more position information, e.g. relevant for errors in generated ML source;
(* Title: Pure/General/symbol_pos.ML
Author: Makarius
Symbols with explicit position information.
*)
signature SYMBOL_POS =
sig
type T = Symbol.symbol * Position.T
val symbol: T -> Symbol.symbol
val $$ : Symbol.symbol -> T list -> T * T list
val ~$$ : Symbol.symbol -> T list -> T * T list
val $$$ : Symbol.symbol -> T list -> T list * T list
val ~$$$ : Symbol.symbol -> T list -> T list * T list
val content: T list -> string
val range: T list -> Position.range
val is_eof: T -> bool
val stopper: T Scan.stopper
val !!! : Scan.message -> (T list -> 'a) -> T list -> 'a
val scan_pos: T list -> Position.T * T list
val scan_string_q: string -> T list -> (Position.T * (T list * Position.T)) * T list
val scan_string_qq: string -> T list -> (Position.T * (T list * Position.T)) * T list
val scan_string_bq: string -> T list -> (Position.T * (T list * Position.T)) * T list
val recover_string_q: T list -> T list * T list
val recover_string_qq: T list -> T list * T list
val recover_string_bq: T list -> T list * T list
val quote_string_q: string -> string
val quote_string_qq: string -> string
val quote_string_bq: string -> string
val scan_cartouche: string -> T list -> T list * T list
val recover_cartouche: T list -> T list * T list
val cartouche_content: T list -> T list
val scan_comment: string -> T list -> T list * T list
val scan_comment_body: string -> T list -> T list * T list
val recover_comment: T list -> T list * T list
val source: Position.T -> (Symbol.symbol, 'a) Source.source ->
(T, Position.T * (Symbol.symbol, 'a) Source.source) Source.source
type text = string
val implode: T list -> text
val implode_range: Position.T -> Position.T -> T list -> text * Position.range
val explode: text * Position.T -> T list
type source = {delimited: bool, text: text, range: Position.range}
val source_explode: source -> T list
val source_content: source -> string * Position.T
val scan_ident: T list -> T list * T list
val is_identifier: string -> bool
end;
structure Symbol_Pos: SYMBOL_POS =
struct
(* type T *)
type T = Symbol.symbol * Position.T;
fun symbol ((s, _): T) = s;
val content = implode o map symbol;
fun range (syms as (_, pos) :: _) =
let val pos' = List.last syms |-> Position.advance
in Position.range pos pos' end
| range [] = Position.no_range;
(* stopper *)
fun mk_eof pos = (Symbol.eof, pos);
val eof = mk_eof Position.none;
val is_eof = Symbol.is_eof o symbol;
val stopper =
Scan.stopper (fn [] => eof | inp => mk_eof (List.last inp |-> Position.advance)) is_eof;
(* basic scanners *)
fun !!! text scan =
let
fun get_pos [] = " (end-of-input)"
| get_pos ((_, pos) :: _) = Position.here pos;
fun err (syms, msg) = fn () =>
text () ^ get_pos syms ^
Markup.markup Markup.no_report (" at " ^ Symbol.beginning 10 (map symbol syms)) ^
(case msg of NONE => "" | SOME m => "\n" ^ m ());
in Scan.!! err scan end;
fun $$ s = Scan.one (fn x => symbol x = s);
fun ~$$ s = Scan.one (fn x => symbol x <> s);
fun $$$ s = Scan.one (fn x => symbol x = s) >> single;
fun ~$$$ s = Scan.one (fn x => symbol x <> s) >> single;
val scan_pos = Scan.ahead (Scan.one (K true)) >> (fn (_, pos): T => pos);
(* scan string literals *)
local
val char_code =
Scan.one (Symbol.is_ascii_digit o symbol) --
Scan.one (Symbol.is_ascii_digit o symbol) --
Scan.one (Symbol.is_ascii_digit o symbol) :|--
(fn (((a, pos), (b, _)), (c, _)) =>
let val (n, _) = Library.read_int [a, b, c]
in if n <= 255 then Scan.succeed [(chr n, pos)] else Scan.fail end);
fun scan_str q err_prefix =
$$$ "\\" |-- !!! (fn () => err_prefix ^ "bad escape character in string")
($$$ q || $$$ "\\" || char_code) ||
Scan.one (fn (s, _) => s <> q andalso s <> "\\" andalso Symbol.not_eof s) >> single;
fun scan_strs q err_prefix =
Scan.ahead ($$ q) |--
!!! (fn () => err_prefix ^ "unclosed string literal")
((scan_pos --| $$$ q) --
((Scan.repeat (scan_str q err_prefix) >> flat) -- ($$$ q |-- scan_pos)));
fun recover_strs q =
$$$ q @@@ (Scan.repeat (Scan.permissive (scan_str q "")) >> flat);
in
val scan_string_q = scan_strs "'";
val scan_string_qq = scan_strs "\"";
val scan_string_bq = scan_strs "`";
val recover_string_q = recover_strs "'";
val recover_string_qq = recover_strs "\"";
val recover_string_bq = recover_strs "`";
end;
(* quote string literals *)
local
fun char_code i =
(if i < 10 then "00" else if i < 100 then "0" else "") ^ string_of_int i;
fun quote_str q s =
if Symbol.is_ascii_control s then "\\" ^ char_code (ord s)
else if s = q orelse s = "\\" then "\\" ^ s
else s;
fun quote_string q = enclose q q o implode o map (quote_str q) o Symbol.explode;
in
val quote_string_q = quote_string "'";
val quote_string_qq = quote_string "\"";
val quote_string_bq = quote_string "`";
end;
(* nested text cartouches *)
val scan_cartouche_depth =
Scan.repeat1 (Scan.depend (fn (d: int) =>
$$ "\\<open>" >> pair (d + 1) ||
(if d > 0 then
Scan.one (fn (s, _) => s <> "\\<close>" andalso Symbol.not_eof s) >> pair d ||
$$ "\\<close>" >> pair (d - 1)
else Scan.fail)));
fun scan_cartouche err_prefix =
Scan.ahead ($$ "\\<open>") |--
!!! (fn () => err_prefix ^ "unclosed text cartouche")
(Scan.provide (fn d => d = 0) 0 scan_cartouche_depth);
val recover_cartouche = Scan.pass 0 scan_cartouche_depth;
fun cartouche_content syms =
let
fun err () =
error ("Malformed text cartouche: "
^ quote (content syms) ^ Position.here (#1 (range syms)));
in
(case syms of
("\\<open>", _) :: rest =>
(case rev rest of
("\\<close>", _) :: rrest => rev rrest
| _ => err ())
| _ => err ())
end;
(* ML-style comments *)
local
val scan_cmt =
Scan.depend (fn (d: int) => $$$ "(" @@@ $$$ "*" >> pair (d + 1)) ||
Scan.depend (fn 0 => Scan.fail | d => $$$ "*" @@@ $$$ ")" >> pair (d - 1)) ||
Scan.lift ($$$ "*" --| Scan.ahead (~$$$ ")")) ||
Scan.lift (Scan.one (fn (s, _) => s <> "*" andalso Symbol.not_eof s)) >> single;
val scan_cmts = Scan.pass 0 (Scan.repeat scan_cmt >> flat);
in
fun scan_comment err_prefix =
Scan.ahead ($$ "(" -- $$ "*") |--
!!! (fn () => err_prefix ^ "unclosed comment")
($$$ "(" @@@ $$$ "*" @@@ scan_cmts @@@ $$$ "*" @@@ $$$ ")");
fun scan_comment_body err_prefix =
Scan.ahead ($$ "(" -- $$ "*") |--
!!! (fn () => err_prefix ^ "unclosed comment")
($$ "(" |-- $$ "*" |-- scan_cmts --| $$ "*" --| $$ ")");
val recover_comment =
$$$ "(" @@@ $$$ "*" @@@ scan_cmts;
end;
(* source *)
fun source pos =
Source.source' pos Symbol.stopper (Scan.bulk (Scan.depend (fn pos =>
Scan.one Symbol.not_eof >> (fn s => (Position.advance s pos, (s, pos))))));
(* compact representation -- with Symbol.DEL padding *)
type text = string;
fun pad [] = []
| pad [(s, _)] = [s]
| pad ((s1, pos1) :: (rest as (_, pos2) :: _)) =
let
val end_pos1 = Position.advance s1 pos1;
val d = Int.max (0, Position.distance_of end_pos1 pos2);
in s1 :: replicate d Symbol.DEL @ pad rest end;
val implode = implode o pad;
fun implode_range pos1 pos2 syms =
let val syms' = (("", pos1) :: syms @ [("", pos2)])
in (implode syms', range syms') end;
fun explode (str, pos) =
let
val (res, _) =
fold (fn s => fn (res, p) => ((s, p) :: res, Position.advance s p))
(Symbol.explode str) ([], Position.reset_range pos);
in fold (fn (s, p) => if s = Symbol.DEL then I else cons (s, p)) res [] end;
(* full source information *)
type source = {delimited: bool, text: text, range: Position.range};
fun source_explode {delimited = _, text, range = (pos, _)} = explode (text, pos);
fun source_content {delimited = _, text, range = (pos, _)} =
let val syms = explode (text, pos) in (content syms, pos) end;
(* identifiers *)
local
val letter = Scan.one (symbol #> Symbol.is_letter);
val letdigs1 = Scan.many1 (symbol #> Symbol.is_letdig);
val sub = Scan.one (symbol #> (fn s => s = "\\<^sub>"));
in
val scan_ident = letter ::: (Scan.repeat (letdigs1 || sub ::: letdigs1) >> flat);
end;
fun is_identifier s =
Symbol.is_ascii_identifier s orelse
(case try (Scan.finite stopper scan_ident) (explode (s, Position.none)) of
SOME (_, []) => true
| _ => false);
end;
structure Basic_Symbol_Pos = (*not open by default*)
struct
val $$ = Symbol_Pos.$$;
val ~$$ = Symbol_Pos.~$$;
val $$$ = Symbol_Pos.$$$;
val ~$$$ = Symbol_Pos.~$$$;
end;