(* 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
+INCLUDE "utils.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_tc_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
+
-module OldBinary =
-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
+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 text_unsorted_contains : tree -> string -> unit = "caml_text_collection_unsorted_contains"
+external get_cached_text : tree -> [`Text] node -> string = "caml_text_collection_get_cached_text"
+let get_cached_text t x =
+ if x == -1 then ""
+ else get_cached_text t x
+
+external tree_serialize : tree -> string -> unit = "caml_xml_tree_serialize"
+
+external tree_unserialize : string -> tree = "caml_xml_tree_unserialize"
- let descr t = t
+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 tree_last_child : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_last_child"
+external tree_is_first_child : tree -> [`Tree] node -> bool = "caml_xml_tree_is_first_child"
+
+(* 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 string = function String s -> s | _ -> failwith "string"
+let tree_is_last t n = equal_node nil (tree_next_sibling t n)
-
- 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"
-
+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"
+
+let text_size tree = int_of_node (snd ( tree_doc_ids tree (Obj.magic 0) ))
+
+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 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
- | _ -> ()
+
+
+
+let text_size t = text_size t.doc
+
+let collect_tags tree =
+ let h_union = Hashtbl.create 511 in
+ let pt_cup s1 s2 =
+ (* special case, since this is a union we want hash(s1,s2) = hash(s2,s1) *)
+ let x = Ptset.hash s1
+ and y = Ptset.hash s2 in
+ let h = if x < y then HASHINT2(x,y) else HASHINT2(y,x) in
+ try
+ Hashtbl.find h_union h
+ with
+ | Not_found -> let s = Ptset.union s1 s2
+ in
+ Hashtbl.add h_union h s;s
+ in
+ let h_add = Hashtbl.create 511 in
+ let pt_add t s =
+ let k = HASHINT2(Tag.hash t,Ptset.hash s) in
+ try
+ Hashtbl.find h_add k
+ with
+ | Not_found -> let r = Ptset.add t s in
+ Hashtbl.add h_add k r;r
in
- loop t
+ let h = Hashtbl.create 511 in
+ let sing = Ptset.singleton Tag.pcdata in
+ let update t sb sa =
+ let sbelow,safter =
+ try
+ Hashtbl.find h t
+ with
+ | Not_found ->
+ (sing,sing)
+ in
+ Hashtbl.replace h t (pt_cup sbelow sb, pt_cup safter sa)
+ in
+ let rec loop id acc =
+ if equal_node id nil
+ then (Ptset.empty,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 tag below1 after2;
+ pt_add tag (pt_cup below1 below2), (pt_add tag after1)
+ in
+ let b,a = loop (tree_root tree) Ptset.empty in
+ update Tag.pcdata b a;
+ h
-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 contains_array = ref [| |]
+let contains_index = Hashtbl.create 4096
+let in_array _ i =
+ try
+ Hashtbl.find contains_index i
+ with
+ Not_found -> false
-
-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 init_contains t s =
+ let a = text_contains t.doc s
+ in
+ Array.fast_sort (compare) a;
+ contains_array := a;
+ Array.iter (fun x -> Hashtbl.add contains_index x true) !contains_array
+
+let count_contains t s = text_count_contains t.doc s
+let unsorted_contains t s = text_unsorted_contains t.doc s
+
+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 = get_cached_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
-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"
+module DocIdSet = struct
+ include Set.Make (struct type t = [`Text] node
+ let compare = compare_node end)
+
+end
+let is_nil t = t.node == Nil
-let right = function Node(_,_,_,r,_) -> r
- | _ -> failwith "right"
+let is_node t = t.node != Nil
-let first_child = left
-let next_sibling = right
+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
+ let i,j = tree_doc_ids t (tree_root t) in
+ Printf.eprintf "%i docs, range from %i to %i\n%!" (Array.length s) i j;
+ Array.iter (fun i -> print_endline (">>>" ^ i ^ "<<<")) s; *)
+ { doc= t;
+ node = Node(tree_root t);
+ ttable = table;
+ }
+let finalize _ = Printf.eprintf "Release the string list !\n%!"
+;;
+
+let parse f str =
+ node_of_t
+ (f str
+ !Options.sample_factor
+ !Options.index_empty_texts
+ !Options.disable_text_collection)
+
+let parse_xml_uri str = parse parse_xml_uri str
+let parse_xml_string str = parse parse_xml_string str
-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
+
+external pool : tree -> Tag.pool = "%identity"
-let is_right n = match n with
- | Node (_,_,_,_,{contents=p}) when not(is_root n) && (right p) == n -> true
- | _ -> false
+let save t str = (save_tree t.doc str)
+;;
+let load ?(sample=64) str =
+ node_of_t (load_tree str sample)
+
-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
+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 n == -1 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 dump_node t = nts t.node
- type t
- type 'a node = int
- type node_kind = [`Text | `Tree ]
+let mk_nil t = { t with node = Nil }
+let root n = { n with node = norm (tree_root n.doc) }
- let compare : 'a node -> 'a node -> int = fun x y -> x - y
- let equal : 'a node -> 'a node -> bool = fun x y -> x == y
+let is_root n = match n.node with
+ | Node(t) -> (int_of_node t) == 0
+ | _ -> false
+
+let is_left n = match n.node with
+ | Node(t) -> (tree_is_first_child n.doc t) && (equal_node nil (tree_prev_text n.doc t))
+ | Text(_,t) -> tree_is_nil t || tree_is_first_child n.doc t
+ | _ -> false
- (* abstract type, values are pointers to a XMLTree C++ object *)
-
-
- external parse_xml_uri : string -> t = "caml_call_shredder_uri"
- let parse_xml_uri uri = parse_xml_uri uri
+let is_below_right t1 t2 =
+ match (t1.node,t2.node) with
+ | Nil,_ | _,Nil -> false
+ | Node(i1), Node(i2) ->
+ tree_is_ancestor t1.doc (tree_parent t1.doc i1) i2
+ && not (tree_is_ancestor t1.doc i1 i2)
+ | Text(_,i1),Node(i2) -> i1 == i2 ||
+ (tree_is_ancestor t1.doc (tree_parent t1.doc i1) i2 && i1 < i2)
+ | Text(_,i1),Text(i,_) ->
+ let x,y = tree_doc_ids t1.doc i1 in
+ i >= x && i <= y
+ | Node(i1), Text(i,_) ->
+ let i2 = tree_next_sibling t1.doc i1 in
+ let x,y = tree_doc_ids t1.doc i2 in
+ i >= x && i <= y
+
+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
- external parse_xml_string : string -> t = "caml_call_shredder_string"
- let parse_xml_string uri = parse_xml_string uri
+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 }
+
- module Text =
- struct
- type t (* pointer to the text collection *)
- (* Todo *)
- external nullt : unit -> [`Text ] node = "caml_xml_tree_nullt"
- let nil = nullt ()
- external get_text1 : t -> [`Text] node -> string = "caml_text_collection_get_text"
+ 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 get_text t n = Printf.printf "@@@@@@%i\n%!" (Obj.magic n);
- if equal nil n then ""
- else get_text1 t n
+
- let is_empty t (n : [`Text] node) = (get_text t n) = ""
- end
+ 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 }
- module Tree =
- struct
+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 }
-
- 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
+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
- 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"
+
+let text_below t =
+ let l = Array.length !contains_array in
+ match t.node with
+ | Node(n) ->
+ let i,j = tree_doc_ids t.doc n in
+ let id = if l == 0 then i else (array_find !contains_array i j)
+ in
+(* Printf.printf "Looking for text below node %i with tag %s in range %i %i, in array : [|\n%!"
+ n (Tag.to_string (tree_tag_id t.doc n)) i j;
+ Array.iter (fun i -> Printf.printf "%i " (int_of_node i )) !contains_array;
+ Printf.printf "|]\nResult is %i\n%!" id; *)
+ 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)) }
+ | _ -> (*Printf.printf "Here\n%!"; *)
+ { t with node = Nil }
+
+let text_next t root =
+ let l = Array.length !contains_array in
+ 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 = if l == 0 then if inf > j then nil else inf
+ else 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}
+
- external text_collection : t -> Text.t = "caml_xml_tree_text_collection"
+(*
+ let subtree_tags t tag =
+ match t with
+ { doc = d; node = Node(NC n) } ->
+ subtree_tags d n tag
+ | _ -> 0
- let is_last t n = equal nil (next_sibling t n)
-
- external prev_text : t -> [`Tree] node -> [`Text ] node = "caml_xml_tree_prev_text"
- let prev_text t id = Printf.eprintf "Calling PrevText for node %i with result" (Obj.magic id);
- let did = if is_nil id then Text.nil else prev_text t id
- in Printf.eprintf " %i!!!\n%!" (Obj.magic did); did
+ 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
+*)
- 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"
+ 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 next_text t id = Printf.eprintf "Calling NextText for node %i with result" (Obj.magic id);
- let did = if is_nil id then Text.nil else next_text t id
- in Printf.eprintf " %i!!!\n%!" (Obj.magic did); did
+ let contains t s =
+ Array.fold_left (fun a i -> DocIdSet.add i a) DocIdSet.empty (Text.contains t.doc s)
- 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"
-
- let print_skel t =
- let rec aux id =
- if (is_nil id)
- then Printf.eprintf "#"
- else
- begin
- Printf.eprintf "%s(" (Tag.to_string (tag t id));
- aux(first_child t id);
- Printf.eprintf ",\n";
- aux(next_sibling t id);
- Printf.eprintf ")\n";
- end
+ 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
- aux (root t)
- end
-
-
- module 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
+ 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
- type doc = t
- type t = { doc : doc;
- node : descr }
-
- let dump { doc=t } = Tree.print_skel t
- 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 =
- Printf.eprintf "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
+ 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
- { n with node = node'}
-
- let next_sibling n =
- Printf.eprintf "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))
+ 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
- { n with node = node'}
+ 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 (get_cached_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,_) -> (get_cached_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 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 "/>"
- | 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 = 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 tags_below t tag =
+ fst(Hashtbl.find t.ttable tag)
- end
+let tags_after t tag =
+ snd(Hashtbl.find t.ttable tag)
-end
+let tags t tag = Hashtbl.find t.ttable tag
+let tagged_lowest t tag =
+ let rec loop_lowest i =
+ let j = tree_tagged_desc t.doc i tag in
+ if tree_is_nil j then i else loop_lowest j
+ in
+ match t.node with
+ | Node i ->
+ let j = loop_lowest i in
+ { t with
+ node = norm(
+ if tree_is_nil j then
+ if (tree_tag_id t.doc i) == tag
+ then i
+ else j
+ else j) }
+ | Nil -> t
+ | _ -> assert false
+
+
+let tagged_next t tag =
+ match t.node with
+ | Node(i) ->
+ let n = tree_tagged_foll_below t.doc i tag (Obj.magic 0)
+ in
+ if tree_is_nil n then mk_nil t
+ else
+ tagged_lowest { t with node = Node n } tag
+ | Nil -> t
+ | _ -> assert false
+
+let rec binary_parent t =
+ let res =
+ match t.node with
+ | Node(0) -> { t with node = Nil }
+ | Node(i) ->
+ let j = tree_prev_sibling t.doc i in
+ if tree_is_nil j then
+ let idoc = tree_prev_text t.doc i in
+ if equal_node nil idoc then
+ { t with node = Node (tree_parent t.doc i) }
+ else
+ { t with node = Text(idoc,i) }
+ else
+ let idoc = tree_prev_text t.doc i in
+ if equal_node nil idoc then
+ { t with node = Node (j) }
+ else { t with node = Text(idoc,i) }
+ | Text(d,i) ->
+ if tree_is_nil i then
+ let n = tree_parent_doc t.doc d in
+ let lc = tree_last_child t.doc n in
+ if tree_is_nil lc then {t with node = Node n }
+ else { t with node = Node lc }
+ else
+ let j = tree_prev_sibling t.doc i in
+ if tree_is_nil j then
+ { t with node = Node (tree_parent t.doc i) }
+ else { t with node = Node j }
+ | Nil -> t
+ in match res.node with
+ | Text(idoc,t) ->
+ if (Array.length !contains_array) != 0
+ then if in_array !contains_array idoc then res
+ else binary_parent res
+ else res
+ | _ -> res
+
+let benchmark_text t =
+ let doc = t.doc in
+ match (root t).node with
+ | Node i -> let _,size = tree_doc_ids doc i in
+ Printf.eprintf "%i will take ~ %i seconds\n%!"
+ size (size/10000) ;
+ let a = Array.create size "" in
+ for i = 0 to size
+ do
+ a.(i) <- text_get_tc_text t.doc (i+1)
+ done; a
+ | _ -> assert false
-let dump = XML.Binary.dump
-include XML
+let doc_ids (t:t) : (int*int) =
+ (Obj.magic (
+ match t.node with
+ | Node i -> tree_doc_ids t.doc i
+ | Text (i,_) -> (i,i)
+ | Nil -> (nil,nil)
+ ))
+
+let subtree_tags t tag = match t.node with
+ | Nil -> 0
+ | Node(i) -> tree_subtree_tags t.doc i tag
+ | Text(_,i) -> tree_subtree_tags t.doc i tag