loop 1 l
in
Hashtbl.iter (fun lhs rhs ->
- let Node( head, _ ) = lhs in
+ let Node( head, args ) = lhs in
let Node( tag1, params) = rhs in
let tag2, pos2 = pos_id2 params in
let id1 = h_find ~msg:"7" tag_to_id tag1
- and id2 = h_find ~msg:"8" tag_to_id tag2
- in
+ and id2 = h_find ~msg:"8" tag_to_id tag2 in
let rule_ = id2 lsl 27 in
let rule_ = (rule_ lor id1) lsl 2 in
let rule_ = (rule_ lor pos2) lsl 2 in
- let rule_ = rule_ lor (List.length params) in
+ let rule_ = (rule_ lor (List.length params)) lsl 2 in
+ let rule_ = rule_ lor (List.length args) in
r_array.((h_find ~msg:"9" tag_to_id head) - rules_offset ) <- rule_
) rules;
- let l = Array.length renum_tags in
- let tag32 = Array32.create l 0 in
+ (*let l = Array.length renum_tags in *)
+ (*let tag32 = Array32.create l 0 in
for i = 0 to l - 1 do
Array32.set tag32 i (renum_tags.(i) land 0x7ffffff);
- done;
+ done; *)
(* Remove the non-terminal names from the hash tables *)
let tag_to_id2 = Hashtbl.create 31 in
Hashtbl.iter (fun s i -> if i < rules_offset then Hashtbl.add tag_to_id2 s i)
tag_to_id;
{ start = bv;
- tags = tag32;
- rules = renum_tags;
+ tags = renum_tags;
+ rules = r_array;
rules_offset = rules_offset;
tag_to_id = tag_to_id2;
tag_of_id = Array.sub old_new_mapping 0 rules_offset
let load f =
let cin = open_in f in
- let pr_pos () =
- Printf.eprintf "Position: %i kiB\n" (pos_in cin / 1024)
- in
let read () = Marshal.from_channel cin in
if read () != _GRAMMAR_MAGIC then failwith "Invalid grammar file";
if read () != _GRAMMAR_VERSION then failwith "Deprecated grammar format";
- pr_pos();
let tags : int array = read () in
- pr_pos();
let rules : int array = read () in
- pr_pos();
let rules_offset : int = read () in
- pr_pos();
let tag_to_id : (string, int) Hashtbl.t = read () in
- pr_pos();
let tag_of_id : string array = read () in
- pr_pos();
let fd = Unix.descr_of_in_channel cin in
let pos = pos_in cin in
ignore(Unix.lseek fd pos Unix.SEEK_SET);
}
-type node = [ `Grammar ] Node.t
+type node = [ `Start ] Node.t
-type p_type = [ `Parameter ]
type n_type = [ `NonTerminal ]
type t_type = [ `Terminal ]
-type any_type = [ p_type | n_type | t_type ]
-type symbol = [ any_type ] Node.t
+type r_type = [ `Rule ]
+type any_type = [ n_type | t_type ]
+type rhs = [ r_type ] Node.t
-type p_symbol = p_type Node.t
type n_symbol = n_type Node.t
type t_symbol = t_type Node.t
-type tn_symbol = [ n_type | t_type ] Node.t
+type tn_symbol = [ any_type ] Node.t
+
+type partial =
+ Leaf of node
+ | Node of tn_symbol * partial array
-let is_nil : (t:t_symbol) =
+
+let is_nil (t : t_symbol) =
(Node.to_int t) == 4
let nil_symbol : t_symbol =
(Node.of_int 4)
let translate_tag _ t = if t == 4 then ~-1 else t
-let to_string t tag = tag_of_id.(Tag.to_int tag)
-let register_tag t tag =
- try Hashtbl.find t.tag_to_id (Tag.to_int tag) with
+let to_string t tag =
+ if tag < Array.length t.tag_of_id then t.tag_of_id.(Tag.to_int tag)
+ else "<!INVALID TAG!>"
+
+let register_tag t s =
+ try Hashtbl.find t.tag_to_id s with
Not_found -> 4
let tag_operations t = {
Tag.translate = (fun s -> translate_tag t s);
}
+let start_root : node = Node.of_int 0
+let start_tag g (idx : node) : [<any_type] Node.t =
+ Node.of_int (g.tags.(Bp.preorder_rank g.start (Node.to_int idx)))
-let rhs_tag t idx =
- t.tags.(Bp.preorder_rank t.start idx)
+;;
-let rhs_first_child t idx =
- Bp.first_child t.start idx
+let start_first_child t (idx : node) =
+ Node.of_int (Bp.first_child t.start (Node.to_int idx))
-let rhs_next_sibling t idx =
- Bp.next_sibling t.start idx
+let start_next_sibling t (idx : node) =
+ Node.of_int (Bp.next_sibling t.start (Node.to_int idx))
let is_non_terminal t (n : [< any_type ] Node.t) =
let n = Node.to_int n in
n >= t.rules_offset
-let is_terminal t (n : [< any_type ] Node.t) = not(is_terminal t n)
+let is_terminal t (n : [< any_type ] Node.t) = not(is_non_terminal t n)
+
+external terminal : [< any_type ] Node.t -> t_symbol = "%identity"
+external non_terminal : [< any_type ] Node.t -> n_symbol = "%identity"
+
let tag (n : t_symbol) : Tag.t = Obj.magic n
+let get_rule g (r : n_symbol) : rhs =
+ Node.of_int (g.rules.((Node.to_int r) - g.rules_offset))
+
+let get_id1 (r : rhs) : tn_symbol =
+ Node.of_int(
+ ((Node.to_int r) lsr 6) land 0x7ffffff)
+
+let get_id2 (r : rhs) : tn_symbol =
+ Node.of_int((Node.to_int r) lsr 33)
+
+let get_rank (r : rhs) : int =
+ (Node.to_int r) land 0b11
+
+let get_id1_rank (r : rhs) : int =
+ ((Node.to_int r) lsr 2) land 0b11
+
+let get_id2_pos (r : rhs) : int =
+ ((Node.to_int r) lsr 4) land 0b11
+
+let get_id2_rank (r : rhs) : int =
+ get_rank r + 1 - get_id1_rank r
projects / SXSI / xpathcomp.git / blobdiff