X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=tree.ml;h=3bfbfceeac181d6976aba24b50b3f8da1b0cc83c;hb=63ca35af9ef5c0b18b3d3217536f3353f77f5465;hp=0c1f10b2ec8e4cf2af2fd453f824de1577e8a8d1;hpb=24fdea81b5506233d139bd7d72364a190bef35b8;p=SXSI%2Fxpathcomp.git

diff --git a/tree.ml b/tree.ml
index 0c1f10b..3bfbfce 100644
--- a/tree.ml
+++ b/tree.ml
@@ -4,6 +4,7 @@
 (*  Copyright NICTA 2008                                                      *)
 (*  Distributed under the terms of the LGPL (see LICENCE)                     *)
 (******************************************************************************)
+INCLUDE "debug.ml"
 module type BINARY =
 sig
   type node_content
@@ -12,6 +13,7 @@ sig
   type t
   val parse_xml_uri : string -> t
   val parse_xml_string : string -> t
+  val tag_pool : t -> Tag.pool
   val string : t -> string
   val descr : t -> descr
   val left : t -> t
@@ -43,16 +45,15 @@ struct
     
   external int_of_node : 'a node -> int = "%identity"
 
-  external parse_xml_uri : string  -> t = "caml_call_shredder_uri"
-  let parse_xml_uri uri = parse_xml_uri uri
-    
-  external parse_xml_string :  string  -> t = "caml_call_shredder_string"
-  let parse_xml_string uri = parse_xml_string uri
+  external parse_xml_uri : string -> int -> bool -> bool -> t = "caml_call_shredder_uri"
+       
     
+  external parse_xml_string :  string -> int -> bool -> bool -> t = "caml_call_shredder_string"
+
 
   module Text =
   struct
-    type t (* pointer to the text collection *)
+
     (* Todo *)
     external nullt : unit -> [`Text ] node = "caml_xml_tree_nullt"
     let nil = nullt ()
@@ -95,10 +96,8 @@ struct
 
     external is_leaf : t  -> [`Tree] node -> bool = "caml_xml_tree_is_leaf"
     
-    external tag : t -> [`Tree ] node -> Tag.t = "caml_xml_tree_tag"
-    external tag_id : t -> [`Tree ] node -> unit = "caml_xml_tree_tag_id"
-
-    external text_collection : t -> Text.t = "caml_xml_tree_text_collection"
+(*    external tag : t -> [`Tree ] node -> T = "caml_xml_tree_tag"*)
+    external tag_id : t -> [`Tree ] node -> Tag.t = "caml_xml_tree_tag_id"
 
     let is_last t n = equal nil (next_sibling t n)
     
@@ -113,7 +112,6 @@ struct
     external is_ancestor : t -> [`Tree ] node -> [`Tree ] node -> bool = "caml_xml_tree_is_ancestor"
 
     let print_skel t =
-      let textcol = text_collection t in
       let rec aux id = 
 	if (is_nil id)
 	then Printf.eprintf "#\n"
@@ -121,14 +119,14 @@ struct
 	  begin
 	    Printf.eprintf "Node %i has tag '%s' DocOrder=%i, DocID of PrevText,MyText,NextText : (%i = %s,%i = %s,%i = %s)\n%!" 
 	      (int_of_node id)
-	      (Tag.to_string (tag t id))
+	      (Tag.to_string (tag_id t id))
 	      (node_xml_id t id)
 	      (int_of_node (prev_text t id))
-	      (Text.get_text textcol (prev_text t id))
+	      (Text.get_text t (prev_text t id))
 	      (int_of_node (my_text t id))
-	      (Text.get_text textcol (my_text t id))
+	      (Text.get_text t (my_text t id))
 	      (int_of_node (next_text t id))
-	      (Text.get_text textcol (next_text t id));
+	      (Text.get_text t (next_text t id));
 	    aux(first_child t id);
 	    aux(next_sibling t id);
 	  end
@@ -136,17 +134,16 @@ struct
 	aux (root t)
 
     let traversal t = 
-      let textcol = text_collection t in
 	let rec aux id =
 	  if not (is_nil id)
 	  then
 	    begin
 	      (* ignore (tag t id);
-	      ignore (Text.get_text textcol (prev_text t id));
+	      ignore (Text.get_text t (prev_text t id));
 	      if (is_leaf t id)
-		then ignore (Text.get_text textcol (my_text t id));
+		then ignore (Text.get_text t (my_text t id));
 	      if (is_last t id)
-		then ignore (Text.get_text textcol (next_text t id)); *)
+		then ignore (Text.get_text t (next_text t id)); *)
 	      aux (first_child t id);
 	      aux (next_sibling t id);
 	    end
@@ -168,8 +165,7 @@ struct
 
     type doc = t
 
-    type t = { doc : doc;
-	       text : Text.t;
+    type t = { doc : doc;	       
 	       node : descr }
 	
     let dump { doc=t } = Tree.print_skel t
@@ -179,12 +175,24 @@ struct
 
     open Tree      	       
     let node_of_t t = { doc= t; 
-			text = text_collection t;
 			node = Node(NC (root t)) }
 
 
-    let parse_xml_uri str = node_of_t (parse_xml_uri str)
-    let parse_xml_string str = node_of_t (parse_xml_string str)
+    let parse_xml_uri str = node_of_t       
+      (MM((parse_xml_uri str 
+	     !Options.sample_factor 
+	     !Options.index_empty_texts
+	     !Options.disable_text_collection),__LOCATION__))
+
+    let parse_xml_string str = node_of_t 
+      (MM((parse_xml_string str
+	 !Options.sample_factor 
+	 !Options.index_empty_texts 
+	 !Options.disable_text_collection),__LOCATION__))
+
+
+    external pool : doc -> Tag.pool = "%identity"
+    let tag_pool t = pool t.doc
 
     let compare a b = match a.node,b.node  with
       | Node(NC i),Node(NC j) -> compare i j
@@ -201,7 +209,7 @@ struct
     let equal a b = (compare a b) == 0
 
     let string t = match t.node with
-      | String i ->  Text.get_text t.text i
+      | String i ->  Text.get_text t.doc i
       | _ -> assert false
 	  
     let norm (n : [`Tree ] node ) =  if is_nil n then Nil else Node (NC n)
@@ -231,13 +239,13 @@ struct
 	match n.node with
 	  | Node (NC t) when is_leaf n.doc t ->
 	      let txt = my_text n.doc t in
-		if Text.is_empty n.text txt
+		if Text.is_empty n.doc txt
 		then Nil
 		else Node(SC (txt,Tree.nil))
 	  | Node (NC t) -> 
 	      let fs = first_child n.doc t in
 	      let txt = prev_text n.doc fs in
-		if Text.is_empty n.text txt
+		if Text.is_empty n.doc txt
 		then norm fs
 		else Node (SC (txt, fs))		  
 	  | Node(SC (i,_)) -> String i
@@ -253,7 +261,7 @@ struct
 	  | Node(NC t) ->
 	      let ns = next_sibling n.doc t in
 	      let txt = next_text n.doc t in
-		if Text.is_empty n.text txt
+		if Text.is_empty n.doc txt
 		then norm ns
 		else Node (SC (txt, ns))
 	  | Nil | String _  -> failwith "next_sibling"
@@ -271,14 +279,14 @@ struct
 	    
     let tag = 
       function { node=Node(SC _) } -> Tag.pcdata
-	| { doc=d; node=Node(NC n)} -> tag d n
-	| _ -> failwith "Tag"
+	| { doc=d; node=Node(NC n)} -> tag_id d n
+	| _ -> failwith "tag"
     
-    let tag_id = 
+(*    let tag_id = 
       function  { node=Node(SC _) } -> ()
 	| { doc=d; node=Node(NC n)} -> tag_id d n
 	| _ -> ()
-
+*)
     let string_below t id =
       let pid = parent_doc t.doc id in
 	match t.node with
@@ -287,7 +295,7 @@ struct
 	  | _ -> false
 	      
     let contains t s = 
-      Array.fold_left (fun a i -> DocIdSet.add i a) DocIdSet.empty (Text.contains t.text 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
@@ -367,20 +375,22 @@ struct
       let rec aux n =
 	match n.node with
 	| Nil -> ()
-	| String i -> () (*ignore(Text.get_text t.text i)  *)
+	| String i -> () (*ignore(Text.get_text t.doc i)  *)
 	| Node(_) -> 
 	    (* tag_id n; *)
 	    aux (first_child n);
 	    aux (next_sibling n)
       in aux t
+
+    let print_stats _ = ()
   end
 
 end
 
 
 
-
-
+IFDEF DEBUG
+THEN
 module DEBUGTREE 
   = struct
     
@@ -438,13 +448,13 @@ module DEBUGTREE
     	match n.node with
     	  | Node (NC t) when is_leaf_ n.doc t ->
     	      let txt = my_text_ n.doc t in
-    		if is_empty_ n.text txt
+    		if is_empty_ n.doc txt
     		then Nil
     		else Node(SC (txt,XML.Tree.nil))
     	  | Node (NC t) ->
     	      let fs = first_child_ n.doc t in
     	      let txt = prev_text_ n.doc fs in
-    		if is_empty_ n.text txt
+    		if is_empty_ n.doc txt
     		then norm fs
     		else Node (SC (txt, fs))
     	  | Node(SC (i,_)) -> String i
@@ -459,8 +469,12 @@ module DEBUGTREE
     	  | Node (SC (_,ns)) -> norm ns
     	  | Node(NC t) ->
     	      let ns = next_sibling_ n.doc t in
-    	      let txt = next_text_ n.doc t in
-    		if is_empty_ n.text txt
+    	      let txt = 
+		if XML.Tree.is_nil ns then
+		  next_text_ n.doc t 
+		else prev_text_ n.doc ns
+	      in
+    		if is_empty_ n.doc txt
     		then norm ns
     		else Node (SC (txt, ns))
     	  | Nil | String _  -> failwith "next_sibling"
@@ -472,7 +486,6 @@ module DEBUGTREE
 	| { doc=d;  node=Node(SC (i,_) )} -> text_xml_id_ d i
 	| _ -> failwith "id"
 	    
-
     (* Wrapper around critical function *)
     let string t = time ("TextCollection.GetText()") (string) t
     let left = first_child
@@ -564,4 +577,6 @@ module DEBUGTREE
 end
 
 module Binary = DEBUGTREE
-
+ELSE
+module Binary = XML.Binary
+END (* IFDEF DEBUG *)