open Printf
+open CicNotationEnv
+open CicNotationPt
+
exception Parse_error of Token.flocation * string
+exception Level_not_found of int
let grammar = Grammar.gcreate CicNotationLexer.notation_lexer
+let min_precedence = 0
+let max_precedence = 100
+let default_precedence = 50
+
let level1_pattern = Grammar.Entry.create grammar "level1_pattern"
let level2_pattern = Grammar.Entry.create grammar "level2_pattern"
let level3_term = Grammar.Entry.create grammar "level3_term"
+let l2_pattern = Grammar.Entry.create grammar "l2_pattern"
let notation = Grammar.Entry.create grammar "notation" (* level1 <-> level 2 *)
let interpretation =
Grammar.Entry.create grammar "interpretation" (* level2 <-> level 3 *)
+let phrase = Grammar.Entry.create grammar "phrase"
let return_term loc term = ()
-let loc_of_floc = function
- | { Lexing.pos_cnum = loc_begin }, { Lexing.pos_cnum = loc_end } ->
- (loc_begin, loc_end)
-
let fail floc msg =
let (x, y) = loc_of_floc floc in
failwith (sprintf "Error at characters %d - %d: %s" x y msg)
with Failure _ ->
failwith (sprintf "Lexer failure: string_of_int \"%s\" failed" s)
-open CicNotationPt
+(** {2 Grammar extension} *)
+
+let symbol s = Gramext.Stoken ("SYMBOL", s)
+let ident s = Gramext.Stoken ("IDENT", s)
+let number s = Gramext.Stoken ("NUMBER", s)
+let term = Gramext.Sself
+
+let g_symbol_of_literal =
+ function
+ | `Symbol s -> symbol s
+ | `Keyword s -> ident s
+ | `Number s -> number s
+
+type binding =
+ | NoBinding
+ | Binding of string * value_type
+ | Env of (string * value_type) list
+
+let make_action action bindings =
+ let rec aux (vl : CicNotationEnv.t) =
+ function
+ [] -> Gramext.action (fun (loc: location) -> action vl loc)
+ | NoBinding :: tl -> Gramext.action (fun _ -> aux vl tl)
+ (* LUCA: DEFCON 5 BEGIN *)
+ | Binding (name, TermType) :: tl ->
+ Gramext.action
+ (fun (v:term) -> aux ((name, (TermType, TermValue v))::vl) tl)
+ | Binding (name, StringType) :: tl ->
+ Gramext.action
+ (fun (v:string) ->
+ aux ((name, (StringType, StringValue v)) :: vl) tl)
+ | Binding (name, NumType) :: tl ->
+ Gramext.action
+ (fun (v:string) -> aux ((name, (NumType, NumValue v)) :: vl) tl)
+ | Binding (name, OptType t) :: tl ->
+ Gramext.action
+ (fun (v:'a option) ->
+ aux ((name, (OptType t, OptValue v)) :: vl) tl)
+ | Binding (name, ListType t) :: tl ->
+ Gramext.action
+ (fun (v:'a list) ->
+ aux ((name, (ListType t, ListValue v)) :: vl) tl)
+ | Env _ :: tl ->
+ Gramext.action (fun (v:CicNotationEnv.t) -> aux (v @ vl) tl)
+ (* LUCA: DEFCON 5 END *)
+ in
+ aux [] (List.rev bindings)
+
+let flatten_opt =
+ let rec aux acc =
+ function
+ [] -> List.rev acc
+ | NoBinding :: tl -> aux acc tl
+ | Env names :: tl -> aux (List.rev names @ acc) tl
+ | Binding (name, ty) :: tl -> aux ((name, ty) :: acc) tl
+ in
+ aux []
+
+ (* given a level 1 pattern computes the new RHS of "term" grammar entry *)
+let extract_term_production pattern =
+ let rec aux = function
+ | AttributedTerm (_, t) -> aux t
+ | Literal l -> aux_literal l
+ | Layout l -> aux_layout l
+ | Magic m -> aux_magic m
+ | Variable v -> aux_variable v
+ | t ->
+ prerr_endline (CicNotationPp.pp_term t);
+ assert false
+ and aux_literal =
+ function
+ | `Symbol s -> [NoBinding, symbol s]
+ | `Keyword s -> [NoBinding, ident s]
+ | `Number s -> [NoBinding, number s]
+ and aux_layout = function
+ | Sub (p1, p2) -> aux p1 @ [NoBinding, symbol "\\SUB"] @ aux p2
+ | Sup (p1, p2) -> aux p1 @ [NoBinding, symbol "\\SUP"] @ aux p2
+ | Below (p1, p2) -> aux p1 @ [NoBinding, symbol "\\BELOW"] @ aux p2
+ | Above (p1, p2) -> aux p1 @ [NoBinding, symbol "\\ABOVE"] @ aux p2
+ | Frac (p1, p2) -> aux p1 @ [NoBinding, symbol "\\FRAC"] @ aux p2
+ | Atop (p1, p2) -> aux p1 @ [NoBinding, symbol "\\ATOP"] @ aux p2
+ | Over (p1, p2) -> aux p1 @ [NoBinding, symbol "\\OVER"] @ aux p2
+ | Root (p1, p2) ->
+ [NoBinding, symbol "\\ROOT"] @ aux p2 @ [NoBinding, symbol "\\OF"]
+ @ aux p1
+ | Sqrt p -> [NoBinding, symbol "\\SQRT"] @ aux p
+ | Break -> []
+ | Box (_, pl) -> List.flatten (List.map aux pl)
+ and aux_magic magic =
+ match magic with
+ | Opt p ->
+ let p_bindings, p_atoms, p_names, p_action = inner_pattern p in
+ let action (env_opt : CicNotationEnv.t option) (loc : location) =
+ match env_opt with
+ | Some env -> List.map opt_binding_some env
+ | None -> List.map opt_binding_of_name p_names
+ in
+ [ Env (List.map opt_declaration p_names),
+ Gramext.srules
+ [ [ Gramext.Sopt (Gramext.srules [ p_atoms, p_action ]) ],
+ Gramext.action action ] ]
+ | List0 (p, _)
+ | List1 (p, _) ->
+ let p_bindings, p_atoms, p_names, p_action = inner_pattern p in
+(* let env0 = List.map list_binding_of_name p_names in
+ let grow_env_entry env n v =
+ List.map
+ (function
+ | (n', (ty, ListValue vl)) as entry ->
+ if n' = n then n', (ty, ListValue (v :: vl)) else entry
+ | _ -> assert false)
+ env
+ in
+ let grow_env env_i env =
+ List.fold_left
+ (fun env (n, (_, v)) -> grow_env_entry env n v)
+ env env_i
+ in *)
+ let action (env_list : CicNotationEnv.t list) (loc : location) =
+ CicNotationEnv.coalesce_env p_names env_list
+ in
+ let g_symbol s =
+ match magic with
+ | List0 (_, None) -> Gramext.Slist0 s
+ | List1 (_, None) -> Gramext.Slist1 s
+ | List0 (_, Some l) -> Gramext.Slist0sep (s, g_symbol_of_literal l)
+ | List1 (_, Some l) -> Gramext.Slist1sep (s, g_symbol_of_literal l)
+ | _ -> assert false
+ in
+ [ Env (List.map list_declaration p_names),
+ Gramext.srules
+ [ [ g_symbol (Gramext.srules [ p_atoms, p_action ]) ],
+ Gramext.action action ] ]
+ | _ -> assert false
+ and aux_variable =
+ function
+ | NumVar s -> [Binding (s, NumType), number ""]
+ | TermVar s -> [Binding (s, TermType), term]
+ | IdentVar s -> [Binding (s, StringType), ident ""]
+ | Ascription (p, s) -> assert false (* TODO *)
+ | FreshVar _ -> assert false
+ and inner_pattern p =
+ let p_bindings, p_atoms = List.split (aux p) in
+ let p_names = flatten_opt p_bindings in
+ let _ = prerr_endline ("inner names: " ^ String.concat " " (List.map fst p_names)) in
+ let action =
+ make_action (fun (env : CicNotationEnv.t) (loc : location) -> env)
+ p_bindings
+ in
+ p_bindings, p_atoms, p_names, action
+ in
+ aux pattern
-let boxify = function
- | [ a ] -> a
- | l -> Layout (Box (H, l))
+let level_of_int precedence =
+ if precedence < min_precedence || precedence > max_precedence then
+ raise (Level_not_found precedence);
+ string_of_int precedence
+
+type rule_id = Token.t Gramext.g_symbol list
+
+let extend level1_pattern ?(precedence = default_precedence)
+ ?associativity action
+=
+ let p_bindings, p_atoms =
+ List.split (extract_term_production level1_pattern)
+ in
+ let level = level_of_int precedence in
+ let p_names = flatten_opt p_bindings in
+ let _ =
+ prerr_endline (string_of_int (List.length p_bindings));
+ Grammar.extend
+ [ Grammar.Entry.obj (l2_pattern: 'a Grammar.Entry.e),
+ Some (Gramext.Level level),
+ [ None,
+ associativity,
+ [ p_atoms,
+ (make_action
+ (fun (env: CicNotationEnv.t) (loc: location) -> (action env loc))
+ p_bindings) ]]]
+ in
+ p_atoms
+
+let delete atoms = Grammar.delete_rule l2_pattern atoms
+
+(** {2 Grammar} *)
let fold_binder binder pt_names body =
- let fold_cluster binder names ty body =
+ let fold_cluster binder terms ty body =
List.fold_right
- (fun name body -> Binder (binder, (Cic.Name name, ty), body))
- names body
+ (fun term body -> Binder (binder, (term, ty), body))
+ terms body (* terms are names: either Ident or FreshVar *)
in
List.fold_right
(fun (names, ty) body -> fold_cluster binder names ty body)
let return_term loc term = AttributedTerm (`Loc loc, term)
+let _ = (* create empty precedence level for "l2_pattern" *)
+ let mk_level_list first last =
+ let rec aux acc = function
+ | i when i < first -> acc
+ | i -> aux ((Some (string_of_int i), None, []) :: acc) (i - 1)
+ in
+ aux [] last
+ in
+ Grammar.extend
+ [ Grammar.Entry.obj (l2_pattern: 'a Grammar.Entry.e),
+ None,
+ mk_level_list min_precedence max_precedence ]
+
EXTEND
GLOBAL: level1_pattern level2_pattern level3_term
- notation interpretation;
+ l2_pattern
+ notation interpretation
+ phrase;
(* {{{ Grammar for concrete syntax patterns, notation level 1 *)
- level1_pattern: [ [ p = l1_pattern; EOI -> boxify p ] ];
- l1_pattern: [ [ p = LIST0 l1_simple_pattern -> p ] ];
+ level1_pattern: [ [ p = l1_simple_pattern -> p ] ];
+ l1_pattern: [ [ p = LIST1 l1_simple_pattern -> p ] ];
literal: [
[ s = SYMBOL -> `Symbol s
| k = KEYWORD -> `Keyword k
]
];
l1_pattern_variable: [
- [ id = IDENT -> TermVar id
- | SYMBOL "\\TERM"; id = IDENT -> TermVar id
+ [ SYMBOL "\\TERM"; id = IDENT -> TermVar id
| SYMBOL "\\NUM"; id = IDENT -> NumVar id
| SYMBOL "\\IDENT"; id = IDENT -> IdentVar id
]
return_term loc (Layout (Below (p1, p2)))
| p1 = SELF; SYMBOL "\\ABOVE"; p2 = SELF ->
return_term loc (Layout (Above (p1, p2)))
- | SYMBOL "["; p1 = l1_pattern; SYMBOL "\\OVER"; p2 = l1_pattern;
- SYMBOL "]" ->
- return_term loc (Layout (Frac (boxify p1, boxify p2)))
- | SYMBOL "["; p1 = l1_pattern; SYMBOL "\\ATOP"; p2 = l1_pattern;
- SYMBOL "]" ->
- return_term loc (Layout (Atop (boxify p1, boxify p2)))
+ | p1 = SELF; SYMBOL "\\OVER"; p2 = SELF ->
+ return_term loc (Layout (Over (p1, p2)))
+ | p1 = SELF; SYMBOL "\\ATOP"; p2 = SELF ->
+ return_term loc (Layout (Atop (p1, p2)))
(* | SYMBOL "\\ARRAY"; p = SELF; csep = OPT field_sep; rsep = OPT row_sep ->
return_term loc (Array (p, csep, rsep)) *)
| SYMBOL "\\FRAC"; p1 = SELF; p2 = SELF ->
return_term loc (Layout (Frac (p1, p2)))
| SYMBOL "\\SQRT"; p = SELF -> return_term loc (Layout (Sqrt p))
- | SYMBOL "\\ROOT"; index = l1_pattern; SYMBOL "\\OF"; arg = SELF ->
- return_term loc (Layout (Root (arg, Layout (Box (H, index)))))
- | SYMBOL "\\HBOX"; SYMBOL "["; p = l1_pattern; SYMBOL "]" ->
+ | SYMBOL "\\ROOT"; index = SELF; SYMBOL "\\OF"; arg = SELF ->
+ return_term loc (Layout (Root (arg, index)));
+ | SYMBOL "\\HBOX"; DELIM "\\["; p = l1_pattern; DELIM "\\]" ->
return_term loc (Layout (Box (H, p)))
- | SYMBOL "\\VBOX"; SYMBOL "["; p = l1_pattern; SYMBOL "]" ->
+ | SYMBOL "\\VBOX"; DELIM "\\["; p = l1_pattern; DELIM "\\]" ->
return_term loc (Layout (Box (V, p)))
| SYMBOL "\\BREAK" -> return_term loc (Layout Break)
- | SYMBOL "["; p = l1_pattern; SYMBOL "]" ->
- return_term loc (Layout (Box (H, p)))
- | SYMBOL "["; p = l1_pattern; SYMBOL "\\AS"; id = IDENT; SYMBOL "]" ->
- return_term loc (Variable (Ascription (Layout (Box (H, p)), id)))
+ | DELIM "\\["; p = l1_pattern; DELIM "\\]" ->
+ return_term loc (CicNotationUtil.boxify p)
+ | p = SELF; SYMBOL "\\AS"; id = IDENT ->
+ return_term loc (Variable (Ascription (p, id)))
]
| "simple" NONA
- [ m = l1_magic_pattern -> return_term loc (Magic m)
+ [ i = IDENT -> return_term loc (Ident (i, None))
+ | m = l1_magic_pattern -> return_term loc (Magic m)
| v = l1_pattern_variable -> return_term loc (Variable v)
| l = literal -> return_term loc (Literal l)
]
]
];
explicit_subst: [
- [ (* TODO explicit substitution *)
+ [ SYMBOL "\\subst"; (* to avoid catching frequent "a [1]" cases *)
+ SYMBOL "[";
+ substs = LIST1 [
+ i = IDENT; SYMBOL <:unicode<Assign>> (* ≔ *); t = l2_pattern -> (i, t)
+ ] SEP SYMBOL ";";
+ SYMBOL "]" ->
+ substs
]
];
meta_subst: [
- [ (* TODO meta substitution *)
- ]
+ [ s = SYMBOL "_" -> None
+ | p = l2_pattern -> Some p ]
+ ];
+ meta_substs: [
+ [ SYMBOL "["; substs = LIST0 meta_subst; SYMBOL "]" -> substs ]
];
possibly_typed_name: [
- [ SYMBOL "("; id = IDENT; SYMBOL ":"; typ = l2_pattern; SYMBOL ")" ->
- Cic.Name id, Some typ
- | id = IDENT -> Cic.Name id, None
+ [ SYMBOL "("; id = bound_name; SYMBOL ":"; typ = l2_pattern; SYMBOL ")" ->
+ id, Some typ
+ | id = bound_name -> id, None
]
];
match_pattern: [
| SYMBOL <:unicode<lambda>> (* λ *) -> `Lambda
]
];
+ bound_name: [
+ [ i = IDENT -> Ident (i, None)
+ | SYMBOL "\\FRESH"; i = IDENT -> Variable (FreshVar i)
+ ]
+ ];
bound_names: [
- [ vars = LIST1 IDENT SEP SYMBOL ",";
+ [ vars = LIST1 bound_name SEP SYMBOL ",";
ty = OPT [ SYMBOL ":"; p = l2_pattern -> p ] ->
[ vars, ty ]
| clusters = LIST1 [
SYMBOL "(";
- vars = LIST1 IDENT SEP SYMBOL ",";
+ vars = LIST1 bound_name SEP SYMBOL ",";
ty = OPT [ SYMBOL ":"; p = l2_pattern -> p ];
SYMBOL ")" ->
vars, ty
];
let_defs: [
[ defs = LIST1 [
- name = IDENT; args = bound_names;
- index_name = OPT [ IDENT "on"; id = IDENT -> id ];
+ name = bound_name; args = bound_names;
+ index_name = OPT [ IDENT "on"; id = bound_name -> id ];
ty = OPT [ SYMBOL ":" ; p = l2_pattern -> p ];
SYMBOL <:unicode<def>> (* ≝ *); body = l2_pattern ->
let body = fold_binder `Lambda args body in
| _ :: tl -> position_of name (p + 1) tl
in
let rec find_arg name n = function
- | [] -> fail loc (sprintf "Argument %s not found" name)
+ | [] ->
+ fail loc (sprintf "Argument %s not found"
+ (CicNotationPp.pp_term name))
| (l,_) :: tl ->
(match position_of name 0 l with
| None, len -> find_arg name (n + len) tl
| None -> 0
| Some name -> find_arg name 0 args
in
- (Cic.Name name, ty), body, index
+ (name, ty), body, index
] SEP IDENT "and" ->
defs
]
];
l2_pattern_variable: [
- [ SYMBOL "\\NUM"; id = IDENT -> NumVar id
+ [ SYMBOL "\\TERM"; id = IDENT -> TermVar id
+ | SYMBOL "\\NUM"; id = IDENT -> NumVar id
| SYMBOL "\\IDENT"; id = IDENT -> IdentVar id
| SYMBOL "\\FRESH"; id = IDENT -> FreshVar id
]
l2_magic_pattern: [
[ SYMBOL "\\FOLD";
kind = [ IDENT "left" -> `Left | IDENT "right" -> `Right ];
- base = l2_pattern;
- SYMBOL "\\LAMBDA"; id = IDENT; recursive = l2_pattern ->
+ DELIM "\\["; base = l2_pattern; DELIM "\\]";
+ SYMBOL "\\LAMBDA"; id = IDENT;
+ DELIM "\\["; recursive = l2_pattern; DELIM "\\]" ->
Fold (kind, base, [id], recursive)
- | SYMBOL "\\DEFAULT"; some = l2_pattern; none = l2_pattern ->
+ | SYMBOL "\\DEFAULT";
+ DELIM "\\["; some = l2_pattern; DELIM "\\]";
+ DELIM "\\["; none = l2_pattern; DELIM "\\]" ->
Default (some, none)
]
];
- l2_pattern:
- [ "letin" NONA
+ l2_pattern: LEVEL "10" (* let in *)
+ [ "10" NONA
[ IDENT "let"; var = possibly_typed_name; SYMBOL <:unicode<def>> (* ≝ *);
p1 = l2_pattern; "in"; p2 = l2_pattern ->
return_term loc (LetIn (var, p1, p2))
body = l2_pattern ->
return_term loc (LetRec (k, defs, body))
]
- | "binder" RIGHTA
+ ];
+ l2_pattern: LEVEL "20" (* binder *)
+ [ "20" RIGHTA
[ b = binder; names = bound_names; SYMBOL "."; body = l2_pattern ->
return_term loc (fold_binder b names body)
]
- | "extension"
- [ ]
- | "apply" LEFTA
+ ];
+ l2_pattern: LEVEL "70" (* apply *)
+ [ "70" LEFTA
[ p1 = l2_pattern; p2 = l2_pattern ->
let rec aux = function
| Appl (hd :: tl)
in
return_term loc (Appl (aux p1 @ [p2]))
]
- | "simple" NONA
+ ];
+ l2_pattern: LEVEL "90" (* simple *)
+ [ "90" NONA
[ id = IDENT -> return_term loc (Ident (id, None))
| id = IDENT; s = explicit_subst -> return_term loc (Ident (id, Some s))
+ | s = CSYMBOL -> return_term loc (Symbol (s, 0))
| u = URI -> return_term loc (Uri (u, None))
| n = NUMBER -> return_term loc (Num (n, 0))
| IMPLICIT -> return_term loc (Implicit)
| m = META -> return_term loc (Meta (int_of_string m, []))
- | m = META; s = meta_subst -> return_term loc (Meta (int_of_string m, s))
+ | m = META; s = meta_substs -> return_term loc (Meta (int_of_string m, s))
| s = sort -> return_term loc (Sort s)
- | s = SYMBOL -> return_term loc (Symbol (s, 0))
| outtyp = OPT [ SYMBOL "["; ty = l2_pattern; SYMBOL "]" -> ty ];
IDENT "match"; t = l2_pattern;
indty_ident = OPT [ SYMBOL ":"; id = IDENT -> id ];
(* }}} *)
(* {{{ Grammar for interpretation, notation level 3 *)
argument: [
- [ id = IDENT -> IdentArg id
- | SYMBOL <:unicode<eta>> (* η *); SYMBOL "."; a = SELF -> EtaArg (None, a)
- | SYMBOL <:unicode<eta>> (* η *); id = IDENT; SYMBOL "."; a = SELF ->
- EtaArg (Some id, a)
+ [ id = IDENT -> IdentArg (0, id)
+ | l = LIST1 [ SYMBOL <:unicode<eta>> (* η *) -> () ] SEP SYMBOL ".";
+ SYMBOL "."; id = IDENT ->
+ IdentArg (List.length l, id)
]
];
level3_term: [
[ u = URI -> UriPattern u
- | a = argument -> ArgPattern a
+ | id = IDENT -> VarPattern id
| SYMBOL "("; terms = LIST1 SELF; SYMBOL ")" ->
(match terms with
| [] -> assert false
(* }}} *)
(* {{{ Notation glues *)
associativity: [
- [ IDENT "left"; IDENT "associative" -> `Left
- | IDENT "right"; IDENT "associative" -> `Right
+ [ IDENT "left"; IDENT "associative" -> Gramext.LeftA
+ | IDENT "right"; IDENT "associative" -> Gramext.RightA
+ | IDENT "non"; IDENT "associative" -> Gramext.NonA
]
];
- precedence: [ [ IDENT "at"; IDENT "precedence"; n = NUMBER -> n ] ];
+ precedence: [
+ [ IDENT "with"; IDENT "precedence"; n = NUMBER -> int_of_string n ]
+ ];
notation: [
- [ IDENT "notation"; p1 = level1_pattern; IDENT "for"; p2 = level2_pattern;
- assoc = OPT associativity; prec = OPT precedence ->
- ()
+ [ p1 = level1_pattern;
+ assoc = OPT associativity; prec = OPT precedence;
+ IDENT "for"; p2 = level2_pattern ->
+ (p1, assoc, prec, p2)
]
];
interpretation: [
- [ IDENT "interpretation"; s = SYMBOL; args = LIST1 argument; IDENT "as";
- t = level3_term ->
- ()
+ [ s = CSYMBOL; args = LIST1 argument; SYMBOL "="; t = level3_term ->
+ (s, args, t)
+ ]
+ ];
+(* }}} *)
+(* {{{ Top-level phrases *)
+ phrase: [
+ [ IDENT "print"; p2 = level2_pattern; SYMBOL "." -> Print p2
+ | IDENT "notation"; (l1, assoc, prec, l2) = notation; SYMBOL "." ->
+ Notation (l1, assoc, prec, l2)
+ | IDENT "interpretation"; (symbol, args, l3) = interpretation; SYMBOL "." ->
+ Interpretation ((symbol, args), l3)
+ | IDENT "render"; u = URI; SYMBOL "." -> Render (UriManager.uri_of_string u)
]
];
(* }}} *)
END
+(** {2 API implementation} *)
+
let exc_located_wrapper f =
try
f ()
let parse_syntax_pattern stream =
exc_located_wrapper (fun () -> Grammar.Entry.parse level1_pattern stream)
-
let parse_ast_pattern stream =
exc_located_wrapper (fun () -> Grammar.Entry.parse level2_pattern stream)
-
let parse_interpretation stream =
exc_located_wrapper (fun () -> Grammar.Entry.parse level3_term stream)
+let parse_phrase stream =
+ exc_located_wrapper (fun () -> Grammar.Entry.parse phrase stream)
+
+(** {2 Debugging} *)
+
+let print_l2_pattern () =
+ Grammar.print_entry Format.std_formatter (Grammar.Entry.obj l2_pattern);
+ Format.pp_print_flush Format.std_formatter ();
+ flush stdout
(* vim:set encoding=utf8 foldmethod=marker: *)
projects / helm.git / blobdiff