(******************************************************************************)
(*  SXSI : XPath evaluator                                                    *)
(*  Kim Nguyen (Kim.Nguyen@nicta.com.au)                                      *)
(*  Copyright NICTA 2008                                                      *)
(*  Distributed under the terms of the LGPL (see LICENCE)                     *)
(******************************************************************************)
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 string : t -> string
  val descr : t -> descr
  val left : t -> t
  val right : 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
end

module Binary = 
struct

  type string_content = string
  type descr = Nil | Node of node_content  | String of string_content
  and node_content = int*Tag.t * descr * descr * (descr ref)
  type t = descr
      
  let descr t = t

  let string = function String s -> s | _ -> failwith "string"
    
 
  external parse_xml_uri : string -> t = "caml_call_shredder_uri"
  external parse_xml_string : string -> t = "caml_call_shredder_string"
       
  let parse_xml_uri s = Node(0,Tag.tag "",parse_xml_uri s,Nil,ref Nil)
  let parse_xml_string s = Node(0,Tag.tag "",parse_xml_string s,Nil,ref Nil)
  let tstring = function Nil -> "Nil"
    | Node (_,_,_,_,_) -> "Node"
    | String _ -> "String"
	

let print_xml fmt t =
  let pp_str = Format.pp_print_string fmt in
  let rec loop = function Nil -> ()
    | String (s) -> pp_str s
    | Node (_,t,l,r,_) when Tag.equal t Tag.pcdata -> loop l;loop r
    | Node (_,t,l,r,_) -> 
	pp_str ("<" ^ (Tag.to_string t));
	( match l with
	      Nil -> pp_str "/>"
	    | Node(_,t',atts,children,_) when Tag.equal t' Tag.attribute -> 
		(loop_attributes atts;
		 match children with
		   | Nil -> pp_str "/>"
		   | _ -> 
		       pp_str ">"; 
		       loop children;
		       pp_str ("</"^ (Tag.to_string t)^">" )
		)
	    | _ -> pp_str ">"; loop l;		
		pp_str ("</"^ (Tag.to_string t)^">" );
	);loop r
  and loop_attributes = function 
    | Node(_,t,Node(_,_,String(s),_,_),r,_) ->
	pp_str (" "^(Tag.to_string t)^"=\""^ s ^"\"") ;
	loop_attributes r
    | _ -> ()

  in
    loop t

let print_xml fmt = 
  function Node(i,t,l,_,_) -> print_xml fmt (Node(i,t,l,Nil,ref Nil))
  | t -> print_xml fmt t


(* a bit ugly but inlining like this makes serialization faster *)

let print_xml_fast outc t =
  let rec loop = function Nil -> ()
    | String (s) -> output_string outc  s
    | Node (_,t,l,r,_) when Tag.equal t Tag.pcdata -> loop l;loop r
    | Node (_,t,l,r,_) -> let t = Tag.to_string t in
	output_char outc  '<';
	output_string outc  t;
	( match l with
	      Nil -> output_string outc  "/>"
	    | Node(_,t',atts,children,_) when Tag.equal t' Tag.attribute -> 
		(loop_attributes atts;
		 match children with
		   | Nil -> output_string outc  "/>"
		   | _ -> 
		       output_char outc  '>'; 
		       loop children;
		       output_string outc  "</";
		       output_string outc  t;
		       output_char outc '>' )
	    | _ ->
		output_char outc  '>'; 
		loop l;		
		output_string outc  "</";
		output_string outc t;
		output_char outc '>'
	);loop r
  and loop_attributes = function 
    | Node(_,t,Node(_,_,String(s),_,_),r,_) -> 
	output_char outc ' ';
	output_string outc (Tag.to_string t);
	output_string outc "=\"";
	output_string outc s;
	output_char outc '"';
	loop_attributes r
    | _ -> ()

  in
    loop t

let print_xml_fast outc = 
  function Node(i,t,l,_,_) -> print_xml_fast outc (Node(i,t,l,Nil,ref Nil))
  | t -> print_xml_fast outc t



let tabs = ref 0

let prtabs fmt = 
  for i = 0 to !tabs 
  do
    Format.fprintf fmt " "
  done

    
let rec dump fmt t = 
  incr tabs;
  let _ = match t with
    | Nil ->  prtabs fmt; Format.fprintf fmt "#" 
    | String s -> prtabs fmt; Format.fprintf fmt "(String %s)" s
    | Node(id,t,l,r,_) -> 
	prtabs fmt;
	Format.fprintf fmt " (tag='";
	Tag.print fmt t;
	Format.fprintf fmt "', id='%i')\n" id;
	prtabs fmt;
	dump fmt l;
	Format.fprintf fmt "\n";
	prtabs fmt;
	dump fmt r;
	Format.fprintf fmt "\n";
	prtabs fmt;prtabs fmt;
	Format.fprintf fmt "(id='%i'end )\n" id
  in decr tabs
	
	  
let dump fmt t = 
  tabs:=0;
  dump fmt t;
  tabs:=0

let id = function Node(i,_,_,_,_) -> i
  | _ -> failwith "id"

let tag = function Node(_,t,_,_,_) -> t
  | _ -> failwith "tag"

let left = function Node(_,_,l,_,_) -> l
  | _ -> failwith "left"

let right = function Node(_,_,_,r,_) -> r
  | _ -> failwith "right"

let first_child = left
let next_sibling = right

let is_root = function Node (_,_,_,_,{contents=Nil}) -> true | _ -> false
let is_left n = match n with
  | Node (_,_,_,_,{contents=p}) when not(is_root n) && (left p) == n -> true 
  | _ -> false

let is_right n = match n with
  | Node (_,_,_,_,{contents=p}) when not(is_root n) && (right p) == n -> true 
  | _ -> false


let compare t1 t2 = match t1,t2 with
  | Nil,Nil -> 0
  | String s1, String s2 -> String.compare s1 s2
  | Nil, String _ -> -1
  | String _, Nil -> 1
  | Node(i1,_,_,_,_), Node(i2,_,_,_,_) -> i1 - i2
  | _, Node _ -> -1
  | Node _ , _ -> 1
let equal t1 t2 = (compare t1 t2) == 0

let int_size = Sys.word_size/8
let ssize s = ((String.length s)/4 +1)*4 
let rec size = 
  function Nil -> (int_size,1,0,0) 
    | String s -> (int_size + (ssize s),0,1,0)
    | Node(_,_,l,r,_) -> 
	let sizel,nl,sl,il = size l 
	and sizer,nr,sr,ir = size r 
	in
	  (sizel+sizer+(7*int_size),nl+nr,sl+sr,il+ir+1)
let size t = 
  let s,n,st,i = size t in
    s/1024,n,st,i
end 


module XML = 
struct

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

  let compare : 'a node -> 'a node -> int = fun x y -> x - y
  let equal : 'a node -> 'a node -> bool = fun x y -> x == y

        (* abstract type, values are pointers to a XMLTree C++ object *)
    
    
  external parse_xml_uri : bool -> float ->  string  -> t = "caml_call_shredder_uri"
  let parse_xml_uri  uri  = parse_xml_uri true 1.0 uri
    
  external parse_xml_string : bool -> float ->  string  -> t = "caml_call_shredder_string"
  let parse_xml_string  = parse_xml_string true 1.0
    

  module Text =
  struct
    type t (* pointer to the text collection *)
    (* Todo *)
    
    external get_text : t -> [`Text] node -> string = "caml_text_collection_get_text"
    let is_empty t (n : [`Text] node) = "" = (get_text t n)

    end
  module Tree = 
  struct

      
    external serialize : 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_first_child"
    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 -> Tag.t = "caml_xml_tree_tag"

    external text_collection : t -> Text.t = "caml_xml_tree_text_collection"

    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"
            
  end
      
      
  module Binary : BINARY = struct
    
    type node_content = 
	[ `Node of [`Tree ] node 
	| `String 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 }
	

    open Tree      	       
    let node_of_t t = { doc= t; node= Node(`Node (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 compare a b = match a.node,b.node  with
      | Node(`Node i),Node(`Node j) -> compare i j
      | _, Node(`Node( _ )) -> 1
      | Node(`String (i,_)),Node(`String (j,_)) -> compare i j
      | Node(`Node( _ )),Node(`String (_,_)) -> -1
      | _, Node(`String (_,_)) -> 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 (text_collection t.doc) i
      | _ -> assert false
	  
    let norm (n : [`Tree ] node ) =  if is_nil n then Nil else Node (`Node n)
	
    let descr t = t.node

    let first_child n = 
      let node' = 
	match n.node with
	  | Nil | String _ -> failwith "first_child"
	  | Node (`Node t) -> 
	      let fs = first_child n.doc t in
	      let txt = prev_text n.doc t in
		if Text.is_empty (text_collection n.doc) txt
		then norm fs
		else Node (`String (txt, fs))
		  
	  | Node(`String (i,_)) -> String i
      in
	{ n with node = node'}
	  
    let next_sibling n = 
      let node' =
	match n.node with
	  | Nil | String _  -> failwith "next_sibling"
	  | Node (`String (_,ns)) -> norm ns
	  | Node(`Node t) ->
	      let ns = next_sibling n.doc t in
	      let txt = next_text n.doc t in
		if Text.is_empty (text_collection n.doc) txt
		then norm ns
		else Node (`String (txt, ns))
      in
	{ n with node = node'}
	  
	  
    let left = first_child
    let right = next_sibling
    let id = 
      function  { doc=d; node=Node(`Node n)}  -> text_xml_id d n
	| { doc=d;  node=Node(`String (i,_) )} -> node_xml_id d i
	| _ -> failwith "id"
	    
    let tag = 
      function { node=Node(`String _) } -> Tag.pcdata
	| { doc=d; node=Node(`Node n)} -> tag d n
	| _ -> failwith "Tag"
	    
	    
	    
    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  "/>"
		  | 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 (left 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 = 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
	  
  end
    
end