src/Pure/Thy/thy_info.ML

deps_thy/load_thy: store compact text to reduce space by factor 12;

(*  Title:      Pure/Thy/thy_info.ML
    Author:     Markus Wenzel, TU Muenchen

Main part of theory loader database, including handling of theory and
file dependencies.
*)

signature THY_INFO =
sig
  datatype action = Update | Outdate | Remove
  val str_of_action: action -> string
  val add_hook: (action -> string -> unit) -> unit
  val get_names: unit -> string list
  val known_thy: string -> bool
  val check_known_thy: string -> bool
  val if_known_thy: (string -> unit) -> string -> unit
  val lookup_theory: string -> theory option
  val get_theory: string -> theory
  val is_finished: string -> bool
  val master_directory: string -> Path.T
  val loaded_files: string -> Path.T list
  val get_parents: string -> string list
  val touch_thy: string -> unit
  val touch_child_thys: string -> unit
  val thy_ord: theory * theory -> order
  val remove_thy: string -> unit
  val kill_thy: string -> unit
  val provide_file: Path.T -> string -> unit
  val load_file: Path.T -> unit
  val exec_file: Path.T -> Context.generic -> Context.generic
  val use: string -> unit
  val use_thys: string list -> unit
  val use_thy: string -> unit
  val begin_theory: string -> string list -> (Path.T * bool) list -> bool -> theory
  val end_theory: theory -> unit
  val register_thy: string -> unit
  val register_theory: theory -> unit
  val finish: unit -> unit
end;

structure Thy_Info: THY_INFO =
struct

(** theory loader actions and hooks **)

datatype action = Update | Outdate | Remove;
val str_of_action = fn Update => "Update" | Outdate => "Outdate" | Remove => "Remove";

local
  val hooks = Unsynchronized.ref ([]: (action -> string -> unit) list);
in
  fun add_hook f = CRITICAL (fn () => Unsynchronized.change hooks (cons f));
  fun perform action name = List.app (fn f => (try (fn () => f action name) (); ())) (! hooks);
end;



(** thy database **)

(* messages *)

fun loader_msg txt [] = "Theory loader: " ^ txt
  | loader_msg txt names = "Theory loader: " ^ txt ^ " " ^ commas_quote names;

val show_path = space_implode " via " o map quote;
fun cycle_msg names = loader_msg ("cyclic dependency of " ^ show_path names) [];


(* derived graph operations *)

fun add_deps name parents G = Graph.add_deps_acyclic (name, parents) G
  handle Graph.CYCLES namess => error (cat_lines (map cycle_msg namess));

fun upd_deps name entry G =
  fold (fn parent => Graph.del_edge (parent, name)) (Graph.imm_preds G name) G
  |> Graph.map_node name (K entry);

fun new_deps name parents entry G =
  (if can (Graph.get_node G) name then upd_deps name entry G else Graph.new_node (name, entry) G)
  |> add_deps name parents;


(* thy database *)

type deps =
  {update_time: int,                      (*symbolic time of update; negative value means outdated*)
    master: (Path.T * File.ident) option, (*master dependencies for thy file*)
    text: string,                         (*source text for thy*)
    parents: string list,                 (*source specification of parents (partially qualified)*)
      (*auxiliary files: source path, physical path + identifier*)
    files: (Path.T * (Path.T * File.ident) option) list};

fun make_deps update_time master text parents files : deps =
  {update_time = update_time, master = master, text = text, parents = parents, files = files};

fun init_deps master text parents files =
  SOME (make_deps ~1 master text parents (map (rpair NONE) files));

fun master_dir NONE = Path.current
  | master_dir (SOME (path, _)) = Path.dir path;

fun master_dir' (d: deps option) = the_default Path.current (Option.map (master_dir o #master) d);
fun master_dir'' d = the_default Path.current (Option.map master_dir' d);

fun base_name s = Path.implode (Path.base (Path.explode s));


type thy = deps option * theory option;

local
  val database = Unsynchronized.ref (Graph.empty: thy Graph.T);
in
  fun get_thys () = ! database;
  fun change_thys f = CRITICAL (fn () => Unsynchronized.change database f);
end;


(* access thy graph *)

fun thy_graph f x = f (get_thys ()) x;

fun get_names () = Graph.topological_order (get_thys ());


(* access thy *)

fun lookup_thy name =
  SOME (thy_graph Graph.get_node name) handle Graph.UNDEF _ => NONE;

val known_thy = is_some o lookup_thy;
fun check_known_thy name = known_thy name orelse (warning ("Unknown theory " ^ quote name); false);
fun if_known_thy f name = if check_known_thy name then f name else ();

fun get_thy name =
  (case lookup_thy name of
    SOME thy => thy
  | NONE => error (loader_msg "nothing known about theory" [name]));

fun change_thy name f = CRITICAL (fn () =>
  (get_thy name; change_thys (Graph.map_node name f)));


(* access deps *)

val lookup_deps = Option.map #1 o lookup_thy;
val get_deps = #1 o get_thy;
fun change_deps name f = change_thy name (fn (deps, x) => (f deps, x));

val is_finished = is_none o get_deps;
val master_directory = master_dir' o get_deps;

fun loaded_files name =
  (case get_deps name of
    NONE => []
  | SOME {master, files, ...} =>
      (case master of SOME (thy_path, _) => [thy_path] | NONE => []) @
      (map_filter (Option.map #1 o #2) files));

fun get_parents name =
  thy_graph Graph.imm_preds name handle Graph.UNDEF _ =>
    error (loader_msg "nothing known about theory" [name]);


(* access theory *)

fun lookup_theory name =
  (case lookup_thy name of
    SOME (_, SOME thy) => SOME thy
  | _ => NONE);

fun get_theory name =
  (case lookup_theory name of
    SOME theory => theory
  | _ => error (loader_msg "undefined theory entry for" [name]));



(** thy operations **)

(* check state *)

fun check_unfinished fail name =
  if known_thy name andalso is_finished name then
    fail (loader_msg "cannot update finished theory" [name])
  else ();

fun check_files name =
  let
    val files = (case get_deps name of SOME {files, ...} => files | NONE => []);
    val missing_files = map_filter (fn (path, NONE) => SOME (Path.implode path) | _ => NONE) files;
    val _ = null missing_files orelse
      error (loader_msg "unresolved dependencies of theory" [name] ^
        " on file(s): " ^ commas_quote missing_files);
  in () end;


(* maintain update_time *)

local

fun is_outdated name =
  (case lookup_deps name of
    SOME (SOME {update_time, ...}) => update_time < 0
  | _ => false);

fun unfinished name =
  if is_finished name then (warning (loader_msg "tried to touch finished theory" [name]); NONE)
  else SOME name;

in

fun outdate_thy name =
  if is_finished name orelse is_outdated name then ()
  else CRITICAL (fn () =>
   (change_deps name (Option.map (fn {master, text, parents, files, ...} =>
    make_deps ~1 master text parents files)); perform Outdate name));

fun touch_thys names =
  List.app outdate_thy (thy_graph Graph.all_succs (map_filter unfinished names));

fun touch_thy name = touch_thys [name];
fun touch_child_thys name = touch_thys (thy_graph Graph.imm_succs name);

end;


(* management data *)

structure Management_Data = Theory_Data
(
  type T =
    Task_Queue.group option *   (*worker thread group*)
    int;                        (*abstract update time*)
  val empty = (NONE, 0);
  fun extend _ = empty;
  fun merge _ = empty;
);

val thy_ord = int_ord o pairself (#2 o Management_Data.get);


(* pending proofs *)

fun join_thy name =
  (case lookup_theory name of
    NONE => ()
  | SOME thy => PureThy.join_proofs thy);

fun cancel_thy name =
  (case lookup_theory name of
    NONE => ()
  | SOME thy =>
      (case #1 (Management_Data.get thy) of
        NONE => ()
      | SOME group => Future.cancel_group group));


(* remove theory *)

fun remove_thy name =
  if is_finished name then error (loader_msg "cannot remove finished theory" [name])
  else
    let
      val succs = thy_graph Graph.all_succs [name];
      val _ = List.app cancel_thy succs;
      val _ = priority (loader_msg "removing" succs);
      val _ = CRITICAL (fn () =>
        (List.app (perform Remove) succs; change_thys (Graph.del_nodes succs)));
    in () end;

val kill_thy = if_known_thy remove_thy;


(* load_file *)

local

fun provide path name info (SOME {update_time, master, text, parents, files}) =
     (if AList.defined (op =) files path then ()
      else warning (loader_msg "undeclared dependency of theory" [name] ^
        " on file: " ^ quote (Path.implode path));
      SOME (make_deps update_time master text parents
        (AList.update (op =) (path, SOME info) files)))
  | provide _ _ _ NONE = NONE;

fun run_file path =
  (case Option.map (Context.theory_name o Context.the_theory) (Context.thread_data ()) of
    NONE => (Thy_Load.load_ml Path.current path; ())
  | SOME name =>
      (case lookup_deps name of
        SOME deps =>
          change_deps name (provide path name (Thy_Load.load_ml (master_dir' deps) path))
      | NONE => (Thy_Load.load_ml Path.current path; ())));

in

fun provide_file path name =
  let
    val dir = master_directory name;
    val _ = check_unfinished error name;
  in
    (case Thy_Load.check_file dir path of
      SOME path_info => change_deps name (provide path name path_info)
    | NONE => error ("Could not find file " ^ quote (Path.implode path)))
  end;

fun load_file path =
  if ! Output.timing then
    let val name = Path.implode path in
      timeit (fn () =>
       (priority ("\n**** Starting file " ^ quote name ^ " ****");
        run_file path;
        priority ("**** Finished file " ^ quote name ^ " ****\n")))
    end
  else run_file path;

fun exec_file path = ML_Context.exec (fn () => load_file path);

val use = load_file o Path.explode;

end;


(* load_thy *)

fun required_by _ [] = ""
  | required_by s initiators = s ^ "(required by " ^ show_path (rev initiators) ^ ")";

fun load_thy upd_time initiators name =
  let
    val _ = priority ("Loading theory " ^ quote name ^ required_by " " initiators);
    fun corrupted () = error (loader_msg "corrupted dependency information" [name]);
    val (pos, text, _) =
      (case get_deps name of
        SOME {master = SOME (master_path, _), text, files, ...} =>
          if text = "" then corrupted ()
          else (Path.position master_path, text, files)
      | _ => corrupted ());
    val _ = touch_thy name;
    val _ = CRITICAL (fn () =>
      change_deps name (Option.map (fn {master, text, parents, files, ...} =>
        make_deps upd_time master text parents files)));
    val after_load = Outer_Syntax.load_thy name pos text;
    val _ =
      CRITICAL (fn () =>
       (change_deps name
          (Option.map (fn {update_time, master, parents, files, ...} =>
            make_deps update_time master "" parents files));
        perform Update name));
  in after_load end;


(* scheduling loader tasks *)

datatype task = Task of (unit -> unit -> unit) | Finished | Running;
fun task_finished Finished = true | task_finished _ = false;

local

fun schedule_futures task_graph = uninterruptible (fn _ => fn () =>
  let
    val tasks = Graph.topological_order task_graph |> map_filter (fn name =>
      (case Graph.get_node task_graph name of Task body => SOME (name, body) | _ => NONE));

    val par_proofs = ! parallel_proofs >= 1;

    fun fork (name, body) tab =
      let
        val deps = Graph.imm_preds task_graph name
          |> map_filter (fn parent =>
            (case Symtab.lookup tab parent of SOME future => SOME (parent, future) | NONE => NONE));
        fun failed (parent, future) = if can Future.join future then NONE else SOME parent;

        val future = Future.fork_deps (map #2 deps) (fn () =>
          (case map_filter failed deps of
            [] => body ()
          | bad => error (loader_msg
              ("failed to load " ^ quote name ^ " (unresolved " ^ commas_quote bad ^ ")") [])));
        val future' =
          if par_proofs then future
          else Future.map (fn after_load => (after_load (); fn () => ())) future;
      in Symtab.update (name, future') tab end;

    val futures = fold fork tasks Symtab.empty;

    val failed = tasks |> maps (fn (name, _) =>
      let
        val after_load = Future.join (the (Symtab.lookup futures name));
        val _ = join_thy name;
        val _ = after_load ();
      in [] end handle exn => [(name, exn)]) |> rev;

    val _ = List.app (kill_thy o #1) failed;
    val _ = Exn.release_all (map (Exn.Exn o #2) failed);
  in () end) ();

fun schedule_seq tasks =
  Graph.topological_order tasks
  |> List.app (fn name =>
    (case Graph.get_node tasks name of
      Task body =>
        let val after_load = body ()
        in after_load () handle exn => (kill_thy name; reraise exn) end
    | _ => ()));

in

fun schedule_tasks tasks =
  if not (Multithreading.enabled ()) then schedule_seq tasks
  else if Multithreading.self_critical () then
     (warning (loader_msg "no multithreading within critical section" []);
      schedule_seq tasks)
  else schedule_futures tasks;

end;


(* require_thy -- checking database entries wrt. the file-system *)

local

fun check_ml master (src_path, info) =
  let val info' =
    (case info of NONE => NONE
    | SOME (_, id) =>
        (case Thy_Load.check_ml (master_dir master) src_path of NONE => NONE
        | SOME (path', id') => if id <> id' then NONE else SOME (path', id')))
  in (src_path, info') end;

fun check_deps dir name =
  (case lookup_deps name of
    SOME NONE => (true, NONE, get_parents name)
  | NONE =>
      let val {master, text, imports = parents, uses = files} = Thy_Load.deps_thy dir name
      in (false, init_deps (SOME master) text parents files, parents) end
  | SOME (SOME {update_time, master, text, parents, files}) =>
      let
        val (thy_path, thy_id) = Thy_Load.check_thy dir name;
        val master' = SOME (thy_path, thy_id);
      in
        if Option.map #2 master <> SOME thy_id then
          let val {text = text', imports = parents', uses = files', ...} =
            Thy_Load.deps_thy dir name;
          in (false, init_deps master' text' parents' files', parents') end
        else
          let
            val files' = map (check_ml master') files;
            val current = update_time >= 0 andalso can get_theory name
              andalso forall (is_some o snd) files';
            val update_time' = if current then update_time else ~1;
            val deps' = SOME (make_deps update_time' master' text parents files');
          in (current, deps', parents) end
      end);

fun read_text (SOME {update_time, master = master as SOME (path, _), text = _, parents, files}) =
  SOME (make_deps update_time master (File.read path) parents files);

in

fun require_thys initiators dir strs tasks =
      fold_map (require_thy initiators dir) strs tasks |>> forall I
and require_thy initiators dir str tasks =
  let
    val path = Path.expand (Path.explode str);
    val name = Path.implode (Path.base path);
    val dir' = Path.append dir (Path.dir path);
    val _ = member (op =) initiators name andalso error (cycle_msg initiators);
  in
    (case try (Graph.get_node tasks) name of
      SOME task => (task_finished task, tasks)
    | NONE =>
        let
          val (current, deps, parents) = check_deps dir' name
            handle ERROR msg => cat_error msg
              (loader_msg "the error(s) above occurred while examining theory" [name] ^
                required_by "\n" initiators);
          val parent_names = map base_name parents;

          val (parents_current, tasks_graph') =
            require_thys (name :: initiators) (Path.append dir (master_dir' deps)) parents tasks;

          val all_current = current andalso parents_current;
          val _ = if not all_current andalso known_thy name then outdate_thy name else ();
          val entry =
            if all_current then (deps, SOME (get_theory name))
            else (read_text deps, NONE);
          val _ = change_thys (new_deps name parent_names entry);

          val upd_time = serial ();
          val tasks_graph'' = tasks_graph' |> new_deps name parent_names
           (if all_current then Finished
            else Task (fn () => load_thy upd_time initiators name));
        in (all_current, tasks_graph'') end)
  end;

end;


(* use_thy etc. *)

fun use_thys_dir dir arg =
  schedule_tasks (snd (require_thys [] dir arg Graph.empty));

val use_thys = use_thys_dir Path.current;

fun use_thy str =
  let
    val name = base_name str;
    val _ = check_unfinished warning name;
  in use_thys [str] end;


(* begin / end theory *)

fun begin_theory name parents uses int =
  let
    val parent_names = map base_name parents;
    val dir = master_dir'' (lookup_deps name);
    val _ = check_unfinished error name;
    val _ = if int then use_thys_dir dir parents else ();

    val theory = Theory.begin_theory name (map get_theory parent_names);

    val deps =
      if known_thy name then get_deps name
      else init_deps NONE "" parents (map #1 uses);
    val _ = change_thys (new_deps name parent_names (deps, NONE));

    val update_time = (case deps of NONE => 0 | SOME {update_time, ...} => update_time);
    val update_time = if update_time > 0 then update_time else serial ();
    val theory' = theory
      |> Management_Data.put (Future.worker_group (), update_time)
      |> Present.begin_theory update_time dir uses;

    val uses_now = map_filter (fn (x, true) => SOME x | _ => NONE) uses;
    val theory'' =
      fold (fn x => Context.theory_map (exec_file x) o Theory.checkpoint) uses_now theory';
  in theory'' end;

fun end_theory theory =
  let
    val name = Context.theory_name theory;
    val _ = check_files name;
    val theory' = Theory.end_theory theory;
    val _ = change_thy name (fn (deps, _) => (deps, SOME theory'));
  in () end;


(* register existing theories *)

fun register_thy name =
  let
    val _ = priority ("Registering theory " ^ quote name);
    val thy = get_theory name;
    val _ = map get_theory (get_parents name);
    val _ = check_unfinished error name;
    val _ = touch_thy name;
    val master = #master (Thy_Load.deps_thy Path.current name);
    val upd_time = #2 (Management_Data.get thy);
  in
    CRITICAL (fn () =>
     (change_deps name (Option.map
       (fn {parents, files, ...} => make_deps upd_time (SOME master) "" parents files));
      perform Update name))
  end;

fun register_theory theory =
  let
    val name = Context.theory_name theory;
    val parents = Theory.parents_of theory;
    val parent_names = map Context.theory_name parents;

    fun err txt bads =
      error (loader_msg txt bads ^ "\ncannot register theory " ^ quote name);

    val nonfinished = filter_out is_finished parent_names;
    fun get_variant (x, y_name) =
      if Theory.eq_thy (x, get_theory y_name) then NONE
      else SOME y_name;
    val variants = map_filter get_variant (parents ~~ parent_names);

    fun register G =
      (Graph.new_node (name, (NONE, SOME theory)) G
        handle Graph.DUP _ => err "duplicate theory entry" [])
      |> add_deps name parent_names;
  in
    if not (null nonfinished) then err "non-finished parent theories" nonfinished
    else if not (null variants) then err "different versions of parent theories" variants
    else CRITICAL (fn () => (change_thys register; perform Update name))
  end;


(* finish all theories *)

fun finish () = change_thys (Graph.map_nodes (fn (_, entry) => (NONE, entry)));

end;