Cosmetic change.

[SXSI/xpathcomp.git] / src / grammar2.ml

diff --git a/src/grammar2.ml b/src/grammar2.ml
index 0c509bf..a7ccb09 100644 (file)
--- a/src/grammar2.ml
+++ b/src/grammar2.ml
@@ -200,16 +200,26 @@ struct
       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
+      and id2 = h_find ~msg:"8" tag_to_id tag2 in
+      let conf =
+       if List.length args = 0 then 0
+       else
+         if List.length args = 1 then
+         if List.length params = 1 then 1
+         else if pos2 = 1 then 2
+         else 3
+         else (* 2 parameters *)
+           if List.length params = 1 then 4
+           else if pos2 = 1 then 5
+           else 6
       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 id1) lsl 3 in
+      let rule_ = rule_ lor conf in
       r_array.((h_find  ~msg:"9" tag_to_id head) - rules_offset ) <- rule_
     ) rules;
     (*let l = Array.length renum_tags in *)
@@ -238,7 +248,7 @@ let parse file =
   g
 
 let _GRAMMAR_MAGIC = 0xaabbcc
-let _GRAMMAR_VERSION = 2
+let _GRAMMAR_VERSION = 3
 
 let save g f =
   let cout = open_out f in
@@ -258,36 +268,30 @@ let save g f =
 
 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);
   let bp = Bp.load fd in
   close_in cin;
-  {
+  let g = {
     start = bp;
     tags = tags;
     rules = rules;
     rules_offset = rules_offset;
     tag_to_id = tag_to_id;
     tag_of_id = tag_of_id;
-  }
+  } in
+  Printf.eprintf "Grammar size:%i kb\n%!"
+    ((Ocaml.size_b g  + Bp.alloc_stats ())/1024);
+  g
 
 
 type node = [ `Start ] Node.t
@@ -303,6 +307,13 @@ type t_symbol = t_type Node.t
 type tn_symbol = [ any_type ] Node.t
 
 
+type partial =
+    Leaf of node
+  | Node0 of tn_symbol (* No parameters *)
+  | Node1 of tn_symbol * partial
+  | Node2 of tn_symbol * partial * partial
+
+
 let is_nil  (t : t_symbol) =
   (Node.to_int t) == 4
 
@@ -312,7 +323,7 @@ let nil_symbol : t_symbol =
 let translate_tag _ t  = if t == 4 then ~-1 else t
 let to_string t tag =
   if tag < Array.length t.tag_of_id then t.tag_of_id.(Tag.to_int tag)
-  else "<!INVALIDTAG!>"
+  else "<!INVALID TAG!>"
 
 let register_tag t s =
   try Hashtbl.find t.tag_to_id s with
@@ -325,14 +336,16 @@ let tag_operations t = {
 }
 
 let start_root : node = Node.of_int 0
-let start_tag t (idx : node) =
-  t.tags.(Bp.preorder_rank t.start (Node.to_int idx))
+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 start_first_child t (idx : node) =
-  Bp.first_child t.start (Node.to_int idx)
+  Node.of_int (Bp.first_child t.start (Node.to_int idx))
 
 let start_next_sibling t (idx : node) =
-  Bp.next_sibling t.start (Node.to_int idx)
+  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
@@ -341,7 +354,7 @@ let is_non_terminal t (n : [< any_type ] Node.t) =
 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 -> t_symbol = "%identity"
+external non_terminal : [< any_type ] Node.t -> n_symbol = "%identity"
 
 
 let tag (n : t_symbol) : Tag.t = Obj.magic n
@@ -350,15 +363,42 @@ 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 4) land 0x7ffffff)
+  Node.of_int(((Node.to_int r) lsr 3) land 0x7ffffff)
 
 let get_id2 (r : rhs) : tn_symbol =
-  Node.of_int((Node.to_int r) lsr 31)
-
-let get_param_pos (r : rhs) : int =
-  ((Node.to_int r) lsr 2) land 0b11
-
-let num_params (r : rhs) : int =
-  (Node.to_int r) land 0b11
+  Node.of_int((Node.to_int r) lsr 30)
+
+type conf = | C0 (* B(C) *)
+           | C1 (* B(C(y0)) *)
+           | C2 (* B(C, y0) *)
+           | C3 (* B(y0, C) *)
+           | C4 (* B(C(y0, y1)) *)
+           | C5 (* B(C(y0), y1) *)
+           | C6 (* B(y0, C(y1)) *)
+
+let get_conf (r : rhs) : conf =
+  (Obj.magic ((Node.to_int r) land 0b111))
+
+
+let get_rank (r : rhs) : int =
+  match get_conf r with
+  | C0 -> 0
+  | C1 | C2 | C3 -> 1
+  | C4 | C5 | C6 -> 2
+
+let get_id1_rank (r : rhs) : int =
+  match get_conf r with
+  | C0 | C1 | C4 -> 1
+  | _ -> 2
+
+let get_id2_pos (r : rhs) : int =
+  match get_conf r with
+  | C0 | C1 |C2 | C4 | C5 -> 1
+  | _ -> 2
+
+let get_id2_rank (r : rhs) : int =
+  match get_conf r with
+  | C0 | C2 | C3 -> 0
+  | C1 | C5 | C6 -> 1
+  | C4 -> 2