111
(* parsers *)
wenzelm@5829
   112
wenzelm@7750
   113
type token = T.token;
wenzelm@5829
   114
type parser_fn = token list -> (Toplevel.transition -> Toplevel.transition) * token list;
wenzelm@5829
   115
wenzelm@5829
   116
datatype parser =
wenzelm@9132
   117
  Parser of string * (string * string * IsarOutput.markup option) * bool * parser_fn;
wenzelm@5829
   118
wenzelm@7750
   119
fun parser int_only markup name comment kind parse =
wenzelm@7750
   120
  Parser (name, (comment, kind, markup), int_only, parse);
wenzelm@5829
   121
wenzelm@5829
   122
wenzelm@5829
   123
(* parse command *)
wenzelm@5829
   124
wenzelm@6860
   125
local
wenzelm@6199
   126
wenzelm@14925
   127
fun terminate false = Scan.succeed ()
wenzelm@14925
   128
  | terminate true = P.group "end of input" (Scan.option P.sync -- P.semicolon >> K ());
wenzelm@14925
   129
wenzelm@14925
   130
fun trace false parse = parse
wenzelm@14925
   131
  | trace true parse = Scan.trace parse >> (fn (f, toks) => f o Toplevel.source toks);
wenzelm@14925
   132
wenzelm@14925
   133
fun body cmd trc (name, _) =
wenzelm@7026
   134
  (case cmd name of
wenzelm@14925
   135
    Some (int_only, parse) =>
wenzelm@14925
   136
      P.!!! (Scan.prompt (name ^ "# ") (trace trc parse >> pair int_only))
wenzelm@7026
   137
  | None => sys_error ("no parser for outer syntax command " ^ quote name));
wenzelm@6860
   138
wenzelm@6860
   139
in
wenzelm@5829
   140
wenzelm@14925
   141
fun command do_terminate do_trace cmd =
wenzelm@9132
   142
  P.semicolon >> K None ||
wenzelm@6860
   143
  P.sync >> K None ||
wenzelm@14925
   144
  (P.position P.command :-- body cmd do_trace) --| terminate do_terminate
wenzelm@6860
   145
    >> (fn ((name, pos), (int_only, f)) =>
wenzelm@6860
   146
      Some (Toplevel.empty |> Toplevel.name name |> Toplevel.position pos |>
wenzelm@6860
   147
        Toplevel.interactive int_only |> f));
wenzelm@5829
   148
wenzelm@6199
   149
end;
wenzelm@6199
   150
wenzelm@5829
   151
wenzelm@5829
   152
wenzelm@9132
   153
(** global outer syntax **)
wenzelm@5829
   154
wenzelm@7026
   155
local
wenzelm@7026
   156
wenzelm@7026
   157
val global_lexicons = ref (Scan.empty_lexicon, Scan.empty_lexicon);
wenzelm@7750
   158
val global_parsers =
wenzelm@9132
   159
  ref (Symtab.empty: (((string * string) * (bool * parser_fn)) * IsarOutput.markup option)
wenzelm@9132
   160
    Symtab.table);
wenzelm@9132
   161
val global_markups = ref ([]: (string * IsarOutput.markup) list);
wenzelm@5952
   162
wenzelm@7026
   163
fun change_lexicons f =
wenzelm@7026
   164
  let val lexs = f (! global_lexicons) in
wenzelm@7026
   165
    (case (op inter_string) (pairself Scan.dest_lexicon lexs) of
wenzelm@7026
   166
      [] => global_lexicons := lexs
wenzelm@7026
   167
    | bads => error ("Clash of outer syntax commands and keywords: " ^ commas_quote bads))
wenzelm@7026
   168
  end;
wenzelm@5829
   169
wenzelm@7789
   170
fun get_markup (ms, (name, (_, Some m))) = (name, m) :: ms
wenzelm@7789
   171
  | get_markup (ms, _) = ms;
wenzelm@7750
   172
wenzelm@7750
   173
fun make_markups () = global_markups := Symtab.foldl get_markup ([], ! global_parsers);
wenzelm@9132
   174
fun change_parsers f = (Library.change global_parsers f; make_markups ());
wenzelm@6722
   175
wenzelm@7026
   176
in
wenzelm@7026
   177
wenzelm@7750
   178
wenzelm@9132
   179
(* access current syntax *)
wenzelm@7026
   180
wenzelm@7026
   181
(*Note: the syntax for files is statically determined at the very
wenzelm@7026
   182
  beginning; for interactive processing it may change dynamically.*)
wenzelm@7026
   183
wenzelm@7026
   184
fun get_lexicons () = ! global_lexicons;
wenzelm@7026
   185
fun get_parsers () = ! global_parsers;
wenzelm@7750
   186
fun get_parser () = apsome (#2 o #1) o curry Symtab.lookup (! global_parsers);
wenzelm@7789
   187
wenzelm@9132
   188
fun is_markup kind name =
wenzelm@9132
   189
  (case assoc (! global_markups, name) of Some k => k = kind | None => false);
wenzelm@9132
   190
fun markup kind = Scan.one (T.is_kind T.Command andf is_markup kind o T.val_of);
wenzelm@5829
   191
wenzelm@5829
   192
wenzelm@5829
   193
(* augment syntax *)
wenzelm@5829
   194
wenzelm@7026
   195
fun add_keywords keywords = change_lexicons (apfst (fn lex =>
wenzelm@7026
   196
  (Scan.extend_lexicon lex (map Symbol.explode keywords))));
wenzelm@5829
   197
wenzelm@7750
   198
fun add_parser (tab, Parser (name, (comment, kind, markup), int_only, parse)) =
wenzelm@5829
   199
 (if is_none (Symtab.lookup (tab, name)) then ()
wenzelm@5829
   200
  else warning ("Redefined outer syntax command " ^ quote name);
wenzelm@7750
   201
  Symtab.update ((name, (((comment, kind), (int_only, parse)), markup)), tab));
wenzelm@5829
   202
wenzelm@5829
   203
fun add_parsers parsers =
wenzelm@7026
   204
  (change_parsers (fn tab => foldl add_parser (tab, parsers));
wenzelm@7026
   205
    change_lexicons (apsnd (fn lex => Scan.extend_lexicon lex
wenzelm@7026
   206
      (map (fn Parser (name, _, _, _) => Symbol.explode name) parsers))));
wenzelm@7026
   207
wenzelm@7026
   208
end;
wenzelm@5829
   209
wenzelm@5829
   210
wenzelm@7026
   211
(* print syntax *)
wenzelm@7026
   212
wenzelm@14687
   213
fun is_keyword s = Scan.is_literal (#1 (get_lexicons ())) (Symbol.explode s);
wenzelm@7026
   214
fun dest_keywords () = Scan.dest_lexicon (#1 (get_lexicons ()));
wenzelm@7026
   215
wenzelm@7026
   216
fun dest_parsers () =
wenzelm@7750
   217
  map (fn (name, (((cmt, kind), (int_only, _)), _)) => (name, cmt, kind, int_only))
wenzelm@7026
   218
    (Symtab.dest (get_parsers ()));
wenzelm@5829
   219
wenzelm@9223
   220
fun print_outer_syntax () =
wenzelm@7026
   221
  let
wenzelm@7026
   222
    fun pretty_cmd (name, comment, _, _) =
wenzelm@7026
   223
      Pretty.block [Pretty.str (name ^ ":"), Pretty.brk 2, Pretty.str comment];
wenzelm@7026
   224
    val (int_cmds, cmds) = partition #4 (dest_parsers ());
wenzelm@7026
   225
  in
wenzelm@8720
   226
    [Pretty.strs ("syntax keywords:" :: map quote (dest_keywords ())),
wenzelm@8720
   227
      Pretty.big_list "proper commands:" (map pretty_cmd cmds),
wenzelm@8720
   228
      Pretty.big_list "improper commands (interactive-only):" (map pretty_cmd int_cmds)]
wenzelm@9223
   229
    |> Pretty.chunks |> Pretty.writeln
wenzelm@7026
   230
  end;
wenzelm@5829
   231
wenzelm@9223
   232
val print_commands = Toplevel.imperative print_outer_syntax;
wenzelm@7367
   233
wenzelm@5829
   234
wenzelm@5829
   235
wenzelm@9132
   236
(** toplevel parsing **)
wenzelm@5829
   237
wenzelm@9132
   238
(* basic sources *)
wenzelm@6860
   239
wenzelm@14925
   240
fun toplevel_source term trc do_recover cmd src =
wenzelm@9132
   241
  let
wenzelm@9132
   242
    val no_terminator =
wenzelm@9132
   243
      Scan.unless P.semicolon (Scan.one (T.not_sync andf T.not_eof));
wenzelm@12876
   244
    fun recover x = (Scan.prompt "recover# " (Scan.repeat no_terminator) >> K [None]) x;
wenzelm@9132
   245
  in
wenzelm@9132
   246
    src
wenzelm@12876
   247
    |> T.source_proper
wenzelm@9132
   248
    |> Source.source T.stopper
wenzelm@12876
   249
      (Scan.bulk (P.$$$ "--" -- P.!!! P.text >> K None || P.not_eof >> Some))
wenzelm@12876
   250
      (if do_recover then Some recover else None)
wenzelm@12876
   251
    |> Source.mapfilter I
wenzelm@14925
   252
    |> Source.source T.stopper (Scan.bulk (fn xs => P.!!! (command term trc (cmd ())) xs))
wenzelm@9132
   253
      (if do_recover then Some recover else None)
wenzelm@9132
   254
    |> Source.mapfilter I
wenzelm@9132
   255
  end;
wenzelm@5829
   256
wenzelm@7746
   257
wenzelm@9132
   258
(* interactive source of toplevel transformers *)
wenzelm@5829
   259
wenzelm@9132
   260
fun isar term no_pos =
wenzelm@9132
   261
  Source.tty
wenzelm@9132
   262
  |> Symbol.source true
wenzelm@9132
   263
  |> T.source true get_lexicons
wenzelm@9132
   264
    (if no_pos then Position.none else Position.line_name 1 "stdin")
wenzelm@14925
   265
  |> toplevel_source term false true get_parser;
wenzelm@6199
   266
wenzelm@6199
   267
aspinall@15144
   268
(* scan text,  read tokens with trace (for Proof General) *)
wenzelm@14925
   269
aspinall@15144
   270
fun scan str =
aspinall@15144
   271
 Source.of_string str
aspinall@15144
   272
 |> Symbol.source false
aspinall@15144
   273
 |> T.source false get_lexicons Position.none
aspinall@15144
   274
 |> Source.exhaust
wenzelm@14925
   275
aspinall@15144
   276
fun read toks =
aspinall@15144
   277
  Source.of_list toks
wenzelm@14925
   278
  |> toplevel_source false true true get_parser
wenzelm@14925
   279
  |> Source.exhaust
wenzelm@14925
   280
  |> map (fn tr => (Toplevel.name_of tr, the (Toplevel.source_of tr), tr));
nipkow@14091
   281
nipkow@14091
   282
wenzelm@9132
   283
(** read theory **)
wenzelm@6247
   284
wenzelm@12943
   285
(* check_text *)
wenzelm@12943
   286
wenzelm@12943
   287
fun check_text s true state = (IsarOutput.eval_antiquote (#1 (get_lexicons ())) state s; ())
wenzelm@12943
   288
  | check_text _ false _ = ();
wenzelm@12943
   289
wenzelm@12943
   290
wenzelm@9132
   291
(* deps_thy *)
wenzelm@6199
   292
wenzelm@7940
   293
fun deps_thy name ml path =
wenzelm@6199
   294
  let
wenzelm@7735
   295
    val src = Source.of_string (File.read path);
wenzelm@7735
   296
    val pos = Path.position path;
wenzelm@9132
   297
    val (name', parents, files) = ThyHeader.scan (src, pos);
wenzelm@6199
   298
    val ml_path = ThyLoad.ml_path name;
wenzelm@7940
   299
    val ml_file = if ml andalso is_some (ThyLoad.check_file ml_path) then [ml_path] else [];
wenzelm@6199
   300
  in
wenzelm@6199
   301
    if name <> name' then
wenzelm@7940
   302
      error ("Filename " ^ quote (Path.pack path) ^
wenzelm@8078
   303
        " does not agree with theory name " ^ quote name')
wenzelm@6247
   304
    else (parents, map (Path.unpack o #1) files @ ml_file)
wenzelm@6199
   305
  end;
wenzelm@6199
   306
wenzelm@7746
   307
wenzelm@9132
   308
(* load_thy *)
wenzelm@6199
   309
wenzelm@7746
   310
local
wenzelm@7746
   311
wenzelm@7940
   312
fun try_ml_file name time =
wenzelm@6199
   313
  let
wenzelm@6199
   314
    val path = ThyLoad.ml_path name;
wenzelm@7940
   315
    val tr = Toplevel.imperative (fn () => ThyInfo.load_file time path);
wenzelm@6247
   316
    val tr_name = if time then "time_use" else "use";
wenzelm@6199
   317
  in
wenzelm@7243
   318
    if is_none (ThyLoad.check_file path) then ()
wenzelm@9132
   319
    else Toplevel.excursion [Toplevel.empty |> Toplevel.name tr_name |> tr]
wenzelm@6199
   320
  end;
wenzelm@6199
   321
wenzelm@9132
   322
fun parse_thy src =
wenzelm@9132
   323
  src
wenzelm@14925
   324
  |> toplevel_source false false false (K (get_parser ()))
wenzelm@7746
   325
  |> Source.exhaust;
wenzelm@5829
   326
wenzelm@6247
   327
fun run_thy name path =
wenzelm@7683
   328
  let
wenzelm@7735
   329
    val pos = Path.position path;
wenzelm@9132
   330
    val text = Library.untabify (explode (File.read path));
wenzelm@9132
   331
    val text_src = Source.of_list text;
wenzelm@9132
   332
    fun present_text () = Source.exhaust (Symbol.source false text_src);
wenzelm@7683
   333
  in
wenzelm@7735
   334
    Present.init_theory name;
wenzelm@9132
   335
    Present.verbatim_source name present_text;
wenzelm@9132
   336
    if ThyHeader.is_old (text_src, pos) then (ThySyn.load_thy name text;
wenzelm@9132
   337
      Present.old_symbol_source name present_text)   (*note: text presented twice*)
wenzelm@9132
   338
    else
wenzelm@9132
   339
      let
wenzelm@12876
   340
        val tok_src = text_src
wenzelm@12876
   341
          |> Symbol.source false
wenzelm@12876
   342
          |> T.source false (K (get_lexicons ())) pos
wenzelm@12876
   343
          |> Source.exhausted;
wenzelm@9132
   344
        val out = Toplevel.excursion_result
wenzelm@9132
   345
          (IsarOutput.parse_thy markup (#1 (get_lexicons ()))
wenzelm@9132
   346
            (parse_thy tok_src) tok_src);
wenzelm@9132
   347
      in Present.theory_output name (Buffer.content out) end
wenzelm@6247
   348
  end;
wenzelm@6199
   349
wenzelm@7746
   350
in
wenzelm@7746
   351
wenzelm@6199
   352
fun load_thy name ml time path =
wenzelm@6247
   353
 (if time then
wenzelm@6247
   354
    timeit (fn () =>
wenzelm@6247
   355
     (writeln ("\n**** Starting theory " ^ quote name ^ " ****");
wenzelm@9036
   356
      run_thy name path;
wenzelm@6247
   357
      writeln ("**** Finished theory " ^ quote name ^ " ****\n")))
wenzelm@6247
   358
  else run_thy name path;
wenzelm@6247
   359
  Context.context (ThyInfo.get_theory name);
wenzelm@7940
   360
  if ml then try_ml_file name time else ());
wenzelm@5829
   361
wenzelm@7746
   362
end;
wenzelm@7746
   363
wenzelm@5829
   364
wenzelm@5829
   365
wenzelm@5829
   366
(** the read-eval-print loop **)
wenzelm@5829
   367
wenzelm@5923
   368
(* main loop *)
wenzelm@5923
   369
wenzelm@7333
   370
fun gen_loop term no_pos =
wenzelm@7333
   371
 (Context.reset_context ();
wenzelm@7333
   372
  Toplevel.loop (isar term no_pos));
wenzelm@5829
   373
wenzelm@7333
   374
fun gen_main term no_pos =
wenzelm@5829
   375
 (Toplevel.set_state Toplevel.toplevel;
wenzelm@6199
   376
  writeln (Session.welcome ());
wenzelm@7333
   377
  gen_loop term no_pos);
wenzelm@6860
   378
wenzelm@7333
   379
fun main () = gen_main false false;
wenzelm@7333
   380
fun loop () = gen_loop false false;
wenzelm@7333
   381
fun sync_main () = gen_main true true;
wenzelm@7333
   382
fun sync_loop () = gen_loop true true;
wenzelm@5829
   383
wenzelm@5829
   384
wenzelm@6373
   385
(*final declarations of this structure!*)
wenzelm@7789
   386
val command = parser false None;
wenzelm@9132
   387
val markup_command = parser false o Some;
wenzelm@7789
   388
val improper_command = parser true None;
wenzelm@6685
   389
wenzelm@6373
   390
wenzelm@5829
   391
end;
wenzelm@5829
   392
wenzelm@6199
   393
(*setup theory syntax dependent operations*)
wenzelm@6199
   394
ThyLoad.deps_thy_fn := OuterSyntax.deps_thy;
wenzelm@6199
   395
ThyLoad.load_thy_fn := OuterSyntax.load_thy;
wenzelm@6199
   396
structure ThyLoad: THY_LOAD = ThyLoad;
wenzelm@6199
   397
wenzelm@5829
   398
structure BasicOuterSyntax: BASIC_OUTER_SYNTAX = OuterSyntax;
wenzelm@5829
   399
open BasicOuterSyntax;