INCLUDE "debug.ml"
INCLUDE "utils.ml"
-
external init_lib : unit -> unit = "sxsi_cpp_init"
exception CPlusPlusError of string
type tree
-
external register_tag : tree -> string -> Tag.t = "caml_xml_tree_register_tag"
external tag_name : tree -> Tag.t -> string = "caml_xml_tree_get_tag_name"
let to_string d = ();
fun t ->
- if t == Tag.pcdata then "<$>"
- else if t == Tag.attribute_data then "<@$>"
- else if t == Tag.attribute then "<@>"
- else if t == Tag.nullt then "<!NIL!>"
- else tag_name d t
+ if t == Tag.pcdata then "<$>"
+ else if t == Tag.attribute_data then "<@$>"
+ else if t == Tag.attribute then "<@>"
+ else if t == Tag.nullt then "<!NIL!>"
+ else tag_name d t
let translate x = x
external create : unit -> t = "caml_xml_tree_builder_create"
external open_document : t -> bool -> int -> bool -> int -> unit = "caml_xml_tree_builder_open_document"
external close_document : t -> tree = "caml_xml_tree_builder_close_document"
- external open_tag : t -> string -> unit = "caml_xml_tree_builder_new_open_tag"
- external close_tag : t -> string -> unit = "caml_xml_tree_builder_new_closing_tag"
- external text : t -> string -> unit = "caml_xml_tree_builder_new_text"
+ external open_tag : t -> string -> unit = "caml_xml_tree_builder_open_tag"
+ external close_tag : t -> string -> unit = "caml_xml_tree_builder_close_tag"
+ external text : t -> string -> unit = "caml_xml_tree_builder_text"
+
+ let is_whitespace s =
+ let rec loop len i =
+ if i < len then
+ let c = s.[i] in
+ (c == '\n' || c == '\t' || c == ' ') && loop len (i+1)
+ else
+ true
+ in
+ loop (String.length s) 0
+
+
+ let display_count =
+ let event_counter = ref 0 in
+ (fun parser_ ->
+ incr event_counter;
+ if !event_counter land 0xffffff == 0 then
+ Logger.print Format.err_formatter "Current position: %i@\n@?" (Expat.get_current_byte_index parser_))
let do_text b t =
if Buffer.length t > 0 then begin
- open_tag b "<$>";
- text b (Buffer.contents t);
- close_tag b "<$>";
+ let s = Buffer.contents t in
+ begin
+ open_tag b "<$>";
+ text b s;
+ close_tag b "<$>";
+ end;
Buffer.clear t
end
let atname = "<@>" ^ name in
open_tag b atname;
open_tag b "<@$>";
- text b value;
- close_tag b "<@$>";
- close_tag b atname
+ text b value;
+ close_tag b "<@$>";
+ close_tag b atname
- let start_element_handler b t tag attr_list =
+ let start_element_handler parser_ b t tag attr_list =
+ display_count parser_;
do_text b t;
open_tag b tag;
match attr_list with
| l ->
open_tag b "<@>";
List.iter (fun (name, value) -> output_attr b name value) l;
- close_tag b "<@>"
+ close_tag b "<@>"
- let end_element_handler b t tag =
+ let end_element_handler parser_ b t tag =
+ display_count parser_;
do_text b t;
close_tag b tag
- let character_data_handler _b t text =
+ let character_data_handler parser_ _ t text =
+ display_count parser_;
Buffer.add_string t text
let create_parser () =
let finalize () =
do_text build buf;
close_tag build "";
- close_document build
+ Logger.print Format.err_formatter "Finished parsing@\n";
+ Logger.print Format.err_formatter "Starting index construction@\n";
+ let r = close_document build in
+ Logger.print Format.err_formatter "Finished index construction@\n";
+ r
in
- Expat.set_start_element_handler parser_ (start_element_handler build buf);
- Expat.set_end_element_handler parser_ (end_element_handler build buf);
- Expat.set_character_data_handler parser_ (character_data_handler build buf);
+ Expat.set_start_element_handler parser_ (start_element_handler parser_ build buf);
+ Expat.set_end_element_handler parser_ (end_element_handler parser_ build buf);
+ Expat.set_character_data_handler parser_ (character_data_handler parser_ build buf);
+ Logger.print Format.err_formatter "Started parsing@\n";
open_document build !Options.index_empty_texts !Options.sample_factor
!Options.disable_text_collection !Options.text_index_type;
open_tag build "";
in
finalizer ()
-
-
end
type bit_vector = string
external bool_of_int : int -> bool = "%identity"
+external int_of_bool : bool -> int = "%identity"
let bit_vector_unsafe_get v i =
bool_of_int
(((Char.code (String.unsafe_get v (i lsr 3))) lsr (i land 7)) land 1)
+let chr (c:int) : char = Obj.magic (c land 0xff)
+let bit_vector_unsafe_set v i b =
+ let j = i lsr 3 in
+ let c = Char.code v.[j] in
+ let bit = int_of_bool b in
+ let mask = bit lsl (i land 7) in
+ if b then v.[j] <- chr (c lor mask) else v.[j] <- (chr (c land (lnot mask)))
+
+let bit_vector_create n =
+ let len = if n <= 0 then 0 else (n - 1) / 8 + 1 in
+ String.make len '\000'
type t = {
doc : tree;
let last_child t n = tree_last_child t.doc n
-
external tree_next_sibling : tree -> [`Tree] Node.t -> [`Tree] Node.t = "caml_xml_tree_next_sibling" "noalloc"
let next_sibling t n = tree_next_sibling t.doc n
external tree_next_element : tree -> [`Tree] Node.t -> [`Tree] Node.t = "caml_xml_tree_next_element" "noalloc"
let next_element t n = tree_next_element t.doc n
-external tree_next_node_before : tree -> [`Tree] Node.t -> [`Tree] Node.t -> [`Tree] Node.t = "caml_xml_tree_next_node_before" "noalloc"
-let next_node_before t n ctx = tree_next_node_before t.doc n ctx
+external tree_tagged_sibling : tree -> [`Tree] Node.t -> Tag.t -> [`Tree] Node.t = "caml_xml_tree_tagged_sibling" "noalloc"
+let tagged_sibling t n tag = tree_tagged_sibling t.doc n tag
-external tree_tagged_following_sibling : tree -> [`Tree] Node.t -> Tag.t -> [`Tree] Node.t = "caml_xml_tree_tagged_following_sibling" "noalloc"
-let tagged_following_sibling t n tag = tree_tagged_following_sibling t.doc n tag
-
-external tree_select_following_sibling : tree -> [`Tree ] Node.t -> unordered_set -> [`Tree] Node.t = "caml_xml_tree_select_following_sibling" "noalloc"
-let select_following_sibling t n tag_set = tree_select_following_sibling t.doc n tag_set
+external tree_select_sibling : tree -> [`Tree ] Node.t -> unordered_set -> [`Tree] Node.t = "caml_xml_tree_select_sibling" "noalloc"
+let select_sibling t n tag_set = tree_select_sibling t.doc n tag_set
external tree_prev_sibling : tree -> [`Tree] Node.t -> [`Tree] Node.t = "caml_xml_tree_prev_sibling" "noalloc"
let prev_sibling t n = tree_prev_sibling t.doc n
external tree_parent : tree -> [`Tree] Node.t -> [`Tree] Node.t = "caml_xml_tree_parent" "noalloc"
let parent t n = tree_parent t.doc n
-external tree_binary_parent : tree -> [`Tree] Node.t -> [`Tree] Node.t = "caml_xml_tree_binary_parent"
- "noalloc"
-let binary_parent t n = tree_binary_parent t.doc n
-
-
external tree_tag : tree -> [`Tree] Node.t -> Tag.t = "caml_xml_tree_tag" "noalloc"
let tag t n = tree_tag t.doc n
open Format
let dump_tag_table t =
- eprintf "Child tags:\n%!";
- Array.iteri
- (fun tag set -> eprintf "%s: %a\n%!"
- (Tag.to_string tag) TagSet.print (TagSet.inj_positive set))
- t.children;
- eprintf "-----------------------------\n%!";
- eprintf "Descendant tags:\n%!";
- Array.iteri
- (fun tag set -> eprintf "%s: %a\n%!"
- (Tag.to_string tag) TagSet.print (TagSet.inj_positive set))
- t.descendants;
- eprintf "-----------------------------\n%!";
- eprintf "Sibling tags:\n%!";
- Array.iteri
- (fun tag set -> eprintf "%s: %a\n%!"
- (Tag.to_string tag) TagSet.print (TagSet.inj_positive set))
- t.siblings;
- eprintf "-----------------------------\n%!";
- eprintf "Following tags:\n%!";
- Array.iteri
- (fun tag set -> eprintf "%s: %a\n%!"
- (Tag.to_string tag) TagSet.print (TagSet.inj_positive set))
- t.followings;
- eprintf "-----------------------------\n%!"
-
+ let tag = ref 0 in
+ let printer ppf set =
+ Logger.print ppf "%s: %a"
+ (Tag.to_string !tag) TagSet.print (TagSet.inj_positive set);
+ incr tag
+ in
+ let set_printer msg set =
+ tag := 0;
+ Logger.print err_formatter "%s :@\n" msg;
+ Pretty.pp_print_array ~sep:pp_force_newline printer err_formatter set;
+ Logger.print err_formatter "-----------------------------@\n";
+ in
+ set_printer "Child tags" t.children;
+ set_printer "Descendant tags" t.descendants;
+ set_printer "Sibling tags" t.siblings;
+ set_printer "Following tags" t.followings
external tree_subtree_tags : tree -> [`Tree] Node.t -> Tag.t -> int = "caml_xml_tree_subtree_tags" "noalloc"
let subtree_tags t n tag = tree_subtree_tags t.doc n tag
in
let a,b = loop true root 0 0 0
in
- Printf.eprintf "Average depth: %f, number of leaves %i\n%!" ((float_of_int a)/. (float_of_int b)) b
+ Logger.print err_formatter "Average depth: %f, number of leaves %i@\n@?" ((float_of_int a)/. (float_of_int b)) b
;;
module TagS =
let is_root t = t == root
let node_of_t t =
+ Logger.print err_formatter "Initializing tag structure@\n";
let _ = Tag.init (mk_tag_ops t) in
+ Logger.print err_formatter "Starting tag table construction@\n";
let f, n, c, d = time collect_labels t ~msg:"Building tag relationship table" in
let c = Array.map TagS.to_ptset c in
let n = Array.map TagS.to_ptset n in
let pos fd =
Unix.lseek fd 0 Unix.SEEK_CUR
-let pr_pos fd = Printf.eprintf "At position %i\n%!" (pos fd)
+let pr_pos fd = Logger.print err_formatter "At position %i@\n" (pos fd)
let write fd s =
let sl = String.length s in
let save t str =
let fd = Unix.openfile str [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 in
let out_c = Unix.out_channel_of_descr fd in
+ let index_prefix = Filename.chop_suffix str ".srx" in
let _ = set_binary_mode_out out_c true in
output_string out_c magic_string;
output_char out_c '\n';
(* we need to move the fd to the correct position *)
flush out_c;
ignore (Unix.lseek fd (pos_out out_c) Unix.SEEK_SET);
- tree_save t.doc fd str;
+ tree_save t.doc fd index_prefix;
close_out out_c
;;
let load_tag_table channel =
let load ?(sample=64) ?(load_text=true) str =
let fd = Unix.openfile str [ Unix.O_RDONLY ] 0o644 in
let in_c = Unix.in_channel_of_descr fd in
+ let index_prefix = Filename.chop_suffix str ".srx" in
let _ = set_binary_mode_in in_c true in
let load_table () =
(let ms = input_line in_c in if ms <> magic_string then failwith "Invalid index file");
in
let c, s, d, f = time ~msg:"Loading tag table"(load_table) () in
ignore(Unix.lseek fd (pos_in in_c) Unix.SEEK_SET);
- let xml_tree = tree_load fd str load_text sample in
+ let xml_tree = tree_load fd index_prefix load_text sample in
let () = Tag.init (Obj.magic xml_tree) in
let attributes = Ptset.Int.add Tag.attribute d.(Tag.attribute) in
let elements = Ptset.Int.add Tag.document_node
external tree_flush : tree -> Unix.file_descr -> unit = "caml_xml_tree_flush"
let flush t fd = tree_flush t.doc fd
-external text_prefix : tree -> string -> query_result = "caml_text_collection_prefix_bv"
-let text_prefix t s = text_prefix t.doc s
+external text_prefix : tree -> string -> bool -> query_result = "caml_text_collection_prefix_bv"
+let text_prefix t s b = text_prefix t.doc s b
-external text_suffix : tree -> string -> query_result = "caml_text_collection_suffix_bv"
-let text_suffix t s = text_suffix t.doc s
+external text_suffix : tree -> string -> bool -> query_result = "caml_text_collection_suffix_bv"
+let text_suffix t s b = text_suffix t.doc s b
-external text_equals : tree -> string -> query_result = "caml_text_collection_equals_bv"
-let text_equals t s = text_equals t.doc s
+external text_equals : tree -> string -> bool -> query_result = "caml_text_collection_equals_bv"
+let text_equals t s b = text_equals t.doc s b
-external text_contains : tree -> string -> query_result = "caml_text_collection_contains_bv"
-let text_contains t s = text_contains t.doc s
+external text_contains : tree -> string -> bool -> query_result = "caml_text_collection_contains_bv"
+let text_contains t s b = text_contains t.doc s b
module Predicate = Hcons.Make (
let _pred_cache = Hashtbl.create 17
;;
let mk_pred query s =
- let f = query_fun query in
+ let f = query_fun query in
let memo = ref (fun _ _ -> failwith "Undefined") in
memo := begin fun tree node ->
let results =
Not_found ->
time ~count:1 ~msg:(Printf.sprintf "Computing text query %s(%s)"
(string_of_query query) s)
- (f tree) s
+ (f tree) s true
in
let bv = results.bv in
memo := begin fun _ n ->
- let b =
- bit_vector_unsafe_get bv (Node.to_int n)
- in
- D_TRACE_(Printf.eprintf "Result of memoized call to query %s is %b for node %i\n" s b (Node.to_int n));
- b
+ bit_vector_unsafe_get bv (Node.to_int n)
end;
- let b = bit_vector_unsafe_get bv (Node.to_int node) in
- D_TRACE_(Printf.eprintf "Result is %b for node %i\n" b (Node.to_int node));
- b
+ bit_vector_unsafe_get bv (Node.to_int node)
end;
Predicate.make memo
-let full_text_prefix t s = (text_prefix t s).pos
+let full_text_prefix t s = (text_prefix t s true).pos
-let full_text_suffix t s = (text_suffix t s).pos
+let full_text_suffix t s = (text_suffix t s true).pos
-let full_text_equals t s = (text_equals t s).pos
+let full_text_equals t s = (text_equals t s true).pos
-let full_text_contains t s = (text_contains t s).pos
+let full_text_contains t s = (text_contains t s true).pos
let full_text_query q t s =
- let res = (query_fun q) t s in
+ let res = (query_fun q) t s true in
Hashtbl.replace _pred_cache (q,s) res;
res.pos
projects / SXSI / xpathcomp.git / blobdiff