X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=tree.ml;h=4958a8281a1ea355298efd932a8a1eb2a932785a;hb=1b4d4c7a0537d30e21068f06535c5d3a1af92f88;hp=4cf40479d4919c580030d491f1bef444a9110ff5;hpb=7489c542a7b7357a1c2bbc436d1d77c601833d3b;p=SXSI%2Fxpathcomp.git

diff --git a/tree.ml b/tree.ml
index 4cf4047..4958a82 100644
--- a/tree.ml
+++ b/tree.ml
@@ -4,480 +4,583 @@
 (*  Copyright NICTA 2008                                                      *)
 (*  Distributed under the terms of the LGPL (see LICENCE)                     *)
 (******************************************************************************)
-(*INCLUDE "debug.ml" *)
+INCLUDE "utils.ml"
+
+
+external init_lib : unit -> unit = "caml_init_lib"
+
+exception CPlusPlusError of string
+
+let () = Callback.register_exception "CPlusPlusError" (CPlusPlusError "")
+
+let () =  init_lib ()
+
 
 type tree
-type 'a node = int
+type 'a node = private int
 type node_kind = [`Text | `Tree ]
-    
-let compare_node : 'a node -> 'a node -> int = (-)
+
+type t = {
+  doc : tree;
+  children : Ptset.Int.t array;
+  siblings : Ptset.Int.t array;
+  descendants: Ptset.Int.t array;
+  followings: Ptset.Int.t array;
+}
+
+external inode : 'a node -> int = "%identity"
+external nodei : int -> 'a node = "%identity"
+let compare_node x y = (inode x) - (inode y)
 let equal_node : 'a node -> 'a node -> bool = (==)
-  
-(* abstract type, values are pointers to a XMLTree C++ object *)
 
-external int_of_node : 'a node -> int = "%identity"
-  
-external parse_xml_uri : string -> int -> bool -> bool -> tree = "caml_call_shredder_uri"         
+
+external parse_xml_uri : string -> int -> bool -> bool -> tree = "caml_call_shredder_uri"
 external parse_xml_string :  string -> int -> bool -> bool -> tree = "caml_call_shredder_string"
-  
-external save_tree : tree -> string -> unit = "caml_xml_tree_save"
-external load_tree : string -> int -> tree = "caml_xml_tree_load"
-  
+external tree_print_xml_fast3 : tree -> [`Tree ] node -> Unix.file_descr ->unit = "caml_xml_tree_print"
+external tree_save : tree -> Unix.file_descr -> string -> unit = "caml_xml_tree_save"
+external tree_load : Unix.file_descr -> string -> bool -> int -> tree = "caml_xml_tree_load"
+
 external nullt : unit -> 'a node = "caml_xml_tree_nullt"
 
-let nil : 'a node = Obj.magic (-1)
+let nil : [`Tree ] node = nodei ~-1
+let nulldoc : [`Text ] node = nodei ~-1
+let root : [`Tree ] node = nodei 0
 
-external text_get_text : tree -> [`Text] node -> string = "caml_text_collection_get_text" 
-		
-external text_is_empty : tree -> [`Text ] node -> bool = "caml_text_collection_empty_text" 
+external text_get_text : tree -> [`Text] node -> string = "caml_text_collection_get_text"
+external text_is_empty : tree -> [`Text ] node -> bool = "caml_text_collection_empty_text"
 
-let text_is_empty t n =
-  (equal_node nil n) || text_is_empty t n
-    
-external get_cached_text : tree -> [`Text ] node -> string = "caml_text_collection_get_cached_text" 
-  
-  
-let text_get_text t n =
-  if equal_node nil n then ""
-  else get_cached_text t n
+let text_is_empty t n = (equal_node nulldoc n) || text_is_empty t n
 
-external text_size : tree -> int = "caml_text_collection_size" 
-external text_is_contains : tree -> string -> bool = "caml_text_collection_is_contains" 
-external text_count_contains : tree -> string -> int = "caml_text_collection_count_contains" 
-external text_count : tree -> string -> int = "caml_text_collection_count" 
-external text_contains : tree -> string -> [`Text ] node array = "caml_text_collection_contains" 
+external text_is_prefix : tree -> string -> bool = "caml_text_collection_is_prefix"
+external text_is_suffix : tree -> string -> bool = "caml_text_collection_is_suffix"
+external text_is_equal : tree -> string -> bool = "caml_text_collection_is_equal"
+external text_is_contains : tree -> string -> bool = "caml_text_collection_is_contains"
+external text_is_lessthan : tree -> string -> bool = "caml_text_collection_is_lessthan"
 
+external text_count : tree -> string -> int = "caml_text_collection_count"
+external text_count_prefix : tree -> string -> int = "caml_text_collection_count_prefix"
+external text_count_suffix : tree -> string -> int = "caml_text_collection_count_suffix"
+external text_count_equal : tree -> string -> int = "caml_text_collection_count_equal"
+external text_count_contains : tree -> string -> int = "caml_text_collection_count_contains"
+external text_count_lessthan : tree -> string -> int = "caml_text_collection_count_lessthan"
 
-external tree_serialize : tree -> string -> unit = "caml_xml_tree_serialize" 
+external text_prefix : tree -> string -> [`Text ] node array = "caml_text_collection_prefix"
+external text_suffix : tree -> string -> [`Text ] node array = "caml_text_collection_suffix"
+external text_equals : tree -> string -> [`Text ] node array = "caml_text_collection_equals"
+external text_contains : tree -> string -> [`Text ] node array = "caml_text_collection_contains"
+external text_lessthan : tree -> string -> [`Text ] node array = "caml_text_collection_lessthan"
 
-external tree_unserialize : string -> tree = "caml_xml_tree_unserialize"
-      
-external tree_root : tree -> [`Tree] node = "caml_xml_tree_root" 
+
+external tree_root : tree -> [`Tree] node = "caml_xml_tree_root"  "noalloc"
+external tree_size : tree -> int = "caml_xml_tree_size" "noalloc"
+external tree_num_tags : tree -> int = "caml_xml_tree_num_tags" "noalloc"
+external tree_subtree_size : tree -> [`Tree] node -> int = "caml_xml_tree_subtree_size" "noalloc"
+external tree_subtree_elements : tree -> [`Tree] node -> int = "caml_xml_tree_subtree_elements" "noalloc"
+external tree_subtree_tags : tree -> [`Tree] node -> Tag.t -> int = "caml_xml_tree_subtree_elements" "noalloc"
 
 let tree_is_nil x = equal_node x nil
+external tree_is_leaf : tree -> [`Tree ] node -> bool = "caml_xml_tree_is_leaf" "noalloc"
+external tree_is_ancestor : tree -> [`Tree ] node -> [`Tree ] node -> bool = "caml_xml_tree_is_ancestor" "noalloc"
+external tree_is_child : tree -> [`Tree ] node -> [`Tree ] node -> bool = "caml_xml_tree_is_child" "noalloc"
+external tree_is_first_child : tree -> [`Tree ] node -> bool = "caml_xml_tree_is_first_child" "noalloc"
+external tree_num_children : tree -> [`Tree ] node -> int = "caml_xml_tree_num_children" "noalloc"
+external tree_child_number : tree -> [`Tree ] node -> int = "caml_xml_tree_child_number" "noalloc"
+external tree_depth : tree -> [`Tree ] node -> int = "caml_xml_tree_depth" "noalloc"
+external tree_preorder : tree -> [`Tree ] node -> int = "caml_xml_tree_preorder" "noalloc"
+external tree_postorder : tree -> [`Tree ] node -> int = "caml_xml_tree_postorder" "noalloc"
+external tree_tag : tree -> [`Tree ] node -> Tag.t = "caml_xml_tree_tag" "noalloc"
+external tree_doc_ids : tree -> [`Tree ] node -> [`Text] node*[`Text] node = "caml_xml_tree_doc_ids"
 
-external tree_parent : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_parent" 
-external tree_parent_doc : tree -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_parent_doc" 
-external tree_prev_doc : tree -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_prev_doc" 
-external tree_first_child : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_first_child" 
-external tree_next_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_next_sibling" 
-external tree_prev_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_prev_sibling" 
-external tree_is_leaf : tree -> [`Tree] node -> bool = "caml_xml_tree_is_leaf" 
-    
-(*    external tag : tree -> [`Tree ] node -> T = "caml_xml_tree_tag"*)
-external tree_tag_id : tree -> [`Tree ] node -> Tag.t = "caml_xml_tree_tag_id" 
-
-let tree_is_last t n = equal_node nil (tree_next_sibling t n)
-    
-external tree_prev_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_prev_text" 
-
-external tree_my_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_my_text" 
-external tree_next_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_next_text" 
-external tree_doc_ids : tree -> [`Tree ] node -> [`Text ] node * [`Text ] node = "caml_xml_tree_doc_ids" 
-external tree_text_xml_id : tree -> [`Text ] node -> int = "caml_xml_tree_text_xml_id" 
-external tree_node_xml_id : tree -> [`Tree ] node -> int = "caml_xml_tree_node_xml_id" 
-external tree_is_ancestor : tree -> [`Tree ] node -> [`Tree ] node -> bool = "caml_xml_tree_is_ancestor" 
-external tree_tagged_desc : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node = "caml_xml_tree_tagged_desc" 
-external tree_tagged_foll_below : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node -> [`Tree ] node = "caml_xml_tree_tagged_foll_below" 
-external tree_subtree_tags : tree -> [`Tree ] node -> Tag.t -> int = "caml_xml_tree_subtree_tags" 
-external tree_select_below : tree -> [`Tree ] node -> Ptset.int_vector -> Ptset.int_vector -> [`Tree ] node = "caml_xml_tree_select_below" 
-external tree_select_desc_only : tree -> [`Tree ] node -> Ptset.int_vector -> [`Tree ] node = "caml_xml_tree_select_desc_only" 
-external tree_select_next : tree -> [`Tree ] node -> Ptset.int_vector -> Ptset.int_vector -> [`Tree ] node -> [`Tree ] node = "caml_xml_tree_select_next" 
-external tree_select_foll_only : tree -> [`Tree ] node -> Ptset.int_vector -> [`Tree ] node -> [`Tree ] node = "caml_xml_tree_select_foll_only" 
-external tree_select_desc_or_foll_only : tree -> [`Tree ] node -> Ptset.int_vector -> [`Tree ] node -> [`Tree ] node = "caml_xml_tree_select_foll_only" 
-  
-type descr = 
-  | Nil 
-  | Node of [`Tree] node
-  | Text of [`Text] node * [`Tree] node
-      
-type t = { doc : tree;	          
-	   node : descr;
-	   ttable : (Tag.t,(Ptset.t*Ptset.t)) Hashtbl.t;	   
-	 }
-	
-
-let update h t sb sa = 
-    let sbelow,safter = 
-      try
-	Hashtbl.find h t 
-      with
-	| Not_found -> Ptset.empty,Ptset.empty
-    in
-      Hashtbl.replace h t (Ptset.union sbelow sb, Ptset.union safter sa)
+external tree_parent : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_parent" "noalloc"
+external tree_child : tree -> [`Tree] node -> int -> [`Tree] node = "caml_xml_tree_child" "noalloc"
+external tree_first_child : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_first_child" "noalloc"
+external tree_first_element : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_first_element" "noalloc"
+external tree_last_child : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_last_child" "noalloc"
+external tree_next_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_next_sibling"  "noalloc"
+external tree_next_element : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_next_element"  "noalloc"
+external tree_prev_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_prev_sibling" "noalloc"
+external tree_tagged_child : tree -> [`Tree] node -> Tag.t -> [`Tree] node = "caml_xml_tree_tagged_child" "noalloc"
 
+type unordered_set
+external unordered_set_alloc : int -> unordered_set = "caml_unordered_set_alloc"
+external unordered_set_length : unordered_set -> int = "caml_unordered_set_length"
+external unordered_set_insert : unordered_set -> int -> unit = "caml_unordered_set_set" "noalloc"
 
+external tree_select_child : tree -> [`Tree ] node -> unordered_set -> [`Tree] node = "caml_xml_tree_select_child" "noalloc"
+external tree_tagged_following_sibling : tree -> [`Tree] node -> Tag.t -> [`Tree] node = "caml_xml_tree_tagged_following_sibling" "noalloc"
+external tree_select_following_sibling : tree -> [`Tree ] node -> unordered_set -> [`Tree] node = "caml_xml_tree_select_following_sibling" "noalloc"
+external tree_tagged_descendant : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node = "caml_xml_tree_tagged_descendant" "noalloc"
+external tree_select_descendant : tree -> [`Tree ] node -> unordered_set -> [`Tree] node = "caml_xml_tree_select_descendant" "noalloc"
+external tree_tagged_following : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node = "caml_xml_tree_tagged_following" "noalloc"
+external tree_tagged_following_below : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node -> [`Tree ] node = "caml_xml_tree_tagged_following_below" "noalloc"
+external tree_select_following_below : tree -> [`Tree ] node -> unordered_set -> [`Tree] node -> [`Tree] node = "caml_xml_tree_select_following_below" "noalloc"
 
 
+external tree_tagged_following_before : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node -> [`Tree ] node = "caml_xml_tree_tagged_following_before" "noalloc"
+external tree_select_following_below : tree -> [`Tree ] node -> unordered_set -> [`Tree] node -> [`Tree] node = "caml_xml_tree_select_following_before" "noalloc"
 
-let collect_tags tree =
-  let h = Hashtbl.create 511 in
-  let rec loop id acc = 
-    if equal_node id nil
-    then (Ptset.singleton Tag.pcdata, Ptset.add Tag.pcdata acc)
+external tree_my_text : tree -> [`Tree ] node -> [`Text] node = "caml_xml_tree_my_text" "noalloc"
+external tree_my_text_unsafe : tree -> [`Tree ] node -> [`Text] node = "caml_xml_tree_my_text_unsafe" "noalloc"
+external tree_text_xml_id : tree -> [`Text ] node -> int = "caml_xml_tree_text_xml_id" "noalloc"
+external tree_node_xml_id : tree -> [`Tree ] node -> int = "caml_xml_tree_node_xml_id" "noalloc"
+
+external tree_parent_node : tree -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_parent_node" "noalloc"
+
+(*external tree_prev_doc : tree -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_prev_doc" "noalloc" *)
+
+external tree_closing : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_closing" "noalloc"
+external tree_is_open : tree -> [`Tree] node -> bool = "caml_xml_tree_is_open" "noalloc"
+
+
+external benchmark_jump : tree -> Tag.t -> int = "caml_benchmark_jump" "noalloc"
+
+let benchmark_jump t s = benchmark_jump t.doc s
+
+external benchmark_fcns : tree -> int = "caml_benchmark_fcns" "noalloc"
+external benchmark_fene : tree -> int = "caml_benchmark_fene" "noalloc"
+external benchmark_iter : tree -> int = "caml_benchmark_iter" "noalloc"
+
+let benchmark_fcns t = benchmark_fcns t.doc
+
+let benchmark_fene t = benchmark_fene t.doc
+
+let benchmark_iter t = benchmark_iter t.doc
+
+external benchmark_lcps : tree -> unit = "caml_benchmark_lcps" "noalloc"
+
+let benchmark_lcps t = benchmark_lcps t.doc
+
+
+
+
+
+
+
+let text_size tree = inode (snd ( tree_doc_ids tree root ))
+
+let text_get_text t (x:[`Text] node) =
+  if x == nulldoc then ""
+  else text_get_text t x
+
+
+
+
+module HPtset = Hashtbl.Make(Ptset.Int)
+
+let vector_htbl = HPtset.create MED_H_SIZE
+
+let ptset_to_vector s =
+  try
+    HPtset.find vector_htbl s
+  with
+      Not_found ->
+	let v = unordered_set_alloc (Ptset.Int.cardinal s) in
+	let _ = Ptset.Int.iter (fun e -> unordered_set_insert v e) s in
+	  HPtset.add vector_htbl s v; v
+
+
+
+let subtree_size t i = tree_subtree_size t.doc i
+let subtree_elements t i = tree_subtree_elements t.doc i
+let text_size t = text_size t.doc
+
+
+let rec fold_siblings tree f node acc =
+  if node == nil then acc else fold_siblings tree f (tree_next_sibling tree node) (f node acc)
+module TS =
+  struct
+    type t = bool array
+    let create n = Array.create n false
+    let add e a = a.(e) <- true; a
+    let merge a b =
+      for i = 0 to Array.length a - 1 do
+	a.(i) <- a.(i) || b.(i)
+      done
+    let clear a =
+      for i = 0 to Array.length a - 1 do
+	a.(i) <- false;
+      done
+
+    let to_ptset a =
+      let r = ref Ptset.Int.empty in
+	for i = 0 to Array.length a - 1 do
+	  r := Ptset.Int.add i !r;
+	done;
+	!r
+  end
+
+
+let collect_children_siblings tree =
+  let ntags = (tree_num_tags tree) in
+  let () =   Printf.eprintf ">>>length: %i\n%!" ntags in
+  let table_c = Array.init (tree_num_tags tree) (fun _ -> TS.create ntags) in
+  let table_n = Array.init (tree_num_tags tree) (fun _ -> TS.create ntags) in
+  let acc_tag n s = TS.add (tree_tag tree n) s in
+  let count = ref 0 in
+  let size = tree_subtree_size tree root in
+  let tmp = TS.create ntags in
+  let rec loop node =
+    if node == nil then ()
     else
-      let below2,after2 = loop (tree_next_sibling tree id) acc in
-      let below1,after1 = loop (tree_first_child tree id) after2 in
-      let tag = tree_tag_id tree id in
-	update h tag below1 after2;
-	Ptset.add tag (Ptset.union below1 below2), (Ptset.add tag after1)
+      let () =   if !count mod 10000 == 0 then
+	Printf.eprintf "Node %i / %i\n%!" !count size;
+      in
+      let () = if !count mod 1000000 == 0 then Gc.compact() in
+      let () = count := !count + 1 in
+      let tag = tree_tag tree node in
+      let () = TS.clear tmp in
+      let children =
+	fold_siblings tree
+	  acc_tag
+	  (tree_first_child tree node) tmp
+      in
+      let () = TS.merge table_c.(tag) children in
+      let () = TS.clear tmp in
+      let siblings =
+	fold_siblings tree
+	  acc_tag
+	  (tree_next_sibling tree node) tmp
+      in
+	TS.merge table_n.(tag) siblings;
+	loop (tree_first_child tree node);
+	loop (tree_next_sibling tree node)
+  in
+    loop root;
+    ( Array.map TS.to_ptset table_c,
+      Array.map TS.to_ptset table_n )
+
+let collect_children_siblings tree =
+  let table_c = Array.create (tree_num_tags tree) Ptset.Int.empty in
+  let table_n = Array.copy table_c in
+  let rec loop node =
+    if node == nil then Ptset.Int.empty
+    else
+      let children = loop (tree_first_child tree node) in
+      let tag = tree_tag tree node in
+      let () = table_c.(tag) <- Ptset.Int.union table_c.(tag) children in
+      let siblings = loop (tree_next_sibling tree node) in
+	Ptset.Int.add tag siblings
   in
-  let b,a = loop (tree_root tree) Ptset.empty in
-    update h Tag.pcdata b a;
-    h
+    ignore (loop root);
+    table_c, table_n
 
 
 
+
+let collect_descendants tree =
+  let table_d = Array.create (tree_num_tags tree) Ptset.Int.empty in
+  let rec loop node =
+    if node == nil then Ptset.Int.empty
+    else
+      let d1 = loop (tree_first_child tree node) in
+      let d2 = loop (tree_next_sibling tree node) in
+      let tag = tree_tag tree node in
+	table_d.(tag) <- Ptset.Int.union table_d.(tag) d1;
+	Ptset.Int.add tag (Ptset.Int.union d1 d2)
+  in
+    ignore (loop root);
+    table_d
+
+let collect_followings tree =
+  let table_f = Array.create (tree_num_tags tree) Ptset.Int.empty in
+  let rec loop node acc =
+    if node == nil then acc else
+      let f1 = loop (tree_next_sibling tree node) acc in
+      let f2 = loop (tree_first_child tree node) f1 in
+      let tag = tree_tag tree node in
+	table_f.(tag) <- Ptset.Int.union table_f.(tag) f1;
+	Ptset.Int.add tag (Ptset.Int.union f1 f2)
+  in
+    ignore (loop root Ptset.Int.empty);
+    table_f
+
+let collect_tags tree =
+  let c,n = time (collect_children_siblings) tree ~msg:"Collecting child and sibling tags" in
+  let d = time collect_descendants tree ~msg:"Collecting descendant tags" in
+  let f = time collect_followings tree ~msg:"Collecting following tags" in
+    c,n,d,f
+
 let contains_array = ref [| |]
-  
-let init_contains t s = 
-  let a = text_contains t.doc s 
+let contains_index = Hashtbl.create 4096
+let in_array _ i =
+  try
+    Hashtbl.find contains_index i
+  with
+      Not_found -> false
+
+let init_textfun f t s =
+  let a = match f with
+    | `CONTAINS -> text_contains t.doc s
+    | `STARTSWITH -> text_prefix t.doc s
+    | `ENDSWITH -> text_suffix t.doc s
+    | `EQUALS -> text_equals t.doc s
   in
-    Array.fast_sort (compare) a;
-    contains_array := a
-      
+    (*Array.fast_sort (compare) a; *)
+    contains_array := a;
+    Array.iter (fun x -> Hashtbl.add contains_index x true) !contains_array
+
+let count_contains t s = text_count_contains t.doc s
+
 let init_naive_contains t s =
   let i,j = tree_doc_ids t.doc (tree_root t.doc)
   in
   let regexp = Str.regexp_string s in
-  let matching arg = 
+  let matching arg =
     try
       let _ = Str.search_forward regexp arg 0;
       in true
     with _ -> false
   in
-  let rec loop n acc l = 
+  let rec loop n acc l =
     if n >= j then acc,l
     else
       let s = text_get_text t.doc n
       in
-	if matching s 
-	then loop (n+1) (n::acc) (l+1) 
-	else loop (n+1) acc l
+	if matching s
+	then loop (nodei ((inode n)+1)) (n::acc) (l+1)
+	else loop (nodei ((inode n)+1)) acc l
   in
   let acc,l = loop i [] 0 in
-  let a = Array.create l nil in
+  let a = Array.create l nulldoc in
   let _ = List.fold_left (fun cpt e -> a.(cpt) <- e; (cpt-1)) (l-1) acc
   in
     contains_array := a
-	  
+
+let last_idx = ref 0
+
+let array_find a i j =
+  let l = Array.length a in
+  let rec loop idx x y =
+    if x > y || idx >= l then nulldoc
+       else
+         if a.(idx) >= x then if a.(idx) > y then nulldoc else (last_idx := idx;a.(idx))
+         else loop (idx+1) x y
+  in
+    if a.(0) > j || a.(l-1) < i then nulldoc
+    else loop !last_idx i j
+
+let text_below tree t =
+  let l = Array.length !contains_array in
+  let i,j = tree_doc_ids tree.doc t in
+  let id = if l == 0 then i else (array_find !contains_array i j) in
+  tree_parent_node tree.doc id
+
+let text_next tree t root =
+  let l = Array.length !contains_array in
+  let inf = nodei((inode(snd(tree_doc_ids tree.doc t)))+1) in
+  let _,j = tree_doc_ids tree.doc root in
+  let id = if l == 0 then if inf > j then nulldoc else  inf
+  else array_find !contains_array inf j
+  in
+  tree_parent_node tree.doc id
+
 
 
 module DocIdSet = struct
   include Set.Make (struct type t = [`Text] node
 			   let compare = compare_node end)
-    
-end
-let is_nil t = match t.node with
-  | Nil -> true
-  | Node(i) -> equal_node i nil
-  | _ -> false
 
-let is_node t =
-match t.node with
-  | Node(i) -> not(equal_node i nil)
-  | _ -> false
+end
+let is_nil t = t == nil
 
+let is_node t = t != nil
+let is_root t = t == root
 
-let node_of_t t =
+let node_of_t t  =
   let _ = Tag.init (Obj.magic t) in
-  let table = collect_tags t 
+  let c,n,d,f = collect_tags t
   in
-(*
-  let _ = Hashtbl.iter (fun t (sb,sa) ->
-			  Printf.eprintf "'%s' -> { " (Tag.to_string t);
-			  Ptset.iter (fun i ->  Printf.eprintf "'%s' " (Tag.to_string i)) sb;
-			  Printf.eprintf "}\n { ";
-			  Ptset.iter (fun i ->  Printf.eprintf "'%s' " (Tag.to_string i)) sa;
-			  Printf.eprintf "} \n----------------------------------\n";
-		       ) table in
-*)
-    { doc= t; 
-      node = Node(tree_root t);
-      ttable = table;
+    { doc= t;
+      children = c;
+      siblings = n;
+      descendants = d;
+      followings = f
+
     }
 
+let finalize _ = Printf.eprintf "Release the string list !\n%!"
+;;
+
+let parse f str =
+  node_of_t
+    (f str
+       !Options.sample_factor
+       !Options.index_empty_texts
+       !Options.disable_text_collection)
+
+let parse_xml_uri str = parse parse_xml_uri str
+let parse_xml_string str =  parse parse_xml_string str
+
+let size t = tree_size t.doc;;
 
-let parse_xml_uri str = node_of_t       
-  (parse_xml_uri str 
-     !Options.sample_factor 
-     !Options.index_empty_texts
-     !Options.disable_text_collection)
-     
-let parse_xml_string str = node_of_t 
-  (parse_xml_string str
-     !Options.sample_factor 
-     !Options.index_empty_texts 
-     !Options.disable_text_collection)
-     
 external pool : tree -> Tag.pool = "%identity"
-let save t str = save_tree t.doc str
-let load ?(sample=64) str = 
-  node_of_t (load_tree str sample)
-    
+
+let magic_string = "SXSI_INDEX"
+let version_string = "3"
+
+let pos fd =
+  Unix.lseek fd 0  Unix.SEEK_CUR
+
+let pr_pos fd = Printf.eprintf "At position %i\n%!" (pos fd)
+
+let write fd s =
+  let sl = String.length s in
+  let ssl = Printf.sprintf "%020i" sl in
+    ignore (Unix.write fd ssl 0 20);
+    ignore (Unix.write fd s 0 (String.length s))
+
+let rec really_read fd buffer start length =
+  if length <= 0 then () else
+    match Unix.read fd buffer start length with
+	0 -> raise End_of_file
+      | r -> really_read fd buffer (start + r) (length - r);;
+
+let read fd =
+  let buffer = String.create 20 in
+  let _ =  really_read fd buffer 0 20 in
+  let size = int_of_string buffer in
+  let buffer = String.create size in
+  let _ =  really_read fd buffer 0 size in
+    buffer
+
+let save_tag_table channel t =
+  let t = Array.map (fun s -> Array.of_list (Ptset.Int.elements s)) t in
+    Marshal.to_channel channel t []
+
+let save t str =
+  let fd = Unix.openfile str [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 in
+  let out_c = Unix.out_channel_of_descr fd in
+  let _ = set_binary_mode_out out_c true in
+    output_string out_c magic_string;
+    output_char out_c '\n';
+    output_string out_c version_string;
+    output_char out_c '\n';
+    save_tag_table out_c t.children;
+    save_tag_table out_c t.siblings;
+    save_tag_table out_c t.descendants;
+    save_tag_table out_c t.followings;
+    (* we need to move the fd to the correct position *)
+    flush out_c;
+    ignore (Unix.lseek fd (pos_out out_c) Unix.SEEK_SET);
+    tree_save t.doc fd str;
+    close_out out_c
+;;
+let load_tag_table channel =
+  let table : int array array = Marshal.from_channel channel in
+    Array.map (fun a -> Ptset.Int.from_list (Array.to_list a)) table
+
+let load ?(sample=64) ?(load_text=true) str =
+  let fd = Unix.openfile str [ Unix.O_RDONLY ] 0o644 in
+  let in_c = Unix.in_channel_of_descr fd in
+  let _ = set_binary_mode_in in_c true in
+  let load_table () =
+    (let ms = input_line in_c in if ms <> magic_string then failwith "Invalid index file");
+    (let vs = input_line in_c in if vs <> version_string then failwith "Invalid version file");
+    let c = load_tag_table in_c in
+    let s = load_tag_table in_c in
+    let d = load_tag_table in_c in
+    let f = load_tag_table in_c in
+      c,s,d,f
+  in
+  let _ = Printf.eprintf "\nLoading tag table : " in
+  let c,s,d,f = time (load_table) () in
+  ignore(Unix.lseek fd (pos_in in_c) Unix.SEEK_SET);
+  let tree = { doc = tree_load fd str load_text sample;
+	       children = c;
+	       siblings = s;
+	       descendants = d;
+	       followings = f
+	     }
+  in close_in in_c;
+  tree
+
 
 
 
 let tag_pool t = pool t.doc
-  
-let compare a b = match a.node,b.node  with
-  | Nil, Nil -> 0
-  | Nil,_ -> 1
-  | _ , Nil -> -1
-  | Node(i),Node(j) -> compare_node i j
-  | Text(i,_), Text(j,_) -> compare_node i j
-  | Node(i), Text(_,j) -> compare_node i j
-  | Text(_,i), Node(j) -> compare_node i j 
-
-let equal a b = (compare a b) == 0
-  
-  
-let norm (n : [`Tree ] node ) =  if tree_is_nil n then Nil else Node (n)
-  
+
+let compare = compare_node
+
+let equal a b = a == b
+
 let nts = function
-    Nil -> "Nil"
-  | Text (i,j) -> Printf.sprintf "Text (%i, %i)" i j
-  | Node (i) -> Printf.sprintf "Node (%i)"  i
-      
-let mk_nil t = { t with node = Nil }		  
-let root n = { n with node = norm (tree_root n.doc) }
-
-let is_root n = match n.node with
-  | Node(t) -> (int_of_node t) == 0 
-  | _ -> false
-      
-let parent n =  
-  let node' = 
-    match n.node with (* inlined parent *)
-      | Node(t) when (int_of_node t)== 0 -> Nil
-      | Node(t) -> 
-	  let txt = tree_prev_text n.doc t in
-	    if text_is_empty n.doc txt then
-	      let ps = tree_prev_sibling n.doc t in
-		if tree_is_nil ps
-		then
-		  Node(tree_parent n.doc t)
-		else Node(ps)
-	    else
-	      Text(txt,t)
-      | Text(i,t) ->
-	  let ps = tree_prev_doc n.doc i in
-	    if tree_is_nil ps
-	    then Node (tree_parent_doc n.doc i)
-	    else Node(ps)
-      | _ -> failwith "parent"
-  in
-    { n with node = node' }
-
-let node_child n =
-  match n.node with
-    | Node i ->  { n with node= norm(tree_first_child n.doc i) }
-    | _ -> { n with node = Nil }
-
-let node_sibling n =
-  match n.node with
-    | Node i ->  { n with node= norm(tree_next_sibling n.doc i) }
-    | _ -> { n with node = Nil }
-
-let node_sibling_ctx n _ = 
-  match n.node with
-    | Node i ->  { n with node= norm(tree_next_sibling n.doc i) }
-    | _ -> { n with node = Nil }
-
-
-let first_child n = 
-  let node' = 
-    match n.node with
-      | Node (t) -> 
-	  let fs = tree_first_child n.doc t in
-	    if equal_node nil fs
-	    then 
-	      let txt = tree_my_text n.doc t in
-		if equal_node nil txt
-		then Nil
-		else Text(txt,nil)
-	    else
-	      let txt = tree_prev_text n.doc fs in
-		if equal_node nil txt
-		then Node(fs)
-		else Text(txt, fs) 
-      | Text(_,_) -> Nil
-      | Nil -> failwith "first_child"
-  in
-    { n with node = node'}
-      
-let next_sibling n = 
-  let node' =
-    match n.node with
-      | Text (_,ns) -> norm ns
-      | Node(t) ->
-	  let ns = tree_next_sibling n.doc t in
-	  let txt = tree_next_text n.doc t in
-	    if equal_node nil txt
-	    then norm ns
-	    else Text(txt, ns)
-      | Nil -> failwith "next_sibling"
-  in
-    { n with node = node'}
-	  
-let next_sibling_ctx n _ = next_sibling n
-	  
-let left = first_child 
-let right = next_sibling
-    
-let id t = 
-  match t.node with
-    | Node(n)  -> tree_node_xml_id t.doc n
-    | Text(i,_)  -> tree_text_xml_id t.doc i
-    | _ ->  -1 
-	
-let tag t =
-  match t.node with 
-  | Text(_) -> Tag.pcdata
-  | Node(n) -> tree_tag_id t.doc n
-  | _ -> failwith "tag"
-    
-(*
-    let string_below t id =
-      let strid = parent_doc t.doc id in
-	match t.node with
-	  | Node(NC(i)) -> 
-	      (Tree.equal i strid) || (is_ancestor t.doc i strid)
-	  | Node(SC(i,_)) -> Text.equal i id
-	  | _ -> false
-
-
-    let tagged_foll t tag =
-      if tag = Tag.attribute || tag = Tag.pcdata then failwith "tagged_foll"
-      else match t with
-	| { doc=d; node=Node(NC n) } -> { t with node = norm (tagged_foll d n tag) }
-	| { doc=d; node=Node(SC (_,n)) } when is_nil n -> { t with node= Nil } 
-	| { doc=d; node=Node(SC (_,n)) } ->
-	    let nnode = 
-	      if tag_id d n == tag then n
-	      else 
-		let n' = tagged_desc d n tag in
-		  if is_nil n' then tagged_foll d n tag
-		  else n'
-	    in {t with node= norm nnode}
-	| _ -> { t with node=Nil }
-	    
-
-    let tagged_desc t tag =
-      if tag = Tag.attribute || tag = Tag.pcdata then failwith "tagged_desc"
-      else match t with
-	| { doc=d; node=Node(NC n) } -> { t with node = norm (tagged_desc d n tag) }
-	| _ -> { t with node=Nil }
-
-*)	    
-let select_next tb tf t s = 
-  match s.node  with
-    | Node (below) -> begin
-	match t.node with
-	  | Node( n)  ->
-	      { t with node = norm (tree_select_next t.doc n (Ptset.to_int_vector tb) (Ptset.to_int_vector tf) below) }
-	  | Text (i,n)  when equal_node nil n ->
-	      let p = tree_parent_doc t.doc i in
-		{ t with node = norm (tree_select_next t.doc p (Ptset.to_int_vector tb) (Ptset.to_int_vector tf) below) }
-	  | Text(_,n)  ->
-	      if Ptset.mem (tree_tag_id t.doc n) (Ptset.union tb tf)
-	      then { t with node=Node(n) }
-	      else
-		let vb = Ptset.to_int_vector tb in
-		let vf = Ptset.to_int_vector tf in
-		let node = 
-		  let dsc = tree_select_below t.doc n vb vf in
-		    if equal_node nil dsc
-		    then tree_select_next t.doc n vb vf below
-		    else dsc
-		in
-		  { t with node = norm node }
-	  | _ -> {t with node = Nil }
-      end
-	
-    | _ -> { t with node = Nil }
-
-  
-
-
-  let select_foll_only  tf t s = 
-    match s.node  with
-      | Node (below)  -> 
-	  begin
-	    match t.node with
-	    | Node(n) ->
-		{ t with node= norm (tree_select_foll_only t.doc n (Ptset.to_int_vector tf) below) }
-	    | Text(i,n)  when equal_node nil n ->
-		let p = tree_parent_doc t.doc i in
-		  { t with node= norm (tree_select_foll_only t.doc p (Ptset.to_int_vector tf) below) }
-	    |  Text(_,n) ->
-		 if Ptset.mem (tree_tag_id t.doc n) tf
-		 then { t with node=Node(n) }
-		 else
-		   let vf = Ptset.to_int_vector tf in
-		   let node = 
-		     let dsc = tree_select_desc_only t.doc n vf in
-		       if tree_is_nil dsc
-		       then tree_select_foll_only t.doc n vf below
-		       else dsc
-		   in
-		     { t with node = norm node }
-	    | _ -> { t with node = Nil }
-	end	    
-      | _ -> {t with node=Nil }	  
-
-let select_below  tc td t=
-  match t.node with
-    | Node( n) -> 
-	let vc = Ptset.to_int_vector tc
-	in
-	let vd = Ptset.to_int_vector td
-	in
-	  { t with node= norm(tree_select_below t.doc n vc vd) }
-    | _ -> { t with node=Nil }
-	
-	
-let select_desc_only  td t =
-  match t.node with
-    | Node(n) -> 
-	let vd = Ptset.to_int_vector td
-	in
-	  { t with node = norm(tree_select_desc_only t.doc n vd) }
-    | _ -> { t with node = Nil }
-
-
-let tagged_desc tag t =
-  match t.node with
-    | Node(n) -> 	
-	  { t with node = norm(tree_tagged_desc t.doc n tag) }
-    | _ -> { t with node = Nil }
-
-
-let tagged_foll_below tag t s =
-    match s.node  with
-      | Node (below)  -> 
-	  begin
-	    match t.node with
-	    | Node(n) ->
-		{ t with node= norm (tree_tagged_foll_below t.doc n tag below) }
-	    | Text(i,n)  when equal_node nil n ->
-		let p = tree_prev_doc t.doc i in
-		  { t with node= norm (tree_tagged_foll_below t.doc p tag below) }
-	    |  Text(_,n) ->
-		 if (tree_tag_id t.doc n) == tag
-		 then { t with node=Node(n) }
-		 else
-		   let node = 
-		     let dsc = tree_tagged_desc t.doc n tag in
-		       if tree_is_nil dsc
-		       then tree_tagged_foll_below t.doc n tag below
-		       else dsc
-		   in
-		     { t with node = norm node }
-	    | _ -> { t with node = Nil }
-	  end	    
-      | _ -> {t with node=Nil }	  
+    -1 -> "Nil"
+  | i -> Printf.sprintf "Node (%i)"  i
+
+let dump_node t = nts (inode t)
+
+let is_left t n = tree_is_first_child t.doc n
+
+
+
+let is_below_right t n1 n2 =
+  tree_is_ancestor t.doc (tree_parent t.doc n1) n2
+  && not (tree_is_ancestor t.doc n1 n2)
+
+let is_binary_ancestor t n1 n2 =
+  let p = tree_parent t.doc n1 in
+  let fin = tree_closing t.doc p in
+  n2 > n1 && n2 < fin
+(*  (is_below_right t n1 n2) ||
+    (tree_is_ancestor t.doc n1 n2) *)
+
+let parent t n = tree_parent t.doc n
+
+let first_child t = let doc = t.doc in ();fun n -> tree_first_child doc n
+let first_element t = let doc = t.doc in (); fun n -> tree_first_element doc n
+let first_element t n = tree_first_element t.doc n
+(* these function will be called in two times: first partial application
+   on the tag, then application of the tag and the tree, then application of
+   the other arguments. We use the trick to let the compiler optimize application
+*)
+
+let tagged_child t tag = () ; fun n -> tree_tagged_child t.doc n tag
+
+let select_child t = fun ts ->
+  let v = ptset_to_vector ts in ();
+    fun n -> tree_select_child t.doc n v
+
+let next_sibling t = let doc = t.doc in (); fun n ->  tree_next_sibling doc n
+let next_element t = let doc = t.doc in (); fun n ->  tree_next_element doc n
+let next_element t n = tree_next_element t.doc n
+
+let tagged_following_sibling t tag = (); fun n -> tree_tagged_following_sibling t.doc n tag
 
+let select_following_sibling t = fun ts ->
+  let v = (ptset_to_vector ts) in ();
+    fun n -> tree_select_following_sibling t.doc n v
+
+let next_sibling_below t = (); fun n _ -> tree_next_sibling t.doc n
+let next_element_below t = (); fun n _ -> tree_next_element t.doc n
+
+let tagged_following_sibling_below t tag = (); fun n  _ -> tree_tagged_following_sibling t.doc n tag
+
+let select_following_sibling_below t = fun ts ->
+  let v = (ptset_to_vector ts) in ();
+     fun n  _ -> tree_select_following_sibling t.doc n v
+
+let id t n = tree_node_xml_id t.doc n
+
+let tag t n = if n == nil then Tag.nullt else tree_tag t.doc n
+
+let tagged_descendant t tag =
+  let doc = t.doc in (); fun n -> tree_tagged_descendant doc n tag
+
+let select_descendant t = fun ts ->
+  let v = (ptset_to_vector ts) in ();
+    fun n -> tree_select_descendant t.doc n v
+
+let tagged_following_below  t tag =
+  let doc = t.doc in
+  (); fun n ctx -> tree_tagged_following_below doc n tag ctx
+
+let select_following_below t = fun ts ->
+  let v = (ptset_to_vector ts) in ();
+    fun n ctx -> tree_select_following_below t.doc n v ctx
+
+let closing t n = tree_closing t.doc n
+let is_open t n = tree_is_open t.doc n
+let get_text_id t n = tree_my_text t.doc n
 
 let last_idx = ref 0
 let array_find a i j =
@@ -489,214 +592,246 @@ let array_find a i j =
 	  else loop (idx+1) x y
   in
     if a.(0) > j || a.(l-1) < i then nil
-    else loop !last_idx i j 
+    else loop !last_idx i j
 
 
-	
-let text_below t = 
-  let l = Array.length !contains_array in
-    if l = 0 then { t with node=Nil }
-    else
-      match t.node with
-	| Node(n)  ->
-	    let i,j = tree_doc_ids t.doc n in
-	    let id = array_find !contains_array i j
+
+  let count t s = text_count t.doc s
+  let stack = ref []
+  let init_stack () = stack := []
+  let push x = stack:= x::!stack
+  let peek () = match !stack with
+     p::_ -> p
+    | _ -> failwith "peek"
+  let pop () = match !stack with
+     p::r -> stack:=r; p
+    | _ -> failwith "pop"
+
+  let next t = nodei ( (inode t) + 1 )
+  let next2 t = nodei ( (inode t) + 2 )
+  let next3 t = nodei ( (inode t) + 3 )
+
+  let print_xml_fast2  =
+    let _ = init_stack () in
+    let h = Hashtbl.create MED_H_SIZE in
+    let tag_str t = try Hashtbl.find h t with
+       Not_found -> let s = Tag.to_string t in
+       Hashtbl.add h t s;s
+    in
+    let h_att = Hashtbl.create MED_H_SIZE in
+    let att_str t = try Hashtbl.find h_att t with
+       Not_found -> let s = Tag.to_string t in
+      let attname = String.sub s 3 ((String.length s) -3) in
+      Hashtbl.add h_att t attname;attname
+    in fun outc tree t ->
+      let tree = tree.doc in
+      let fin = tree_closing tree t in
+      let rec loop_tag t tag =
+	if t <= fin then
+	if tree_is_open tree t then
+	(* opening tag *)
+	if tag == Tag.pcdata then
+	begin
+	  output_string outc (text_get_text tree (tree_my_text_unsafe tree t));
+	  loop (next2 t) (* skip closing $ *)
+	end
+	else
+	let tagstr = tag_str tag in
+	let _ = output_char outc '<';
+	output_string outc tagstr in
+	let t' = next t in
+	if tree_is_open tree t' then
+	let _ = push tagstr in
+	let tag' = tree_tag tree t' in
+	if tag' == Tag.attribute then let t'' = loop_attr (next t') 0 in
+	output_string outc ">";	loop t'' else (output_string outc ">";loop_tag t' tag')
+	else (* closing with no content *)
+	let _ = output_string outc "/>" in
+	loop (next t')
+	else
+	begin
+	(* closing tag *)
+	  output_string outc "</";
+	  output_string outc (pop());
+	  output_char outc '>';
+	  loop (next t);
+	end
+      and loop t = loop_tag t (tree_tag tree t)
+      and loop_attr t n =
+	if tree_is_open tree t then
+	let attname = att_str (tree_tag tree t) in
+	output_char outc ' ';
+	output_string outc attname;
+	output_string outc "=\"";
+	let t = next t in (* open $@ *)
+	output_string outc (text_get_text tree (tree_my_text_unsafe tree t));
+	output_char outc '"';
+	loop_attr (next3 t) (n+1)
+	else
+	next t (* close @ *)
+      in loop t
+
+  let print_xml_fast  =
+    let h = Hashtbl.create MED_H_SIZE in
+    let tag_str t = try Hashtbl.find h t with
+       Not_found -> let s = Tag.to_string t in
+       Hashtbl.add h t s;s
+    in
+    let h_att = Hashtbl.create MED_H_SIZE in
+    let att_str t = try Hashtbl.find h_att t with
+       Not_found -> let s = Tag.to_string t in
+      let attname = String.sub s 3 ((String.length s) -3) in
+      Hashtbl.add h_att t attname;attname
+    in fun outc tree t ->
+    let rec loop ?(print_right=true) t =
+      if t != nil
+      then
+	let tagid = tree_tag tree.doc t in
+	  if tagid==Tag.pcdata
+	  then
+	    begin
+ 	      let tid =  tree_my_text_unsafe tree.doc t in
+	      output_string outc (text_get_text tree.doc tid);
+	      if print_right
+	      then loop (next_sibling tree t);
+	    end
+	  else
+	    let tagstr = tag_str tagid in
+	    let l = first_child tree t
+	    and r = next_sibling tree t
 	    in
-	      if id == nil then  
-		{ t with  node=Nil }
+	      output_char outc  '<';
+	      output_string outc tagstr;
+	      if l == nil then output_string outc  "/>"
 	      else
-		{ t with  node = Text(id, tree_next_sibling t.doc (tree_prev_doc t.doc id)) }
-	| _ -> { t with node = Nil }
-	    
-let text_next t root =
-  let l = Array.length !contains_array in
-    if l = 0 then { t with node=Nil }
-    else
-      let inf = match t.node with
-	|  Node(n)  -> snd(tree_doc_ids t.doc n)+1
-	| Text(i,_)  -> i+1
-	| _ -> assert false
-      in
-	match root.node with
-	  | Node (n)  ->
-	      let _,j = tree_doc_ids t.doc n in		    
-	      let id = array_find !contains_array inf j
-	      in
-		if id == nil then  { t with node= Nil }
+		if (tag tree l) == Tag.attribute then
+		  begin
+		    loop_attributes (first_child tree l);
+		    if (next_sibling tree l) == nil then output_string outc  "/>"
+		    else
+		      begin
+			output_char outc  '>';
+			loop (next_sibling tree l);
+			output_string outc  "</";
+			output_string outc  tagstr;
+			output_char outc '>';
+		      end;
+		  end
 		else
-		  { t with node = Text(id,tree_next_sibling t.doc (tree_prev_doc t.doc id)) }
-	  | _ -> { t with node = Nil}
-		  
-
-(*		  
-    let subtree_tags t tag =
-      match t with 
-	  { doc = d; node = Node(NC n) } -> 
-	    subtree_tags d n tag
-	| _ -> 0
-
-    let select_desc_array = ref [| |]
-    let idx = ref 0
-
-    let init_tagged_next t tagid =
-      let l = subtree_tags (root t) tagid
-      in
-	tagged_desc_array := Array.create l { t with node= Nil };
-	let i = ref 0 in
-	  let rec collect t =
-	    if is_node t then begin
-	      if tag t == tagid then
-		begin
-		  !tagged_desc_array.(!i) <- t;
-		  incr i;
-		end;
-	      collect (first_child t);
-	      collect (next_sibling t)
-	    end;
-	  in
-	    collect t;
-	    idx := 0
-
-    let print_id ppf v = 
-      let pr x= Format.fprintf ppf x in
-	match v with
-	    { node=Nil } -> pr "NULLT: -1"
-	  | { node=String(i) } | { node=Node(SC(i,_)) } -> pr "DocID: %i" (int_of_node i)
-	  | { node=Node(NC(i)) } -> pr "Node: %i" (int_of_node i)
-	      
-	      
-	  
-(*    let tagged_next t tag = 
-      if !idx >= Array.length !tagged_desc_array 
-      then {t with node=Nil}
-      else
-	let r = !tagged_desc_array.(!idx) 
-	in
-	  incr idx; r
-*)		  
-
-
-    let has_tagged_foll t tag = is_node (tagged_foll t tag)
-    let has_tagged_desc t tag = is_node (tagged_desc t tag)
-
-    let contains t s = 
-      Array.fold_left (fun a i -> DocIdSet.add i a) DocIdSet.empty (Text.contains t.doc s)
-
-
-    let contains_old t s = 
-      let regexp = Str.regexp_string s in
-      let matching arg = 
-	try
-	  let _ = Str.search_forward regexp arg 0;
-	  in true
-	with _ -> false
-      in
-      let rec find t acc = match t.node with
-	| Nil -> acc
-	| String i ->
-	    if  matching (string t) then DocIdSet.add i acc else acc
-	| Node(_) ->  (find (left t )) ((find (right t))  acc)
-      in
-	find t DocIdSet.empty
+		  begin
+		    output_char outc  '>';
+		    loop l;
+		    output_string outc "</";
+		    output_string outc tagstr;
+		    output_char outc '>';
+		  end;
+	      if print_right then loop r
+    and loop_attributes a =
+      if a != nil
+      then
+      let attname = att_str (tag tree a) in
+      let fsa = first_child tree a in
+      let tid =  tree_my_text_unsafe tree.doc fsa in
+	output_char outc ' ';
+	output_string outc attname;
+	output_string outc "=\"";
+	output_string outc (text_get_text tree.doc tid);
+	output_char outc '"';
+	loop_attributes (next_sibling tree a)
+    in
+	loop ~print_right:false t
 
 
-    let contains_iter t s = 
-      let regexp = Str.regexp_string s in
-      let matching arg = 
-	try
-	  let _ = Str.search_forward regexp arg 0;
-	  in true
-	with _ -> false
-      in
-      let size = Text.size t.doc in
-      let rec find acc n = 
-	if n == size then acc
-	else
-	  find 
-	    (if matching (Text.get_cached_text t.doc (Obj.magic n)) then 
-	     DocIdSet.add (Obj.magic n) acc
-	   else acc) (n+1)
-      in
-	find DocIdSet.empty 0
+    let print_xml_fast outc tree t =
+      if (tag tree t) = Tag.document_node then
+	print_xml_fast outc tree (first_child tree t)
+      else print_xml_fast outc tree t
 
+let tags_children t tag = t.children.(tag)
 
+let tags_below t tag = t.descendants.(tag)
+
+let tags_siblings t tag = t.siblings.(tag)
+
+let tags_after t tag = t.followings.(tag)
+
+
+
+let tags t tag =
+  t.children.(tag),
+  t.descendants.(tag),
+  t.siblings.(tag),
+  t.followings.(tag)
+
+
+let rec binary_parent t n =
+  let r =
+  if tree_is_first_child t.doc n
+  then tree_parent t.doc n
+  else tree_prev_sibling t.doc n
+  in if tree_tag t.doc r = Tag.pcdata then
+  binary_parent t r
+  else r
+
+let doc_ids t n = tree_doc_ids t.doc n
+
+let subtree_tags t tag = ();
+  fun n -> if n == nil then 0 else
+    tree_subtree_tags t.doc n tag
+
+let get_text t n =
+  let tid = tree_my_text t.doc n in
+    if tid == nulldoc then "" else
+      text_get_text t.doc tid
+
+
+let dump_tree fmt tree =
+  let rec loop t n =
+    if t != nil then
+      let tag = (tree_tag tree.doc t ) in
+      let tagstr = Tag.to_string tag in
+	let tab = String.make n ' ' in
+
+	  if tag == Tag.pcdata || tag == Tag.attribute_data
+	  then
+	    Format.fprintf fmt "%s<%s>%s</%s>\n"
+	      tab tagstr (text_get_text tree.doc (tree_my_text tree.doc t)) tagstr
+	  else begin
+	    Format.fprintf fmt "%s<%s>\n" tab tagstr;
+	    loop (tree_first_child tree.doc t) (n+2);
+	    Format.fprintf fmt "%s</%s>\n%!" tab tagstr;
+	  end;
+	  loop (tree_next_sibling tree.doc t) n
+  in
+    loop root 0
+;;
+
+
+let print_xml_fast3 t = tree_print_xml_fast3 t.doc
+
+
+
+
+let stats t =
+  let tree = t.doc in
+  let rec loop left node acc_d total_d num_leaves =
+    if node == nil then
+    (acc_d+total_d,if left then num_leaves+1 else num_leaves)
+    else
+    let d,td = loop true (tree_first_child tree node) (acc_d+1) total_d num_leaves in
+    loop false (tree_next_sibling tree  node) (acc_d)  d td
+  in
+  let a,b = loop true root 0 0 0
+  in
+  Printf.eprintf "Average depth: %f, number of leaves %i\n%!" ((float_of_int a)/. (float_of_int b)) b
+;;
 
 
-    let count_contains t s =   Text.count_contains t.doc s
-*)
 
-  let count t s = text_count t.doc s
-(*
-    let is_left t =
-      if is_root t then false
-      else
-      if tag (parent t) == Tag.pcdata then false
-      else
-	let u = left (parent t) in
-	  (id t) == (id u)
-*)
-  let print_xml_fast outc t =
-    let rec loop ?(print_right=true) t = 
-      match t.node with 
-      | Nil -> ()    
-      | Text(i,n) -> output_string outc (text_get_text t.doc i);
-	  if print_right
-	  then loop (left t)
-      | Node (n) -> 
-	  let tg = Tag.to_string (tag t) in
-	  let l = left t 
-	  and r = right t 
-	  in
-	    output_char outc  '<';
-	    output_string outc  tg;
-	    ( match l.node with
-		  Nil -> output_string outc  "/>"
-		| Node(_) when Tag.equal (tag l) Tag.attribute -> 
-		    (loop_attributes (left l);
-		     match (right l).node with
-		       | Nil -> output_string outc  "/>"
-		       | _ -> 
-			   output_char outc  '>'; 
-			   loop (right l);
-			   output_string outc  "</";
-			   output_string outc  tg;
-			   output_char outc '>' )
-		| _ ->
-		    output_char outc  '>'; 
-		    loop l;
-		    output_string outc "</";
-		    output_string outc tg;
-		    output_char outc '>'
-	    );if print_right then loop r
-    and loop_attributes a =      
-	match a.node with 
-	  | Node(_) ->
-	      let value =
-		match (left a).node with
-		  | Text(i,_) -> text_get_text a.doc i
-		  | _ -> assert false
-	      in
-	 	output_char outc ' ';
-		output_string outc (Tag.to_string (tag a));
-		output_string outc "=\"";
-		output_string outc value;
-		output_char outc '"';
-		loop_attributes (right a)
-	  | _ -> ()
-    in
-	loop ~print_right:false t
-	  
-	  
-    let print_xml_fast outc t = 
-      if Tag.to_string (tag t) = "" then
-	print_xml_fast outc (first_child t)
-      else print_xml_fast outc t
-	
 
-let tags_below t tag = 
-  fst(Hashtbl.find t.ttable tag)
 
-let tags_after t tag = 
-  snd(Hashtbl.find t.ttable tag)
 
-let tags t tag = Hashtbl.find t.ttable tag
+let test_prefix t s = Array.length (text_prefix t.doc s)
+let test_suffix t s = Array.length (text_suffix t.doc s)
+let test_contains t s = Array.length (text_contains t.doc s)
+let test_equals t s = Array.length (text_equals t.doc s)