diff options
Diffstat (limited to 'gramlib')
| -rw-r--r-- | gramlib/LICENSE | 29 | ||||
| -rw-r--r-- | gramlib/dune | 4 | ||||
| -rw-r--r-- | gramlib/gramext.ml | 461 | ||||
| -rw-r--r-- | gramlib/gramext.mli | 65 | ||||
| -rw-r--r-- | gramlib/gramlib.mllib | 4 | ||||
| -rw-r--r-- | gramlib/grammar.ml | 864 | ||||
| -rw-r--r-- | gramlib/grammar.mli | 83 | ||||
| -rw-r--r-- | gramlib/plexing.ml | 18 | ||||
| -rw-r--r-- | gramlib/plexing.mli | 37 | ||||
| -rw-r--r-- | gramlib/ploc.ml | 26 | ||||
| -rw-r--r-- | gramlib/ploc.mli | 40 |
11 files changed, 1631 insertions, 0 deletions
diff --git a/gramlib/LICENSE b/gramlib/LICENSE new file mode 100644 index 0000000000..b696affde7 --- /dev/null +++ b/gramlib/LICENSE @@ -0,0 +1,29 @@ +gramlib was derived from Daniel de Rauglaudre's camlp5 +(https://github.com/camlp5/camlp5) whose licence follows: + +* Copyright (c) 2007-2017, INRIA (Institut National de Recherches en +* Informatique et Automatique). All rights reserved. +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions are met: +* +* * Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* * Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* * Neither the name of INRIA, nor the names of its contributors may be +* used to endorse or promote products derived from this software without +* specific prior written permission. +* +* THIS SOFTWARE IS PROVIDED BY INRIA AND CONTRIBUTORS ``AS IS'' AND +* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, +* THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A +* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INRIA AND +* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF +* USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND +* ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT +* OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF +* SUCH DAMAGE. diff --git a/gramlib/dune b/gramlib/dune new file mode 100644 index 0000000000..8ca6aff25a --- /dev/null +++ b/gramlib/dune @@ -0,0 +1,4 @@ +(library + (name gramlib) + (public_name coq.gramlib) + (libraries coq.lib)) diff --git a/gramlib/gramext.ml b/gramlib/gramext.ml new file mode 100644 index 0000000000..46c2688f05 --- /dev/null +++ b/gramlib/gramext.ml @@ -0,0 +1,461 @@ +(* camlp5r *) +(* gramext.ml,v *) +(* Copyright (c) INRIA 2007-2017 *) + +open Printf + +type 'a parser_t = 'a Stream.t -> Obj.t + +type 'te grammar = + { gtokens : (Plexing.pattern, int ref) Hashtbl.t; + glexer : 'te Plexing.lexer } + +type 'te g_entry = + { egram : 'te grammar; + ename : string; + elocal : bool; + mutable estart : int -> 'te parser_t; + mutable econtinue : int -> int -> Obj.t -> 'te parser_t; + mutable edesc : 'te g_desc } +and 'te g_desc = + Dlevels of 'te g_level list + | Dparser of 'te parser_t +and 'te g_level = + { assoc : g_assoc; + lname : string option; + lsuffix : 'te g_tree; + lprefix : 'te g_tree } +and g_assoc = NonA | RightA | LeftA +and 'te g_symbol = + | Snterm of 'te g_entry + | Snterml of 'te g_entry * string + | Slist0 of 'te g_symbol + | Slist0sep of 'te g_symbol * 'te g_symbol * bool + | Slist1 of 'te g_symbol + | Slist1sep of 'te g_symbol * 'te g_symbol * bool + | Sopt of 'te g_symbol + | Sself + | Snext + | Stoken of Plexing.pattern + | Stree of 'te g_tree +and g_action = Obj.t +and 'te g_tree = + Node of 'te g_node + | LocAct of g_action * g_action list + | DeadEnd +and 'te g_node = + { node : 'te g_symbol; son : 'te g_tree; brother : 'te g_tree } +and err_fun = unit -> string + +type position = + First + | Last + | Before of string + | After of string + | Level of string + +let rec derive_eps = + function + Slist0 _ -> true + | Slist0sep (_, _, _) -> true + | Sopt _ -> true + | Stree t -> tree_derive_eps t + | Slist1 _ | Slist1sep (_, _, _) | Snterm _ | + Snterml (_, _) | Snext | Sself | Stoken _ -> + false +and tree_derive_eps = + function + LocAct (_, _) -> true + | Node {node = s; brother = bro; son = son} -> + derive_eps s && tree_derive_eps son || tree_derive_eps bro + | DeadEnd -> false + +let rec eq_symbol s1 s2 = + match s1, s2 with + Snterm e1, Snterm e2 -> e1 == e2 + | Snterml (e1, l1), Snterml (e2, l2) -> e1 == e2 && l1 = l2 + | Slist0 s1, Slist0 s2 -> eq_symbol s1 s2 + | Slist0sep (s1, sep1, b1), Slist0sep (s2, sep2, b2) -> + eq_symbol s1 s2 && eq_symbol sep1 sep2 && b1 = b2 + | Slist1 s1, Slist1 s2 -> eq_symbol s1 s2 + | Slist1sep (s1, sep1, b1), Slist1sep (s2, sep2, b2) -> + eq_symbol s1 s2 && eq_symbol sep1 sep2 && b1 = b2 + | Sopt s1, Sopt s2 -> eq_symbol s1 s2 + | Stree _, Stree _ -> false + | _ -> s1 = s2 + +let is_before s1 s2 = + match s1, s2 with + Stoken ("ANY", _), _ -> false + | _, Stoken ("ANY", _) -> true + | Stoken (_, s), Stoken (_, "") when s <> "" -> true + | Stoken _, Stoken _ -> false + | Stoken _, _ -> true + | _ -> false + +let insert_tree ~warning entry_name gsymbols action tree = + let rec insert symbols tree = + match symbols with + s :: sl -> insert_in_tree s sl tree + | [] -> + match tree with + Node {node = s; son = son; brother = bro} -> + Node {node = s; son = son; brother = insert [] bro} + | LocAct (old_action, action_list) -> + begin match warning with + | None -> () + | Some warn_fn -> + let msg = + "<W> Grammar extension: " ^ + (if entry_name <> "" then "" else "in ["^entry_name^"%s], ") ^ + "some rule has been masked" in + warn_fn msg + end; + LocAct (action, old_action :: action_list) + | DeadEnd -> LocAct (action, []) + and insert_in_tree s sl tree = + match try_insert s sl tree with + Some t -> t + | None -> Node {node = s; son = insert sl DeadEnd; brother = tree} + and try_insert s sl tree = + match tree with + Node {node = s1; son = son; brother = bro} -> + if eq_symbol s s1 then + let t = Node {node = s1; son = insert sl son; brother = bro} in + Some t + else if is_before s1 s || derive_eps s && not (derive_eps s1) then + let bro = + match try_insert s sl bro with + Some bro -> bro + | None -> Node {node = s; son = insert sl DeadEnd; brother = bro} + in + let t = Node {node = s1; son = son; brother = bro} in Some t + else + begin match try_insert s sl bro with + Some bro -> + let t = Node {node = s1; son = son; brother = bro} in Some t + | None -> None + end + | LocAct (_, _) | DeadEnd -> None + in + insert gsymbols tree + +let srules ~warning rl = + let t = + List.fold_left + (fun tree (symbols, action) -> insert_tree ~warning "" symbols action tree) + DeadEnd rl + in + Stree t + +let is_level_labelled n lev = + match lev.lname with + Some n1 -> n = n1 + | None -> false + +let insert_level ~warning entry_name e1 symbols action slev = + match e1 with + true -> + {assoc = slev.assoc; lname = slev.lname; + lsuffix = insert_tree ~warning entry_name symbols action slev.lsuffix; + lprefix = slev.lprefix} + | false -> + {assoc = slev.assoc; lname = slev.lname; lsuffix = slev.lsuffix; + lprefix = insert_tree ~warning entry_name symbols action slev.lprefix} + +let empty_lev lname assoc = + let assoc = + match assoc with + Some a -> a + | None -> LeftA + in + {assoc = assoc; lname = lname; lsuffix = DeadEnd; lprefix = DeadEnd} + +let change_lev ~warning lev n lname assoc = + let a = + match assoc with + None -> lev.assoc + | Some a -> + if a <> lev.assoc then + begin + match warning with + | None -> () + | Some warn_fn -> + warn_fn ("<W> Changing associativity of level \""^n^"\"") + end; + a + in + begin match lname with + Some n -> + if lname <> lev.lname then + begin match warning with + | None -> () + | Some warn_fn -> + warn_fn ("<W> Level label \""^n^"\" ignored") + end; + | None -> () + end; + {assoc = a; lname = lev.lname; lsuffix = lev.lsuffix; lprefix = lev.lprefix} + +let get_level ~warning entry position levs = + match position with + Some First -> [], empty_lev, levs + | Some Last -> levs, empty_lev, [] + | Some (Level n) -> + let rec get = + function + [] -> + eprintf "No level labelled \"%s\" in entry \"%s\"\n" n + entry.ename; + flush stderr; + failwith "Grammar.extend" + | lev :: levs -> + if is_level_labelled n lev then [], change_lev ~warning lev n, levs + else + let (levs1, rlev, levs2) = get levs in lev :: levs1, rlev, levs2 + in + get levs + | Some (Before n) -> + let rec get = + function + [] -> + eprintf "No level labelled \"%s\" in entry \"%s\"\n" n + entry.ename; + flush stderr; + failwith "Grammar.extend" + | lev :: levs -> + if is_level_labelled n lev then [], empty_lev, lev :: levs + else + let (levs1, rlev, levs2) = get levs in lev :: levs1, rlev, levs2 + in + get levs + | Some (After n) -> + let rec get = + function + [] -> + eprintf "No level labelled \"%s\" in entry \"%s\"\n" n + entry.ename; + flush stderr; + failwith "Grammar.extend" + | lev :: levs -> + if is_level_labelled n lev then [lev], empty_lev, levs + else + let (levs1, rlev, levs2) = get levs in lev :: levs1, rlev, levs2 + in + get levs + | None -> + match levs with + lev :: levs -> [], change_lev ~warning lev "<top>", levs + | [] -> [], empty_lev, [] + +let change_to_self entry = + function + Snterm e when e == entry -> Sself + | x -> x + +let get_initial entry = + function + Sself :: symbols -> true, symbols + | symbols -> false, symbols + +let insert_tokens gram symbols = + let rec insert = + function + | Slist0 s -> insert s + | Slist1 s -> insert s + | Slist0sep (s, t, _) -> insert s; insert t + | Slist1sep (s, t, _) -> insert s; insert t + | Sopt s -> insert s + | Stree t -> tinsert t + | Stoken ("ANY", _) -> () + | Stoken tok -> + gram.glexer.Plexing.tok_using tok; + let r = + try Hashtbl.find gram.gtokens tok with + Not_found -> let r = ref 0 in Hashtbl.add gram.gtokens tok r; r + in + incr r + | Snterm _ | Snterml (_, _) | Snext | Sself -> () + and tinsert = + function + Node {node = s; brother = bro; son = son} -> + insert s; tinsert bro; tinsert son + | LocAct (_, _) | DeadEnd -> () + in + List.iter insert symbols + +let levels_of_rules ~warning entry position rules = + let elev = + match entry.edesc with + Dlevels elev -> elev + | Dparser _ -> + eprintf "Error: entry not extensible: \"%s\"\n" entry.ename; + flush stderr; + failwith "Grammar.extend" + in + if rules = [] then elev + else + let (levs1, make_lev, levs2) = get_level ~warning entry position elev in + let (levs, _) = + List.fold_left + (fun (levs, make_lev) (lname, assoc, level) -> + let lev = make_lev lname assoc in + let lev = + List.fold_left + (fun lev (symbols, action) -> + let symbols = List.map (change_to_self entry) symbols in + let (e1, symbols) = get_initial entry symbols in + insert_tokens entry.egram symbols; + insert_level ~warning entry.ename e1 symbols action lev) + lev level + in + lev :: levs, empty_lev) + ([], make_lev) rules + in + levs1 @ List.rev levs @ levs2 + +let logically_eq_symbols entry = + let rec eq_symbols s1 s2 = + match s1, s2 with + Snterm e1, Snterm e2 -> e1.ename = e2.ename + | Snterm e1, Sself -> e1.ename = entry.ename + | Sself, Snterm e2 -> entry.ename = e2.ename + | Snterml (e1, l1), Snterml (e2, l2) -> e1.ename = e2.ename && l1 = l2 + | Slist0 s1, Slist0 s2 -> eq_symbols s1 s2 + | Slist0sep (s1, sep1, b1), Slist0sep (s2, sep2, b2) -> + eq_symbols s1 s2 && eq_symbols sep1 sep2 && b1 = b2 + | Slist1 s1, Slist1 s2 -> eq_symbols s1 s2 + | Slist1sep (s1, sep1, b1), Slist1sep (s2, sep2, b2) -> + eq_symbols s1 s2 && eq_symbols sep1 sep2 && b1 = b2 + | Sopt s1, Sopt s2 -> eq_symbols s1 s2 + | Stree t1, Stree t2 -> eq_trees t1 t2 + | _ -> s1 = s2 + and eq_trees t1 t2 = + match t1, t2 with + Node n1, Node n2 -> + eq_symbols n1.node n2.node && eq_trees n1.son n2.son && + eq_trees n1.brother n2.brother + | (LocAct (_, _) | DeadEnd), (LocAct (_, _) | DeadEnd) -> true + | _ -> false + in + eq_symbols + +(* [delete_rule_in_tree] returns + [Some (dsl, t)] if success + [dsl] = + Some (list of deleted nodes) if branch deleted + None if action replaced by previous version of action + [t] = remaining tree + [None] if failure *) + +let delete_rule_in_tree entry = + let rec delete_in_tree symbols tree = + match symbols, tree with + s :: sl, Node n -> + if logically_eq_symbols entry s n.node then delete_son sl n + else + begin match delete_in_tree symbols n.brother with + Some (dsl, t) -> + Some (dsl, Node {node = n.node; son = n.son; brother = t}) + | None -> None + end + | s :: sl, _ -> None + | [], Node n -> + begin match delete_in_tree [] n.brother with + Some (dsl, t) -> + Some (dsl, Node {node = n.node; son = n.son; brother = t}) + | None -> None + end + | [], DeadEnd -> None + | [], LocAct (_, []) -> Some (Some [], DeadEnd) + | [], LocAct (_, action :: list) -> Some (None, LocAct (action, list)) + and delete_son sl n = + match delete_in_tree sl n.son with + Some (Some dsl, DeadEnd) -> Some (Some (n.node :: dsl), n.brother) + | Some (Some dsl, t) -> + let t = Node {node = n.node; son = t; brother = n.brother} in + Some (Some (n.node :: dsl), t) + | Some (None, t) -> + let t = Node {node = n.node; son = t; brother = n.brother} in + Some (None, t) + | None -> None + in + delete_in_tree + +let rec decr_keyw_use gram = + function + Stoken tok -> + let r = Hashtbl.find gram.gtokens tok in + decr r; + if !r == 0 then + begin + Hashtbl.remove gram.gtokens tok; + gram.glexer.Plexing.tok_removing tok + end + | Slist0 s -> decr_keyw_use gram s + | Slist1 s -> decr_keyw_use gram s + | Slist0sep (s1, s2, _) -> decr_keyw_use gram s1; decr_keyw_use gram s2 + | Slist1sep (s1, s2, _) -> decr_keyw_use gram s1; decr_keyw_use gram s2 + | Sopt s -> decr_keyw_use gram s + | Stree t -> decr_keyw_use_in_tree gram t + | Sself | Snext | Snterm _ | Snterml (_, _) -> () +and decr_keyw_use_in_tree gram = + function + DeadEnd | LocAct (_, _) -> () + | Node n -> + decr_keyw_use gram n.node; + decr_keyw_use_in_tree gram n.son; + decr_keyw_use_in_tree gram n.brother + +let rec delete_rule_in_suffix entry symbols = + function + lev :: levs -> + begin match delete_rule_in_tree entry symbols lev.lsuffix with + Some (dsl, t) -> + begin match dsl with + Some dsl -> List.iter (decr_keyw_use entry.egram) dsl + | None -> () + end; + begin match t with + DeadEnd when lev.lprefix == DeadEnd -> levs + | _ -> + let lev = + {assoc = lev.assoc; lname = lev.lname; lsuffix = t; + lprefix = lev.lprefix} + in + lev :: levs + end + | None -> + let levs = delete_rule_in_suffix entry symbols levs in lev :: levs + end + | [] -> raise Not_found + +let rec delete_rule_in_prefix entry symbols = + function + lev :: levs -> + begin match delete_rule_in_tree entry symbols lev.lprefix with + Some (dsl, t) -> + begin match dsl with + Some dsl -> List.iter (decr_keyw_use entry.egram) dsl + | None -> () + end; + begin match t with + DeadEnd when lev.lsuffix == DeadEnd -> levs + | _ -> + let lev = + {assoc = lev.assoc; lname = lev.lname; lsuffix = lev.lsuffix; + lprefix = t} + in + lev :: levs + end + | None -> + let levs = delete_rule_in_prefix entry symbols levs in lev :: levs + end + | [] -> raise Not_found + +let delete_rule_in_level_list entry symbols levs = + match symbols with + Sself :: symbols -> delete_rule_in_suffix entry symbols levs + | Snterm e :: symbols when e == entry -> + delete_rule_in_suffix entry symbols levs + | _ -> delete_rule_in_prefix entry symbols levs diff --git a/gramlib/gramext.mli b/gramlib/gramext.mli new file mode 100644 index 0000000000..f1e294fb4c --- /dev/null +++ b/gramlib/gramext.mli @@ -0,0 +1,65 @@ +(* camlp5r *) +(* gramext.mli,v *) +(* Copyright (c) INRIA 2007-2017 *) + +type 'a parser_t = 'a Stream.t -> Obj.t + +type 'te grammar = + { gtokens : (Plexing.pattern, int ref) Hashtbl.t; + glexer : 'te Plexing.lexer } + +type 'te g_entry = + { egram : 'te grammar; + ename : string; + elocal : bool; + mutable estart : int -> 'te parser_t; + mutable econtinue : int -> int -> Obj.t -> 'te parser_t; + mutable edesc : 'te g_desc } +and 'te g_desc = + Dlevels of 'te g_level list + | Dparser of 'te parser_t +and 'te g_level = + { assoc : g_assoc; + lname : string option; + lsuffix : 'te g_tree; + lprefix : 'te g_tree } +and g_assoc = NonA | RightA | LeftA +and 'te g_symbol = + | Snterm of 'te g_entry + | Snterml of 'te g_entry * string + | Slist0 of 'te g_symbol + | Slist0sep of 'te g_symbol * 'te g_symbol * bool + | Slist1 of 'te g_symbol + | Slist1sep of 'te g_symbol * 'te g_symbol * bool + | Sopt of 'te g_symbol + | Sself + | Snext + | Stoken of Plexing.pattern + | Stree of 'te g_tree +and g_action = Obj.t +and 'te g_tree = + Node of 'te g_node + | LocAct of g_action * g_action list + | DeadEnd +and 'te g_node = + { node : 'te g_symbol; son : 'te g_tree; brother : 'te g_tree } +and err_fun = unit -> string + +type position = + First + | Last + | Before of string + | After of string + | Level of string + +val levels_of_rules : warning:(string -> unit) option -> + 'te g_entry -> position option -> + (string option * g_assoc option * ('te g_symbol list * g_action) list) + list -> + 'te g_level list + +val srules : warning:(string -> unit) option -> ('te g_symbol list * g_action) list -> 'te g_symbol +val eq_symbol : 'a g_symbol -> 'a g_symbol -> bool + +val delete_rule_in_level_list : + 'te g_entry -> 'te g_symbol list -> 'te g_level list -> 'te g_level list diff --git a/gramlib/gramlib.mllib b/gramlib/gramlib.mllib new file mode 100644 index 0000000000..4c915b2b05 --- /dev/null +++ b/gramlib/gramlib.mllib @@ -0,0 +1,4 @@ +Ploc +Plexing +Gramext +Grammar diff --git a/gramlib/grammar.ml b/gramlib/grammar.ml new file mode 100644 index 0000000000..0ad11d075f --- /dev/null +++ b/gramlib/grammar.ml @@ -0,0 +1,864 @@ +(* camlp5r *) +(* grammar.ml,v *) +(* Copyright (c) INRIA 2007-2017 *) + +open Gramext +open Format + +external gramext_action : 'a -> g_action = "%identity" + +let rec flatten_tree = + function + DeadEnd -> [] + | LocAct (_, _) -> [[]] + | Node {node = n; brother = b; son = s} -> + List.map (fun l -> n :: l) (flatten_tree s) @ flatten_tree b + +let utf8_print = ref true + +let utf8_string_escaped s = + let b = Buffer.create (String.length s) in + let rec loop i = + if i = String.length s then Buffer.contents b + else + begin + begin match s.[i] with + '"' -> Buffer.add_string b "\\\"" + | '\\' -> Buffer.add_string b "\\\\" + | '\n' -> Buffer.add_string b "\\n" + | '\t' -> Buffer.add_string b "\\t" + | '\r' -> Buffer.add_string b "\\r" + | '\b' -> Buffer.add_string b "\\b" + | c -> Buffer.add_char b c + end; + loop (i + 1) + end + in + loop 0 + +let string_escaped s = + if !utf8_print then utf8_string_escaped s else String.escaped s + +let print_str ppf s = fprintf ppf "\"%s\"" (string_escaped s) + +let rec print_symbol ppf = + function + | Slist0 s -> fprintf ppf "LIST0 %a" print_symbol1 s + | Slist0sep (s, t, osep) -> + fprintf ppf "LIST0 %a SEP %a%s" print_symbol1 s print_symbol1 t + (if osep then " OPT_SEP" else "") + | Slist1 s -> fprintf ppf "LIST1 %a" print_symbol1 s + | Slist1sep (s, t, osep) -> + fprintf ppf "LIST1 %a SEP %a%s" print_symbol1 s print_symbol1 t + (if osep then " OPT_SEP" else "") + | Sopt s -> fprintf ppf "OPT %a" print_symbol1 s + | Stoken (con, prm) when con <> "" && prm <> "" -> + fprintf ppf "%s@ %a" con print_str prm + | Snterml (e, l) -> + fprintf ppf "%s%s@ LEVEL@ %a" e.ename (if e.elocal then "*" else "") + print_str l + | Snterm _ | Snext | Sself | Stoken _ | Stree _ as s -> + print_symbol1 ppf s +and print_symbol1 ppf = + function + | Snterm e -> fprintf ppf "%s%s" e.ename (if e.elocal then "*" else "") + | Sself -> pp_print_string ppf "SELF" + | Snext -> pp_print_string ppf "NEXT" + | Stoken ("", s) -> print_str ppf s + | Stoken (con, "") -> pp_print_string ppf con + | Stree t -> print_level ppf pp_print_space (flatten_tree t) + | Snterml (_, _) | Slist0 _ | Slist0sep (_, _, _) | + Slist1 _ | Slist1sep (_, _, _) | Sopt _ | Stoken _ as s -> + fprintf ppf "(%a)" print_symbol s +and print_rule ppf symbols = + fprintf ppf "@[<hov 0>"; + let _ = + List.fold_left + (fun sep symbol -> + fprintf ppf "%t%a" sep print_symbol symbol; + fun ppf -> fprintf ppf ";@ ") + (fun ppf -> ()) symbols + in + fprintf ppf "@]" +and print_level ppf pp_print_space rules = + fprintf ppf "@[<hov 0>[ "; + let _ = + List.fold_left + (fun sep rule -> + fprintf ppf "%t%a" sep print_rule rule; + fun ppf -> fprintf ppf "%a| " pp_print_space ()) + (fun ppf -> ()) rules + in + fprintf ppf " ]@]" + +let print_levels ppf elev = + let _ = + List.fold_left + (fun sep lev -> + let rules = + List.map (fun t -> Sself :: t) (flatten_tree lev.lsuffix) @ + flatten_tree lev.lprefix + in + fprintf ppf "%t@[<hov 2>" sep; + begin match lev.lname with + Some n -> fprintf ppf "%a@;<1 2>" print_str n + | None -> () + end; + begin match lev.assoc with + LeftA -> fprintf ppf "LEFTA" + | RightA -> fprintf ppf "RIGHTA" + | NonA -> fprintf ppf "NONA" + end; + fprintf ppf "@]@;<1 2>"; + print_level ppf pp_force_newline rules; + fun ppf -> fprintf ppf "@,| ") + (fun ppf -> ()) elev + in + () + +let print_entry ppf e = + fprintf ppf "@[<v 0>[ "; + begin match e.edesc with + Dlevels elev -> print_levels ppf elev + | Dparser _ -> fprintf ppf "<parser>" + end; + fprintf ppf " ]@]" + +let floc = ref (fun _ -> failwith "internal error when computing location") + +let loc_of_token_interval bp ep = + if bp == ep then + if bp == 0 then Ploc.dummy else Ploc.after (!floc (bp - 1)) 0 1 + else + let loc1 = !floc bp in let loc2 = !floc (pred ep) in Loc.merge loc1 loc2 + +let name_of_symbol entry = + function + Snterm e -> "[" ^ e.ename ^ "]" + | Snterml (e, l) -> "[" ^ e.ename ^ " level " ^ l ^ "]" + | Sself | Snext -> "[" ^ entry.ename ^ "]" + | Stoken tok -> entry.egram.glexer.Plexing.tok_text tok + | _ -> "???" + +let rec get_token_list entry rev_tokl last_tok tree = + match tree with + Node {node = Stoken tok; son = son; brother = DeadEnd} -> + get_token_list entry (last_tok :: rev_tokl) tok son + | _ -> if rev_tokl = [] then None else Some (rev_tokl, last_tok, tree) + +let rec name_of_symbol_failed entry = + function + | Slist0 s -> name_of_symbol_failed entry s + | Slist0sep (s, _, _) -> name_of_symbol_failed entry s + | Slist1 s -> name_of_symbol_failed entry s + | Slist1sep (s, _, _) -> name_of_symbol_failed entry s + | Sopt s -> name_of_symbol_failed entry s + | Stree t -> name_of_tree_failed entry t + | s -> name_of_symbol entry s +and name_of_tree_failed entry = + function + Node {node = s; brother = bro; son = son} -> + let tokl = + match s with + Stoken tok -> get_token_list entry [] tok son + | _ -> None + in + begin match tokl with + None -> + let txt = name_of_symbol_failed entry s in + let txt = + match s, son with + Sopt _, Node _ -> txt ^ " or " ^ name_of_tree_failed entry son + | _ -> txt + in + let txt = + match bro with + DeadEnd | LocAct (_, _) -> txt + | Node _ -> txt ^ " or " ^ name_of_tree_failed entry bro + in + txt + | Some (rev_tokl, last_tok, son) -> + List.fold_left + (fun s tok -> + (if s = "" then "" else s ^ " ") ^ + entry.egram.glexer.Plexing.tok_text tok) + "" (List.rev (last_tok :: rev_tokl)) + end + | DeadEnd | LocAct (_, _) -> "???" + +let tree_failed entry prev_symb_result prev_symb tree = + let txt = name_of_tree_failed entry tree in + let txt = + match prev_symb with + Slist0 s -> + let txt1 = name_of_symbol_failed entry s in + txt1 ^ " or " ^ txt ^ " expected" + | Slist1 s -> + let txt1 = name_of_symbol_failed entry s in + txt1 ^ " or " ^ txt ^ " expected" + | Slist0sep (s, sep, _) -> + begin match Obj.magic prev_symb_result with + [] -> + let txt1 = name_of_symbol_failed entry s in + txt1 ^ " or " ^ txt ^ " expected" + | _ -> + let txt1 = name_of_symbol_failed entry sep in + txt1 ^ " or " ^ txt ^ " expected" + end + | Slist1sep (s, sep, _) -> + begin match Obj.magic prev_symb_result with + [] -> + let txt1 = name_of_symbol_failed entry s in + txt1 ^ " or " ^ txt ^ " expected" + | _ -> + let txt1 = name_of_symbol_failed entry sep in + txt1 ^ " or " ^ txt ^ " expected" + end + | Sopt _ | Stree _ -> txt ^ " expected" + | _ -> txt ^ " expected after " ^ name_of_symbol_failed entry prev_symb + in + txt ^ " (in [" ^ entry.ename ^ "])" + +let symb_failed entry prev_symb_result prev_symb symb = + let tree = Node {node = symb; brother = DeadEnd; son = DeadEnd} in + tree_failed entry prev_symb_result prev_symb tree + +external app : Obj.t -> 'a = "%identity" + +let is_level_labelled n lev = + match lev.lname with + Some n1 -> n = n1 + | None -> false + +let level_number entry lab = + let rec lookup levn = + function + [] -> failwith ("unknown level " ^ lab) + | lev :: levs -> + if is_level_labelled lab lev then levn else lookup (succ levn) levs + in + match entry.edesc with + Dlevels elev -> lookup 0 elev + | Dparser _ -> raise Not_found + +let rec top_symb entry = + function + Sself | Snext -> Snterm entry + | Snterml (e, _) -> Snterm e + | Slist1sep (s, sep, b) -> Slist1sep (top_symb entry s, sep, b) + | _ -> raise Stream.Failure + +let entry_of_symb entry = + function + Sself | Snext -> entry + | Snterm e -> e + | Snterml (e, _) -> e + | _ -> raise Stream.Failure + +let top_tree entry = + function + Node {node = s; brother = bro; son = son} -> + Node {node = top_symb entry s; brother = bro; son = son} + | LocAct (_, _) | DeadEnd -> raise Stream.Failure + +let skip_if_empty bp p strm = + if Stream.count strm == bp then gramext_action (fun a -> p strm) + else raise Stream.Failure + +let continue entry bp a s son p1 (strm__ : _ Stream.t) = + let a = (entry_of_symb entry s).econtinue 0 bp a strm__ in + let act = + try p1 strm__ with + Stream.Failure -> raise (Stream.Error (tree_failed entry a s son)) + in + gramext_action (fun _ -> app act a) + +let do_recover parser_of_tree entry nlevn alevn bp a s son + (strm__ : _ Stream.t) = + try parser_of_tree entry nlevn alevn (top_tree entry son) strm__ with + Stream.Failure -> + try + skip_if_empty bp (fun (strm__ : _ Stream.t) -> raise Stream.Failure) + strm__ + with Stream.Failure -> + continue entry bp a s son (parser_of_tree entry nlevn alevn son) + strm__ + +let recover parser_of_tree entry nlevn alevn bp a s son strm = + do_recover parser_of_tree entry nlevn alevn bp a s son strm + +let token_count = ref 0 + +let peek_nth n strm = + let list = Stream.npeek n strm in + token_count := Stream.count strm + n; + let rec loop list n = + match list, n with + x :: _, 1 -> Some x + | _ :: l, n -> loop l (n - 1) + | [], _ -> None + in + loop list n + +let item_skipped = ref false + +let call_and_push ps al strm = + item_skipped := false; + let a = ps strm in + let al = if !item_skipped then al else a :: al in item_skipped := false; al + +let token_ematch gram tok = + let tematch = gram.glexer.Plexing.tok_match tok in + fun tok -> Obj.repr (tematch tok : string) + +let rec parser_of_tree entry nlevn alevn = + function + DeadEnd -> (fun (strm__ : _ Stream.t) -> raise Stream.Failure) + | LocAct (act, _) -> (fun (strm__ : _ Stream.t) -> act) + | Node {node = Sself; son = LocAct (act, _); brother = DeadEnd} -> + (fun (strm__ : _ Stream.t) -> + let a = entry.estart alevn strm__ in app act a) + | Node {node = Sself; son = LocAct (act, _); brother = bro} -> + let p2 = parser_of_tree entry nlevn alevn bro in + (fun (strm__ : _ Stream.t) -> + match + try Some (entry.estart alevn strm__) with Stream.Failure -> None + with + Some a -> app act a + | _ -> p2 strm__) + | Node {node = s; son = son; brother = DeadEnd} -> + let tokl = + match s with + Stoken tok -> get_token_list entry [] tok son + | _ -> None + in + begin match tokl with + None -> + let ps = parser_of_symbol entry nlevn s in + let p1 = parser_of_tree entry nlevn alevn son in + let p1 = parser_cont p1 entry nlevn alevn s son in + (fun (strm__ : _ Stream.t) -> + let bp = Stream.count strm__ in + let a = ps strm__ in + let act = + try p1 bp a strm__ with + Stream.Failure -> + raise (Stream.Error (tree_failed entry a s son)) + in + app act a) + | Some (rev_tokl, last_tok, son) -> + let lt = Stoken last_tok in + let p1 = parser_of_tree entry nlevn alevn son in + let p1 = parser_cont p1 entry nlevn alevn lt son in + parser_of_token_list entry s son p1 + (fun (strm__ : _ Stream.t) -> raise Stream.Failure) rev_tokl + last_tok + end + | Node {node = s; son = son; brother = bro} -> + let tokl = + match s with + Stoken tok -> get_token_list entry [] tok son + | _ -> None + in + match tokl with + None -> + let ps = parser_of_symbol entry nlevn s in + let p1 = parser_of_tree entry nlevn alevn son in + let p1 = parser_cont p1 entry nlevn alevn s son in + let p2 = parser_of_tree entry nlevn alevn bro in + (fun (strm : _ Stream.t) -> + let bp = Stream.count strm in + match try Some (ps strm) with Stream.Failure -> None with + Some a -> + begin match + (try Some (p1 bp a strm) with Stream.Failure -> None) + with + Some act -> app act a + | None -> raise (Stream.Error (tree_failed entry a s son)) + end + | None -> p2 strm) + | Some (rev_tokl, last_tok, son) -> + let lt = Stoken last_tok in + let p2 = parser_of_tree entry nlevn alevn bro in + let p1 = parser_of_tree entry nlevn alevn son in + let p1 = parser_cont p1 entry nlevn alevn lt son in + let p1 = + parser_of_token_list entry lt son p1 p2 rev_tokl last_tok + in + fun (strm__ : _ Stream.t) -> + try p1 strm__ with Stream.Failure -> p2 strm__ +and parser_cont p1 entry nlevn alevn s son bp a (strm__ : _ Stream.t) = + try p1 strm__ with + Stream.Failure -> + recover parser_of_tree entry nlevn alevn bp a s son strm__ +and parser_of_token_list entry s son p1 p2 rev_tokl last_tok = + let plast = + let n = List.length rev_tokl + 1 in + let tematch = token_ematch entry.egram last_tok in + let ps strm = + match peek_nth n strm with + Some tok -> + let r = tematch tok in + for _i = 1 to n do Stream.junk strm done; Obj.repr r + | None -> raise Stream.Failure + in + fun (strm : _ Stream.t) -> + let bp = Stream.count strm in + let a = ps strm in + match try Some (p1 bp a strm) with Stream.Failure -> None with + Some act -> app act a + | None -> raise (Stream.Error (tree_failed entry a s son)) + in + match List.rev rev_tokl with + [] -> (fun (strm__ : _ Stream.t) -> plast strm__) + | tok :: tokl -> + let tematch = token_ematch entry.egram tok in + let ps strm = + match peek_nth 1 strm with + Some tok -> tematch tok + | None -> raise Stream.Failure + in + let p1 = + let rec loop n = + function + [] -> plast + | tok :: tokl -> + let tematch = token_ematch entry.egram tok in + let ps strm = + match peek_nth n strm with + Some tok -> tematch tok + | None -> raise Stream.Failure + in + let p1 = loop (n + 1) tokl in + fun (strm__ : _ Stream.t) -> + let a = ps strm__ in let act = p1 strm__ in app act a + in + loop 2 tokl + in + fun (strm__ : _ Stream.t) -> + let a = ps strm__ in let act = p1 strm__ in app act a +and parser_of_symbol entry nlevn = + function + | Slist0 s -> + let ps = call_and_push (parser_of_symbol entry nlevn s) in + let rec loop al (strm__ : _ Stream.t) = + match try Some (ps al strm__) with Stream.Failure -> None with + Some al -> loop al strm__ + | _ -> al + in + (fun (strm__ : _ Stream.t) -> + let a = loop [] strm__ in Obj.repr (List.rev a)) + | Slist0sep (symb, sep, false) -> + let ps = call_and_push (parser_of_symbol entry nlevn symb) in + let pt = parser_of_symbol entry nlevn sep in + let rec kont al (strm__ : _ Stream.t) = + match try Some (pt strm__) with Stream.Failure -> None with + Some v -> + let al = + try ps al strm__ with + Stream.Failure -> + raise (Stream.Error (symb_failed entry v sep symb)) + in + kont al strm__ + | _ -> al + in + (fun (strm__ : _ Stream.t) -> + match try Some (ps [] strm__) with Stream.Failure -> None with + Some al -> let a = kont al strm__ in Obj.repr (List.rev a) + | _ -> Obj.repr []) + | Slist0sep (symb, sep, true) -> + let ps = call_and_push (parser_of_symbol entry nlevn symb) in + let pt = parser_of_symbol entry nlevn sep in + let rec kont al (strm__ : _ Stream.t) = + match try Some (pt strm__) with Stream.Failure -> None with + Some v -> + begin match + (try Some (ps al strm__) with Stream.Failure -> None) + with + Some al -> kont al strm__ + | _ -> al + end + | _ -> al + in + (fun (strm__ : _ Stream.t) -> + match try Some (ps [] strm__) with Stream.Failure -> None with + Some al -> let a = kont al strm__ in Obj.repr (List.rev a) + | _ -> Obj.repr []) + | Slist1 s -> + let ps = call_and_push (parser_of_symbol entry nlevn s) in + let rec loop al (strm__ : _ Stream.t) = + match try Some (ps al strm__) with Stream.Failure -> None with + Some al -> loop al strm__ + | _ -> al + in + (fun (strm__ : _ Stream.t) -> + let al = ps [] strm__ in + let a = loop al strm__ in Obj.repr (List.rev a)) + | Slist1sep (symb, sep, false) -> + let ps = call_and_push (parser_of_symbol entry nlevn symb) in + let pt = parser_of_symbol entry nlevn sep in + let rec kont al (strm__ : _ Stream.t) = + match try Some (pt strm__) with Stream.Failure -> None with + Some v -> + let al = + try ps al strm__ with + Stream.Failure -> + let a = + try parse_top_symb entry symb strm__ with + Stream.Failure -> + raise (Stream.Error (symb_failed entry v sep symb)) + in + a :: al + in + kont al strm__ + | _ -> al + in + (fun (strm__ : _ Stream.t) -> + let al = ps [] strm__ in + let a = kont al strm__ in Obj.repr (List.rev a)) + | Slist1sep (symb, sep, true) -> + let ps = call_and_push (parser_of_symbol entry nlevn symb) in + let pt = parser_of_symbol entry nlevn sep in + let rec kont al (strm__ : _ Stream.t) = + match try Some (pt strm__) with Stream.Failure -> None with + Some v -> + begin match + (try Some (ps al strm__) with Stream.Failure -> None) + with + Some al -> kont al strm__ + | _ -> + match + try Some (parse_top_symb entry symb strm__) with + Stream.Failure -> None + with + Some a -> kont (a :: al) strm__ + | _ -> al + end + | _ -> al + in + (fun (strm__ : _ Stream.t) -> + let al = ps [] strm__ in + let a = kont al strm__ in Obj.repr (List.rev a)) + | Sopt s -> + let ps = parser_of_symbol entry nlevn s in + (fun (strm__ : _ Stream.t) -> + match try Some (ps strm__) with Stream.Failure -> None with + Some a -> Obj.repr (Some a) + | _ -> Obj.repr None) + | Stree t -> + let pt = parser_of_tree entry 1 0 t in + (fun (strm__ : _ Stream.t) -> + let bp = Stream.count strm__ in + let a = pt strm__ in + let ep = Stream.count strm__ in + let loc = loc_of_token_interval bp ep in app a loc) + | Snterm e -> (fun (strm__ : _ Stream.t) -> e.estart 0 strm__) + | Snterml (e, l) -> + (fun (strm__ : _ Stream.t) -> e.estart (level_number e l) strm__) + | Sself -> (fun (strm__ : _ Stream.t) -> entry.estart 0 strm__) + | Snext -> (fun (strm__ : _ Stream.t) -> entry.estart nlevn strm__) + | Stoken tok -> parser_of_token entry tok +and parser_of_token entry tok = + let f = entry.egram.glexer.Plexing.tok_match tok in + fun strm -> + match Stream.peek strm with + Some tok -> let r = f tok in Stream.junk strm; Obj.repr r + | None -> raise Stream.Failure +and parse_top_symb entry symb = parser_of_symbol entry 0 (top_symb entry symb) + +let rec start_parser_of_levels entry clevn = + function + [] -> (fun levn (strm__ : _ Stream.t) -> raise Stream.Failure) + | lev :: levs -> + let p1 = start_parser_of_levels entry (succ clevn) levs in + match lev.lprefix with + DeadEnd -> p1 + | tree -> + let alevn = + match lev.assoc with + LeftA | NonA -> succ clevn + | RightA -> clevn + in + let p2 = parser_of_tree entry (succ clevn) alevn tree in + match levs with + [] -> + (fun levn strm -> + (* this code should be there but is commented to preserve + compatibility with previous versions... with this code, + the grammar entry e: [[ "x"; a = e | "y" ]] should fail + because it should be: e: [RIGHTA[ "x"; a = e | "y" ]]... + if levn > clevn then match strm with parser [] + else + *) + let (strm__ : _ Stream.t) = strm in + let bp = Stream.count strm__ in + let act = p2 strm__ in + let ep = Stream.count strm__ in + let a = app act (loc_of_token_interval bp ep) in + entry.econtinue levn bp a strm) + | _ -> + fun levn strm -> + if levn > clevn then p1 levn strm + else + let (strm__ : _ Stream.t) = strm in + let bp = Stream.count strm__ in + match try Some (p2 strm__) with Stream.Failure -> None with + Some act -> + let ep = Stream.count strm__ in + let a = app act (loc_of_token_interval bp ep) in + entry.econtinue levn bp a strm + | _ -> p1 levn strm__ + +let rec continue_parser_of_levels entry clevn = + function + [] -> (fun levn bp a (strm__ : _ Stream.t) -> raise Stream.Failure) + | lev :: levs -> + let p1 = continue_parser_of_levels entry (succ clevn) levs in + match lev.lsuffix with + DeadEnd -> p1 + | tree -> + let alevn = + match lev.assoc with + LeftA | NonA -> succ clevn + | RightA -> clevn + in + let p2 = parser_of_tree entry (succ clevn) alevn tree in + fun levn bp a strm -> + if levn > clevn then p1 levn bp a strm + else + let (strm__ : _ Stream.t) = strm in + try p1 levn bp a strm__ with + Stream.Failure -> + let act = p2 strm__ in + let ep = Stream.count strm__ in + let a = app act a (loc_of_token_interval bp ep) in + entry.econtinue levn bp a strm + +let continue_parser_of_entry entry = + match entry.edesc with + Dlevels elev -> + let p = continue_parser_of_levels entry 0 elev in + (fun levn bp a (strm__ : _ Stream.t) -> + try p levn bp a strm__ with Stream.Failure -> a) + | Dparser p -> fun levn bp a (strm__ : _ Stream.t) -> raise Stream.Failure + +let empty_entry ename levn strm = + raise (Stream.Error ("entry [" ^ ename ^ "] is empty")) + +let start_parser_of_entry entry = + match entry.edesc with + Dlevels [] -> empty_entry entry.ename + | Dlevels elev -> start_parser_of_levels entry 0 elev + | Dparser p -> fun levn strm -> p strm + +(* Extend syntax *) + +let init_entry_functions entry = + entry.estart <- + (fun lev strm -> + let f = start_parser_of_entry entry in entry.estart <- f; f lev strm); + entry.econtinue <- + (fun lev bp a strm -> + let f = continue_parser_of_entry entry in + entry.econtinue <- f; f lev bp a strm) + +let extend_entry ~warning entry position rules = + try + let elev = Gramext.levels_of_rules ~warning entry position rules in + entry.edesc <- Dlevels elev; init_entry_functions entry + with Plexing.Error s -> + Printf.eprintf "Lexer initialization error:\n- %s\n" s; + flush stderr; + failwith "Grammar.extend" + +(* Deleting a rule *) + +let delete_rule entry sl = + match entry.edesc with + Dlevels levs -> + let levs = Gramext.delete_rule_in_level_list entry sl levs in + entry.edesc <- Dlevels levs; + entry.estart <- + (fun lev strm -> + let f = start_parser_of_entry entry in + entry.estart <- f; f lev strm); + entry.econtinue <- + (fun lev bp a strm -> + let f = continue_parser_of_entry entry in + entry.econtinue <- f; f lev bp a strm) + | Dparser _ -> () + +(* Normal interface *) + +let create_toktab () = Hashtbl.create 301 +let gcreate glexer = + {gtokens = create_toktab (); glexer = glexer } + +let tokens g con = + let list = ref [] in + Hashtbl.iter + (fun (p_con, p_prm) c -> if p_con = con then list := (p_prm, !c) :: !list) + g.gtokens; + !list + +type 'te gen_parsable = + { pa_chr_strm : char Stream.t; + pa_tok_strm : 'te Stream.t; + pa_loc_func : Plexing.location_function } + +let parse_parsable entry p = + let efun = entry.estart 0 in + let ts = p.pa_tok_strm in + let cs = p.pa_chr_strm in + let fun_loc = p.pa_loc_func in + let restore = + let old_floc = !floc in + let old_tc = !token_count in + fun () -> floc := old_floc; token_count := old_tc + in + let get_loc () = + try + let cnt = Stream.count ts in + let loc = fun_loc cnt in + if !token_count - 1 <= cnt then loc + else Loc.merge loc (fun_loc (!token_count - 1)) + with Failure _ -> Ploc.make_unlined (Stream.count cs, Stream.count cs + 1) + in + floc := fun_loc; + token_count := 0; + try let r = efun ts in restore (); r with + Stream.Failure -> + let loc = get_loc () in + restore (); + Ploc.raise loc (Stream.Error ("illegal begin of " ^ entry.ename)) + | Stream.Error _ as exc -> + let loc = get_loc () in restore (); Ploc.raise loc exc + | exc -> + let loc = Stream.count cs, Stream.count cs + 1 in + restore (); Ploc.raise (Ploc.make_unlined loc) exc + +(* Unsafe *) + +let clear_entry e = + e.estart <- (fun _ (strm__ : _ Stream.t) -> raise Stream.Failure); + e.econtinue <- (fun _ _ _ (strm__ : _ Stream.t) -> raise Stream.Failure); + match e.edesc with + Dlevels _ -> e.edesc <- Dlevels [] + | Dparser _ -> () + +(* Functorial interface *) + +module type GLexerType = sig type te val lexer : te Plexing.lexer end + +module type S = + sig + type te + type parsable + val parsable : char Stream.t -> parsable + val tokens : string -> (string * int) list + module Entry : + sig + type 'a e + val create : string -> 'a e + val parse : 'a e -> parsable -> 'a + val name : 'a e -> string + val of_parser : string -> (te Stream.t -> 'a) -> 'a e + val parse_token_stream : 'a e -> te Stream.t -> 'a + val print : Format.formatter -> 'a e -> unit + end + type ('self, 'a) ty_symbol + type ('self, 'f, 'r) ty_rule + type 'a ty_production + val s_nterm : 'a Entry.e -> ('self, 'a) ty_symbol + val s_nterml : 'a Entry.e -> string -> ('self, 'a) ty_symbol + val s_list0 : ('self, 'a) ty_symbol -> ('self, 'a list) ty_symbol + val s_list0sep : + ('self, 'a) ty_symbol -> ('self, 'b) ty_symbol -> bool -> + ('self, 'a list) ty_symbol + val s_list1 : ('self, 'a) ty_symbol -> ('self, 'a list) ty_symbol + val s_list1sep : + ('self, 'a) ty_symbol -> ('self, 'b) ty_symbol -> bool -> + ('self, 'a list) ty_symbol + val s_opt : ('self, 'a) ty_symbol -> ('self, 'a option) ty_symbol + val s_self : ('self, 'self) ty_symbol + val s_next : ('self, 'self) ty_symbol + val s_token : Plexing.pattern -> ('self, string) ty_symbol + val s_rules : warning:(string -> unit) option -> 'a ty_production list -> ('self, 'a) ty_symbol + val r_stop : ('self, 'r, 'r) ty_rule + val r_next : + ('self, 'a, 'r) ty_rule -> ('self, 'b) ty_symbol -> + ('self, 'b -> 'a, 'r) ty_rule + val production : ('a, 'f, Loc.t -> 'a) ty_rule * 'f -> 'a ty_production + module Unsafe : + sig + val clear_entry : 'a Entry.e -> unit + end + val safe_extend : warning:(string -> unit) option -> + 'a Entry.e -> Gramext.position option -> + (string option * Gramext.g_assoc option * 'a ty_production list) + list -> + unit + val safe_delete_rule : 'a Entry.e -> ('a, 'r, 'f) ty_rule -> unit + end + +module GMake (L : GLexerType) = + struct + type te = L.te + type parsable = te gen_parsable + let gram = gcreate L.lexer + let parsable cs = + let (ts, lf) = L.lexer.Plexing.tok_func cs in + {pa_chr_strm = cs; pa_tok_strm = ts; pa_loc_func = lf} + let tokens = tokens gram + module Entry = + struct + type 'a e = te g_entry + let create n = + {egram = gram; ename = n; elocal = false; estart = empty_entry n; + econtinue = + (fun _ _ _ (strm__ : _ Stream.t) -> raise Stream.Failure); + edesc = Dlevels []} + external obj : 'a e -> te Gramext.g_entry = "%identity" + let parse (e : 'a e) p : 'a = + Obj.magic (parse_parsable e p : Obj.t) + let parse_token_stream (e : 'a e) ts : 'a = + Obj.magic (e.estart 0 ts : Obj.t) + let name e = e.ename + let of_parser n (p : te Stream.t -> 'a) : 'a e = + {egram = gram; ename = n; elocal = false; + estart = (fun _ -> (Obj.magic p : te Stream.t -> Obj.t)); + econtinue = + (fun _ _ _ (strm__ : _ Stream.t) -> raise Stream.Failure); + edesc = Dparser (Obj.magic p : te Stream.t -> Obj.t)} + let print ppf e = fprintf ppf "%a@." print_entry (obj e) + end + type ('self, 'a) ty_symbol = te Gramext.g_symbol + type ('self, 'f, 'r) ty_rule = ('self, Obj.t) ty_symbol list + type 'a ty_production = ('a, Obj.t, Obj.t) ty_rule * Gramext.g_action + let s_nterm e = Snterm e + let s_nterml e l = Snterml (e, l) + let s_list0 s = Slist0 s + let s_list0sep s sep b = Slist0sep (s, sep, b) + let s_list1 s = Slist1 s + let s_list1sep s sep b = Slist1sep (s, sep, b) + let s_opt s = Sopt s + let s_self = Sself + let s_next = Snext + let s_token tok = Stoken tok + let s_rules ~warning (t : Obj.t ty_production list) = Gramext.srules ~warning (Obj.magic t) + let r_stop = [] + let r_next r s = r @ [s] + let production + (p : ('a, 'f, Loc.t -> 'a) ty_rule * 'f) : 'a ty_production = + Obj.magic p + module Unsafe = + struct + let clear_entry = clear_entry + end + let safe_extend ~warning e pos + (r : + (string option * Gramext.g_assoc option * Obj.t ty_production list) + list) = + extend_entry ~warning e pos (Obj.magic r) + let safe_delete_rule e r = delete_rule (Entry.obj e) r + end diff --git a/gramlib/grammar.mli b/gramlib/grammar.mli new file mode 100644 index 0000000000..bde07ddc48 --- /dev/null +++ b/gramlib/grammar.mli @@ -0,0 +1,83 @@ +(* camlp5r *) +(* grammar.mli,v *) +(* Copyright (c) INRIA 2007-2017 *) + +(** Extensible grammars. + + This module implements the Camlp5 extensible grammars system. + Grammars entries can be extended using the [EXTEND] statement, + added by loading the Camlp5 [pa_extend.cmo] file. *) + +(** {6 Functorial interface} *) + + (** Alternative for grammars use. Grammars are no more Ocaml values: + there is no type for them. Modules generated preserve the + rule "an entry cannot call an entry of another grammar" by + normal OCaml typing. *) + +module type GLexerType = sig type te val lexer : te Plexing.lexer end + (** The input signature for the functor [Grammar.GMake]: [te] is the + type of the tokens. *) + +module type S = + sig + type te + type parsable + val parsable : char Stream.t -> parsable + val tokens : string -> (string * int) list + module Entry : + sig + type 'a e + val create : string -> 'a e + val parse : 'a e -> parsable -> 'a + val name : 'a e -> string + val of_parser : string -> (te Stream.t -> 'a) -> 'a e + val parse_token_stream : 'a e -> te Stream.t -> 'a + val print : Format.formatter -> 'a e -> unit + end + type ('self, 'a) ty_symbol + type ('self, 'f, 'r) ty_rule + type 'a ty_production + val s_nterm : 'a Entry.e -> ('self, 'a) ty_symbol + val s_nterml : 'a Entry.e -> string -> ('self, 'a) ty_symbol + val s_list0 : ('self, 'a) ty_symbol -> ('self, 'a list) ty_symbol + val s_list0sep : + ('self, 'a) ty_symbol -> ('self, 'b) ty_symbol -> bool -> + ('self, 'a list) ty_symbol + val s_list1 : ('self, 'a) ty_symbol -> ('self, 'a list) ty_symbol + val s_list1sep : + ('self, 'a) ty_symbol -> ('self, 'b) ty_symbol -> bool -> + ('self, 'a list) ty_symbol + val s_opt : ('self, 'a) ty_symbol -> ('self, 'a option) ty_symbol + val s_self : ('self, 'self) ty_symbol + val s_next : ('self, 'self) ty_symbol + val s_token : Plexing.pattern -> ('self, string) ty_symbol + val s_rules : warning:(string -> unit) option -> 'a ty_production list -> ('self, 'a) ty_symbol + val r_stop : ('self, 'r, 'r) ty_rule + val r_next : + ('self, 'a, 'r) ty_rule -> ('self, 'b) ty_symbol -> + ('self, 'b -> 'a, 'r) ty_rule + val production : ('a, 'f, Loc.t -> 'a) ty_rule * 'f -> 'a ty_production + + module Unsafe : + sig + val clear_entry : 'a Entry.e -> unit + end + val safe_extend : warning:(string -> unit) option -> + 'a Entry.e -> Gramext.position option -> + (string option * Gramext.g_assoc option * 'a ty_production list) + list -> + unit + val safe_delete_rule : 'a Entry.e -> ('a, 'f, 'r) ty_rule -> unit + end + (** Signature type of the functor [Grammar.GMake]. The types and + functions are almost the same than in generic interface, but: +- Grammars are not values. Functions holding a grammar as parameter + do not have this parameter yet. +- The type [parsable] is used in function [parse] instead of + the char stream, avoiding the possible loss of tokens. +- The type of tokens (expressions and patterns) can be any + type (instead of (string * string)); the module parameter + must specify a way to show them as (string * string) *) + +module GMake (L : GLexerType) : S with type te = L.te diff --git a/gramlib/plexing.ml b/gramlib/plexing.ml new file mode 100644 index 0000000000..f99a3c2480 --- /dev/null +++ b/gramlib/plexing.ml @@ -0,0 +1,18 @@ +(* camlp5r *) +(* plexing.ml,v *) +(* Copyright (c) INRIA 2007-2017 *) + +type pattern = string * string + +exception Error of string + +type location_function = int -> Loc.t +type 'te lexer_func = char Stream.t -> 'te Stream.t * location_function + +type 'te lexer = + { tok_func : 'te lexer_func; + tok_using : pattern -> unit; + tok_removing : pattern -> unit; + tok_match : pattern -> 'te -> string; + tok_text : pattern -> string; + } diff --git a/gramlib/plexing.mli b/gramlib/plexing.mli new file mode 100644 index 0000000000..eed4082e00 --- /dev/null +++ b/gramlib/plexing.mli @@ -0,0 +1,37 @@ +(* camlp5r *) +(* plexing.mli,v *) +(* Copyright (c) INRIA 2007-2017 *) + +(** Lexing for Camlp5 grammars. + + This module defines the Camlp5 lexer type to be used in extensible + grammars (see module [Grammar]). It also provides some useful functions + to create lexers. *) + +type pattern = string * string + (* Type for values used by the generated code of the EXTEND + statement to represent terminals in entry rules. +- The first string is the constructor name (must start with + an uppercase character). When it is empty, the second string + is supposed to be a keyword. +- The second string is the constructor parameter. Empty if it + has no parameter (corresponding to the 'wildcard' pattern). +- The way tokens patterns are interpreted to parse tokens is done + by the lexer, function [tok_match] below. *) + +exception Error of string + (** A lexing error exception to be used by lexers. *) + +(** Lexer type *) + +type 'te lexer = + { tok_func : 'te lexer_func; + tok_using : pattern -> unit; + tok_removing : pattern -> unit; + tok_match : pattern -> 'te -> string; + tok_text : pattern -> string; + } +and 'te lexer_func = char Stream.t -> 'te Stream.t * location_function +and location_function = int -> Loc.t + (** The type of a function giving the location of a token in the + source from the token number in the stream (starting from zero). *) diff --git a/gramlib/ploc.ml b/gramlib/ploc.ml new file mode 100644 index 0000000000..9342fc6c1d --- /dev/null +++ b/gramlib/ploc.ml @@ -0,0 +1,26 @@ +(* camlp5r *) +(* ploc.ml,v *) +(* Copyright (c) INRIA 2007-2017 *) + +open Loc + +let make_unlined (bp, ep) = + {fname = InFile ""; line_nb = 1; bol_pos = 0; line_nb_last = -1; bol_pos_last = 0; + bp = bp; ep = ep; comm = ""; ecomm = ""} + +let dummy = + {fname = InFile ""; line_nb = 1; bol_pos = 0; line_nb_last = -1; bol_pos_last = 0; + bp = 0; ep = 0; comm = ""; ecomm = ""} + +(* *) + +let sub loc sh len = {loc with bp = loc.bp + sh; ep = loc.bp + sh + len} +let after loc sh len = {loc with bp = loc.ep + sh; ep = loc.ep + sh + len} +let with_comment loc comm = {loc with comm = comm} + +exception Exc of Loc.t * exn + +let raise loc exc = + match exc with + Exc (_, _) -> raise exc + | _ -> raise (Exc (loc, exc)) diff --git a/gramlib/ploc.mli b/gramlib/ploc.mli new file mode 100644 index 0000000000..100fbc7271 --- /dev/null +++ b/gramlib/ploc.mli @@ -0,0 +1,40 @@ +(* camlp5r *) +(* ploc.mli,v *) +(* Copyright (c) INRIA 2007-2017 *) + +(* located exceptions *) + +exception Exc of Loc.t * exn + (** [Ploc.Exc loc e] is an encapsulation of the exception [e] with + the input location [loc]. To be used to specify a location + for an error. This exception must not be raised by [raise] but + rather by [Ploc.raise] (see below), to prevent the risk of several + encapsulations of [Ploc.Exc]. *) + +val raise : Loc.t -> exn -> 'a + (** [Ploc.raise loc e], if [e] is already the exception [Ploc.Exc], + re-raise it (ignoring the new location [loc]), else raise the + exception [Ploc.Exc loc e]. *) + +val make_unlined : int * int -> Loc.t + (** [Ploc.make_unlined] is like [Ploc.make] except that the line number + is not provided (to be used e.g. when the line number is unknown. *) + +val dummy : Loc.t + (** [Ploc.dummy] is a dummy location, used in situations when location + has no meaning. *) + +(* combining locations *) + +val sub : Loc.t -> int -> int -> Loc.t + (** [Ploc.sub loc sh len] is the location [loc] shifted with [sh] + characters and with length [len]. The previous ending position + of the location is lost. *) + +val after : Loc.t -> int -> int -> Loc.t + (** [Ploc.after loc sh len] is the location just after loc (starting at + the end position of [loc]) shifted with [sh] characters and of length + [len]. *) + +val with_comment : Loc.t -> string -> Loc.t + (** Change the comment part of the given location *) |
