wenzelm@5829: (*  Title:      Pure/Isar/outer_syntax.ML
wenzelm@5829:     ID:         $Id$
wenzelm@5829:     Author:     Markus Wenzel, TU Muenchen
wenzelm@5829: 
wenzelm@5829: The global Isabelle/Isar outer syntax.
wenzelm@6247: 
wenzelm@6247: TODO:
wenzelm@6247:   - cleanup;
wenzelm@6247:   - avoid string constants;
wenzelm@5829: *)
wenzelm@5829: 
wenzelm@5829: signature BASIC_OUTER_SYNTAX =
wenzelm@5829: sig
wenzelm@5829:   val main: unit -> unit
wenzelm@5883:   val loop: unit -> unit
wenzelm@5829:   val help: unit -> unit
wenzelm@5829: end;
wenzelm@5829: 
wenzelm@5829: signature OUTER_SYNTAX =
wenzelm@5829: sig
wenzelm@5829:   include BASIC_OUTER_SYNTAX
wenzelm@5829:   type token
wenzelm@5829:   type parser
wenzelm@6373:   val command: string -> string ->
wenzelm@6373:     (token list -> (Toplevel.transition -> Toplevel.transition) * token list) -> parser
wenzelm@6373:   val improper_command: string -> string ->
wenzelm@6373:     (token list -> (Toplevel.transition -> Toplevel.transition) * token list) -> parser
wenzelm@5883:   val print_outer_syntax: unit -> unit
wenzelm@5952:   val commands: unit -> string list
wenzelm@5829:   val add_keywords: string list -> unit
wenzelm@5829:   val add_parsers: parser list -> unit
wenzelm@6247:   val theory_header: token list -> (string * string list * (string * bool) list) * token list
wenzelm@6199:   val deps_thy: string -> bool -> Path.T -> string list * Path.T list
wenzelm@6199:   val load_thy: string -> bool -> bool -> Path.T -> unit
wenzelm@5829:   val isar: Toplevel.isar
wenzelm@5829: end;
wenzelm@5829: 
wenzelm@5829: structure OuterSyntax: OUTER_SYNTAX =
wenzelm@5829: struct
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (** outer syntax **)
wenzelm@5829: 
wenzelm@5829: (* parsers *)
wenzelm@5829: 
wenzelm@5829: type token = OuterLex.token;
wenzelm@5829: type parser_fn = token list -> (Toplevel.transition -> Toplevel.transition) * token list;
wenzelm@5829: 
wenzelm@5829: datatype parser =
wenzelm@5829:   Parser of string * string * bool * parser_fn;
wenzelm@5829: 
wenzelm@5829: fun parser int_only name comment parse = Parser (name, comment, int_only, parse);
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (* parse command *)
wenzelm@5829: 
wenzelm@6199: local open OuterParse in
wenzelm@6199: 
wenzelm@5829: fun command_name cmd =
wenzelm@5829:   group "command"
wenzelm@5829:     (position (Scan.one (OuterLex.keyword_pred (is_some o cmd)) >> OuterLex.val_of));
wenzelm@5829: 
wenzelm@5829: fun command_body cmd (name, _) =
wenzelm@5829:   let val (int_only, parse) = the (cmd name)
wenzelm@5829:   in !!! (Scan.prompt (name ^ "# ") (parse >> pair int_only)) end;
wenzelm@5829: 
wenzelm@5829: fun command cmd =
wenzelm@5829:   $$$ ";" >> K None ||
wenzelm@5829:   command_name cmd :-- command_body cmd >> (fn ((name, pos), (int_only, f)) =>
wenzelm@5829:     Some (Toplevel.empty |> Toplevel.name name |> Toplevel.position pos |>
wenzelm@5829:       Toplevel.interactive int_only |> f));
wenzelm@5829: 
wenzelm@6199: end;
wenzelm@6199: 
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (** global syntax state **)
wenzelm@5829: 
wenzelm@5829: val global_lexicon = ref Scan.empty_lexicon;
wenzelm@5829: val global_parsers = ref (Symtab.empty: (string * (bool * parser_fn)) Symtab.table);
wenzelm@5829: 
wenzelm@5952: fun commands () = Symtab.keys (! global_parsers);
wenzelm@5952: 
wenzelm@5829: 
wenzelm@5829: (* print syntax *)
wenzelm@5829: 
wenzelm@5829: fun print_outer_syntax () =
wenzelm@5829:   let
wenzelm@5829:     val keywords = map implode (Scan.dest_lexicon (! global_lexicon));
wenzelm@5829:     fun pretty_cmd (name, (comment, _)) =
wenzelm@5829:       Pretty.block [Pretty.str (name ^ ":"), Pretty.brk 2, Pretty.str comment];
wenzelm@5829:     val (int_cmds, cmds) = partition (#1 o #2 o #2) (Symtab.dest (! global_parsers));
wenzelm@5829:   in
wenzelm@6107:     Pretty.writeln (Pretty.strs ("syntax keywords:" :: map quote keywords));
wenzelm@6095:     Pretty.writeln (Pretty.big_list "proper commands:" (map pretty_cmd cmds));
wenzelm@6095:     Pretty.writeln (Pretty.big_list "improper commands (interactive-only):"
wenzelm@6095:       (map pretty_cmd int_cmds))
wenzelm@5829:   end;
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (* augment syntax *)
wenzelm@5829: 
wenzelm@5829: fun add_keywords keywords =
wenzelm@5829:   global_lexicon := Scan.extend_lexicon (! global_lexicon) (map Symbol.explode keywords);
wenzelm@5829: 
wenzelm@5829: fun add_parser (tab, Parser (name, comment, int_only, parse)) =
wenzelm@5829:  (if is_none (Symtab.lookup (tab, name)) then ()
wenzelm@5829:   else warning ("Redefined outer syntax command " ^ quote name);
wenzelm@5829:   Symtab.update ((name, (comment, (int_only, parse))), tab));
wenzelm@5829: 
wenzelm@5829: fun add_parsers parsers =
wenzelm@5829:   (global_parsers := foldl add_parser (! global_parsers, parsers);
wenzelm@5829:     add_keywords (map (fn Parser (name, _, _, _) => name) parsers));
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (* get current lexer / parser *)
wenzelm@5829: 
wenzelm@5829: (*Note: the syntax for files is statically determined at the very
wenzelm@5829:   beginning; for interactive processing it may change dynamically.*)
wenzelm@5829: 
wenzelm@5829: fun get_lexicon () = ! global_lexicon;
wenzelm@5829: fun get_parser () = apsome snd o curry Symtab.lookup (! global_parsers);
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (** read theory **)
wenzelm@5829: 
wenzelm@6247: (* theory keyword *)
wenzelm@6247: 
wenzelm@6247: val theoryN = "theory";
wenzelm@6247: val theory_keyword = OuterParse.$$$ theoryN;
wenzelm@6247: 
wenzelm@6247: 
wenzelm@5829: (* source *)
wenzelm@5829: 
wenzelm@5829: fun no_command cmd =
wenzelm@5829:   Scan.one ((not o OuterLex.keyword_pred ((is_some o cmd) orf equal ";")) andf OuterLex.not_eof);
wenzelm@5829: 
wenzelm@5829: fun recover cmd =
wenzelm@5829:   Scan.prompt "recover# " (Scan.one OuterLex.not_eof -- Scan.repeat (no_command cmd));
wenzelm@5829: 
wenzelm@5829: fun source do_recover cmd src =
wenzelm@5829:   src
wenzelm@6199:   |> Source.source OuterLex.stopper (Scan.bulk (fn xs => OuterParse.!!! (command (cmd ())) xs))
wenzelm@5829:     (if do_recover then Some (fn xs => recover (cmd ()) xs) else None)
wenzelm@5829:   |> Source.mapfilter I;
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (* detect header *)
wenzelm@5829: 
wenzelm@6199: fun scan_header get_lexicon scan (src, pos) =
wenzelm@5829:   src
wenzelm@5829:   |> Symbol.source false
wenzelm@6199:   |> OuterLex.source false get_lexicon pos
wenzelm@6199:   |> Source.source OuterLex.stopper (Scan.single scan) None
wenzelm@5829:   |> (fst o the o Source.get_single);
wenzelm@5829: 
wenzelm@6247: val check_header_lexicon = Scan.make_lexicon [Symbol.explode theoryN];
wenzelm@5829: 
wenzelm@6199: fun is_old_theory src =
wenzelm@6247:   is_none (scan_header (K check_header_lexicon) (Scan.option theory_keyword) src);
wenzelm@6199: 
wenzelm@6199: fun warn_theory_style path is_old =
wenzelm@6199:   let
wenzelm@6199:     val style = if is_old then "old" else "new";
wenzelm@6199:     val _ = warning ("Assuming " ^ style ^ "-style theory format for " ^ quote (Path.pack path));
wenzelm@6199:   in is_old end;
wenzelm@6199: 
wenzelm@6199: 
wenzelm@6199: (* deps_thy --- inspect theory header *)
wenzelm@6199: 
wenzelm@6247: val header_lexicon =
wenzelm@6247:   Scan.make_lexicon (map Symbol.explode ["(", ")", "+", ":", "=", "files", theoryN]);
wenzelm@6199: 
wenzelm@6199: local open OuterParse in
wenzelm@6199: 
wenzelm@6247: val file_name = ($$$ "(" |-- !!! (name --| $$$ ")")) >> rpair false || name >> rpair true;
wenzelm@6247: 
wenzelm@6247: val theory_head =
wenzelm@6247:   (name -- ($$$ "=" |-- enum1 "+" name) --
wenzelm@6247:     Scan.optional ($$$ "files" |-- !!! (Scan.repeat1 file_name)) [])
wenzelm@6247:   >> (fn ((A, Bs), files) => (A, Bs, files));
wenzelm@6247: 
wenzelm@6247: val theory_header = theory_head --| (Scan.ahead eof || $$$ ":");
wenzelm@6247: val only_header = theory_keyword |-- theory_head --| Scan.ahead eof;
wenzelm@6247: val new_header = theory_keyword |-- !!! theory_header;
wenzelm@6199: 
wenzelm@6199: val old_header =
wenzelm@6199:   name -- ($$$ "=" |-- name -- Scan.repeat ($$$ "+" |-- name))
wenzelm@6247:   >> (fn (A, (B, Bs)) => (A, B :: Bs, []: (string * bool) list));
wenzelm@6199: 
wenzelm@6199: end;
wenzelm@5829: 
wenzelm@6199: fun deps_thy name ml path =
wenzelm@6199:   let
wenzelm@6641:     val src = File.source path;
wenzelm@6199:     val is_old = warn_theory_style path (is_old_theory src);
wenzelm@6247:     val (name', parents, files) =
wenzelm@6199:       (*Note: old style headers dynamically depend on the current lexicon :-( *)
wenzelm@6199:       if is_old then scan_header ThySyn.get_lexicon (Scan.error old_header) src
wenzelm@6247:       else scan_header (K header_lexicon) (Scan.error new_header) src;
wenzelm@6199: 
wenzelm@6199:     val ml_path = ThyLoad.ml_path name;
wenzelm@6199:     val ml_file = if not ml orelse is_none (ThyLoad.check_file ml_path) then [] else [ml_path];
wenzelm@6199:   in
wenzelm@6199:     if name <> name' then
wenzelm@6199:       error ("Filename " ^ quote (Path.pack path) ^ " does not match theory name " ^ quote name)
wenzelm@6247:     else (parents, map (Path.unpack o #1) files @ ml_file)
wenzelm@6199:   end;
wenzelm@6199: 
wenzelm@6199: 
wenzelm@6199: (* load_thy --- read text (including header) *)
wenzelm@6199: 
wenzelm@6247: fun try_ml_file name ml time =
wenzelm@6199:   let
wenzelm@6199:     val path = ThyLoad.ml_path name;
wenzelm@6247:     val tr = Toplevel.imperative (fn () => ThyInfo.load_file time path);
wenzelm@6247:     val tr_name = if time then "time_use" else "use";
wenzelm@6199:   in
wenzelm@6199:     if not ml orelse is_none (ThyLoad.check_file path) then ()
wenzelm@6247:     else Toplevel.excursion [Toplevel.empty |> Toplevel.name tr_name |> tr]
wenzelm@6199:   end;
wenzelm@6199: 
wenzelm@6199: fun parse_thy (src, pos) =
wenzelm@6247:   let
wenzelm@6247:     val lex_src =
wenzelm@6247:       src
wenzelm@6247:       |> Symbol.source false
wenzelm@6247:       |> OuterLex.source false (K (get_lexicon ())) pos;
wenzelm@6247:     val only_head =
wenzelm@6247:       lex_src
wenzelm@6247:       |> Source.source OuterLex.stopper (Scan.single (Scan.option only_header)) None
wenzelm@6247:       |> (fst o the o Source.get_single);
wenzelm@6247:   in
wenzelm@6247:     (case only_head of
wenzelm@6247:       None =>
wenzelm@6247:         lex_src
wenzelm@6247:         |> source false (K (get_parser ()))
wenzelm@6247:         |> Source.exhaust
wenzelm@6247:     | Some spec =>
wenzelm@6247:         [Toplevel.empty |> Toplevel.name theoryN |> IsarThy.theory spec,
wenzelm@6247:           Toplevel.empty |> Toplevel.name "end" |> Toplevel.exit])
wenzelm@6247:   end;
wenzelm@5829: 
wenzelm@6247: fun run_thy name path =
wenzelm@6641:   let val (src, pos) = File.source path in
wenzelm@6332:     Present.theory_source name src;
wenzelm@6247:     if is_old_theory (src, pos) then ThySyn.load_thy name (Source.exhaust src)
wenzelm@6247:     else (Toplevel.excursion (parse_thy (src, pos))
wenzelm@6247:       handle exn => error (Toplevel.exn_message exn))
wenzelm@6247:   end;
wenzelm@6199: 
wenzelm@6199: fun load_thy name ml time path =
wenzelm@6247:  (if time then
wenzelm@6247:     timeit (fn () =>
wenzelm@6247:      (writeln ("\n**** Starting theory " ^ quote name ^ " ****");
wenzelm@6247:       setmp Goals.proof_timing true (run_thy name) path;
wenzelm@6247:       writeln ("**** Finished theory " ^ quote name ^ " ****\n")))
wenzelm@6247:   else run_thy name path;
wenzelm@6247:   Context.context (ThyInfo.get_theory name);
wenzelm@6247:   try_ml_file name ml time);
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (* interactive source of state transformers *)
wenzelm@5829: 
wenzelm@5829: val isar =
wenzelm@5829:   Source.tty
wenzelm@5829:   |> Symbol.source true
wenzelm@5829:   |> OuterLex.source true get_lexicon (Position.line_name 1 "stdin")
wenzelm@5829:   |> source true get_parser;
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (** the read-eval-print loop **)
wenzelm@5829: 
wenzelm@5923: (* main loop *)
wenzelm@5923: 
wenzelm@5883: fun loop () = (Context.reset_context (); Toplevel.loop isar);
wenzelm@5829: 
wenzelm@5829: fun main () =
wenzelm@5829:  (Toplevel.set_state Toplevel.toplevel;
wenzelm@5829:   ml_prompts "ML> " "ML# ";
wenzelm@6199:   writeln (Session.welcome ());
wenzelm@5883:   loop ());
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (* help *)
wenzelm@5829: 
wenzelm@5829: fun help () =
wenzelm@5829:   writeln ("This is Isabelle's underlying ML system (" ^ ml_system ^ ");\n\
wenzelm@5883:     \invoke 'loop();' to enter the Isar loop.");
wenzelm@5829: 
wenzelm@5829: 
wenzelm@6373: (*final declarations of this structure!*)
wenzelm@6685: val command = parser false;
wenzelm@6685: val improper_command = parser true;
wenzelm@6685: 
wenzelm@6373: 
wenzelm@5829: end;
wenzelm@5829: 
wenzelm@6199: (*setup theory syntax dependent operations*)
wenzelm@6199: ThyLoad.deps_thy_fn := OuterSyntax.deps_thy;
wenzelm@6199: ThyLoad.load_thy_fn := OuterSyntax.load_thy;
wenzelm@6199: structure ThyLoad: THY_LOAD = ThyLoad;
wenzelm@6199: 
wenzelm@5829: structure BasicOuterSyntax: BASIC_OUTER_SYNTAX = OuterSyntax;
wenzelm@5829: open BasicOuterSyntax;