src/Tools/Code/code_namespace.ML

formal markup for public ingredients

(*  Title:      Tools/Code/code_namespace.ML
    Author:     Florian Haftmann, TU Muenchen

Mastering target language namespaces.
*)

signature CODE_NAMESPACE =
sig
  type flat_program
  val flat_program: Proof.context
    -> { module_prefix: string, module_name: string,
    reserved: Name.context, identifiers: Code_Target.identifier_data, empty_nsp: 'a,
    namify_stmt: Code_Thingol.stmt -> string -> 'a -> string * 'a,
    modify_stmt: Code_Thingol.stmt -> Code_Thingol.stmt option }
      -> Code_Thingol.program
      -> { deresolver: string -> Code_Symbol.T -> string,
           flat_program: flat_program }

  datatype ('a, 'b) node =
      Dummy
    | Stmt of bool * 'a
    | Module of ('b * (string * ('a, 'b) node) Code_Symbol.Graph.T)
  type ('a, 'b) hierarchical_program
  val hierarchical_program: Proof.context
    -> { module_name: string,
    reserved: Name.context, identifiers: Code_Target.identifier_data,
    empty_nsp: 'c, namify_module: string -> 'c -> string * 'c,
    namify_stmt: Code_Thingol.stmt -> string -> 'c -> string * 'c,
    cyclic_modules: bool, empty_data: 'b, memorize_data: Code_Symbol.T -> 'b -> 'b,
    modify_stmts: (Code_Symbol.T * Code_Thingol.stmt) list -> 'a option list }
      -> Code_Thingol.program
      -> { deresolver: string list -> Code_Symbol.T -> string,
           hierarchical_program: ('a, 'b) hierarchical_program }
  val print_hierarchical: { print_module: string list -> string -> 'b -> 'c list -> 'c,
    print_stmt: string list -> Code_Symbol.T * (bool * 'a) -> 'c,
    lift_markup: (Pretty.T -> Pretty.T) -> 'c -> 'c }
      -> ('a, 'b) hierarchical_program -> 'c list
end;

structure Code_Namespace : CODE_NAMESPACE =
struct

(** fundamental module name hierarchy **)

fun module_fragments' { identifiers, reserved } name =
  case Code_Symbol.lookup_module_data identifiers name of
      SOME (fragments, _) => fragments
    | NONE => map (fn fragment => fst (Name.variant fragment reserved)) (Long_Name.explode name);

fun module_fragments { module_name, identifiers, reserved } =
  if module_name = ""
  then module_fragments' { identifiers = identifiers, reserved = reserved }
  else K (Long_Name.explode module_name);

fun build_module_namespace ctxt { module_prefix, module_name, identifiers, reserved } program =
  let
    val module_names = Code_Symbol.Graph.fold (insert (op =) o Code_Symbol.default_prefix ctxt o fst) program [];
    val module_fragments' = module_fragments
      { module_name = module_name, identifiers = identifiers, reserved = reserved };
  in
    fold (fn name => Symtab.update (name, Long_Name.explode module_prefix @ module_fragments' name))
      module_names Symtab.empty
  end;

fun prep_symbol ctxt { module_namespace, force_module, identifiers } sym =
  case Code_Symbol.lookup identifiers sym of
      NONE => ((the o Symtab.lookup module_namespace o Code_Symbol.default_prefix ctxt) sym,
        Code_Symbol.default_base sym)
    | SOME prefix_name => if null force_module then prefix_name
        else (force_module, snd prefix_name);

fun has_priority identifiers = is_some o Code_Symbol.lookup identifiers;

fun build_proto_program { empty, add_stmt, add_dep } program =
  empty
  |> Code_Symbol.Graph.fold (fn (sym, (stmt, _)) => add_stmt sym stmt) program
  |> Code_Symbol.Graph.fold (fn (sym, (_, (_, syms))) =>
      Code_Symbol.Graph.Keys.fold (add_dep sym) syms) program;

fun prioritize has_priority = uncurry append o List.partition has_priority;


(** flat program structure **)

type flat_program = ((string * (bool * Code_Thingol.stmt) option) Code_Symbol.Graph.T * (string * Code_Symbol.T list) list) Graph.T;

fun flat_program ctxt { module_prefix, module_name, reserved,
    identifiers, empty_nsp, namify_stmt, modify_stmt } program =
  let

    (* building module name hierarchy *)
    val module_namespace = build_module_namespace ctxt { module_prefix = module_prefix,
      module_name = module_name, identifiers = identifiers, reserved = reserved } program;
    val prep_sym = prep_symbol ctxt { module_namespace = module_namespace,
      force_module = Long_Name.explode module_name, identifiers = identifiers }
      #>> Long_Name.implode;
    val sym_priority = has_priority identifiers;

    (* distribute statements over hierarchy *)
    fun add_stmt sym stmt =
      let
        val (module_name, base) = prep_sym sym;
      in
        Graph.default_node (module_name, (Code_Symbol.Graph.empty, []))
        #> (Graph.map_node module_name o apfst) (Code_Symbol.Graph.new_node (sym, (base, (true, stmt))))
      end;
    fun add_dep sym sym' =
      let
        val (module_name, _) = prep_sym sym;
        val (module_name', _) = prep_sym sym';
      in if module_name = module_name'
        then (Graph.map_node module_name o apfst) (Code_Symbol.Graph.add_edge (sym, sym'))
        else (Graph.map_node module_name o apsnd) (AList.map_default (op =) (module_name', []) (insert (op =) sym'))
      end;
    val proto_program = build_proto_program
      { empty = Graph.empty, add_stmt = add_stmt, add_dep = add_dep } program;

    (* name declarations and statement modifications *)
    fun declare sym (base, (_, stmt)) (gr, nsp) = 
      let
        val (base', nsp') = namify_stmt stmt base nsp;
        val gr' = (Code_Symbol.Graph.map_node sym o apfst) (K base') gr;
      in (gr', nsp') end;
    fun declarations gr = (gr, empty_nsp)
      |> fold (fn sym => declare sym (Code_Symbol.Graph.get_node gr sym))
          (prioritize sym_priority (Code_Symbol.Graph.keys gr))
      |> fst
      |> (Code_Symbol.Graph.map o K o apsnd)
        (fn (export, stmt) => Option.map (pair export) (modify_stmt stmt));
    val flat_program = proto_program
      |> (Graph.map o K o apfst) declarations;

    (* qualified and unqualified imports, deresolving *)
    fun base_deresolver sym = fst (Code_Symbol.Graph.get_node
      (fst (Graph.get_node flat_program (fst (prep_sym sym)))) sym);
    fun classify_names gr imports =
      let
        val import_tab = maps
          (fn (module_name, syms) => map (rpair module_name) syms) imports;
        val imported_syms = map fst import_tab;
        val here_syms = Code_Symbol.Graph.keys gr;
      in
        Code_Symbol.Table.empty
        |> fold (fn sym => Code_Symbol.Table.update (sym, base_deresolver sym)) here_syms
        |> fold (fn sym => Code_Symbol.Table.update (sym,
            Long_Name.append (the (AList.lookup (op =) import_tab sym))
              (base_deresolver sym))) imported_syms
      end;
    val deresolver_tab = Symtab.make (AList.make
      (uncurry classify_names o Graph.get_node flat_program)
        (Graph.keys flat_program));
    fun deresolver "" sym =
          Long_Name.append (fst (prep_sym sym)) (base_deresolver sym)
      | deresolver module_name sym =
          the (Code_Symbol.Table.lookup (the (Symtab.lookup deresolver_tab module_name)) sym)
          handle Option.Option => error ("Unknown statement name: "
            ^ Code_Symbol.quote ctxt sym);

  in { deresolver = deresolver, flat_program = flat_program } end;


(** hierarchical program structure **)

datatype ('a, 'b) node =
    Dummy
  | Stmt of bool * 'a
  | Module of ('b * (string * ('a, 'b) node) Code_Symbol.Graph.T);

type ('a, 'b) hierarchical_program = (string * ('a, 'b) node) Code_Symbol.Graph.T;

fun the_stmt (Stmt (export, stmt)) = (export, stmt);

fun map_module_content f (Module content) = Module (f content);

fun map_module [] = I
  | map_module (name_fragment :: name_fragments) =
      apsnd o Code_Symbol.Graph.map_node (Code_Symbol.Module name_fragment) o apsnd o map_module_content
        o map_module name_fragments;

fun hierarchical_program ctxt { module_name, reserved, identifiers, empty_nsp,
      namify_module, namify_stmt, cyclic_modules, empty_data, memorize_data, modify_stmts } program =
  let

    (* building module name hierarchy *)
    val module_namespace = build_module_namespace ctxt { module_prefix = "",
      module_name = module_name, identifiers = identifiers, reserved = reserved } program;
    val prep_sym = prep_symbol ctxt { module_namespace = module_namespace,
      force_module = Long_Name.explode module_name, identifiers = identifiers }
    val sym_priority = has_priority identifiers;

    (* building empty module hierarchy *)
    val empty_module = (empty_data, Code_Symbol.Graph.empty);
    fun ensure_module name_fragment (data, nodes) =
      if can (Code_Symbol.Graph.get_node nodes) (Code_Symbol.Module name_fragment) then (data, nodes)
      else (data,
        nodes |> Code_Symbol.Graph.new_node (Code_Symbol.Module name_fragment, (name_fragment, Module empty_module)));
    fun allocate_module [] = I
      | allocate_module (name_fragment :: name_fragments) =
          ensure_module name_fragment
          #> (apsnd o Code_Symbol.Graph.map_node (Code_Symbol.Module name_fragment) o apsnd o map_module_content o allocate_module) name_fragments;
    val empty_program =
      empty_module
      |> Symtab.fold (fn (_, fragments) => allocate_module fragments) module_namespace
      |> Code_Symbol.Graph.fold (allocate_module o these o Option.map fst
          o Code_Symbol.lookup identifiers o fst) program;

    (* distribute statements over hierarchy *)
    fun add_stmt sym stmt =
      let
        val (name_fragments, base) = prep_sym sym;
      in
        (map_module name_fragments o apsnd) (Code_Symbol.Graph.new_node (sym, (base, Stmt (true, stmt))))
      end;
    fun add_dep sym sym' =
      let
        val (name_fragments, _) = prep_sym sym;
        val (name_fragments', _) = prep_sym sym';
        val (name_fragments_common, (diff, diff')) =
          chop_prefix (op =) (name_fragments, name_fragments');
        val is_module = not (null diff andalso null diff');
        val dep = pairself hd (map Code_Symbol.Module diff @ [sym], map Code_Symbol.Module diff' @ [sym']);
        val add_edge = if is_module andalso not cyclic_modules
          then (fn node => Code_Symbol.Graph.add_edge_acyclic dep node
            handle Graph.CYCLES _ => error ("Dependency "
              ^ Code_Symbol.quote ctxt sym ^ " -> "
              ^ Code_Symbol.quote ctxt sym'
              ^ " would result in module dependency cycle"))
          else Code_Symbol.Graph.add_edge dep
      in (map_module name_fragments_common o apsnd) add_edge end;
    val proto_program = build_proto_program
      { empty = empty_program, add_stmt = add_stmt, add_dep = add_dep } program;

    (* name declarations, data and statement modifications *)
    fun make_declarations nsps (data, nodes) =
      let
        val (module_fragments, stmt_syms) =
          Code_Symbol.Graph.keys nodes
          |> List.partition
              (fn sym => case Code_Symbol.Graph.get_node nodes sym
                of (_, Module _) => true | _ => false)
          |> pairself (prioritize sym_priority)
        fun declare namify sym (nsps, nodes) =
          let
            val (base, node) = Code_Symbol.Graph.get_node nodes sym;
            val (base', nsps') = namify node base nsps;
            val nodes' = Code_Symbol.Graph.map_node sym (K (base', node)) nodes;
          in (nsps', nodes') end;
        val (nsps', nodes') = (nsps, nodes)
          |> fold (declare (K namify_module)) module_fragments
          |> fold (declare (namify_stmt o snd o the_stmt)) stmt_syms;
        fun select_syms syms = case filter (member (op =) stmt_syms) syms
         of [] => NONE
          | syms => SOME syms;
        fun select_stmt (sym, (export, SOME stmt)) = SOME (sym, (export, stmt))
          | select_stmt _ = NONE;
        fun modify_stmts' syms =
          let
            val (exports, stmts) = map_split (the_stmt o snd o Code_Symbol.Graph.get_node nodes) syms;
            val stmts' = modify_stmts (syms ~~ stmts);
            val stmts'' = stmts' @ replicate (length syms - length stmts') NONE;
          in map_filter select_stmt (syms ~~ (exports ~~ stmts'')) end;
        val stmtss' = (maps modify_stmts' o map_filter select_syms o Code_Symbol.Graph.strong_conn) nodes;
        val nodes'' = Code_Symbol.Graph.map (fn sym => apsnd (fn Module content => Module (make_declarations nsps' content)
            | _ => case AList.lookup (op =) stmtss' sym of SOME stmt => Stmt stmt | _ => Dummy)) nodes';
        val data' = fold memorize_data stmt_syms data;
      in (data', nodes'') end;
    val (_, hierarchical_program) = make_declarations empty_nsp proto_program;

    (* deresolving *)
    fun deresolver prefix_fragments sym =
      let
        val (name_fragments, _) = prep_sym sym;
        val (_, (_, remainder)) = chop_prefix (op =) (prefix_fragments, name_fragments);
        val nodes = fold (fn name_fragment => fn nodes => case Code_Symbol.Graph.get_node nodes (Code_Symbol.Module name_fragment)
         of (_, Module (_, nodes)) => nodes) name_fragments hierarchical_program;
        val (base', _) = Code_Symbol.Graph.get_node nodes sym;
      in Long_Name.implode (remainder @ [base']) end
        handle Code_Symbol.Graph.UNDEF _ => error ("Unknown statement name: "
          ^ Code_Symbol.quote ctxt sym);

  in { deresolver = deresolver, hierarchical_program = hierarchical_program } end;

fun print_hierarchical { print_module, print_stmt, lift_markup } =
  let
    fun print_node _ (_, Dummy) =
          NONE
      | print_node prefix_fragments (sym, Stmt stmt) =
          SOME (lift_markup (Code_Printer.markup_stmt sym)
            (print_stmt prefix_fragments (sym, stmt)))
      | print_node prefix_fragments (Code_Symbol.Module name_fragment, Module (data, nodes)) =
          let
            val prefix_fragments' = prefix_fragments @ [name_fragment]
          in
            Option.map (print_module prefix_fragments'
              name_fragment data) (print_nodes prefix_fragments' nodes)
          end
    and print_nodes prefix_fragments nodes =
      let
        val xs = (map_filter (fn sym => print_node prefix_fragments
          (sym, snd (Code_Symbol.Graph.get_node nodes sym))) o rev o flat o Code_Symbol.Graph.strong_conn) nodes
      in if null xs then NONE else SOME xs end;
  in these o print_nodes [] end;

end;