1 (* Copyright (C) 2005, HELM Team.
3 * This file is part of HELM, an Hypertextual, Electronic
4 * Library of Mathematics, developed at the Computer Science
5 * Department, University of Bologna, Italy.
7 * HELM is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation; either version 2
10 * of the License, or (at your option) any later version.
12 * HELM is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with HELM; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place - Suite 330, Boston,
22 * For details, see the HELM World-Wide-Web page,
23 * http://helm.cs.unibo.it/
30 module Ast = CicNotationPt
31 module Env = CicNotationEnv
33 exception Parse_error of string
34 exception Level_not_found of int
36 let level1_pattern_grammar =
37 Grammar.gcreate CicNotationLexer.level1_pattern_lexer
38 let level2_ast_grammar = Grammar.gcreate CicNotationLexer.level2_ast_lexer
39 let level2_meta_grammar = Grammar.gcreate CicNotationLexer.level2_meta_lexer
41 let min_precedence = 0
42 let max_precedence = 100
45 Grammar.Entry.create level1_pattern_grammar "level1_pattern"
46 let level2_ast = Grammar.Entry.create level2_ast_grammar "level2_ast"
47 let term = Grammar.Entry.create level2_ast_grammar "term"
48 let let_defs = Grammar.Entry.create level2_ast_grammar "let_defs"
49 let level2_meta = Grammar.Entry.create level2_meta_grammar "level2_meta"
53 Pervasives.int_of_string s
55 failwith (sprintf "Lexer failure: string_of_int \"%s\" failed" s)
57 (** {2 Grammar extension} *)
59 let gram_symbol s = Gramext.Stoken ("SYMBOL", s)
60 let gram_ident s = Gramext.Stoken ("IDENT", s)
61 let gram_number s = Gramext.Stoken ("NUMBER", s)
62 let gram_keyword s = Gramext.Stoken ("", s)
63 let gram_term = Gramext.Sself
67 | `Symbol s -> gram_symbol s
68 | `Keyword s -> gram_keyword s
69 | `Number s -> gram_number s
73 | Binding of string * Env.value_type
74 | Env of (string * Env.value_type) list
76 let make_action action bindings =
77 let rec aux (vl : CicNotationEnv.t) =
79 [] -> Gramext.action (fun (loc: Ast.location) -> action vl loc)
80 | NoBinding :: tl -> Gramext.action (fun _ -> aux vl tl)
81 (* LUCA: DEFCON 3 BEGIN *)
82 | Binding (name, Env.TermType) :: tl ->
85 aux ((name, (Env.TermType, Env.TermValue v))::vl) tl)
86 | Binding (name, Env.StringType) :: tl ->
89 aux ((name, (Env.StringType, Env.StringValue v)) :: vl) tl)
90 | Binding (name, Env.NumType) :: tl ->
93 aux ((name, (Env.NumType, Env.NumValue v)) :: vl) tl)
94 | Binding (name, Env.OptType t) :: tl ->
97 aux ((name, (Env.OptType t, Env.OptValue v)) :: vl) tl)
98 | Binding (name, Env.ListType t) :: tl ->
101 aux ((name, (Env.ListType t, Env.ListValue v)) :: vl) tl)
103 Gramext.action (fun (v:CicNotationEnv.t) -> aux (v @ vl) tl)
104 (* LUCA: DEFCON 3 END *)
106 aux [] (List.rev bindings)
112 | NoBinding :: tl -> aux acc tl
113 | Env names :: tl -> aux (List.rev names @ acc) tl
114 | Binding (name, ty) :: tl -> aux ((name, ty) :: acc) tl
118 (* given a level 1 pattern computes the new RHS of "term" grammar entry *)
119 let extract_term_production pattern =
120 let rec aux = function
121 | Ast.AttributedTerm (_, t) -> aux t
122 | Ast.Literal l -> aux_literal l
123 | Ast.Layout l -> aux_layout l
124 | Ast.Magic m -> aux_magic m
125 | Ast.Variable v -> aux_variable v
127 prerr_endline (CicNotationPp.pp_term t);
131 | `Symbol s -> [NoBinding, gram_symbol s]
133 (* assumption: s will be registered as a keyword with the lexer *)
134 [NoBinding, gram_keyword s]
135 | `Number s -> [NoBinding, gram_number s]
136 and aux_layout = function
137 | Ast.Sub (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\sub"] @ aux p2
138 | Ast.Sup (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\sup"] @ aux p2
139 | Ast.Below (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\below"] @ aux p2
140 | Ast.Above (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\above"] @ aux p2
141 | Ast.Frac (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\frac"] @ aux p2
142 | Ast.Atop (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\atop"] @ aux p2
143 | Ast.Over (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\over"] @ aux p2
144 | Ast.Root (p1, p2) ->
145 [NoBinding, gram_symbol "\\root"] @ aux p2
146 @ [NoBinding, gram_symbol "\\of"] @ aux p1
147 | Ast.Sqrt p -> [NoBinding, gram_symbol "\\sqrt"] @ aux p
149 | Ast.Box (_, pl) -> List.flatten (List.map aux pl)
150 | Ast.Group pl -> List.flatten (List.map aux pl)
151 and aux_magic magic =
154 let p_bindings, p_atoms, p_names, p_action = inner_pattern p in
155 let action (env_opt : CicNotationEnv.t option) (loc : Ast.location) =
157 | Some env -> List.map Env.opt_binding_some env
158 | None -> List.map Env.opt_binding_of_name p_names
160 [ Env (List.map Env.opt_declaration p_names),
162 [ [ Gramext.Sopt (Gramext.srules [ p_atoms, p_action ]) ],
163 Gramext.action action ] ]
165 | Ast.List1 (p, _) ->
166 let p_bindings, p_atoms, p_names, p_action = inner_pattern p in
167 (* let env0 = List.map list_binding_of_name p_names in
168 let grow_env_entry env n v =
171 | (n', (ty, ListValue vl)) as entry ->
172 if n' = n then n', (ty, ListValue (v :: vl)) else entry
176 let grow_env env_i env =
178 (fun env (n, (_, v)) -> grow_env_entry env n v)
181 let action (env_list : CicNotationEnv.t list) (loc : Ast.location) =
182 CicNotationEnv.coalesce_env p_names env_list
186 | Ast.List0 (_, None) -> Gramext.Slist0 s
187 | Ast.List1 (_, None) -> Gramext.Slist1 s
188 | Ast.List0 (_, Some l) -> Gramext.Slist0sep (s, gram_of_literal l)
189 | Ast.List1 (_, Some l) -> Gramext.Slist1sep (s, gram_of_literal l)
192 [ Env (List.map Env.list_declaration p_names),
194 [ [ gram_of_list (Gramext.srules [ p_atoms, p_action ]) ],
195 Gramext.action action ] ]
199 | Ast.NumVar s -> [Binding (s, Env.NumType), gram_number ""]
200 | Ast.TermVar s -> [Binding (s, Env.TermType), gram_term]
201 | Ast.IdentVar s -> [Binding (s, Env.StringType), gram_ident ""]
202 | Ast.Ascription (p, s) -> assert false (* TODO *)
203 | Ast.FreshVar _ -> assert false
204 and inner_pattern p =
205 let p_bindings, p_atoms = List.split (aux p) in
206 let p_names = flatten_opt p_bindings in
208 make_action (fun (env : CicNotationEnv.t) (loc : Ast.location) -> env)
211 p_bindings, p_atoms, p_names, action
215 let level_of precedence associativity =
216 if precedence < min_precedence || precedence > max_precedence then
217 raise (Level_not_found precedence);
219 match associativity with
220 | Gramext.NonA -> "N"
221 | Gramext.LeftA -> "L"
222 | Gramext.RightA -> "R"
224 string_of_int precedence ^ assoc_string
226 type rule_id = Token.t Gramext.g_symbol list
228 (* mapping: rule_id -> owned keywords. (rule_id, string list) Hashtbl.t *)
229 let owned_keywords = Hashtbl.create 23
231 let extend level1_pattern ~precedence ~associativity action =
232 let p_bindings, p_atoms =
233 List.split (extract_term_production level1_pattern)
235 let level = level_of precedence associativity in
236 (* let p_names = flatten_opt p_bindings in *)
239 [ Grammar.Entry.obj (term: 'a Grammar.Entry.e),
240 Some (Gramext.Level level),
245 (fun (env: CicNotationEnv.t) (loc: Ast.location) ->
249 let keywords = CicNotationUtil.keywords_of_term level1_pattern in
250 let rule_id = p_atoms in
251 List.iter CicNotationLexer.add_level2_ast_keyword keywords;
252 Hashtbl.add owned_keywords rule_id keywords; (* keywords may be [] *)
256 let atoms = rule_id in
258 let keywords = Hashtbl.find owned_keywords rule_id in
259 List.iter CicNotationLexer.remove_level2_ast_keyword keywords
260 with Not_found -> assert false);
261 Grammar.delete_rule term atoms
265 let parse_level1_pattern_ref = ref (fun _ -> assert false)
266 let parse_level2_ast_ref = ref (fun _ -> assert false)
267 let parse_level2_meta_ref = ref (fun _ -> assert false)
269 let fold_cluster binder terms ty body =
271 (fun term body -> Ast.Binder (binder, (term, ty), body))
272 terms body (* terms are names: either Ident or FreshVar *)
274 let fold_exists terms ty body =
277 let lambda = Ast.Binder (`Lambda, (term, ty), body) in
278 Ast.Appl [ Ast.Symbol ("exists", 0); lambda ])
281 let fold_binder binder pt_names body =
283 (fun (names, ty) body -> fold_cluster binder names ty body)
286 let return_term loc term = Ast.AttributedTerm (`Loc loc, term)
288 (* create empty precedence level for "term" *)
291 Gramext.action (fun _ ->
292 failwith "internal error, lexer generated a dummy token")
294 (* Needed since campl4 on "delete_rule" remove the precedence level if it gets
295 * empty after the deletion. The lexer never generate the Stoken below. *)
296 let dummy_prod = [ [ Gramext.Stoken ("DUMMY", "") ], dummy_action ] in
297 let mk_level_list first last =
298 let rec aux acc = function
299 | i when i < first -> acc
302 ((Some (string_of_int i ^ "N"), Some Gramext.NonA, dummy_prod)
303 :: (Some (string_of_int i ^ "L"), Some Gramext.LeftA, dummy_prod)
304 :: (Some (string_of_int i ^ "R"), Some Gramext.RightA, dummy_prod)
311 [ Grammar.Entry.obj (term: 'a Grammar.Entry.e),
313 mk_level_list min_precedence max_precedence ]
315 (* {{{ Grammar for concrete syntax patterns, notation level 1 *)
317 GLOBAL: level1_pattern;
319 level1_pattern: [ [ p = l1_pattern; EOI -> CicNotationUtil.boxify p ] ];
320 l1_pattern: [ [ p = LIST1 l1_simple_pattern -> p ] ];
322 [ s = SYMBOL -> `Symbol s
323 | k = QKEYWORD -> `Keyword k
324 | n = NUMBER -> `Number n
327 sep: [ [ "sep"; sep = literal -> sep ] ];
328 (* row_sep: [ [ "rowsep"; sep = literal -> sep ] ];
329 field_sep: [ [ "fieldsep"; sep = literal -> sep ] ]; *)
331 [ "list0"; p = l1_simple_pattern; sep = OPT sep -> Ast.List0 (p, sep)
332 | "list1"; p = l1_simple_pattern; sep = OPT sep -> Ast.List1 (p, sep)
333 | "opt"; p = l1_simple_pattern -> Ast.Opt p
336 l1_pattern_variable: [
337 [ "term"; id = IDENT -> Ast.TermVar id
338 | "number"; id = IDENT -> Ast.NumVar id
339 | "ident"; id = IDENT -> Ast.IdentVar id
344 [ p1 = SELF; SYMBOL "\\sub"; p2 = SELF ->
345 return_term loc (Ast.Layout (Ast.Sub (p1, p2)))
346 | p1 = SELF; SYMBOL "\\sup"; p2 = SELF ->
347 return_term loc (Ast.Layout (Ast.Sup (p1, p2)))
348 | p1 = SELF; SYMBOL "\\below"; p2 = SELF ->
349 return_term loc (Ast.Layout (Ast.Below (p1, p2)))
350 | p1 = SELF; SYMBOL "\\above"; p2 = SELF ->
351 return_term loc (Ast.Layout (Ast.Above (p1, p2)))
352 | p1 = SELF; SYMBOL "\\over"; p2 = SELF ->
353 return_term loc (Ast.Layout (Ast.Over (p1, p2)))
354 | p1 = SELF; SYMBOL "\\atop"; p2 = SELF ->
355 return_term loc (Ast.Layout (Ast.Atop (p1, p2)))
356 (* | "array"; p = SELF; csep = OPT field_sep; rsep = OPT row_sep ->
357 return_term loc (Array (p, csep, rsep)) *)
358 | SYMBOL "\\frac"; p1 = SELF; p2 = SELF ->
359 return_term loc (Ast.Layout (Ast.Frac (p1, p2)))
360 | SYMBOL "\\sqrt"; p = SELF -> return_term loc (Ast.Layout (Ast.Sqrt p))
361 | SYMBOL "\\root"; index = SELF; SYMBOL "\\of"; arg = SELF ->
362 return_term loc (Ast.Layout (Ast.Root (arg, index)))
363 | "hbox"; LPAREN; p = l1_pattern; RPAREN ->
364 return_term loc (Ast.Layout (Ast.Box ((Ast.H, false, false), p)))
365 | "vbox"; LPAREN; p = l1_pattern; RPAREN ->
366 return_term loc (Ast.Layout (Ast.Box ((Ast.V, false, false), p)))
367 | "hvbox"; LPAREN; p = l1_pattern; RPAREN ->
368 return_term loc (Ast.Layout (Ast.Box ((Ast.HV, false, false), p)))
369 | "hovbox"; LPAREN; p = l1_pattern; RPAREN ->
370 return_term loc (Ast.Layout (Ast.Box ((Ast.HOV, false, false), p)))
371 | "break" -> return_term loc (Ast.Layout Ast.Break)
372 (* | SYMBOL "\\SPACE" -> return_term loc (Layout Space) *)
373 | LPAREN; p = l1_pattern; RPAREN ->
374 return_term loc (CicNotationUtil.group p)
377 [ i = IDENT -> return_term loc (Ast.Variable (Ast.TermVar i))
378 | m = l1_magic_pattern -> return_term loc (Ast.Magic m)
379 | v = l1_pattern_variable -> return_term loc (Ast.Variable v)
380 | l = literal -> return_term loc (Ast.Literal l)
386 (* {{{ Grammar for ast magics, notation level 2 *)
390 [ "term"; id = IDENT -> Ast.TermVar id
391 | "number"; id = IDENT -> Ast.NumVar id
392 | "ident"; id = IDENT -> Ast.IdentVar id
393 | "fresh"; id = IDENT -> Ast.FreshVar id
394 | "anonymous" -> Ast.TermVar "_"
395 | id = IDENT -> Ast.TermVar id
399 [ "fold"; kind = [ "left" -> `Left | "right" -> `Right ];
400 base = level2_meta; "rec"; id = IDENT; recursive = level2_meta ->
401 Ast.Fold (kind, base, [id], recursive)
402 | "default"; some = level2_meta; none = level2_meta ->
403 Ast.Default (some, none)
404 | "if"; p_test = level2_meta;
405 "then"; p_true = level2_meta;
406 "else"; p_false = level2_meta ->
407 Ast.If (p_test, p_true, p_false)
412 [ magic = l2_magic -> Ast.Magic magic
413 | var = l2_variable -> Ast.Variable var
414 | blob = UNPARSED_AST ->
415 !parse_level2_ast_ref (Ulexing.from_utf8_string blob)
421 (* {{{ Grammar for ast patterns, notation level 2 *)
423 GLOBAL: level2_ast term let_defs;
424 level2_ast: [ [ p = term -> p ] ];
428 | "Type" -> `Type (CicUniv.fresh ())
433 [ SYMBOL "\\subst"; (* to avoid catching frequent "a [1]" cases *)
436 i = IDENT; SYMBOL <:unicode<Assign>> (* ≔ *); t = term -> (i, t)
443 [ s = SYMBOL "_" -> None
444 | p = term -> Some p ]
447 [ SYMBOL "["; substs = LIST0 meta_subst; SYMBOL "]" -> substs ]
449 possibly_typed_name: [
450 [ LPAREN; id = single_arg; SYMBOL ":"; typ = term; RPAREN ->
452 | arg = single_arg -> arg, None
456 [ id = IDENT -> id, None, []
457 | LPAREN; id = IDENT; vars = LIST1 possibly_typed_name; RPAREN ->
462 [ SYMBOL <:unicode<Pi>> (* Π *) -> `Pi
463 (* | SYMBOL <:unicode<exists>> |+ ∃ +| -> `Exists *)
464 | SYMBOL <:unicode<forall>> (* ∀ *) -> `Forall
465 | SYMBOL <:unicode<lambda>> (* λ *) -> `Lambda
469 [ LPAREN; names = LIST1 IDENT SEP SYMBOL ",";
470 SYMBOL ":"; ty = term; RPAREN ->
471 List.map (fun n -> Ast.Ident (n, None)) names, Some ty
472 | name = IDENT -> [Ast.Ident (name, None)], None
473 | blob = UNPARSED_META ->
474 let meta = !parse_level2_meta_ref (Ulexing.from_utf8_string blob) in
476 | Ast.Variable (Ast.FreshVar _) -> [meta], None
477 | Ast.Variable (Ast.TermVar "_") -> [Ast.Ident ("_", None)], None
478 | _ -> failwith "Invalid bound name."
482 [ name = IDENT -> Ast.Ident (name, None)
483 | blob = UNPARSED_META ->
484 let meta = !parse_level2_meta_ref (Ulexing.from_utf8_string blob) in
486 | Ast.Variable (Ast.FreshVar _)
487 | Ast.Variable (Ast.IdentVar _) -> meta
488 | Ast.Variable (Ast.TermVar "_") -> Ast.Ident ("_", None)
489 | _ -> failwith "Invalid index name."
493 [ "rec" -> `Inductive
494 | "corec" -> `CoInductive
501 index_name = OPT [ "on"; id = single_arg -> id ];
502 ty = OPT [ SYMBOL ":" ; p = term -> p ];
503 SYMBOL <:unicode<def>> (* ≝ *); body = term ->
504 let body = fold_binder `Lambda args body in
508 | Some ty -> Some (fold_binder `Pi args ty)
510 let rec position_of name p = function
512 | n :: _ when n = name -> Some p, p
513 | _ :: tl -> position_of name (p + 1) tl
515 let rec find_arg name n = function
517 Ast.fail loc (sprintf "Argument %s not found"
518 (CicNotationPp.pp_term name))
520 (match position_of name 0 l with
521 | None, len -> find_arg name (n + len) tl
522 | Some where, len -> n + where)
525 match index_name with
527 | Some index_name -> find_arg index_name 0 args
529 (name, ty), body, index
536 l = LIST1 single_arg SEP SYMBOL "," -> l
537 | SYMBOL "_" -> [Ast.Ident ("_", None)] ];
538 typ = OPT [ SYMBOL ":"; t = term -> t ] -> (vars, typ)
541 l = LIST1 single_arg SEP SYMBOL "," -> l
542 | SYMBOL "_" -> [Ast.Ident ("_", None)] ];
543 typ = OPT [ SYMBOL ":"; t = term -> t ];
544 RPAREN -> (vars, typ)
547 term: LEVEL "10N" [ (* let in *)
548 [ "let"; var = possibly_typed_name; SYMBOL <:unicode<def>> (* ≝ *);
549 p1 = term; "in"; p2 = term ->
550 return_term loc (Ast.LetIn (var, p1, p2))
551 | "let"; k = induction_kind; defs = let_defs; "in";
553 return_term loc (Ast.LetRec (k, defs, body))
556 term: LEVEL "20R" (* binder *)
558 [ b = binder; (vars, typ) = binder_vars; SYMBOL "."; body = term ->
559 return_term loc (fold_cluster b vars typ body)
560 | SYMBOL <:unicode<exists>> (* ∃ *);
561 (vars, typ) = binder_vars; SYMBOL "."; body = term ->
562 return_term loc (fold_exists vars typ body)
565 term: LEVEL "70L" (* apply *)
567 [ p1 = term; p2 = term ->
568 let rec aux = function
569 | Ast.Appl (hd :: tl)
570 | Ast.AttributedTerm (_, Ast.Appl (hd :: tl)) ->
574 return_term loc (Ast.Appl (aux p1 @ [p2]))
577 term: LEVEL "90N" (* simple *)
579 [ id = IDENT -> return_term loc (Ast.Ident (id, None))
580 | id = IDENT; s = explicit_subst ->
581 return_term loc (Ast.Ident (id, Some s))
582 | s = CSYMBOL -> return_term loc (Ast.Symbol (s, 0))
583 | u = URI -> return_term loc (Ast.Uri (u, None))
584 | n = NUMBER -> return_term loc (Ast.Num (n, 0))
585 | IMPLICIT -> return_term loc (Ast.Implicit)
586 | PLACEHOLDER -> return_term loc Ast.UserInput
587 | m = META -> return_term loc (Ast.Meta (int_of_string m, []))
588 | m = META; s = meta_substs ->
589 return_term loc (Ast.Meta (int_of_string m, s))
590 | s = sort -> return_term loc (Ast.Sort s)
592 indty_ident = OPT [ "in"; id = IDENT -> id, None ];
593 outtyp = OPT [ "return"; ty = term -> ty ];
596 lhs = match_pattern; SYMBOL <:unicode<Rightarrow>> (* ⇒ *);
601 return_term loc (Ast.Case (t, indty_ident, outtyp, patterns))
602 | LPAREN; p1 = term; SYMBOL ":"; p2 = term; RPAREN ->
603 return_term loc (Ast.Cast (p1, p2))
604 | LPAREN; p = term; RPAREN -> p
605 | blob = UNPARSED_META ->
606 !parse_level2_meta_ref (Ulexing.from_utf8_string blob)
612 (** {2 API implementation} *)
614 let exc_located_wrapper f =
618 | Stdpp.Exc_located (floc, Stream.Error msg) ->
619 raise (HExtlib.Localized (floc, Parse_error msg))
620 | Stdpp.Exc_located (floc, exn) ->
621 raise (HExtlib.Localized (floc, (Parse_error (Printexc.to_string exn))))
623 let parse_level1_pattern lexbuf =
625 (fun () -> Grammar.Entry.parse level1_pattern (Obj.magic lexbuf))
627 let parse_level2_ast lexbuf =
629 (fun () -> Grammar.Entry.parse level2_ast (Obj.magic lexbuf))
631 let parse_level2_meta lexbuf =
633 (fun () -> Grammar.Entry.parse level2_meta (Obj.magic lexbuf))
636 parse_level1_pattern_ref := parse_level1_pattern;
637 parse_level2_ast_ref := parse_level2_ast;
638 parse_level2_meta_ref := parse_level2_meta
642 let print_l2_pattern () =
643 Grammar.print_entry Format.std_formatter (Grammar.Entry.obj term);
644 Format.pp_print_flush Format.std_formatter ();
647 (* vim:set encoding=utf8 foldmethod=marker: *)