(******************************************************************************)
(*  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" *)

type tree
type 'a node = int
type node_kind = [`Text | `Tree ]
    
let compare_node : 'a node -> 'a node -> int = (-)
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_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 nullt : unit -> 'a node = "caml_xml_tree_nullt"

let nil : 'a node = Obj.magic (-1)

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

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 tree_serialize : tree -> string -> unit = "caml_xml_tree_serialize" 

external tree_unserialize : string -> tree = "caml_xml_tree_unserialize"
      
external tree_root : tree -> [`Tree] node = "caml_xml_tree_root" 

let tree_is_nil x = equal_node x nil

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)





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)
    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)
  in
  let b,a = loop (tree_root tree) Ptset.empty in
    update h Tag.pcdata b a;
    h



let contains_array = ref [| |]
  
let init_contains t s = 
  let a = text_contains t.doc s 
  in
    Array.fast_sort (compare) a;
    contains_array := a
      
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 = 
    try
      let _ = Str.search_forward regexp arg 0;
      in true
    with _ -> false
  in
  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
  in
  let acc,l = loop i [] 0 in
  let a = Array.create l nil in
  let _ = List.fold_left (fun cpt e -> a.(cpt) <- e; (cpt-1)) (l-1) acc
  in
    contains_array := a
	  


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


let node_of_t t =
  let _ = Tag.init (Obj.magic t) in
  let table = 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;
    }


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 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 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 }	  


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 nil
	else
	  if a.(idx) >= x then if a.(idx) > y then nil else (last_idx := idx;a.(idx))
	  else loop (idx+1) x y
  in
    if a.(0) > j || a.(l-1) < i then nil
    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
	    in
	      if id == nil then  
		{ t with  node=Nil }
	      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 }
		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


    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 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 (right 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