src/Pure/General/symbol_pos.ML
author wenzelm
Sat Jan 18 19:15:12 2014 +0100 (2014-01-18)
changeset 55033 8e8243975860
parent 54732 b01bb3d09928
child 55035 68afbb5ce4ff
permissions -rw-r--r--
support for nested text cartouches;
clarified Symbol.is_symbolic: exclude \<open> and \<close>;
wenzelm@27763
     1
(*  Title:      Pure/General/symbol_pos.ML
wenzelm@27763
     2
    Author:     Makarius
wenzelm@27763
     3
wenzelm@27763
     4
Symbols with explicit position information.
wenzelm@27763
     5
*)
wenzelm@27763
     6
wenzelm@53016
     7
val legacy_isub_isup = Unsynchronized.ref false;
wenzelm@53016
     8
wenzelm@36957
     9
signature SYMBOL_POS =
wenzelm@27763
    10
sig
wenzelm@27763
    11
  type T = Symbol.symbol * Position.T
wenzelm@27763
    12
  val symbol: T -> Symbol.symbol
wenzelm@27763
    13
  val $$$ : Symbol.symbol -> T list -> T list * T list
wenzelm@27763
    14
  val ~$$$ : Symbol.symbol -> T list -> T list * T list
wenzelm@27797
    15
  val content: T list -> string
wenzelm@55033
    16
  val range: T list -> Position.range
wenzelm@27763
    17
  val is_eof: T -> bool
wenzelm@27763
    18
  val stopper: T Scan.stopper
wenzelm@43947
    19
  val !!! : Scan.message -> (T list -> 'a) -> T list -> 'a
wenzelm@30586
    20
  val change_prompt: ('a -> 'b) -> 'a -> 'b
wenzelm@27778
    21
  val scan_pos: T list -> Position.T * T list
wenzelm@48764
    22
  val scan_string_q: string -> T list -> (Position.T * (T list * Position.T)) * T list
wenzelm@48764
    23
  val scan_string_qq: string -> T list -> (Position.T * (T list * Position.T)) * T list
wenzelm@48764
    24
  val scan_string_bq: string -> T list -> (Position.T * (T list * Position.T)) * T list
wenzelm@48743
    25
  val recover_string_q: T list -> T list * T list
wenzelm@48743
    26
  val recover_string_qq: T list -> T list * T list
wenzelm@48743
    27
  val recover_string_bq: T list -> T list * T list
wenzelm@43773
    28
  val quote_string_q: string -> string
wenzelm@43773
    29
  val quote_string_qq: string -> string
wenzelm@43773
    30
  val quote_string_bq: string -> string
wenzelm@55033
    31
  val scan_cartouche: (string -> (T list -> T list * T list) -> T list -> T list * T list) ->
wenzelm@55033
    32
    T list -> T list * T list
wenzelm@55033
    33
  val scan_cartouche_body: (string -> (T list -> T list * T list) -> T list -> T list * T list) ->
wenzelm@55033
    34
    T list -> T list * T list
wenzelm@55033
    35
  val recover_cartouche: T list -> T list * T list
wenzelm@55033
    36
  val cartouche_content: T list -> T list
wenzelm@27763
    37
  val scan_comment: (string -> (T list -> T list * T list) -> T list -> T list * T list) ->
wenzelm@27763
    38
    T list -> T list * T list
wenzelm@27763
    39
  val scan_comment_body: (string -> (T list -> T list * T list) -> T list -> T list * T list) ->
wenzelm@27763
    40
    T list -> T list * T list
wenzelm@48743
    41
  val recover_comment: T list -> T list * T list
wenzelm@27763
    42
  val source: Position.T -> (Symbol.symbol, 'a) Source.source ->
wenzelm@27763
    43
    (T, Position.T * (Symbol.symbol, 'a) Source.source) Source.source
wenzelm@27778
    44
  type text = string
wenzelm@27797
    45
  val implode: T list -> text
wenzelm@27797
    46
  val implode_range: Position.T -> Position.T -> T list -> text * Position.range
wenzelm@27778
    47
  val explode: text * Position.T -> T list
wenzelm@50239
    48
  val scan_ident: T list -> T list * T list
wenzelm@50239
    49
  val is_identifier: string -> bool
wenzelm@27763
    50
end;
wenzelm@27763
    51
wenzelm@30573
    52
structure Symbol_Pos: SYMBOL_POS =
wenzelm@27763
    53
struct
wenzelm@27763
    54
wenzelm@27763
    55
(* type T *)
wenzelm@27763
    56
wenzelm@27763
    57
type T = Symbol.symbol * Position.T;
wenzelm@27763
    58
wenzelm@27763
    59
fun symbol ((s, _): T) = s;
wenzelm@27852
    60
wenzelm@27797
    61
val content = implode o map symbol;
wenzelm@27763
    62
wenzelm@55033
    63
fun range (syms as (_, pos) :: _) =
wenzelm@55033
    64
      let val pos' = List.last syms |-> Position.advance
wenzelm@55033
    65
      in Position.range pos pos' end
wenzelm@55033
    66
  | range [] = Position.no_range;
wenzelm@55033
    67
wenzelm@27763
    68
wenzelm@27763
    69
(* stopper *)
wenzelm@27763
    70
wenzelm@27763
    71
fun mk_eof pos = (Symbol.eof, pos);
wenzelm@27763
    72
val eof = mk_eof Position.none;
wenzelm@27763
    73
wenzelm@27763
    74
val is_eof = Symbol.is_eof o symbol;
wenzelm@27763
    75
wenzelm@27763
    76
val stopper =
wenzelm@27763
    77
  Scan.stopper (fn [] => eof | inp => mk_eof (List.last inp |-> Position.advance)) is_eof;
wenzelm@27763
    78
wenzelm@27763
    79
wenzelm@27763
    80
(* basic scanners *)
wenzelm@27763
    81
wenzelm@27763
    82
fun !!! text scan =
wenzelm@27763
    83
  let
wenzelm@48911
    84
    fun get_pos [] = " (end-of-input)"
wenzelm@48992
    85
      | get_pos ((_, pos) :: _) = Position.here pos;
wenzelm@27763
    86
wenzelm@43947
    87
    fun err (syms, msg) = fn () =>
wenzelm@48770
    88
      text () ^ get_pos syms ^
wenzelm@50201
    89
      Markup.markup Markup.no_report (" at " ^ Symbol.beginning 10 (map symbol syms)) ^
wenzelm@43947
    90
      (case msg of NONE => "" | SOME m => "\n" ^ m ());
wenzelm@27763
    91
  in Scan.!! err scan end;
wenzelm@27763
    92
wenzelm@30586
    93
fun change_prompt scan = Scan.prompt "# " scan;
wenzelm@30586
    94
wenzelm@55033
    95
fun $$ s = Scan.one (fn x => symbol x = s);
wenzelm@27763
    96
fun $$$ s = Scan.one (fn x => symbol x = s) >> single;
wenzelm@27763
    97
fun ~$$$ s = Scan.one (fn x => symbol x <> s) >> single;
wenzelm@27763
    98
wenzelm@27778
    99
val scan_pos = Scan.ahead (Scan.one (K true)) >> (fn (_, pos): T => pos);
wenzelm@27763
   100
wenzelm@27763
   101
wenzelm@43773
   102
(* scan string literals *)
wenzelm@30586
   103
wenzelm@30586
   104
local
wenzelm@30586
   105
wenzelm@30586
   106
val char_code =
wenzelm@30586
   107
  Scan.one (Symbol.is_ascii_digit o symbol) --
wenzelm@30586
   108
  Scan.one (Symbol.is_ascii_digit o symbol) --
wenzelm@30586
   109
  Scan.one (Symbol.is_ascii_digit o symbol) :|--
wenzelm@30586
   110
  (fn (((a, pos), (b, _)), (c, _)) =>
wenzelm@30586
   111
    let val (n, _) = Library.read_int [a, b, c]
wenzelm@30586
   112
    in if n <= 255 then Scan.succeed [(chr n, pos)] else Scan.fail end);
wenzelm@30586
   113
wenzelm@48764
   114
fun scan_str q err_prefix =
wenzelm@48764
   115
  $$$ "\\" |-- !!! (fn () => err_prefix ^ "bad escape character in string")
wenzelm@48764
   116
    ($$$ q || $$$ "\\" || char_code) ||
wenzelm@30586
   117
  Scan.one (fn (s, _) => s <> q andalso s <> "\\" andalso Symbol.is_regular s) >> single;
wenzelm@30586
   118
wenzelm@48764
   119
fun scan_strs q err_prefix =
wenzelm@48764
   120
  (scan_pos --| $$$ q) -- !!! (fn () => err_prefix ^ "missing quote at end of string")
wenzelm@48764
   121
    (change_prompt ((Scan.repeat (scan_str q err_prefix) >> flat) -- ($$$ q |-- scan_pos)));
wenzelm@30586
   122
wenzelm@48743
   123
fun recover_strs q =
wenzelm@48764
   124
  $$$ q @@@ (Scan.repeat (Scan.permissive (scan_str q "")) >> flat);
wenzelm@48743
   125
wenzelm@30586
   126
in
wenzelm@30586
   127
wenzelm@42503
   128
val scan_string_q = scan_strs "'";
wenzelm@42503
   129
val scan_string_qq = scan_strs "\"";
wenzelm@42503
   130
val scan_string_bq = scan_strs "`";
wenzelm@30586
   131
wenzelm@48743
   132
val recover_string_q = recover_strs "'";
wenzelm@48743
   133
val recover_string_qq = recover_strs "\"";
wenzelm@48743
   134
val recover_string_bq = recover_strs "`";
wenzelm@48743
   135
wenzelm@30586
   136
end;
wenzelm@30586
   137
wenzelm@30586
   138
wenzelm@43773
   139
(* quote string literals *)
wenzelm@43773
   140
wenzelm@43773
   141
local
wenzelm@43773
   142
wenzelm@43773
   143
fun char_code i =
wenzelm@43773
   144
  (if i < 10 then "00" else if i < 100 then "0" else "") ^ string_of_int i;
wenzelm@43773
   145
wenzelm@43773
   146
fun quote_str q s =
wenzelm@43773
   147
  if Symbol.is_ascii_control s then "\\" ^ char_code (ord s)
wenzelm@43773
   148
  else if s = q orelse s = "\\" then "\\" ^ s
wenzelm@43773
   149
  else s;
wenzelm@43773
   150
wenzelm@43773
   151
fun quote_string q = enclose q q o implode o map (quote_str q) o Symbol.explode;
wenzelm@43773
   152
wenzelm@43773
   153
in
wenzelm@43773
   154
wenzelm@43773
   155
val quote_string_q = quote_string "'";
wenzelm@43773
   156
val quote_string_qq = quote_string "\"";
wenzelm@43773
   157
val quote_string_bq = quote_string "`";
wenzelm@43773
   158
wenzelm@43773
   159
end;
wenzelm@43773
   160
wenzelm@43773
   161
wenzelm@55033
   162
(* nested text cartouches *)
wenzelm@55033
   163
wenzelm@55033
   164
local
wenzelm@55033
   165
wenzelm@55033
   166
val scan_cart =
wenzelm@55033
   167
  Scan.depend (fn (d: int) => $$ "\\<open>" >> pair (d + 1)) ||
wenzelm@55033
   168
  Scan.depend (fn 0 => Scan.fail | d => $$ "\\<close>" >> pair (d - 1)) ||
wenzelm@55033
   169
  Scan.lift (Scan.one (fn (s, _) => s <> "\\<close>" andalso Symbol.is_regular s));
wenzelm@55033
   170
wenzelm@55033
   171
val scan_carts = Scan.pass 0 (Scan.repeat scan_cart);
wenzelm@55033
   172
wenzelm@55033
   173
val scan_body = change_prompt scan_carts;
wenzelm@55033
   174
wenzelm@55033
   175
in
wenzelm@55033
   176
wenzelm@55033
   177
fun scan_cartouche cut =
wenzelm@55033
   178
  $$ "\\<open>" ::: cut "unclosed text cartouche" (scan_body @@@ $$$ "\\<close>");
wenzelm@55033
   179
wenzelm@55033
   180
fun scan_cartouche_body cut =
wenzelm@55033
   181
  $$ "\\<open>" |-- cut "unclosed text cartouche" (scan_body --| $$ "\\<close>");
wenzelm@55033
   182
wenzelm@55033
   183
val recover_cartouche =
wenzelm@55033
   184
  $$$ "\\<open>" @@@ scan_carts;
wenzelm@55033
   185
wenzelm@55033
   186
end;
wenzelm@55033
   187
wenzelm@55033
   188
fun cartouche_content syms =
wenzelm@55033
   189
  let
wenzelm@55033
   190
    fun err () =
wenzelm@55033
   191
      error ("Malformed text cartouche: " ^ quote (content syms) ^
wenzelm@55033
   192
        Position.here (Position.set_range (range syms)));
wenzelm@55033
   193
  in
wenzelm@55033
   194
    (case syms of
wenzelm@55033
   195
      ("\\<open>", _) :: rest =>
wenzelm@55033
   196
        (case rev rest of
wenzelm@55033
   197
          ("\\<close>", _) :: rrest => rev rrest
wenzelm@55033
   198
        | _ => err ())
wenzelm@55033
   199
    | _ => err ())
wenzelm@55033
   200
  end;
wenzelm@55033
   201
wenzelm@55033
   202
wenzelm@27763
   203
(* ML-style comments *)
wenzelm@27763
   204
wenzelm@27763
   205
local
wenzelm@27763
   206
wenzelm@27763
   207
val scan_cmt =
wenzelm@27763
   208
  Scan.depend (fn (d: int) => $$$ "(" @@@ $$$ "*" >> pair (d + 1)) ||
wenzelm@27763
   209
  Scan.depend (fn 0 => Scan.fail | d => $$$ "*" @@@ $$$ ")" >> pair (d - 1)) ||
wenzelm@27763
   210
  Scan.lift ($$$ "*" --| Scan.ahead (~$$$ ")")) ||
wenzelm@27763
   211
  Scan.lift (Scan.one (fn (s, _) => s <> "*" andalso Symbol.is_regular s)) >> single;
wenzelm@27763
   212
wenzelm@48743
   213
val scan_cmts = Scan.pass 0 (Scan.repeat scan_cmt >> flat);
wenzelm@48743
   214
wenzelm@48743
   215
val scan_body = change_prompt scan_cmts;
wenzelm@27763
   216
wenzelm@27763
   217
in
wenzelm@27763
   218
wenzelm@27763
   219
fun scan_comment cut =
wenzelm@27763
   220
  $$$ "(" @@@ $$$ "*" @@@ cut "missing end of comment" (scan_body @@@ $$$ "*" @@@ $$$ ")");
wenzelm@27763
   221
wenzelm@27763
   222
fun scan_comment_body cut =
wenzelm@27763
   223
  $$$ "(" |-- $$$ "*" |-- cut "missing end of comment" (scan_body --| $$$ "*" --| $$$ ")");
wenzelm@27763
   224
wenzelm@48743
   225
val recover_comment =
wenzelm@48743
   226
  $$$ "(" @@@ $$$ "*" @@@ scan_cmts;
wenzelm@48743
   227
wenzelm@27763
   228
end;
wenzelm@27763
   229
wenzelm@27763
   230
wenzelm@27763
   231
(* source *)
wenzelm@27763
   232
wenzelm@27763
   233
fun source pos =
wenzelm@27763
   234
  Source.source' pos Symbol.stopper (Scan.bulk (Scan.depend (fn pos =>
wenzelm@27763
   235
    Scan.one Symbol.not_eof >> (fn s => (Position.advance s pos, (s, pos)))))) NONE;
wenzelm@27763
   236
wenzelm@27763
   237
wenzelm@27763
   238
(* compact representation -- with Symbol.DEL padding *)
wenzelm@27763
   239
wenzelm@27778
   240
type text = string;
wenzelm@27778
   241
wenzelm@27763
   242
fun pad [] = []
wenzelm@27763
   243
  | pad [(s, _)] = [s]
wenzelm@32784
   244
  | pad ((s1, pos1) :: (rest as (_, pos2) :: _)) =
wenzelm@27763
   245
      let
wenzelm@27763
   246
        val end_pos1 = Position.advance s1 pos1;
wenzelm@27797
   247
        val d = Int.max (0, Position.distance_of end_pos1 pos2);
wenzelm@27763
   248
      in s1 :: replicate d Symbol.DEL @ pad rest end;
wenzelm@27763
   249
wenzelm@27797
   250
val implode = implode o pad;
wenzelm@27763
   251
wenzelm@27797
   252
fun implode_range pos1 pos2 syms =
wenzelm@27797
   253
  let val syms' = (("", pos1) :: syms @ [("", pos2)])
wenzelm@27797
   254
  in (implode syms', range syms') end;
wenzelm@27778
   255
wenzelm@27763
   256
fun explode (str, pos) =
wenzelm@41416
   257
  let
wenzelm@41416
   258
    val (res, _) =
wenzelm@41416
   259
      fold (fn s => fn (res, p) => ((s, p) :: res, Position.advance s p))
wenzelm@41416
   260
        (Symbol.explode str) ([], Position.reset_range pos);
wenzelm@41416
   261
  in fold (fn (s, p) => if s = Symbol.DEL then I else cons (s, p)) res [] end;
wenzelm@27763
   262
wenzelm@50239
   263
wenzelm@50239
   264
(* identifiers *)
wenzelm@50239
   265
wenzelm@50242
   266
local
wenzelm@50242
   267
wenzelm@52616
   268
val letter = Scan.one (symbol #> Symbol.is_letter);
wenzelm@52616
   269
val letdigs1 = Scan.many1 (symbol #> Symbol.is_letdig);
wenzelm@53016
   270
wenzelm@54732
   271
val sub = Scan.one (symbol #> (fn s => s = "\\<^sub>"));
wenzelm@50242
   272
wenzelm@50242
   273
in
wenzelm@50242
   274
wenzelm@52920
   275
val scan_ident = letter ::: (Scan.repeat (letdigs1 || sub ::: letdigs1) >> flat);
wenzelm@50242
   276
wenzelm@50242
   277
end;
wenzelm@50242
   278
wenzelm@50239
   279
fun is_identifier s =
wenzelm@50295
   280
  Symbol.is_ascii_identifier s orelse
wenzelm@50295
   281
    (case try (Scan.finite stopper scan_ident) (explode (s, Position.none)) of
wenzelm@50295
   282
      SOME (_, []) => true
wenzelm@50295
   283
    | _ => false);
wenzelm@50239
   284
wenzelm@27763
   285
end;
wenzelm@27763
   286
wenzelm@36957
   287
structure Basic_Symbol_Pos =   (*not open by default*)
wenzelm@36957
   288
struct
wenzelm@36957
   289
  val $$$ = Symbol_Pos.$$$;
wenzelm@36957
   290
  val ~$$$ = Symbol_Pos.~$$$;
wenzelm@36957
   291
end;
wenzelm@27763
   292