(* Copyright (C) 2005, HELM Team. * * This file is part of HELM, an Hypertextual, Electronic * Library of Mathematics, developed at the Computer Science * Department, University of Bologna, Italy. * * HELM is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License * as published by the Free Software Foundation; either version 2 * of the License, or (at your option) any later version. * * HELM is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with HELM; if not, write to the Free Software * Foundation, Inc., 59 Temple Place - Suite 330, Boston, * MA 02111-1307, USA. * * For details, see the HELM World-Wide-Web page, * http://helm.cs.unibo.it/ *) open Printf module Ast = CicNotationPt module Env = CicNotationEnv exception Parse_error of Token.flocation * string exception Level_not_found of int let level1_pattern_grammar = Grammar.gcreate CicNotationLexer.level1_pattern_lexer let level2_ast_grammar = Grammar.gcreate CicNotationLexer.level2_ast_lexer let level2_meta_grammar = Grammar.gcreate CicNotationLexer.level2_meta_lexer let min_precedence = 0 let max_precedence = 100 let level1_pattern = Grammar.Entry.create level1_pattern_grammar "level1_pattern" let level2_ast = Grammar.Entry.create level2_ast_grammar "level2_ast" let term = Grammar.Entry.create level2_ast_grammar "term" let let_defs = Grammar.Entry.create level2_ast_grammar "let_defs" let level2_meta = Grammar.Entry.create level2_meta_grammar "level2_meta" let int_of_string s = try Pervasives.int_of_string s with Failure _ -> failwith (sprintf "Lexer failure: string_of_int \"%s\" failed" s) (** {2 Grammar extension} *) let gram_symbol s = Gramext.Stoken ("SYMBOL", s) let gram_ident s = Gramext.Stoken ("IDENT", s) let gram_number s = Gramext.Stoken ("NUMBER", s) let gram_keyword s = Gramext.Stoken ("", s) let gram_term = Gramext.Sself let gram_of_literal = function | `Symbol s -> gram_symbol s | `Keyword s -> gram_keyword s | `Number s -> gram_number s type binding = | NoBinding | Binding of string * Env.value_type | Env of (string * Env.value_type) list let make_action action bindings = let rec aux (vl : CicNotationEnv.t) = function [] -> Gramext.action (fun (loc: Ast.location) -> action vl loc) | NoBinding :: tl -> Gramext.action (fun _ -> aux vl tl) (* LUCA: DEFCON 3 BEGIN *) | Binding (name, Env.TermType) :: tl -> Gramext.action (fun (v:Ast.term) -> aux ((name, (Env.TermType, Env.TermValue v))::vl) tl) | Binding (name, Env.StringType) :: tl -> Gramext.action (fun (v:string) -> aux ((name, (Env.StringType, Env.StringValue v)) :: vl) tl) | Binding (name, Env.NumType) :: tl -> Gramext.action (fun (v:string) -> aux ((name, (Env.NumType, Env.NumValue v)) :: vl) tl) | Binding (name, Env.OptType t) :: tl -> Gramext.action (fun (v:'a option) -> aux ((name, (Env.OptType t, Env.OptValue v)) :: vl) tl) | Binding (name, Env.ListType t) :: tl -> Gramext.action (fun (v:'a list) -> aux ((name, (Env.ListType t, Env.ListValue v)) :: vl) tl) | Env _ :: tl -> Gramext.action (fun (v:CicNotationEnv.t) -> aux (v @ vl) tl) (* LUCA: DEFCON 3 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 | Ast.AttributedTerm (_, t) -> aux t | Ast.Literal l -> aux_literal l | Ast.Layout l -> aux_layout l | Ast.Magic m -> aux_magic m | Ast.Variable v -> aux_variable v | t -> prerr_endline (CicNotationPp.pp_term t); assert false and aux_literal = function | `Symbol s -> [NoBinding, gram_symbol s] | `Keyword s -> (* assumption: s will be registered as a keyword with the lexer *) [NoBinding, gram_keyword s] | `Number s -> [NoBinding, gram_number s] and aux_layout = function | Ast.Sub (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\sub"] @ aux p2 | Ast.Sup (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\sup"] @ aux p2 | Ast.Below (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\below"] @ aux p2 | Ast.Above (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\above"] @ aux p2 | Ast.Frac (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\frac"] @ aux p2 | Ast.Atop (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\atop"] @ aux p2 | Ast.Over (p1, p2) -> aux p1 @ [NoBinding, gram_symbol "\\over"] @ aux p2 | Ast.Root (p1, p2) -> [NoBinding, gram_symbol "\\root"] @ aux p2 @ [NoBinding, gram_symbol "\\of"] @ aux p1 | Ast.Sqrt p -> [NoBinding, gram_symbol "\\sqrt"] @ aux p | Ast.Break -> [] | Ast.Box (_, pl) -> List.flatten (List.map aux pl) | Ast.Group pl -> List.flatten (List.map aux pl) and aux_magic magic = match magic with | Ast.Opt p -> let p_bindings, p_atoms, p_names, p_action = inner_pattern p in let action (env_opt : CicNotationEnv.t option) (loc : Ast.location) = match env_opt with | Some env -> List.map Env.opt_binding_some env | None -> List.map Env.opt_binding_of_name p_names in [ Env (List.map Env.opt_declaration p_names), Gramext.srules [ [ Gramext.Sopt (Gramext.srules [ p_atoms, p_action ]) ], Gramext.action action ] ] | Ast.List0 (p, _) | Ast.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 : Ast.location) = CicNotationEnv.coalesce_env p_names env_list in let gram_of_list s = match magic with | Ast.List0 (_, None) -> Gramext.Slist0 s | Ast.List1 (_, None) -> Gramext.Slist1 s | Ast.List0 (_, Some l) -> Gramext.Slist0sep (s, gram_of_literal l) | Ast.List1 (_, Some l) -> Gramext.Slist1sep (s, gram_of_literal l) | _ -> assert false in [ Env (List.map Env.list_declaration p_names), Gramext.srules [ [ gram_of_list (Gramext.srules [ p_atoms, p_action ]) ], Gramext.action action ] ] | _ -> assert false and aux_variable = function | Ast.NumVar s -> [Binding (s, Env.NumType), gram_number ""] | Ast.TermVar s -> [Binding (s, Env.TermType), gram_term] | Ast.IdentVar s -> [Binding (s, Env.StringType), gram_ident ""] | Ast.Ascription (p, s) -> assert false (* TODO *) | Ast.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 action = make_action (fun (env : CicNotationEnv.t) (loc : Ast.location) -> env) p_bindings in p_bindings, p_atoms, p_names, action in aux pattern let level_of precedence associativity = if precedence < min_precedence || precedence > max_precedence then raise (Level_not_found precedence); let assoc_string = match associativity with | Gramext.NonA -> "N" | Gramext.LeftA -> "L" | Gramext.RightA -> "R" in string_of_int precedence ^ assoc_string type rule_id = Token.t Gramext.g_symbol list (* mapping: rule_id -> owned keywords. (rule_id, string list) Hashtbl.t *) let owned_keywords = Hashtbl.create 23 let extend level1_pattern ~precedence ~associativity action = let p_bindings, p_atoms = List.split (extract_term_production level1_pattern) in let level = level_of precedence associativity in let p_names = flatten_opt p_bindings in let _ = Grammar.extend [ Grammar.Entry.obj (term: 'a Grammar.Entry.e), Some (Gramext.Level level), [ None, Some associativity, [ p_atoms, (make_action (fun (env: CicNotationEnv.t) (loc: Ast.location) -> (action env loc)) p_bindings) ]]] in let keywords = CicNotationUtil.keywords_of_term level1_pattern in let rule_id = p_atoms in List.iter CicNotationLexer.add_level2_ast_keyword keywords; Hashtbl.add owned_keywords rule_id keywords; (* keywords may be [] *) rule_id let delete rule_id = let atoms = rule_id in (try let keywords = Hashtbl.find owned_keywords rule_id in List.iter CicNotationLexer.remove_level2_ast_keyword keywords with Not_found -> assert false); Grammar.delete_rule term atoms (** {2 Grammar} *) let parse_level1_pattern_ref = ref (fun _ -> assert false) let parse_level2_ast_ref = ref (fun _ -> assert false) let parse_level2_meta_ref = ref (fun _ -> assert false) let fold_cluster binder terms ty body = List.fold_right (fun term body -> Ast.Binder (binder, (term, ty), body)) terms body (* terms are names: either Ident or FreshVar *) let fold_exists terms ty body = List.fold_right (fun term body -> let lambda = Ast.Binder (`Lambda, (term, ty), body) in Ast.Appl [ Ast.Symbol ("exists", 0); lambda ]) terms body let fold_binder binder pt_names body = List.fold_right (fun (names, ty) body -> fold_cluster binder names ty body) pt_names body let return_term loc term = Ast.AttributedTerm (`Loc loc, term) (* create empty precedence level for "term" *) let _ = let dummy_action = Gramext.action (fun _ -> failwith "internal error, lexer generated a dummy token") in (* Needed since campl4 on "delete_rule" remove the precedence level if it gets * empty after the deletion. The lexer never generate the Stoken below. *) let dummy_prod = [ [ Gramext.Stoken ("DUMMY", "") ], dummy_action ] in let mk_level_list first last = let rec aux acc = function | i when i < first -> acc | i -> aux ((Some (string_of_int i ^ "N"), Some Gramext.NonA, dummy_prod) :: (Some (string_of_int i ^ "L"), Some Gramext.LeftA, dummy_prod) :: (Some (string_of_int i ^ "R"), Some Gramext.RightA, dummy_prod) :: acc) (i - 1) in aux [] last in Grammar.extend [ Grammar.Entry.obj (term: 'a Grammar.Entry.e), None, mk_level_list min_precedence max_precedence ] (* {{{ Grammar for concrete syntax patterns, notation level 1 *) EXTEND GLOBAL: level1_pattern; level1_pattern: [ [ p = l1_pattern; EOI -> CicNotationUtil.boxify p ] ]; l1_pattern: [ [ p = LIST1 l1_simple_pattern -> p ] ]; literal: [ [ s = SYMBOL -> `Symbol s | k = QKEYWORD -> `Keyword k | n = NUMBER -> `Number n ] ]; sep: [ [ "sep"; sep = literal -> sep ] ]; (* row_sep: [ [ "rowsep"; sep = literal -> sep ] ]; field_sep: [ [ "fieldsep"; sep = literal -> sep ] ]; *) l1_magic_pattern: [ [ "list0"; p = l1_simple_pattern; sep = OPT sep -> Ast.List0 (p, sep) | "list1"; p = l1_simple_pattern; sep = OPT sep -> Ast.List1 (p, sep) | "opt"; p = l1_simple_pattern -> Ast.Opt p ] ]; l1_pattern_variable: [ [ "term"; id = IDENT -> Ast.TermVar id | "number"; id = IDENT -> Ast.NumVar id | "ident"; id = IDENT -> Ast.IdentVar id ] ]; l1_simple_pattern: [ "layout" LEFTA [ p1 = SELF; SYMBOL "\\sub"; p2 = SELF -> return_term loc (Ast.Layout (Ast.Sub (p1, p2))) | p1 = SELF; SYMBOL "\\sup"; p2 = SELF -> return_term loc (Ast.Layout (Ast.Sup (p1, p2))) | p1 = SELF; SYMBOL "\\below"; p2 = SELF -> return_term loc (Ast.Layout (Ast.Below (p1, p2))) | p1 = SELF; SYMBOL "\\above"; p2 = SELF -> return_term loc (Ast.Layout (Ast.Above (p1, p2))) | p1 = SELF; SYMBOL "\\over"; p2 = SELF -> return_term loc (Ast.Layout (Ast.Over (p1, p2))) | p1 = SELF; SYMBOL "\\atop"; p2 = SELF -> return_term loc (Ast.Layout (Ast.Atop (p1, p2))) (* | "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 (Ast.Layout (Ast.Frac (p1, p2))) | SYMBOL "\\sqrt"; p = SELF -> return_term loc (Ast.Layout (Ast.Sqrt p)) | SYMBOL "\\root"; index = SELF; SYMBOL "\\of"; arg = SELF -> return_term loc (Ast.Layout (Ast.Root (arg, index))) | "hbox"; LPAREN; p = l1_pattern; RPAREN -> return_term loc (Ast.Layout (Ast.Box ((Ast.H, false, false), p))) | "vbox"; LPAREN; p = l1_pattern; RPAREN -> return_term loc (Ast.Layout (Ast.Box ((Ast.V, false, false), p))) | "hvbox"; LPAREN; p = l1_pattern; RPAREN -> return_term loc (Ast.Layout (Ast.Box ((Ast.HV, false, false), p))) | "hovbox"; LPAREN; p = l1_pattern; RPAREN -> return_term loc (Ast.Layout (Ast.Box ((Ast.HOV, false, false), p))) | "break" -> return_term loc (Ast.Layout Ast.Break) (* | SYMBOL "\\SPACE" -> return_term loc (Layout Space) *) | LPAREN; p = l1_pattern; RPAREN -> return_term loc (CicNotationUtil.group p) ] | "simple" NONA [ i = IDENT -> return_term loc (Ast.Variable (Ast.TermVar i)) | m = l1_magic_pattern -> return_term loc (Ast.Magic m) | v = l1_pattern_variable -> return_term loc (Ast.Variable v) | l = literal -> return_term loc (Ast.Literal l) ] ]; END (* }}} *) (* {{{ Grammar for ast magics, notation level 2 *) EXTEND GLOBAL: level2_meta; l2_variable: [ [ "term"; id = IDENT -> Ast.TermVar id | "number"; id = IDENT -> Ast.NumVar id | "ident"; id = IDENT -> Ast.IdentVar id | "fresh"; id = IDENT -> Ast.FreshVar id | "anonymous" -> Ast.TermVar "_" | id = IDENT -> Ast.TermVar id ] ]; l2_magic: [ [ "fold"; kind = [ "left" -> `Left | "right" -> `Right ]; base = level2_meta; "rec"; id = IDENT; recursive = level2_meta -> Ast.Fold (kind, base, [id], recursive) | "default"; some = level2_meta; none = level2_meta -> Ast.Default (some, none) | "if"; p_test = level2_meta; "then"; p_true = level2_meta; "else"; p_false = level2_meta -> Ast.If (p_test, p_true, p_false) | "fail" -> Ast.Fail ] ]; level2_meta: [ [ magic = l2_magic -> Ast.Magic magic | var = l2_variable -> Ast.Variable var | blob = UNPARSED_AST -> !parse_level2_ast_ref (Ulexing.from_utf8_string blob) ] ]; END (* }}} *) (* {{{ Grammar for ast patterns, notation level 2 *) EXTEND GLOBAL: level2_ast term let_defs; level2_ast: [ [ p = term -> p ] ]; sort: [ [ "Prop" -> `Prop | "Set" -> `Set | "Type" -> `Type (CicUniv.fresh ()) | "CProp" -> `CProp ] ]; explicit_subst: [ [ SYMBOL "\\subst"; (* to avoid catching frequent "a [1]" cases *) SYMBOL "["; substs = LIST1 [ i = IDENT; SYMBOL <:unicode> (* ≔ *); t = term -> (i, t) ] SEP SYMBOL ";"; SYMBOL "]" -> substs ] ]; meta_subst: [ [ s = SYMBOL "_" -> None | p = term -> Some p ] ]; meta_substs: [ [ SYMBOL "["; substs = LIST0 meta_subst; SYMBOL "]" -> substs ] ]; possibly_typed_name: [ [ LPAREN; id = single_arg; SYMBOL ":"; typ = term; RPAREN -> id, Some typ | arg = single_arg -> arg, None ] ]; match_pattern: [ [ id = IDENT -> id, None, [] | LPAREN; id = IDENT; vars = LIST1 possibly_typed_name; RPAREN -> id, None, vars ] ]; binder: [ [ SYMBOL <:unicode> (* Π *) -> `Pi (* | SYMBOL <:unicode> |+ ∃ +| -> `Exists *) | SYMBOL <:unicode> (* ∀ *) -> `Forall | SYMBOL <:unicode> (* λ *) -> `Lambda ] ]; arg: [ [ LPAREN; names = LIST1 IDENT SEP SYMBOL ","; SYMBOL ":"; ty = term; RPAREN -> List.map (fun n -> Ast.Ident (n, None)) names, Some ty | name = IDENT -> [Ast.Ident (name, None)], None | blob = UNPARSED_META -> let meta = !parse_level2_meta_ref (Ulexing.from_utf8_string blob) in match meta with | Ast.Variable (Ast.FreshVar _) -> [meta], None | Ast.Variable (Ast.TermVar "_") -> [Ast.Ident ("_", None)], None | _ -> failwith "Invalid bound name." ] ]; single_arg: [ [ name = IDENT -> Ast.Ident (name, None) | blob = UNPARSED_META -> let meta = !parse_level2_meta_ref (Ulexing.from_utf8_string blob) in match meta with | Ast.Variable (Ast.FreshVar _) | Ast.Variable (Ast.IdentVar _) -> meta | Ast.Variable (Ast.TermVar "_") -> Ast.Ident ("_", None) | _ -> failwith "Invalid index name." ] ]; induction_kind: [ [ "rec" -> `Inductive | "corec" -> `CoInductive ] ]; let_defs: [ [ defs = LIST1 [ name = single_arg; args = LIST1 arg; index_name = OPT [ "on"; id = single_arg -> id ]; ty = OPT [ SYMBOL ":" ; p = term -> p ]; SYMBOL <:unicode> (* ≝ *); body = term -> let body = fold_binder `Lambda args body in let ty = match ty with | None -> None | Some ty -> Some (fold_binder `Pi args ty) in let rec position_of name p = function | [] -> None, p | n :: _ when n = name -> Some p, p | _ :: tl -> position_of name (p + 1) tl in let rec find_arg name n = function | [] -> Ast.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 | Some where, len -> n + where) in let index = match index_name with | None -> 0 | Some index_name -> find_arg index_name 0 args in (name, ty), body, index ] SEP "and" -> defs ] ]; binder_vars: [ [ vars = [ l = LIST1 single_arg SEP SYMBOL "," -> l | SYMBOL "_" -> [Ast.Ident ("_", None)] ]; typ = OPT [ SYMBOL ":"; t = term -> t ] -> (vars, typ) | LPAREN; vars = [ l = LIST1 single_arg SEP SYMBOL "," -> l | SYMBOL "_" -> [Ast.Ident ("_", None)] ]; typ = OPT [ SYMBOL ":"; t = term -> t ]; RPAREN -> (vars, typ) ] ]; term: LEVEL "10N" [ (* let in *) [ "let"; var = possibly_typed_name; SYMBOL <:unicode> (* ≝ *); p1 = term; "in"; p2 = term -> return_term loc (Ast.LetIn (var, p1, p2)) | "let"; k = induction_kind; defs = let_defs; "in"; body = term -> return_term loc (Ast.LetRec (k, defs, body)) ] ]; term: LEVEL "20R" (* binder *) [ [ b = binder; (vars, typ) = binder_vars; SYMBOL "."; body = term -> return_term loc (fold_cluster b vars typ body) | SYMBOL <:unicode> (* ∃ *); (vars, typ) = binder_vars; SYMBOL "."; body = term -> return_term loc (fold_exists vars typ body) ] ]; term: LEVEL "70L" (* apply *) [ [ p1 = term; p2 = term -> let rec aux = function | Ast.Appl (hd :: tl) | Ast.AttributedTerm (_, Ast.Appl (hd :: tl)) -> aux hd @ tl | term -> [term] in return_term loc (Ast.Appl (aux p1 @ [p2])) ] ]; term: LEVEL "90N" (* simple *) [ [ id = IDENT -> return_term loc (Ast.Ident (id, None)) | id = IDENT; s = explicit_subst -> return_term loc (Ast.Ident (id, Some s)) | s = CSYMBOL -> return_term loc (Ast.Symbol (s, 0)) | u = URI -> return_term loc (Ast.Uri (u, None)) | n = NUMBER -> return_term loc (Ast.Num (n, 0)) | IMPLICIT -> return_term loc (Ast.Implicit) | PLACEHOLDER -> return_term loc Ast.UserInput | m = META -> return_term loc (Ast.Meta (int_of_string m, [])) | m = META; s = meta_substs -> return_term loc (Ast.Meta (int_of_string m, s)) | s = sort -> return_term loc (Ast.Sort s) | "match"; t = term; indty_ident = OPT [ "in"; id = IDENT -> id, None ]; outtyp = OPT [ "return"; ty = term -> ty ]; "with"; SYMBOL "["; patterns = LIST0 [ lhs = match_pattern; SYMBOL <:unicode> (* ⇒ *); rhs = term -> lhs, rhs ] SEP SYMBOL "|"; SYMBOL "]" -> return_term loc (Ast.Case (t, indty_ident, outtyp, patterns)) | LPAREN; p1 = term; SYMBOL ":"; p2 = term; RPAREN -> return_term loc (Ast.Cast (p1, p2)) | LPAREN; p = term; RPAREN -> p | blob = UNPARSED_META -> !parse_level2_meta_ref (Ulexing.from_utf8_string blob) ] ]; END (* }}} *) (** {2 API implementation} *) let exc_located_wrapper f = try f () with | Stdpp.Exc_located (floc, Stream.Error msg) -> raise (Parse_error (floc, msg)) | Stdpp.Exc_located (floc, exn) -> raise (Parse_error (floc, (Printexc.to_string exn))) let parse_level1_pattern lexbuf = exc_located_wrapper (fun () -> Grammar.Entry.parse level1_pattern (Obj.magic lexbuf)) let parse_level2_ast lexbuf = exc_located_wrapper (fun () -> Grammar.Entry.parse level2_ast (Obj.magic lexbuf)) let parse_level2_meta lexbuf = exc_located_wrapper (fun () -> Grammar.Entry.parse level2_meta (Obj.magic lexbuf)) let _ = parse_level1_pattern_ref := parse_level1_pattern; parse_level2_ast_ref := parse_level2_ast; parse_level2_meta_ref := parse_level2_meta (** {2 Debugging} *) let print_l2_pattern () = Grammar.print_entry Format.std_formatter (Grammar.Entry.obj term); Format.pp_print_flush Format.std_formatter (); flush stdout (* vim:set encoding=utf8 foldmethod=marker: *)