X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=tree.ml;h=16b32a30c6f9699205d4086a7502955b4b1fbded;hb=4680fa5b41156d70f0fde69981f0d241184b19d9;hp=1b4ce2ecb26d79d1377b652d207173de8d63d508;hpb=c10ce35cd399aff15a49f3b24a31b38cb2191da0;p=SXSI%2Fxpathcomp.git

diff --git a/tree.ml b/tree.ml
index 1b4ce2e..16b32a3 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
@@ -12,22 +13,38 @@ sig
   type t
   val parse_xml_uri : string -> t
   val parse_xml_string : string -> t
+  val save : t -> string -> unit
+  val load : ?sample:int -> string -> t
   val tag_pool : t -> Tag.pool
   val string : t -> string
   val descr : t -> descr
+  val is_node : t -> bool
   val left : t -> t
   val right : t -> t
+  val first_child : t -> t
+  val next_sibling : t -> t
   val parent : t -> t
   val id : t -> int
   val tag : t -> Tag.t
   val print_xml_fast : out_channel -> t -> unit
   val compare : t -> t -> int
   val equal : t -> t -> bool
-  module DocIdSet : Set.S with type elt = string_content
+  module DocIdSet :
+  sig 
+    include Set.S 
+  end
+    with type elt = string_content
   val string_below : t -> string_content -> bool
   val contains : t -> string -> DocIdSet.t
   val contains_old : t -> string -> bool
   val dump : t -> unit
+  val get_string : t -> string_content -> string
+  val has_tagged_desc : t -> Tag.t -> bool
+  val has_tagged_foll : t -> Tag.t -> bool
+  val tagged_desc : t -> Tag.t -> t
+  val tagged_foll : t -> Tag.t -> t
+  val tagged_next : t -> Tag.t -> t
+  val subtree_tags : t -> Tag.t -> int
 end
 
 module XML = 
@@ -44,29 +61,37 @@ 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"
+
+  external save_tree : t -> string -> unit = "caml_xml_tree_save"
+  external load_tree : string -> int -> t = "caml_xml_tree_load"
+
 
   module Text =
   struct
-
+    let equal : [`Text] node -> [`Text] node -> bool = equal
+      
     (* Todo *)
     external nullt : unit -> [`Text ] node = "caml_xml_tree_nullt"
     let nil = nullt ()
     external get_text : t -> [`Text] node -> string = "caml_text_collection_get_text"
 
-    let get_text t n = 
+(*    let get_text t n = 
       if equal nil n then "" 
       else  get_text t n
+*)
 		
     external is_empty : t -> [`Text ] node -> bool = "caml_text_collection_empty_text"
     let is_empty t n =
       (equal nil n) || is_empty t n
 
+    external get_cached_text : t -> [`Text ] node -> string = "caml_text_collection_get_cached_text"
+      
+    let get_text t n =
+      if (equal nil n) || is_empty t n then ""
+      else get_cached_text t n
+
     external is_contains : t -> string -> bool = "caml_text_collection_is_contains"
     external count_contains : t -> string -> int = "caml_text_collection_count_contains"
     external contains : t -> string -> [`Text ] node array = "caml_text_collection_contains"
@@ -76,7 +101,7 @@ struct
   module Tree = 
   struct
 
-      
+    let equal : [`Tree ] node -> [`Tree] node -> bool = equal
     external serialize : t -> string -> unit = "caml_xml_tree_serialize"
     external unserialize : string -> t = "caml_xml_tree_unserialize"
       
@@ -110,6 +135,10 @@ struct
     external text_xml_id : t -> [`Text ] node -> int = "caml_xml_tree_text_xml_id"
     external node_xml_id : t -> [`Tree ] node -> int = "caml_xml_tree_node_xml_id"
     external is_ancestor : t -> [`Tree ] node -> [`Tree ] node -> bool = "caml_xml_tree_is_ancestor"
+    external tagged_desc : t -> [`Tree ] node -> Tag.t -> [`Tree ] node = "caml_xml_tree_tagged_desc"
+    external tagged_foll : t -> [`Tree ] node -> Tag.t -> [`Tree ] node = "caml_xml_tree_tagged_foll"
+    external tagged_next : t -> [`Tree ] node -> Tag.t -> [`Tree ] node = "caml_xml_tree_tagged_next"
+    external subtree_tags : t -> [`Tree ] node -> Tag.t -> int = "caml_xml_tree_subtree_tags"
 
     let print_skel t =
       let rec aux id = 
@@ -117,7 +146,7 @@ struct
 	then Printf.eprintf "#\n"
 	else 
 	  begin
-	    Printf.eprintf "Node %i has tag '%s' DocOrder=%i, DocID of PrevText,MyText,NextText : (%i = %s,%i = %s,%i = %s)\n%!" 
+	    Printf.eprintf "Node %i has tag '%s' DocOrder=%i, DocID of PrevText,MyText,NextText : (%i = %s,%i = %s,%i = %s) parent_doc(my_text)=%i\n%!" 
 	      (int_of_node id)
 	      (Tag.to_string (tag_id t id))
 	      (node_xml_id t id)
@@ -126,7 +155,9 @@ struct
 	      (int_of_node (my_text t id))
 	      (Text.get_text t (my_text t id))
 	      (int_of_node (next_text t id))
-	      (Text.get_text t (next_text t id));
+	      (Text.get_text t (next_text t id))
+	      (int_of_node(parent_doc t (my_text t id)));
+    
 	    aux(first_child t id);
 	    aux(next_sibling t id);
 	  end
@@ -169,17 +200,35 @@ struct
 	       node : descr }
 	
     let dump { doc=t } = Tree.print_skel t
-    module DocIdSet = Set.Make (struct type t = string_content
-				       let compare = (-) end)
+    module DocIdSet = struct
+      include Set.Make (struct type t = string_content
+			       let compare = (-) end)
+			
+    end
+    let is_node = function { node=Node(_) } -> true | _ -> false
       
-
+    let get_string t (i:string_content) = Text.get_text t.doc i
     open Tree      	       
     let node_of_t t = { doc= 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__))
+
+
+    let save t str = save_tree t.doc str
+
+    let load ?(sample=64) str = node_of_t (load_tree str sample)
 
 
     external pool : doc -> Tag.pool = "%identity"
@@ -279,12 +328,54 @@ struct
 	| _ -> ()
 *)
     let string_below t id =
-      let pid = parent_doc t.doc id in
+      let strid = parent_doc t.doc id in
 	match t.node with
-	  | Node(NC(i)) -> (is_ancestor t.doc i pid)
-	  | Node(SC(i,_)) -> (is_ancestor t.doc (parent_doc t.doc i) pid)
+	  | 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 tagged_next t tag =
+      if tag = Tag.attribute || tag = Tag.pcdata then failwith "tagged_next"
+      else match t with
+	| { doc=d; node=Node(NC n) } -> { t with node = norm (tagged_next d n tag) }
+	| { doc=d; node=Node(SC (_,n)) } -> { t with node = norm (tagged_next d n tag) }
+	| _ -> { 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 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)
 
@@ -307,7 +398,9 @@ struct
       let rec loop ?(print_right=true) t = match t.node with 
 	| Nil -> ()
 	| String (s) -> output_string outc (string t)
-	| Node _ when Tag.equal (tag t) Tag.pcdata -> loop (left t); loop (right t)
+	| Node _ when Tag.equal (tag t) Tag.pcdata -> 
+	    loop (left t); 
+	    if print_right then loop (right t)
 	    
 	| Node (_) -> 
 	    let tg = Tag.to_string (tag t) in
@@ -372,14 +465,16 @@ struct
 	    aux (first_child n);
 	    aux (next_sibling n)
       in aux t
+
+    let print_stats _ = ()
   end
 
 end
 
 
 
-
-
+IFDEF DEBUG
+THEN
 module DEBUGTREE 
   = struct
     
@@ -566,3 +661,6 @@ module DEBUGTREE
 end
 
 module Binary = DEBUGTREE
+ELSE
+module Binary = XML.Binary
+END (* IFDEF DEBUG *)