(* 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 root : t -> t
- val is_root : t -> bool
- 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 -> DocIdSet.t
- val contains_iter : t -> string -> DocIdSet.t
- val count_contains : t -> string -> int
- val count : t -> string -> int
- 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
- val is_left : t -> bool
-end
-
-module XML =
-struct
-
- type t
- type 'a node = int
- type node_kind = [`Text | `Tree ]
+INCLUDE "utils.ml"
- let compare : 'a node -> 'a node -> int = (-)
- let equal : 'a node -> 'a node -> bool = (==)
-
- (* abstract type, values are pointers to a XMLTree C++ object *)
+type tree
+type 'a node = int
+type node_kind = [`Text | `Tree ]
- 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
-*)
+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 is_empty : t -> [`Text ] node -> bool = "caml_text_collection_empty_text"
- let is_empty t n =
- (equal nil n) || is_empty t n
+external text_is_empty : tree -> [`Text ] node -> bool = "caml_text_collection_empty_text"
- external get_cached_text : t -> [`Text ] node -> string = "caml_text_collection_get_cached_text"
-
+let text_is_empty t n =
+ (equal_node nil n) || text_is_empty t n
+
- let get_text t n =
- if (equal nil n) || is_empty t n then ""
- else get_cached_text t n
- external size : t -> int = "caml_text_collection_size"
- external is_contains : t -> string -> bool = "caml_text_collection_is_contains"
- external count_contains : t -> string -> int = "caml_text_collection_count_contains"
- external count : t -> string -> int = "caml_text_collection_count"
- external contains : t -> string -> [`Text ] node array = "caml_text_collection_contains"
- end
+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 text_get_cached_text : tree -> [`Text] node -> string = "caml_text_collection_get_cached_text"
- module Tree =
- struct
+external tree_serialize : tree -> string -> unit = "caml_xml_tree_serialize"
- 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 tree_unserialize : string -> tree = "caml_xml_tree_unserialize"
- external root : t -> [`Tree] node = "caml_xml_tree_root"
- external nullt : unit -> [`Tree ] node = "caml_xml_tree_nullt"
+external tree_root : tree -> [`Tree] node = "caml_xml_tree_root"
- let nil = nullt ()
- let is_nil x = equal x nil
+let tree_is_nil x = equal_node 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 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_tagged_child : tree -> [`Tree] node -> Tag.t -> [`Tree] node = "caml_xml_tree_tagged_child"
+external tree_next_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_next_sibling"
+external tree_tagged_sibling : tree -> [`Tree] node -> Tag.t -> [`Tree] node = "caml_xml_tree_tagged_sibling"
-
- external next_sibling : t -> [`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 is_leaf : t -> [`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"
-(* 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"
- 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 =
- 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)));
+let tree_is_last t n = equal_node nil (tree_next_sibling t n)
- 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
+external tree_prev_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_prev_text"
- 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
+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"
- 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 text_size tree = int_of_node (snd ( tree_doc_ids tree (Obj.magic 0) ))
+let text_get_cached_text t x =
+ if x == -1 then ""
+ else
+ text_get_cached_text t x
- 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__))
+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"
- let save t str = save_tree t.doc str
- let load ?(sample=64) str = node_of_t (load_tree str sample)
+type int_vector
+external int_vector_alloc : int -> int_vector = "caml_int_vector_alloc"
+external int_vector_length : int_vector -> int = "caml_int_vector_length"
+external int_vector_set : int_vector -> int -> int -> unit = "caml_int_vector_set"
+external tree_select_child : tree -> [`Tree ] node -> int_vector -> [`Tree] node = "caml_xml_tree_select_child"
+external tree_select_foll_sibling : tree -> [`Tree ] node -> int_vector -> [`Tree] node = "caml_xml_tree_select_foll_sibling"
+external tree_select_desc : tree -> [`Tree ] node -> int_vector -> [`Tree] node = "caml_xml_tree_select_desc"
+external tree_select_foll_below : tree -> [`Tree ] node -> int_vector -> [`Tree] node -> [`Tree] node = "caml_xml_tree_select_foll_below"
- 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
+module HPtset = Hashtbl.Make(Ptset.Int)
- let equal a b = (compare a b) == 0
+let vector_htbl = HPtset.create MED_H_SIZE
- 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 root n = { n with node = norm (Tree.root n.doc) }
- let is_root n = match n.node with
- | Node(NC t) when (Tree.root n.doc) == t -> true
- | _ -> false
-
- let parent n =
- let node' =
- match n.node with
- | Node(NC t) ->
- let txt = prev_text n.doc t in
- if Text.is_empty n.doc txt then
- Node(NC (Tree.parent n.doc t))
- else
- Node(SC (txt,t))
- | Node(SC(t,_)) -> Node (NC(parent_doc n.doc t))
- | _ -> failwith "parent"
- 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 ptset_to_vector s =
+ try
+ HPtset.find vector_htbl s
+ with
+ Not_found ->
+ let v = int_vector_alloc (Ptset.Int.cardinal s) in
+ let _ = Ptset.Int.fold (fun e i -> int_vector_set v i e;i+1) s 0 in
+ HPtset.add vector_htbl s v; v
-
- 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"
+
+type t = { doc : tree;
+ node : [`Tree] node;
+ ttable : (Tag.t,(Ptset.Int.t*Ptset.Int.t)) Hashtbl.t;
+ }
+
+let text_size t = text_size t.doc
+
+module MemUnion = Hashtbl.Make (struct
+ type t = Ptset.Int.t*Ptset.Int.t
+ let equal (x,y) (z,t) = (Ptset.Int.equal x z)&&(Ptset.Int.equal y t)
+ let equal a b = equal a b || equal b a
+ let hash (x,y) = (* commutative hash *)
+ let x = Ptset.Int.hash x
+ and y = Ptset.Int.hash y
+ in
+ if x < y then HASHINT2(x,y) else HASHINT2(y,x)
+ end)
+
+let collect_tags tree =
+ let h_union = MemUnion.create BIG_H_SIZE in
+ let pt_cup s1 s2 =
+ try
+ MemUnion.find h_union (s1,s2)
+ with
+ | Not_found -> let s = Ptset.Int.union s1 s2
+ in
+ MemUnion.add h_union (s1,s2) s;s
+ in
+ let h_add = Hashtbl.create BIG_H_SIZE in
+ let pt_add t s =
+ let k = HASHINT2(Tag.hash t,Ptset.Int.hash s) in
+ try
+ Hashtbl.find h_add k
+ with
+ | Not_found -> let r = Ptset.Int.add t s in
+ Hashtbl.add h_add k r;r
+ in
+ let h = Hashtbl.create BIG_H_SIZE in
+ let update t sb sa =
+ let sbelow,safter =
+ try
+ Hashtbl.find h t
+ with
+ | Not_found ->
+ (Ptset.Int.empty,Ptset.Int.empty)
+ 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.Int.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 _ = loop (tree_root tree) Ptset.Int.empty in h
+
+
+
+
+
+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 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 = text_get_cached_text t.doc n
in
- { n with node = node'}
+ 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 left = first_child
- let right = next_sibling
+
+
+module DocIdSet = struct
+ include Set.Make (struct type t = [`Text] node
+ let compare = compare_node end)
- 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"
+end
+let is_nil t = t.node == nil
+
+let is_node t = t.node != nil
+
+let node_of_t t =
+ let _ = Tag.init (Obj.magic t) in
+ let table = collect_tags t
+ in
+ { doc= t;
+ 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 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 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 parse_xml_uri str = parse parse_xml_uri str
+let parse_xml_string str = parse parse_xml_string str
- 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 }
+
+external pool : tree -> Tag.pool = "%identity"
+let save t str = (save_tree t.doc str)
+;;
- 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 load ?(sample=64) str =
+ node_of_t (load_tree str sample)
+
- let subtree_tags t tag =
- match t with
- { doc = d; node = Node(NC n) } -> subtree_tags d n tag
- | _ -> 0
+let tag_pool t = pool t.doc
+
+let compare a b = a.node - b.node
- 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 equal a b = a.node == b.node
+
+let nts = function
+ -1 -> "Nil"
+ | i -> Printf.sprintf "Node (%i)" i
+
+let dump_node t = nts t.node
- let contains t s =
- Array.fold_left (fun a i -> DocIdSet.add i a) DocIdSet.empty (Text.contains t.doc s)
+let mk_nil t = { t with node = nil }
+let root n = { n with node = tree_root n.doc }
+let is_root n = n.node == (tree_root n.doc)
+
+let is_left n = tree_is_first_child n.doc n.node
- 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 is_below_right t1 t2 = tree_is_ancestor t1.doc (tree_parent t1.doc t1.node) t2.node
+let parent n = { n with node = tree_parent n.doc n.node }
- 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 first_child n = { n with node = tree_first_child n.doc n.node }
+let tagged_child tag n = { n with node = tree_tagged_child n.doc n.node tag }
+let select_child ts n = { n with node = tree_select_child n.doc n.node (ptset_to_vector ts) }
+
+let next_sibling n = { n with node = tree_next_sibling n.doc n.node }
+let tagged_sibling tag n = { n with node = tree_tagged_sibling n.doc n.node tag }
+let select_sibling ts n = { n with node = tree_select_foll_sibling n.doc n.node (ptset_to_vector ts) }
+
+let next_sibling_ctx n _ = next_sibling n
+let tagged_sibling_ctx tag n _ = tagged_sibling tag n
+let select_sibling_ctx ts n _ = select_sibling ts n
+
+let id t = tree_node_xml_id t.doc t.node
+
+let tag t = if t.node == nil then Tag.nullt else tree_tag_id t.doc t.node
+
+let tagged_desc tag n = { n with node = tree_tagged_desc n.doc n.node tag }
+let select_desc ts n = { n with node = tree_select_desc n.doc n.node (ptset_to_vector ts) }
+let tagged_foll_ctx tag t ctx =
+ { t with node = tree_tagged_foll_below t.doc t.node tag ctx.node }
+let select_foll_ctx ts n ctx = { n with node = tree_select_foll_below n.doc n.node (ptset_to_vector ts) ctx.node }
+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 count_contains t s = Text.count_contains t.doc s
- let count t s = Text.count t.doc s
- let is_left t =
- let u = left (parent t) in
- (id t) == (id u)
+ let count t s = text_count t.doc s
- 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)
+ let print_xml_fast outc t =
+ let rec loop ?(print_right=true) t =
+ if t.node != nil
+ then
+ let tagid = tree_tag_id t.doc t.node in
+ if tagid==Tag.pcdata
+ then output_string outc (text_get_cached_text t.doc t.node);
+ if print_right
+ then loop (next_sibling t)
- | Node (_) ->
- let tg = Tag.to_string (tag t) in
- let l = left t
- and r = right t
+ else
+ let tagstr = Tag.to_string tagid in
+ let l = first_child t
+ and r = next_sibling 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
+ output_string outc tagstr;
+ if l.node == nil then output_string outc "/>"
+ else
+ if (tag l) == Tag.attribute then
+ begin
+ loop_attributes (first_child l);
+ if (next_sibling l).node == nil then output_string outc "/>"
+ else
+ begin
+ output_char outc '>';
+ loop (next_sibling l);
+ output_string outc "</";
+ output_string outc tagstr;
+ output_char outc '>';
+ end;
+ end
+ else
+ begin
+ output_char outc '>';
+ loop l;
+ output_string outc "</";
+ output_string outc tagstr;
+ output_char outc '>';
+ end;
+ if print_right then loop r
+ and loop_attributes a =
+ let s = (Tag.to_string (tag a)) in
+ let attname = String.sub s 3 ((String.length s) -3) in
+ output_char outc ' ';
+ output_string outc attname;
+ output_string outc "=\"";
+ output_string outc (text_get_cached_text t.doc
+ (tree_my_text a.doc (first_child a).node));
+ output_char outc '"';
+ loop_attributes (next_sibling a)
+ in
loop ~print_right:false t
-
-
+
+
let print_xml_fast outc t =
- if Tag.to_string (tag t) = "" then
+ if (tag t) = Tag.document_node then
print_xml_fast outc (first_child t)
- else print_xml_fast outc 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 tags_below t tag =
+ fst(Hashtbl.find t.ttable tag)
+let tags_after t tag =
+ snd(Hashtbl.find t.ttable tag)
- 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 tags t tag = Hashtbl.find t.ttable tag
- 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 rec binary_parent t =
+ if tree_is_first_child t.doc t.node
+ then { t with node = tree_parent t.doc t.node }
+ else { t with node = tree_prev_sibling t.doc t.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 doc_ids (t:t) : (int*int) =
+ (Obj.magic (tree_doc_ids t.doc t.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 subtree_tags t tag =
+ if t.node == nil then 0 else
+ tree_subtree_tags t.doc t.node tag
-
- 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 get_text t =
+ let tid = tree_my_text t.doc t.node in
+ if tid == nil then "" else
+ let a, b = tree_doc_ids t.doc (tree_root t.doc) in
+ let _ = Printf.eprintf "Trying to take text %i of node %i in %i %i\n%!" tid t.node a b in
+ text_get_cached_text t.doc tid
- 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 dump_tree fmt t =
+ let rec loop tree n =
+ if tree != nil then
+ let tag = (tree_tag_id t.doc tree ) in
+ let tagstr = Tag.to_string tag in
+ let tab = String.make n ' ' in
-
- 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
+ if tag == Tag.pcdata || tag == Tag.attribute_data
+ then
+ Format.fprintf fmt "%s<%s>%s</%s>\n"
+ tab tagstr (text_get_cached_text t.doc (tree_my_text t.doc tree)) tagstr
+ else begin
+ Format.fprintf fmt "%s<%s>\n" tab tagstr;
+ loop (tree_first_child t.doc tree) (n+2);
+ Format.fprintf fmt "%s</%s>\n%!" tab tagstr;
+ end;
+ loop (tree_next_sibling t.doc tree) n
+ in
+ loop (tree_root t.doc) 0
+;;
-
-
-end
-
-module Binary = DEBUGTREE
-ELSE
-module Binary = XML.Binary
-END (* IFDEF DEBUG *)