src/Pure/Isar/token.ML
author wenzelm
Fri Apr 08 16:34:14 2011 +0200 (2011-04-08)
changeset 42290 b1f544c84040
parent 40958 755f8fe7ced9
child 42503 27514b6fbe93
permissions -rw-r--r--
discontinued special treatment of structure Lexicon;
wenzelm@36959
     1
(*  Title:      Pure/Isar/token.ML
wenzelm@5825
     2
    Author:     Markus Wenzel, TU Muenchen
wenzelm@5825
     3
wenzelm@36959
     4
Outer token syntax for Isabelle/Isar.
wenzelm@5825
     5
*)
wenzelm@5825
     6
wenzelm@36959
     7
signature TOKEN =
wenzelm@5825
     8
sig
wenzelm@36959
     9
  datatype kind =
wenzelm@27814
    10
    Command | Keyword | Ident | LongIdent | SymIdent | Var | TypeIdent | TypeVar |
wenzelm@40290
    11
    Nat | Float | String | AltString | Verbatim | Space | Comment | InternalValue |
wenzelm@40958
    12
    Error of string | Sync | EOF
wenzelm@27814
    13
  datatype value =
wenzelm@27814
    14
    Text of string | Typ of typ | Term of term | Fact of thm list |
wenzelm@27814
    15
    Attribute of morphism -> attribute
wenzelm@36959
    16
  type T
wenzelm@36959
    17
  val str_of_kind: kind -> string
wenzelm@36959
    18
  val position_of: T -> Position.T
wenzelm@36959
    19
  val end_position_of: T -> Position.T
wenzelm@36959
    20
  val pos_of: T -> string
wenzelm@36959
    21
  val eof: T
wenzelm@36959
    22
  val is_eof: T -> bool
wenzelm@36959
    23
  val not_eof: T -> bool
wenzelm@36959
    24
  val not_sync: T -> bool
wenzelm@36959
    25
  val stopper: T Scan.stopper
wenzelm@36959
    26
  val kind_of: T -> kind
wenzelm@36959
    27
  val is_kind: kind -> T -> bool
wenzelm@36959
    28
  val keyword_with: (string -> bool) -> T -> bool
wenzelm@36959
    29
  val ident_with: (string -> bool) -> T -> bool
wenzelm@36959
    30
  val is_proper: T -> bool
wenzelm@36959
    31
  val is_semicolon: T -> bool
wenzelm@36959
    32
  val is_comment: T -> bool
wenzelm@36959
    33
  val is_begin_ignore: T -> bool
wenzelm@36959
    34
  val is_end_ignore: T -> bool
wenzelm@36959
    35
  val is_blank: T -> bool
wenzelm@36959
    36
  val is_newline: T -> bool
wenzelm@36959
    37
  val source_of: T -> string
wenzelm@36959
    38
  val source_position_of: T -> Symbol_Pos.text * Position.T
wenzelm@36959
    39
  val content_of: T -> string
wenzelm@36959
    40
  val unparse: T -> string
wenzelm@36959
    41
  val text_of: T -> string * string
wenzelm@36959
    42
  val get_value: T -> value option
wenzelm@36959
    43
  val map_value: (value -> value) -> T -> T
wenzelm@36959
    44
  val mk_text: string -> T
wenzelm@36959
    45
  val mk_typ: typ -> T
wenzelm@36959
    46
  val mk_term: term -> T
wenzelm@36959
    47
  val mk_fact: thm list -> T
wenzelm@36959
    48
  val mk_attribute: (morphism -> attribute) -> T
wenzelm@36959
    49
  val assignable: T -> T
wenzelm@36959
    50
  val assign: value option -> T -> unit
wenzelm@36959
    51
  val closure: T -> T
wenzelm@27814
    52
  val ident_or_symbolic: string -> bool
wenzelm@30573
    53
  val !!! : string -> (Symbol_Pos.T list -> 'a) -> Symbol_Pos.T list -> 'a
wenzelm@36959
    54
  val source_proper: (T, 'a) Source.source -> (T, (T, 'a) Source.source) Source.source
wenzelm@27835
    55
  val source': {do_recover: bool Option.option} -> (unit -> Scan.lexicon * Scan.lexicon) ->
wenzelm@36959
    56
    (Symbol_Pos.T, 'a) Source.source -> (T, (Symbol_Pos.T, 'a) Source.source) Source.source
wenzelm@27835
    57
  val source: {do_recover: bool Option.option} -> (unit -> Scan.lexicon * Scan.lexicon) ->
wenzelm@36959
    58
    Position.T -> (Symbol.symbol, 'a) Source.source -> (T,
wenzelm@30573
    59
      (Symbol_Pos.T, Position.T * (Symbol.symbol, 'a) Source.source) Source.source) Source.source
wenzelm@36959
    60
  val read_antiq: Scan.lexicon -> (T list -> 'a * T list) -> Symbol_Pos.T list * Position.T -> 'a
wenzelm@5825
    61
end;
wenzelm@5825
    62
wenzelm@36959
    63
structure Token: TOKEN =
wenzelm@5825
    64
struct
wenzelm@5825
    65
wenzelm@5825
    66
(** tokens **)
wenzelm@5825
    67
wenzelm@27814
    68
(* token values *)
wenzelm@27814
    69
wenzelm@27814
    70
(*The value slot assigns an (optional) internal value to a token,
wenzelm@27814
    71
  usually as a side-effect of special scanner setup (see also
wenzelm@27814
    72
  args.ML).  Note that an assignable ref designates an intermediate
wenzelm@27814
    73
  state of internalization -- it is NOT meant to persist.*)
wenzelm@27814
    74
wenzelm@27814
    75
datatype value =
wenzelm@27814
    76
  Text of string |
wenzelm@27814
    77
  Typ of typ |
wenzelm@27814
    78
  Term of term |
wenzelm@27814
    79
  Fact of thm list |
wenzelm@27814
    80
  Attribute of morphism -> attribute;
wenzelm@27814
    81
wenzelm@27814
    82
datatype slot =
wenzelm@27814
    83
  Slot |
wenzelm@27814
    84
  Value of value option |
wenzelm@32738
    85
  Assignable of value option Unsynchronized.ref;
wenzelm@27814
    86
wenzelm@27814
    87
wenzelm@5825
    88
(* datatype token *)
wenzelm@5825
    89
wenzelm@36959
    90
datatype kind =
wenzelm@27814
    91
  Command | Keyword | Ident | LongIdent | SymIdent | Var | TypeIdent | TypeVar |
wenzelm@40290
    92
  Nat | Float | String | AltString | Verbatim | Space | Comment | InternalValue |
wenzelm@40958
    93
  Error of string | Sync | EOF;
wenzelm@5825
    94
wenzelm@36959
    95
datatype T = Token of (Symbol_Pos.text * Position.range) * (kind * string) * slot;
wenzelm@5825
    96
wenzelm@5825
    97
val str_of_kind =
wenzelm@7026
    98
 fn Command => "command"
wenzelm@7026
    99
  | Keyword => "keyword"
wenzelm@5825
   100
  | Ident => "identifier"
wenzelm@5825
   101
  | LongIdent => "long identifier"
wenzelm@5825
   102
  | SymIdent => "symbolic identifier"
wenzelm@5825
   103
  | Var => "schematic variable"
wenzelm@5825
   104
  | TypeIdent => "type variable"
wenzelm@5825
   105
  | TypeVar => "schematic type variable"
wenzelm@40290
   106
  | Nat => "natural number"
wenzelm@40290
   107
  | Float => "floating-point number"
wenzelm@5825
   108
  | String => "string"
wenzelm@17164
   109
  | AltString => "back-quoted string"
wenzelm@5825
   110
  | Verbatim => "verbatim text"
wenzelm@7682
   111
  | Space => "white space"
wenzelm@7682
   112
  | Comment => "comment text"
wenzelm@27814
   113
  | InternalValue => "internal value"
wenzelm@23729
   114
  | Error _ => "bad input"
wenzelm@23788
   115
  | Sync => "sync marker"
wenzelm@5825
   116
  | EOF => "end-of-file";
wenzelm@5825
   117
wenzelm@5825
   118
wenzelm@27733
   119
(* position *)
wenzelm@5825
   120
wenzelm@27814
   121
fun position_of (Token ((_, (pos, _)), _, _)) = pos;
wenzelm@27814
   122
fun end_position_of (Token ((_, (_, pos)), _, _)) = pos;
wenzelm@27663
   123
wenzelm@5825
   124
val pos_of = Position.str_of o position_of;
wenzelm@5825
   125
wenzelm@5825
   126
wenzelm@27733
   127
(* control tokens *)
wenzelm@27733
   128
wenzelm@27814
   129
fun mk_eof pos = Token (("", (pos, Position.none)), (EOF, ""), Slot);
wenzelm@27733
   130
val eof = mk_eof Position.none;
wenzelm@27733
   131
wenzelm@27814
   132
fun is_eof (Token (_, (EOF, _), _)) = true
wenzelm@27733
   133
  | is_eof _ = false;
wenzelm@27733
   134
wenzelm@27733
   135
val not_eof = not o is_eof;
wenzelm@27733
   136
wenzelm@27814
   137
fun not_sync (Token (_, (Sync, _), _)) = false
wenzelm@27733
   138
  | not_sync _ = true;
wenzelm@27733
   139
wenzelm@27752
   140
val stopper =
wenzelm@27752
   141
  Scan.stopper (fn [] => eof | toks => mk_eof (end_position_of (List.last toks))) is_eof;
wenzelm@27733
   142
wenzelm@27733
   143
wenzelm@5825
   144
(* kind of token *)
wenzelm@5825
   145
wenzelm@27814
   146
fun kind_of (Token (_, (k, _), _)) = k;
wenzelm@27814
   147
fun is_kind k (Token (_, (k', _), _)) = k = k';
wenzelm@5825
   148
wenzelm@27814
   149
fun keyword_with pred (Token (_, (Keyword, x), _)) = pred x
wenzelm@7026
   150
  | keyword_with _ _ = false;
wenzelm@5825
   151
wenzelm@27814
   152
fun ident_with pred (Token (_, (Ident, x), _)) = pred x
wenzelm@16029
   153
  | ident_with _ _ = false;
wenzelm@16029
   154
wenzelm@27814
   155
fun is_proper (Token (_, (Space, _), _)) = false
wenzelm@27814
   156
  | is_proper (Token (_, (Comment, _), _)) = false
wenzelm@5825
   157
  | is_proper _ = true;
wenzelm@5825
   158
wenzelm@27814
   159
fun is_semicolon (Token (_, (Keyword, ";"), _)) = true
wenzelm@9130
   160
  | is_semicolon _ = false;
wenzelm@9130
   161
wenzelm@27814
   162
fun is_comment (Token (_, (Comment, _), _)) = true
wenzelm@17069
   163
  | is_comment _ = false;
wenzelm@17069
   164
wenzelm@27814
   165
fun is_begin_ignore (Token (_, (Comment, "<"), _)) = true
wenzelm@8580
   166
  | is_begin_ignore _ = false;
wenzelm@8580
   167
wenzelm@27814
   168
fun is_end_ignore (Token (_, (Comment, ">"), _)) = true
wenzelm@8580
   169
  | is_end_ignore _ = false;
wenzelm@8580
   170
wenzelm@8651
   171
wenzelm@17069
   172
(* blanks and newlines -- space tokens obey lines *)
wenzelm@8651
   173
wenzelm@27814
   174
fun is_blank (Token (_, (Space, x), _)) = not (String.isSuffix "\n" x)
wenzelm@17069
   175
  | is_blank _ = false;
wenzelm@17069
   176
wenzelm@27814
   177
fun is_newline (Token (_, (Space, x), _)) = String.isSuffix "\n" x
wenzelm@8651
   178
  | is_newline _ = false;
wenzelm@8651
   179
wenzelm@5825
   180
wenzelm@14991
   181
(* token content *)
wenzelm@9155
   182
wenzelm@27814
   183
fun source_of (Token ((source, (pos, _)), _, _)) =
wenzelm@38229
   184
  YXML.string_of (XML.Elem (Markup.token (Position.properties_of pos), [XML.Text source]));
wenzelm@25642
   185
wenzelm@27885
   186
fun source_position_of (Token ((source, (pos, _)), _, _)) = (source, pos);
wenzelm@27873
   187
wenzelm@27814
   188
fun content_of (Token (_, (_, x), _)) = x;
wenzelm@27747
   189
wenzelm@27747
   190
wenzelm@27747
   191
(* unparse *)
wenzelm@27747
   192
wenzelm@18547
   193
fun escape q =
wenzelm@18547
   194
  implode o map (fn s => if s = q orelse s = "\\" then "\\" ^ s else s) o Symbol.explode;
wenzelm@18547
   195
wenzelm@27814
   196
fun unparse (Token (_, (kind, x), _)) =
wenzelm@14991
   197
  (case kind of
wenzelm@18547
   198
    String => x |> quote o escape "\""
wenzelm@18547
   199
  | AltString => x |> enclose "`" "`" o escape "`"
wenzelm@14991
   200
  | Verbatim => x |> enclose "{*" "*}"
wenzelm@14991
   201
  | Comment => x |> enclose "(*" "*)"
wenzelm@23729
   202
  | Sync => ""
wenzelm@23729
   203
  | EOF => ""
wenzelm@14991
   204
  | _ => x);
wenzelm@14991
   205
wenzelm@23788
   206
fun text_of tok =
wenzelm@23788
   207
  if is_semicolon tok then ("terminator", "")
wenzelm@23729
   208
  else
wenzelm@23788
   209
    let
wenzelm@23788
   210
      val k = str_of_kind (kind_of tok);
wenzelm@40523
   211
      val s = unparse tok;
wenzelm@23788
   212
    in
wenzelm@23788
   213
      if s = "" then (k, "")
wenzelm@23788
   214
      else if size s < 40 andalso not (exists_string (fn c => c = "\n") s) then (k ^ " " ^ s, "")
wenzelm@23788
   215
      else (k, s)
wenzelm@23788
   216
    end;
wenzelm@23729
   217
wenzelm@5825
   218
wenzelm@5825
   219
wenzelm@27814
   220
(** associated values **)
wenzelm@27814
   221
wenzelm@27814
   222
(* access values *)
wenzelm@27814
   223
wenzelm@27814
   224
fun get_value (Token (_, _, Value v)) = v
wenzelm@27814
   225
  | get_value _ = NONE;
wenzelm@27814
   226
wenzelm@27814
   227
fun map_value f (Token (x, y, Value (SOME v))) = Token (x, y, Value (SOME (f v)))
wenzelm@27814
   228
  | map_value _ tok = tok;
wenzelm@27814
   229
wenzelm@27814
   230
wenzelm@27814
   231
(* make values *)
wenzelm@27814
   232
wenzelm@27814
   233
fun mk_value k v = Token ((k, Position.no_range), (InternalValue, k), Value (SOME v));
wenzelm@27814
   234
wenzelm@27814
   235
val mk_text = mk_value "<text>" o Text;
wenzelm@27814
   236
val mk_typ = mk_value "<typ>" o Typ;
wenzelm@27814
   237
val mk_term = mk_value "<term>" o Term;
wenzelm@27814
   238
val mk_fact = mk_value "<fact>" o Fact;
wenzelm@27814
   239
val mk_attribute = mk_value "<attribute>" o Attribute;
wenzelm@27814
   240
wenzelm@27814
   241
wenzelm@27814
   242
(* static binding *)
wenzelm@27814
   243
wenzelm@27814
   244
(*1st stage: make empty slots assignable*)
wenzelm@32738
   245
fun assignable (Token (x, y, Slot)) = Token (x, y, Assignable (Unsynchronized.ref NONE))
wenzelm@27814
   246
  | assignable tok = tok;
wenzelm@27814
   247
wenzelm@27814
   248
(*2nd stage: assign values as side-effect of scanning*)
wenzelm@27814
   249
fun assign v (Token (_, _, Assignable r)) = r := v
wenzelm@27814
   250
  | assign _ _ = ();
wenzelm@27814
   251
wenzelm@27814
   252
(*3rd stage: static closure of final values*)
wenzelm@32738
   253
fun closure (Token (x, y, Assignable (Unsynchronized.ref v))) = Token (x, y, Value v)
wenzelm@27814
   254
  | closure tok = tok;
wenzelm@27814
   255
wenzelm@27814
   256
wenzelm@27814
   257
wenzelm@5825
   258
(** scanners **)
wenzelm@5825
   259
wenzelm@30573
   260
open Basic_Symbol_Pos;
wenzelm@5825
   261
wenzelm@30573
   262
fun !!! msg = Symbol_Pos.!!! ("Outer lexical error: " ^ msg);
wenzelm@5825
   263
wenzelm@5825
   264
wenzelm@5825
   265
(* scan symbolic idents *)
wenzelm@5825
   266
wenzelm@40627
   267
val is_sym_char = member (op =) (raw_explode "!#$%&*+-/<=>?@^_|~");
wenzelm@5825
   268
wenzelm@8231
   269
val scan_symid =
wenzelm@40525
   270
  Scan.many1 (is_sym_char o Symbol_Pos.symbol) ||
wenzelm@40525
   271
  Scan.one (Symbol.is_symbolic o Symbol_Pos.symbol) >> single;
wenzelm@5825
   272
wenzelm@8231
   273
fun is_symid str =
wenzelm@8231
   274
  (case try Symbol.explode str of
skalberg@15531
   275
    SOME [s] => Symbol.is_symbolic s orelse is_sym_char s
skalberg@15531
   276
  | SOME ss => forall is_sym_char ss
wenzelm@8231
   277
  | _ => false);
wenzelm@8231
   278
wenzelm@27814
   279
fun ident_or_symbolic "begin" = false
wenzelm@27814
   280
  | ident_or_symbolic ":" = true
wenzelm@27814
   281
  | ident_or_symbolic "::" = true
wenzelm@42290
   282
  | ident_or_symbolic s = Lexicon.is_identifier s orelse is_symid s;
wenzelm@5825
   283
wenzelm@5825
   284
wenzelm@5825
   285
(* scan verbatim text *)
wenzelm@5825
   286
wenzelm@5825
   287
val scan_verb =
wenzelm@27769
   288
  $$$ "*" --| Scan.ahead (~$$$ "}") ||
wenzelm@27769
   289
  Scan.one (fn (s, _) => s <> "*" andalso Symbol.is_regular s) >> single;
wenzelm@5825
   290
wenzelm@5825
   291
val scan_verbatim =
wenzelm@30573
   292
  (Symbol_Pos.scan_pos --| $$$ "{" --| $$$ "*") -- !!! "missing end of verbatim text"
wenzelm@30586
   293
    (Symbol_Pos.change_prompt
wenzelm@30586
   294
      ((Scan.repeat scan_verb >> flat) -- ($$$ "*" |-- $$$ "}" |-- Symbol_Pos.scan_pos)));
wenzelm@5825
   295
wenzelm@5825
   296
wenzelm@5825
   297
(* scan space *)
wenzelm@5825
   298
wenzelm@19305
   299
fun is_space s = Symbol.is_blank s andalso s <> "\n";
wenzelm@5825
   300
wenzelm@5825
   301
val scan_space =
wenzelm@40525
   302
  Scan.many1 (is_space o Symbol_Pos.symbol) @@@ Scan.optional ($$$ "\n") [] ||
wenzelm@40525
   303
  Scan.many (is_space o Symbol_Pos.symbol) @@@ $$$ "\n";
wenzelm@5825
   304
wenzelm@5825
   305
wenzelm@27780
   306
(* scan comment *)
wenzelm@5825
   307
wenzelm@5825
   308
val scan_comment =
wenzelm@30573
   309
  Symbol_Pos.scan_pos -- (Symbol_Pos.scan_comment_body !!! -- Symbol_Pos.scan_pos);
wenzelm@5825
   310
wenzelm@5825
   311
wenzelm@27663
   312
wenzelm@27769
   313
(** token sources **)
wenzelm@5825
   314
wenzelm@27769
   315
fun source_proper src = src |> Source.filter is_proper;
wenzelm@5825
   316
wenzelm@23678
   317
local
wenzelm@23678
   318
wenzelm@27769
   319
fun token_leq ((_, syms1), (_, syms2)) = length syms1 <= length syms2;
wenzelm@27780
   320
wenzelm@27799
   321
fun token k ss =
wenzelm@30573
   322
  Token ((Symbol_Pos.implode ss, Symbol_Pos.range ss), (k, Symbol_Pos.untabify_content ss), Slot);
wenzelm@27799
   323
wenzelm@27799
   324
fun token_range k (pos1, (ss, pos2)) =
wenzelm@30573
   325
  Token (Symbol_Pos.implode_range pos1 pos2 ss, (k, Symbol_Pos.untabify_content ss), Slot);
wenzelm@23678
   326
wenzelm@27769
   327
fun scan (lex1, lex2) = !!! "bad input"
wenzelm@30586
   328
  (Symbol_Pos.scan_string >> token_range String ||
wenzelm@30586
   329
    Symbol_Pos.scan_alt_string >> token_range AltString ||
wenzelm@27799
   330
    scan_verbatim >> token_range Verbatim ||
wenzelm@27799
   331
    scan_comment >> token_range Comment ||
wenzelm@27780
   332
    scan_space >> token Space ||
wenzelm@40525
   333
    Scan.one (Symbol.is_sync o Symbol_Pos.symbol) >> (token Sync o single) ||
wenzelm@27780
   334
    (Scan.max token_leq
wenzelm@27769
   335
      (Scan.max token_leq
wenzelm@27769
   336
        (Scan.literal lex2 >> pair Command)
wenzelm@27769
   337
        (Scan.literal lex1 >> pair Keyword))
wenzelm@42290
   338
      (Lexicon.scan_longid >> pair LongIdent ||
wenzelm@42290
   339
        Lexicon.scan_id >> pair Ident ||
wenzelm@42290
   340
        Lexicon.scan_var >> pair Var ||
wenzelm@42290
   341
        Lexicon.scan_tid >> pair TypeIdent ||
wenzelm@42290
   342
        Lexicon.scan_tvar >> pair TypeVar ||
wenzelm@42290
   343
        Lexicon.scan_float >> pair Float ||
wenzelm@42290
   344
        Lexicon.scan_nat >> pair Nat ||
wenzelm@27780
   345
        scan_symid >> pair SymIdent) >> uncurry token));
wenzelm@27769
   346
wenzelm@27769
   347
fun recover msg =
wenzelm@40525
   348
  Scan.many ((Symbol.is_regular andf (not o Symbol.is_blank)) o Symbol_Pos.symbol)
wenzelm@27780
   349
  >> (single o token (Error msg));
wenzelm@23678
   350
wenzelm@23678
   351
in
wenzelm@5825
   352
wenzelm@27835
   353
fun source' {do_recover} get_lex =
wenzelm@30573
   354
  Source.source Symbol_Pos.stopper (Scan.bulk (fn xs => scan (get_lex ()) xs))
wenzelm@27780
   355
    (Option.map (rpair recover) do_recover);
wenzelm@27780
   356
wenzelm@5825
   357
fun source do_recover get_lex pos src =
wenzelm@30573
   358
  Symbol_Pos.source pos src
wenzelm@27780
   359
  |> source' do_recover get_lex;
wenzelm@23678
   360
wenzelm@23678
   361
end;
wenzelm@5825
   362
wenzelm@30586
   363
wenzelm@30586
   364
(* read_antiq *)
wenzelm@30586
   365
wenzelm@30586
   366
fun read_antiq lex scan (syms, pos) =
wenzelm@30586
   367
  let
wenzelm@30586
   368
    fun err msg = cat_error msg ("Malformed antiquotation" ^ Position.str_of pos ^ ":\n" ^
wenzelm@30586
   369
      "@{" ^ Symbol_Pos.content syms ^ "}");
wenzelm@30586
   370
wenzelm@30586
   371
    val res =
wenzelm@30586
   372
      Source.of_list syms
wenzelm@30586
   373
      |> source' {do_recover = NONE} (K (lex, Scan.empty_lexicon))
wenzelm@30586
   374
      |> source_proper
wenzelm@30586
   375
      |> Source.source stopper (Scan.error (Scan.bulk scan)) NONE
wenzelm@30586
   376
      |> Source.exhaust;
wenzelm@30586
   377
  in (case res of [x] => x | _ => err "") handle ERROR msg => err msg end;
wenzelm@30586
   378
wenzelm@5825
   379
end;