(******************************************************************************)
(*  SXSI : XPath evaluator                                                    *)
(*  Kim Nguyen (Kim.Nguyen@nicta.com.au)                                      *)
(*  Copyright NICTA 2008                                                      *)
(*  Distributed under the terms of the LGPL (see LICENCE)                     *)
(******************************************************************************)
INCLUDE "debug.ml"
module type BINARY =
sig
  type node_content
  type string_content
  type descr = Nil | Node of node_content  |String of string_content 
  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 :
  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
end

module XML = 
struct

  type t
  type 'a node = int
  type node_kind = [`Text | `Tree ]

  let compare : 'a node -> 'a node -> int = (-)
  let equal : '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 -> 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 = 
      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 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"
  end


  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"
      
    external root : t -> [`Tree] node = "caml_xml_tree_root"
    external nullt : unit -> [`Tree ] node = "caml_xml_tree_nullt"

    let nil = nullt ()
    let is_nil x = equal x nil

    external parent : t -> [`Tree] node -> [`Tree] node = "caml_xml_tree_parent"
    external parent_doc : t -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_parent_doc"
    external first_child : t -> [`Tree] node -> [`Tree] node = "caml_xml_tree_first_child"
      

      
    external next_sibling : t -> [`Tree] node -> [`Tree] node = "caml_xml_tree_next_sibling"

    external is_leaf : t  -> [`Tree] node -> bool = "caml_xml_tree_is_leaf"
    
(*    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)
    
    external prev_text : t -> [`Tree] node -> [`Text ] node = "caml_xml_tree_prev_text" 


    external my_text : t -> [`Tree] node -> [`Text ] node = "caml_xml_tree_my_text"
    external next_text : t -> [`Tree] node -> [`Text ] node = "caml_xml_tree_next_text"

    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"

    let print_skel t =
      let rec aux id = 
	if (is_nil id)
	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) parent_doc(my_text)=%i\n%!" 
	      (int_of_node id)
	      (Tag.to_string (tag_id t id))
	      (node_xml_id t id)
	      (int_of_node (prev_text t id))
	      (Text.get_text t (prev_text t id))
	      (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))
	      (int_of_node(parent_doc t (my_text t id)));
    
	    aux(first_child t id);
	    aux(next_sibling t id);
	  end
      in
	aux (root t)

    let traversal t = 
	let rec aux id =
	  if not (is_nil id)
	  then
	    begin
	      (* ignore (tag t id);
	      ignore (Text.get_text t (prev_text t id));
	      if (is_leaf t id)
		then ignore (Text.get_text t (my_text t id));
	      if (is_last t id)
		then ignore (Text.get_text t (next_text t id)); *)
	      aux (first_child t id);
	      aux (next_sibling t id);
	    end
	in
	  aux (root t)
  end
      
      
  module Binary  = struct

    type node_content = 
	NC of [`Tree ] node 
      | SC of [`Text ] node * [`Tree ] node 
    type string_content = [ `Text ] node
    type descr = 
      | Nil 
      | Node of node_content
      | String of string_content

    type doc = t

    type t = { doc : doc;	       
	       node : descr }
	
    let dump { doc=t } = Tree.print_skel t
    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       
      (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"
    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
      | _, Node(NC( _ )) -> 1
      | Node(SC (i,_)),Node(SC (j,_)) -> compare i j
      | Node(NC( _ )),Node(SC (_,_)) -> -1
      | _, Node(SC (_,_)) -> 1
      | String i, String j -> compare i j
      | Node _ , String _ -> -1
      | _ , String _ -> 1
      | Nil, Nil -> 0
      | _,Nil -> -1

    let equal a b = (compare a b) == 0

    let string t = match t.node with
      | 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)
	
    let descr t = t.node

    let nts = function
	Nil -> "Nil"
      | String i -> Printf.sprintf "String %i" i
      | Node (NC t) -> Printf.sprintf "Node (NC %i)"  (int_of_node t)
      | Node (SC (t,i)) -> Printf.sprintf "Node (SC (%i,%i))"  (int_of_node t) (int_of_node i)


    let parent n = 
      let node' =
	match n.node with
	  | Node(NC t) | Node(SC (_,t)) -> 
	      if (Tree.root n.doc) == t
	      then Nil
	      else Node(NC(Tree.parent n.doc t)) (* A parent node can never be a SC *)
	  | _ -> assert false
      in
	{ n with node = node' }

    let first_child n = 
      let node' = 
	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.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.doc txt
		then norm fs
		else Node (SC (txt, fs))		  
	  | Node(SC (i,_)) -> String i
	  | Nil | String _ -> failwith "first_child"
      in
	{ n with node = node'}

	  
    let next_sibling n = 
      let node' =
	match n.node with
	  | 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 Text.is_empty n.doc txt
		then norm ns
		else Node (SC (txt, ns))
	  | Nil | String _  -> failwith "next_sibling"
      in
	{ n with node = node'}
	  
	  
    let left = first_child 
    let right = next_sibling
    
    let id = 
      function  { doc=d; node=Node(NC n)}  -> node_xml_id d n
	| { doc=d;  node=Node(SC (i,_) )} -> text_xml_id d i
	| _ -> failwith "id"
	    
    let tag = 
      function { node=Node(SC _) } -> Tag.pcdata
	| { doc=d; node=Node(NC n)} -> tag_id d n
	| _ -> failwith "tag"
    
(*    let tag_id = 
      function  { node=Node(SC _) } -> ()
	| { doc=d; node=Node(NC n)} -> tag_id d n
	| _ -> ()
*)
    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 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 = match t.node with
	| Nil -> false
	| String _ -> matching (string t)
	| Node(_) -> (find (left t )) || (find (right t)) 
      in
	find t 

    let print_xml_fast outc t =
      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); 
	    if print_right then loop (right t)
	    
	| Node (_) -> 
	    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  "/>"
		  | String _ -> assert false
		  | 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
		  | Nil -> ""
		  | _ -> string (left(left a)) 
	      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 traversal t = Tree.traversal t.doc
    let full_traversal t = 
      let rec aux n =
	match n.node with
	| Nil -> ()
	| 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
    
    let _timings = Hashtbl.create 107
    

    let time _ref f arg = 
      let t1 = Unix.gettimeofday () in
      let r = f arg in
      let t2 = Unix.gettimeofday () in 
      let t = (1000. *.(t2 -. t1)) in

      let (time,count) = try 
	Hashtbl.find _timings _ref
      with
	| Not_found -> 0.,0
      in
      let time = time+. t 
      and count = count + 1
      in
	Hashtbl.replace _timings _ref (time,count);r

    include XML.Binary


    let first_child_ doc node = 
     time ("XMLTree.FirstChild()") (XML.Tree.first_child doc)  node
    let next_sibling_ doc node = 
      time ("XMLTree.NextSibling()") (XML.Tree.next_sibling doc) node

    let is_empty_ text node = 
      time ("TextCollection.IsEmpty()") (XML.Text.is_empty text) node

    let prev_text_ doc node = 
      time ("XMLTree.PrevText()") (XML.Tree.prev_text doc) node

    let my_text_ doc node = 
      time ("XMLTree.MyText()") (XML.Tree.my_text doc) node
	
    let next_text_ doc node = 
      time ("XMLTree.NextText()") (XML.Tree.next_text doc) node

    let is_leaf_ doc node =  
      time ("XMLTree.IsLeaf()") (XML.Tree.is_leaf doc ) node
	
    let node_xml_id_ doc node =  
      time ("XMLTree.NodeXMLId()") (XML.Tree.node_xml_id doc ) node
	
    let text_xml_id_ doc node =  
      time ("XMLTree.TextXMLId()") (XML.Tree.text_xml_id doc ) node


    let first_child n =
      let node' =
    	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.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.doc txt
    		then norm fs
    		else Node (SC (txt, fs))
    	  | Node(SC (i,_)) -> String i
    	  | Nil | String _ -> failwith "first_child"
      in
    	{ n with node = node'}

	  
    let next_sibling n =
      let node' =
    	match n.node with
    	  | Node (SC (_,ns)) -> norm ns
    	  | Node(NC t) ->
    	      let ns = next_sibling_ n.doc t in
    	      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"
      in
    	{ n with node = node'}

    let id = 
      function  { doc=d; node=Node(NC n)}  -> node_xml_id_ d n
	| { 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
    let right = next_sibling
    let tag t =  time ("XMLTree.GetTag()") (tag) t
      
    let print_stats ppf = 
      let total_time,total_calls =
	Hashtbl.fold  (fun _ (t,c) (tacc,cacc) ->
			 tacc+. t, cacc + c)  _timings (0.,0)

      in
	Format.fprintf ppf
	  "Timing : Function Name, number of calls,%% of total calls, mean time, total time, %% of total time\n%!";
	Hashtbl.iter (fun name (time,count) ->
			Format.fprintf ppf  "%-27s% 8d\t% 4.2f%%\t% 4.6f ms\t% 4.6f ms\t%04.2f%%\n%!"
			  name 
			  count 
			  (100. *. (float_of_int count)/.(float_of_int total_calls))
			  (time /. (float_of_int count))
			  time
			  (100. *. time /.  total_time)) _timings;
	Format.fprintf ppf  "-------------------------------------------------------------------\n";
	Format.fprintf ppf "%-27s% 8d\t% 4.0f%%\t########## ms\t% 4.6f ms\t% 4.0f%%\n%!"
	  "Total" total_calls 100. total_time 100.
			  

    let print_xml_fast outc t =
      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 (_) -> 
	    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  "/>"
		  | String _ -> assert false
		  | 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
		  | Nil -> ""
		  | _ -> string (left(left a)) 
	      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

	


end

module Binary = DEBUGTREE
ELSE
module Binary = XML.Binary
END (* IFDEF DEBUG *)