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@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@6860:   val sync_main: unit -> unit
wenzelm@6860:   val sync_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@6722:   structure Keyword:
wenzelm@6722:     sig
wenzelm@6722:       val control: string
wenzelm@6722:       val diag: string
wenzelm@6722:       val thy_begin: string
wenzelm@6722:       val thy_end: string
wenzelm@6722:       val thy_heading: string
wenzelm@6722:       val thy_decl: string
wenzelm@6722:       val thy_goal: string
wenzelm@6722:       val qed: string
wenzelm@6733:       val qed_block: string
wenzelm@6722:       val prf_goal: string
wenzelm@6722:       val prf_block: string
wenzelm@6722:       val prf_chain: string
wenzelm@6722:       val prf_decl: string
wenzelm@6868:       val prf_asm: string
wenzelm@6722:       val prf_script: string
wenzelm@6722:       val kinds: string list
wenzelm@6722:     end
wenzelm@5829:   type token
wenzelm@5829:   type parser
wenzelm@6722:   val command: string -> string -> string ->
wenzelm@6373:     (token list -> (Toplevel.transition -> Toplevel.transition) * token list) -> parser
wenzelm@6722:   val improper_command: string -> string -> string ->
wenzelm@6373:     (token list -> (Toplevel.transition -> Toplevel.transition) * token list) -> parser
wenzelm@7026:   val dest_keywords: unit -> string list
wenzelm@7026:   val dest_parsers: unit -> (string * string * string * bool) list
wenzelm@5883:   val print_outer_syntax: unit -> unit
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@6860:   val isar: bool -> Toplevel.isar
wenzelm@5829: end;
wenzelm@5829: 
wenzelm@5829: structure OuterSyntax: OUTER_SYNTAX =
wenzelm@5829: struct
wenzelm@5829: 
wenzelm@6860: structure P = OuterParse;
wenzelm@6860: 
wenzelm@5829: 
wenzelm@5829: (** outer syntax **)
wenzelm@5829: 
wenzelm@6722: (* command keyword classification *)
wenzelm@6722: 
wenzelm@6722: structure Keyword =
wenzelm@6722: struct
wenzelm@6722:   val control = "control";
wenzelm@6722:   val diag = "diag";
wenzelm@6722:   val thy_begin = "theory-begin";
wenzelm@6722:   val thy_end = "theory-end";
wenzelm@6722:   val thy_heading = "theory-heading";
wenzelm@6722:   val thy_decl = "theory-decl";
wenzelm@6722:   val thy_goal = "theory-goal";
wenzelm@6722:   val qed = "qed";
wenzelm@6733:   val qed_block = "qed-block";
wenzelm@6722:   val prf_goal = "proof-goal";
wenzelm@6722:   val prf_block = "proof-block";
wenzelm@6722:   val prf_chain = "proof-chain";
wenzelm@6722:   val prf_decl = "proof-decl";
wenzelm@6868:   val prf_asm = "proof-asm";
wenzelm@6722:   val prf_script = "proof-script";
wenzelm@6722: 
wenzelm@6722:   val kinds = [control, diag, thy_begin, thy_end, thy_heading, thy_decl, thy_goal, qed,
wenzelm@6868:     qed_block, prf_goal, prf_block, prf_chain, prf_decl, prf_asm, prf_script];
wenzelm@6722: end;
wenzelm@6722: 
wenzelm@6722: 
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@6722:   Parser of string * (string * string) * bool * parser_fn;
wenzelm@5829: 
wenzelm@6722: fun parser int_only name comment kind parse = Parser (name, (comment, kind), int_only, parse);
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (* parse command *)
wenzelm@5829: 
wenzelm@6860: local
wenzelm@6199: 
wenzelm@5829: fun command_body cmd (name, _) =
wenzelm@7026:   (case cmd name of
wenzelm@7026:     Some (int_only, parse) => P.!!! (Scan.prompt (name ^ "# ") (parse >> pair int_only))
wenzelm@7026:   | None => sys_error ("no parser for outer syntax command " ^ quote name));
wenzelm@6860: 
wenzelm@6860: fun terminator false = Scan.succeed ()
wenzelm@6860:   | terminator true = P.group "terminator" (Scan.option P.sync -- P.$$$ ";" >> K ());
wenzelm@6860: 
wenzelm@6860: in
wenzelm@5829: 
wenzelm@6860: fun command term cmd =
wenzelm@6860:   P.$$$ ";" >> K None ||
wenzelm@6860:   P.sync >> K None ||
wenzelm@7026:   (P.position P.command :-- command_body cmd) --| terminator term
wenzelm@6860:     >> (fn ((name, pos), (int_only, f)) =>
wenzelm@6860:       Some (Toplevel.empty |> Toplevel.name name |> Toplevel.position pos |>
wenzelm@6860:         Toplevel.interactive int_only |> f));
wenzelm@5829: 
wenzelm@6199: end;
wenzelm@6199: 
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (** global syntax state **)
wenzelm@5829: 
wenzelm@7026: local
wenzelm@7026: 
wenzelm@7026: val global_lexicons = ref (Scan.empty_lexicon, Scan.empty_lexicon);
wenzelm@6722: val global_parsers = ref (Symtab.empty: ((string * string) * (bool * parser_fn)) Symtab.table);
wenzelm@5952: 
wenzelm@7026: fun change_lexicons f =
wenzelm@7026:   let val lexs = f (! global_lexicons) in
wenzelm@7026:     (case (op inter_string) (pairself Scan.dest_lexicon lexs) of
wenzelm@7026:       [] => global_lexicons := lexs
wenzelm@7026:     | bads => error ("Clash of outer syntax commands and keywords: " ^ commas_quote bads))
wenzelm@7026:   end;
wenzelm@5829: 
wenzelm@7026: fun change_parsers f = global_parsers := f (! global_parsers);
wenzelm@6722: 
wenzelm@7026: in
wenzelm@7026: 
wenzelm@7026: (* get current lexers / parsers *)
wenzelm@7026: 
wenzelm@7026: (*Note: the syntax for files is statically determined at the very
wenzelm@7026:   beginning; for interactive processing it may change dynamically.*)
wenzelm@7026: 
wenzelm@7026: fun get_lexicons () = ! global_lexicons;
wenzelm@7026: fun get_parsers () = ! global_parsers;
wenzelm@7026: fun get_parser () = apsome snd o curry Symtab.lookup (! global_parsers);
wenzelm@5829: 
wenzelm@5829: 
wenzelm@5829: (* augment syntax *)
wenzelm@5829: 
wenzelm@7026: fun add_keywords keywords = change_lexicons (apfst (fn lex =>
wenzelm@7026:   (Scan.extend_lexicon lex (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@7026:   (change_parsers (fn tab => foldl add_parser (tab, parsers));
wenzelm@7026:     change_lexicons (apsnd (fn lex => Scan.extend_lexicon lex
wenzelm@7026:       (map (fn Parser (name, _, _, _) => Symbol.explode name) parsers))));
wenzelm@7026: 
wenzelm@7026: end;
wenzelm@5829: 
wenzelm@5829: 
wenzelm@7026: (* print syntax *)
wenzelm@7026: 
wenzelm@7026: fun dest_keywords () = Scan.dest_lexicon (#1 (get_lexicons ()));
wenzelm@7026: 
wenzelm@7026: fun dest_parsers () =
wenzelm@7026:   map (fn (name, ((cmt, kind), (int_only, _))) => (name, cmt, kind, int_only))
wenzelm@7026:     (Symtab.dest (get_parsers ()));
wenzelm@5829: 
wenzelm@7026: fun print_outer_syntax () =
wenzelm@7026:   let
wenzelm@7026:     fun pretty_cmd (name, comment, _, _) =
wenzelm@7026:       Pretty.block [Pretty.str (name ^ ":"), Pretty.brk 2, Pretty.str comment];
wenzelm@7026:     val (int_cmds, cmds) = partition #4 (dest_parsers ());
wenzelm@7026:   in
wenzelm@7026:     Pretty.writeln (Pretty.strs ("syntax keywords:" :: map quote (dest_keywords ())));
wenzelm@7026:     Pretty.writeln (Pretty.big_list "proper commands:" (map pretty_cmd cmds));
wenzelm@7026:     Pretty.writeln (Pretty.big_list "improper commands (interactive-only):"
wenzelm@7026:       (map pretty_cmd int_cmds))
wenzelm@7026:   end;
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.$$$ theoryN;
wenzelm@6247: 
wenzelm@6247: 
wenzelm@5829: (* source *)
wenzelm@5829: 
wenzelm@6860: local
wenzelm@5829: 
wenzelm@6860: val no_terminator =
wenzelm@6860:   Scan.unless (P.$$$ ";") (Scan.one (OuterLex.not_sync andf OuterLex.not_eof));
wenzelm@6860: 
wenzelm@6860: val recover = Scan.prompt "recover# " (Scan.repeat no_terminator);
wenzelm@6860: 
wenzelm@6860: in
wenzelm@5829: 
wenzelm@6860: fun source term do_recover cmd src =
wenzelm@5829:   src
wenzelm@6860:   |> Source.source OuterLex.stopper
wenzelm@6860:     (Scan.bulk (fn xs => OuterParse.!!! (command term (cmd ())) xs))
wenzelm@6860:     (if do_recover then Some recover else None)
wenzelm@5829:   |> Source.mapfilter I;
wenzelm@5829: 
wenzelm@6860: end;
wenzelm@6860: 
wenzelm@5829: 
wenzelm@5829: (* detect header *)
wenzelm@5829: 
wenzelm@7026: fun scan_header get_lex scan (src, pos) =
wenzelm@5829:   src
wenzelm@5829:   |> Symbol.source false
wenzelm@7026:   |> OuterLex.source false (fn () => (get_lex (), Scan.empty_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@6860: local
wenzelm@6199: 
wenzelm@7026: val file_name =
wenzelm@7026:   (P.$$$ "(" |-- P.!!! (P.name --| P.$$$ ")")) >> rpair false || P.name >> rpair true;
wenzelm@6247: 
wenzelm@6247: val theory_head =
wenzelm@6860:   (P.name -- (P.$$$ "=" |-- P.enum1 "+" P.name) --
wenzelm@6860:     Scan.optional (P.$$$ "files" |-- P.!!! (Scan.repeat1 file_name)) [])
wenzelm@6247:   >> (fn ((A, Bs), files) => (A, Bs, files));
wenzelm@6247: 
wenzelm@6860: in
wenzelm@6860: 
wenzelm@6860: val theory_header = theory_head --| (Scan.ahead P.eof || P.$$$ ":");
wenzelm@6860: val only_header = theory_keyword |-- theory_head --| Scan.ahead P.eof;
wenzelm@6860: val new_header = theory_keyword |-- P.!!! theory_header;
wenzelm@6199: 
wenzelm@6199: val old_header =
wenzelm@6860:   P.name -- (P.$$$ "=" |-- P.name -- Scan.repeat (P.$$$ "+" |-- P.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@7026:       |> OuterLex.source false (K (get_lexicons ())) 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@6860:         |> source false 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@6860: fun isar term =
wenzelm@5829:   Source.tty
wenzelm@5829:   |> Symbol.source true
wenzelm@7026:   |> OuterLex.source true get_lexicons (Position.line_name 1 "stdin")
wenzelm@6860:   |> source term 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@6860: fun gen_loop term = (Context.reset_context (); Toplevel.loop (isar term));
wenzelm@5829: 
wenzelm@6860: fun gen_main term =
wenzelm@5829:  (Toplevel.set_state Toplevel.toplevel;
wenzelm@5829:   ml_prompts "ML> " "ML# ";
wenzelm@6199:   writeln (Session.welcome ());
wenzelm@6860:   gen_loop term);
wenzelm@6860: 
wenzelm@6860: fun main () = gen_main false;
wenzelm@6860: fun loop () = gen_loop false;
wenzelm@6860: fun sync_main () = gen_main true;
wenzelm@6860: fun sync_loop () = gen_loop true;
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;