src/Pure/Tools/codegen_names.ML

new SML serializer

(*  Title:      Pure/Tools/codegen_names.ML
    ID:         $Id$
    Author:     Florian Haftmann, TU Muenchen

Name policies for code generation: prefixing any name by corresponding theory name,
conversion to alphanumeric representation, shallow name spaces.
Mappings are incrementally cached.
*)

signature CODEGEN_NAMES =
sig
  type tyco = string
  type const = string
  val class: theory -> class -> class
  val class_rev: theory -> class -> class option
  val tyco: theory -> tyco -> tyco
  val tyco_rev: theory -> tyco -> tyco option
  val tyco_force: tyco * string -> theory -> theory
  val instance: theory -> class * tyco -> string
  val instance_rev: theory -> string -> (class * tyco) option
  val instance_force: (class * tyco) * string -> theory -> theory
  val const: theory -> CodegenConsts.const -> const
  val const_rev: theory -> const -> CodegenConsts.const option
  val const_force: CodegenConsts.const * const -> theory -> theory
  val purify_var: string -> string
  val check_modulename: string -> string
  val has_nsp: string -> string -> bool
  val nsp_module: string
  val nsp_class: string
  val nsp_tyco: string
  val nsp_inst: string
  val nsp_fun: string
  val nsp_dtco: string
  val nsp_eval: string
end;

structure CodegenNames: CODEGEN_NAMES =
struct


(* theory data *)

type tyco = string;
type const = string;
val inst_ord = prod_ord fast_string_ord fast_string_ord;
val eq_inst = is_equal o inst_ord;
structure Consttab = CodegenConsts.Consttab;

structure Insttab =
  TableFun(
    type key = class * tyco;
    val ord = inst_ord;
  );

datatype names = Names of {
  class: class Symtab.table * class Symtab.table,
  tyco: tyco Symtab.table * tyco Symtab.table,
  instance: string Insttab.table * (class * tyco) Symtab.table,
  const: const Consttab.table * (string * typ list) Symtab.table
}

val empty_names = Names {
  class = (Symtab.empty, Symtab.empty),
  tyco = (Symtab.empty, Symtab.empty),
  instance = (Insttab.empty, Symtab.empty),
  const = (Consttab.empty, Symtab.empty)
};

local
  fun mk_names (class, tyco, instance, const) =
    Names { class = class, tyco = tyco, instance = instance, const = const};
  fun map_names f (Names { class, tyco, instance, const }) =
    mk_names (f (class, tyco, instance, const));
  val eq_string = op = : string * string -> bool;
in
  fun merge_names (Names { class = (class1, classrev1), tyco = (tyco1, tycorev1),
      instance = (instance1, instancerev1), const = (const1, constrev1) },
    Names { class = (class2, classrev2), tyco = (tyco2, tycorev2),
      instance = (instance2, instancerev2), const = (const2, constrev2) }) =
    mk_names ((Symtab.merge eq_string (class1, class2), Symtab.merge eq_string (classrev1, classrev2)),
      (Symtab.merge eq_string (tyco1, tyco2), Symtab.merge eq_string (tycorev1, tycorev2)),
      (Insttab.merge eq_string (instance1, instance2), Symtab.merge eq_inst (instancerev1, instancerev2)),
      (Consttab.merge eq_string (const1, const2), Symtab.merge CodegenConsts.eq_const (constrev1, constrev2)));
  fun map_class f = map_names
    (fn (class, tyco, inst, const) => (f class, tyco, inst, const));
  fun map_tyco f = map_names
    (fn (class, tyco, inst, const) => (class, f tyco, inst, const));
  fun map_inst f = map_names
    (fn (class, tyco, inst, const) => (class, tyco, f inst, const));
  fun map_const f = map_names
    (fn (class, tyco, inst, const) => (class, tyco, inst, f const));

end; (*local*)

structure CodeName = TheoryDataFun
(struct
  val name = "Pure/codegen_names";
  type T = names ref;
  val empty = ref empty_names;
  fun copy (ref names) = ref names;
  val extend = copy;
  fun merge _ (ref names1, ref names2) = ref (merge_names (names1, names2));
  fun print _ _ = ();
end);

val _ = Context.add_setup CodeName.init;


(* forward lookup with cache update *)

fun get thy get_tabs get upd_names upd policy x =
  let
    val names_ref = CodeName.get thy
    val (Names names) = ! names_ref;
    fun mk_unique used name =
      let
        fun mk_name 0 = name
          | mk_name i = name ^ "_" ^ string_of_int i
        fun find_name i =
          let
            val name = mk_name i
          in
            if used name
            then find_name (i+1)
            else name
          end;
        val name = find_name 0;
      in name end;
    val tabs = get_tabs names;
    fun declare name =
      let
        val names' = upd_names (K (upd (x, name) (fst tabs),
          Symtab.update_new (name, x) (snd tabs))) (Names names)
      in (names_ref := names'; name) end;
  in case get (fst tabs) x
   of SOME name => name
    | NONE => let
        val used = Symtab.defined (snd tabs);
        val raw_name = policy thy x;
        val name = mk_unique used raw_name;
      in declare name end
  end;


(* backward lookup *)

fun rev thy get_tabs errname name =
  let
    val names_ref = CodeName.get thy
    val (Names names) = ! names_ref;
    val tab = (snd o get_tabs) names;
  in case Symtab.lookup tab name
   of SOME x => x
    | NONE => error ("No such " ^ errname ^ ": " ^ quote name)
  end;


(* theory name lookup *)

fun thyname_of thy f errmsg x =
  let
    fun thy_of thy =
      if f thy x then case get_first thy_of (Theory.parents_of thy)
        of NONE => SOME thy
         | thy => thy
      else NONE;
  in case thy_of thy
   of SOME thy => Context.theory_name thy
    | NONE => error (errmsg x) end;

fun thyname_of_class thy =
  thyname_of thy (fn thy => member (op =) (Sign.classes thy))
    (fn class => "thyname_of_class: no such class: " ^ quote class);

fun thyname_of_tyco thy =
  thyname_of thy Sign.declared_tyname
    (fn tyco => "thyname_of_tyco: no such type constructor: " ^ quote tyco);

fun thyname_of_instance thy =
  let
    fun test_instance thy (class, tyco) =
      can (Sorts.mg_domain (Sign.classes_of thy) tyco) [class]
  in thyname_of thy test_instance
    (fn (class, tyco) => "thyname_of_instance: no such instance: " ^ quote class ^ ", " ^ quote tyco)
  end;

fun thyname_of_const thy =
  thyname_of thy Sign.declared_const
    (fn c => "thyname_of_const: no such constant: " ^ quote c);


(* purification of identifiers *)

val purify_name =
  let
    fun is_valid s = Symbol.is_ascii_letter s orelse Symbol.is_ascii_digit s orelse s = "'";
    val is_junk = not o is_valid andf Symbol.not_eof;
    val junk = Scan.any is_junk;
    val scan_valids = Symbol.scanner "Malformed input"
      ((junk |--
        (Scan.optional (Scan.one Symbol.is_ascii_letter) "x" ^^ (Scan.any is_valid >> implode)
        --| junk))
      -- Scan.repeat ((Scan.any1 is_valid >> implode) --| junk) >> op ::);
  in explode #> scan_valids #> space_implode "_" end;

fun purify_op "0" = "zero"
  | purify_op "1" = "one"
  | purify_op s =
      let
        fun rep_op "+" = SOME "sum"
          | rep_op "-" = SOME "diff"
          | rep_op "*" = SOME "prod"
          | rep_op "/" = SOME "quotient"
          | rep_op "&" = SOME "conj"
          | rep_op "|" = SOME "disj"
          | rep_op "=" = SOME "eq"
          | rep_op "~" = SOME "inv"
          | rep_op "@" = SOME "append"
          | rep_op s = NONE;
        val scan_valids = Symbol.scanner "Malformed input"
           (Scan.repeat (Scan.some rep_op || Scan.one Symbol.not_eof));
      in (explode #> scan_valids #> implode) s end;

val purify_lower =
  explode
  #> (fn cs => (if forall Symbol.is_ascii_upper cs
        then map else nth_map 0) Symbol.to_ascii_lower cs)
  #> implode;

val purify_upper = explode #> nth_map 0 Symbol.to_ascii_upper #> implode;

fun check_modulename mn =
  let
    val mns = NameSpace.unpack mn;
    val mns' = map purify_upper mns;
  in
    if mns' = mns then mn else error ("Invalid module name: " ^ quote mn ^ "\n"
      ^ "perhaps try " ^ quote (NameSpace.pack mns'))
  end


fun purify_var "" = "x"
  | purify_var v =
      if nth (explode v) 0 = "'"
      then (unprefix "'" #> purify_name #> purify_lower #> prefix "'") v
      else (purify_name #> purify_lower) v;

val purify_idf = purify_op #> purify_name;
val purify_prefix = map (purify_idf #> purify_upper);
val purify_base = purify_idf #> purify_lower;


(* explicit given names with cache update *)

fun force get defined upd_names upd errname string_of (x, name) thy =
  let
    val names = NameSpace.unpack name;
    val (prefix, base) = split_last (NameSpace.unpack name);
    val prefix' = purify_prefix prefix;
    val base' = purify_base base;
    val _ = if (base' = base) andalso forall (op =) (prefix' ~~ prefix)
      then ()
      else
        error ("Name violates naming conventions: " ^ quote name
          ^ "; perhaps try with " ^ quote (NameSpace.pack (prefix' @ [base'])))
    val names_ref = CodeName.get thy;
    val (Names names) = ! names_ref;
    val (tab, rev) = get names;
    val _ = if defined tab x
      then error ("Already named " ^ errname  ^ ": " ^ string_of thy x)
      else ();
    val _ = if Symtab.defined rev name
      then error ("Name already given to other " ^ errname ^ ": " ^ quote name)
      else ();
    val _ = names_ref := upd_names (K (upd (x, name) tab,
          Symtab.update_new (name, x) rev)) (Names names);
  in
    thy
  end;

val tyco_force = force #tyco Symtab.defined map_tyco Symtab.update_new
  "type constructor" (K quote);
val instance_force = force #instance Insttab.defined map_inst Insttab.update_new
  "instance" (fn _ => fn (class, tyco) => quote class ^ ", " ^ quote tyco);
val const_force = force #const Consttab.defined map_const Consttab.update_new
  "constant" (quote oo CodegenConsts.string_of_const);


(* naming policies *)

fun policy thy get_basename get_thyname name =
  let
    val prefix = (purify_prefix o NameSpace.unpack o get_thyname thy) name;
    val base = (purify_base o get_basename) name;
  in NameSpace.pack (prefix @ [base]) end;

fun class_policy thy = policy thy NameSpace.base thyname_of_class;
fun tyco_policy thy = policy thy NameSpace.base thyname_of_tyco;
fun instance_policy thy = policy thy (fn (class, tyco) => 
  NameSpace.base class ^ "_" ^ NameSpace.base tyco) thyname_of_instance;

fun force_thyname thy (c, tys) =
  case AxClass.class_of_param thy c
   of SOME class => (case tys
       of [Type (tyco, _)] => SOME (thyname_of_instance thy (class, tyco))
        | _ => NONE)
    | NONE => (case CodegenConsts.find_def thy (c, tys)
   of SOME ((thyname, _), _) => SOME thyname
    | NONE => NONE);

fun const_policy thy (c, tys) =
  case force_thyname thy (c, tys)
   of NONE => policy thy NameSpace.base thyname_of_const c
    | SOME thyname => let
        val prefix = (purify_prefix o NameSpace.unpack) thyname;
        val tycos = map_filter (fn Type (tyco, _) => SOME tyco | _ => NONE) tys;
        val base = map (purify_base o NameSpace.base) (c :: tycos);
      in NameSpace.pack (prefix @ [space_implode "_" base]) end;


(* shallow name spaces *)

val nsp_module = ""; (*a dummy by convention*)
val nsp_class = "class";
val nsp_tyco = "tyco";
val nsp_inst = "inst";
val nsp_fun = "fun";
val nsp_dtco = "dtco";
val nsp_eval = "EVAL"; (*only for evaluation*)

fun add_nsp shallow name =
  name
  |> NameSpace.unpack
  |> split_last
  |> apsnd (single #> cons shallow)
  |> (op @)
  |> NameSpace.pack;

fun dest_nsp nsp nspname =
  let
    val xs = NameSpace.unpack nspname;
    val (ys, base) = split_last xs;
    val (module, shallow) = split_last ys;
  in
    if nsp = shallow
   then (SOME o NameSpace.pack) (module @ [base])
    else NONE
  end;

val has_nsp = is_some oo dest_nsp;

fun if_nsp nsp f idf =
  Option.map f (dest_nsp nsp idf);


(* external interfaces *)

fun class thy =
  get thy #class Symtab.lookup map_class Symtab.update class_policy
  #> add_nsp nsp_class;
fun tyco thy =
  get thy #tyco Symtab.lookup map_tyco Symtab.update tyco_policy
  #> add_nsp nsp_tyco;
fun instance thy =
  get thy #instance Insttab.lookup map_inst Insttab.update instance_policy
  #> add_nsp nsp_inst;
fun const thy c_ty = case CodegenConsts.norm thy c_ty
 of (c_tys as (c, tys)) => add_nsp (if (is_some o CodegenData.get_datatype_of_constr thy) c_tys
     then nsp_dtco
   else nsp_fun)
  (get thy #const Consttab.lookup map_const Consttab.update const_policy c_tys);

fun class_rev thy =
  dest_nsp nsp_class
  #> Option.map (rev thy #class "class");
fun tyco_rev thy =
  dest_nsp nsp_tyco
  #> Option.map (rev thy #tyco "type constructor");
fun instance_rev thy =
  dest_nsp nsp_inst
  #> Option.map (rev thy #instance "instance");
fun const_rev thy nspname =
  (case dest_nsp nsp_fun nspname
   of name as SOME _ => name
    | _ => (case dest_nsp nsp_dtco nspname
   of name as SOME _ => name
    | _ => NONE))
  |> Option.map (rev thy #const "constant");


(* outer syntax *)

local

structure P = OuterParse
and K = OuterKeyword

fun const_force_e (raw_c, name) thy =
  const_force (CodegenConsts.read_const thy raw_c, name) thy;

fun tyco_force_e (raw_tyco, name) thy =
  tyco_force (Sign.intern_type thy raw_tyco, name) thy;

fun instance_force_e ((raw_tyco, raw_class), name) thy =
  instance_force ((Sign.intern_class thy raw_class, Sign.intern_type thy raw_tyco), name) thy;

val (constnameK, typenameK, instnameK) = ("code_constname", "code_typename", "code_instname");

val constnameP =
  OuterSyntax.improper_command constnameK "declare code name for constant" K.thy_decl (
    Scan.repeat1 (P.term -- P.name)
    >> (fn aliasses =>
          Toplevel.theory (fold const_force_e aliasses))
  );

val typenameP =
  OuterSyntax.improper_command typenameK "declare code name for type constructor" K.thy_decl (
    Scan.repeat1 (P.xname -- P.name)
    >> (fn aliasses =>
          Toplevel.theory (fold tyco_force_e aliasses))
  );

val instnameP =
  OuterSyntax.improper_command instnameK "declare code name for instance relation" K.thy_decl (
    Scan.repeat1 ((P.xname --| P.$$$ "::" -- P.xname) -- P.name)
    >> (fn aliasses =>
          Toplevel.theory (fold instance_force_e aliasses))
  );

in

val _ = OuterSyntax.add_parsers [constnameP, typenameP, instnameP];


end; (*local*)

end;