speedup range/position of loc

language-server/dm/document.ml

@@ -742,7 +742,6 @@ let validate_document ({ parsed_loc; raw_doc; cancel_handle } as document) =
 let handle_invalidate {parsed; errors; parsed_comments; stop; top_id; started; previous_document} document =
   let end_ = Unix.gettimeofday ()in
   let time = end_ -. started in
-  (* log (fun () -> Format.sprintf "Parsing phase ended in %5.3f" time); *)
   log (fun () -> Format.sprintf "Parsing phase ended in %5.3f\n%!" time);
   let new_sentences = List.rev parsed in
   let new_comments = List.rev parsed_comments in

language-server/dm/rawDocument.ml

@@ -31,11 +31,11 @@ let text t = t.text
 
 let line_text raw i =
   if i + 1 < Array.length raw.lines then
-    String.sub raw.text (raw.lines.(i)) (raw.lines.(i+1) - raw.lines.(i))
+   (raw.lines.(i), raw.lines.(i+1) - raw.lines.(i))
   else
-    String.sub raw.text (raw.lines.(i)) (String.length raw.text - raw.lines.(i))
+   (raw.lines.(i), String.length raw.text - raw.lines.(i))
 
-let get_character_pos linestr loc =
+let get_character_pos raw (i,e) loc =
   let rec loop d =
     if Uutf.decoder_byte_count d >= loc then
       Uutf.decoder_count d
@@ -44,20 +44,48 @@ let get_character_pos linestr loc =
       | `Uchar _ -> loop d
       | `Malformed _ -> loop d
       | `End -> Uutf.decoder_count d
-      | `Await -> assert false
+      | `Await -> Uutf.Manual.src d (Bytes.unsafe_of_string "") 0 0; loop d
   in
   let nln = `Readline (Uchar.of_int 0x000A) in
   let encoding = `UTF_8 in
-  loop (Uutf.decoder ~nln ~encoding (`String linestr))
+  let d = Uutf.decoder ~nln ~encoding `Manual in
+  (* Printf.eprintf "lookup %d|%s\n" loc (String.sub raw.text i e); *)
+  Uutf.Manual.src d (Bytes.unsafe_of_string raw.text) i e;
+  loop d
 
-let position_of_loc raw loc =
+let rec position_of_loc_bisect raw loc i m n len =
+  (* Printf.eprintf "%i | %i(%d) <= %i(%d) < %i(%d)\n" loc m raw.lines.(m) i raw.lines.(i) n raw.lines.(n); *)
+  if i = len || i = n || raw.lines.(i) <= loc && loc < raw.lines.(i+1) then i
+  else if loc < raw.lines.(i) then position_of_loc_bisect raw loc (m + (max 1 ((i - m) / 2))) m i len
+  else position_of_loc_bisect raw loc (i + (max 1 ((n - i) / 2))) i n len
+
+
+let position_of_loc_bisect raw loc =
+  let nlines = Array.length raw.lines in
+  let line = if loc = 0 then 0 else position_of_loc_bisect raw loc (nlines/2) 0 (nlines-1) (nlines-1) in
+  (* Printf.eprintf "line bisect: %d\n" line; *)
+  let char = get_character_pos raw (line_text raw line) (loc - raw.lines.(line)) in
+  Position.{ line = line; character = char }
+
+let position_of_loc = position_of_loc_bisect
+
+(* let position_of_loc raw loc =
   let i = ref 0 in
   while (!i < Array.length raw.lines && raw.lines.(!i) <= loc) do incr(i) done;
   let line = !i - 1 in
-  let char = get_character_pos (line_text raw line) (loc - raw.lines.(line)) in
-  Position.{ line = line; character = char }
+  (* Printf.eprintf "line linear: %d\n" line; *)
+  let char = get_character_pos raw (line_text raw line) (loc - raw.lines.(line)) in
+   Position.{ line = line; character = char }
 
-let get_character_loc linestr pos =
+let position_of_loc r l =
+  assert(l>=0);
+  let p1 = position_of_loc_bisect r l in
+  let p2 = position_of_loc r l in
+  (* Printf.eprintf "%d=%d, %d=%d\n" p1.Position.character p2.Position.character p1.Position.line p2.Position.line; *)
+  assert(p1=p2);
+  p1
+ *)
+let get_character_loc raw (i,e) pos =
   let rec loop d =
     if Uutf.decoder_count d >= pos then
       Uutf.decoder_byte_count d
@@ -66,24 +94,26 @@ let get_character_loc linestr pos =
       | `Uchar _ -> loop d
       | `Malformed _ -> loop d
       | `End -> Uutf.decoder_byte_count d
-      | `Await -> assert false
+      | `Await -> Uutf.Manual.src d (Bytes.unsafe_of_string "") 0 0; loop d
   in
   let nln = `Readline (Uchar.of_int 0x000A) in
   let encoding = `UTF_8 in
-  loop (Uutf.decoder ~nln ~encoding (`String linestr))
+  let d = Uutf.decoder ~nln ~encoding `Manual in
+  Uutf.Manual.src d (Bytes.unsafe_of_string raw.text) i e;
+  loop d
 
 let loc_of_position raw Position.{ line; character } =
   let linestr = line_text raw line in
-  let charloc = get_character_loc linestr character in
+  let charloc = get_character_loc raw linestr character in
   raw.lines.(line) + charloc
 
 let end_loc raw =
   String.length raw.text
 
 let range_of_loc raw loc =
   let open Range in
-  { start = position_of_loc raw loc.Loc.bp;
-    end_ = position_of_loc raw loc.Loc.ep;
+  { start = position_of_loc_bisect raw loc.Loc.bp;
+    end_  = position_of_loc_bisect raw loc.Loc.ep;
   }
 
 let word_at_position raw pos : string option =

language-server/tests/d_perf.ml

@@ -12,7 +12,7 @@ let edit_text doc ~start ~stop text =
   let range = Lsp.Types.Range.{ start; end_ } in
   Document.apply_text_edits doc [range, text]
 
-let size = 3_000  let adjust = 5 + size
+let size = 30_000  let adjust = 5 + size
 
 let validate_document parsed_document =
   let doc, events = Document.validate_document parsed_document in
@@ -27,16 +27,16 @@ let checkpoint () =
 let%test_unit "diff: huge doc" =
   let t0 = checkpoint () in
   let text = Stdlib.String.concat ".\n" (Stdlib.List.init size (fun _ -> "Definition x := 3")) in
-  let edit = Random.int (String.length text) in
+  let edit = String.length text / 2 in
   let Document.{parsed_document} = init_and_parse_test_doc ~steps:(size+adjust) ~text () in
   let t1 = checkpoint () in
   let parsed_document = edit_text parsed_document ~start:edit ~stop:edit "x" in
-  let _ = validate_document parsed_document in
+  let { Document.parsed_document } = validate_document parsed_document in
   let t2 = checkpoint () in
-  let parsed_document = edit_text parsed_document ~start:edit ~stop:edit "" in
-  let _ = validate_document parsed_document in
+  let o = Document.outline parsed_document in
   let t3 = checkpoint () in
+  [%test_pred: int] (fun x -> x + 3 >= size && x <=size) (List.length o); (* the error *)
   Stdlib.Printf.eprintf "full parse: %5.3f\n" (t1 -. t0);
-  Stdlib.Printf.eprintf "edit: %5.3f\n" (t2 -. t1);
-  Stdlib.Printf.eprintf "edit: %5.3f\n" (t3 -. t2);
+  Stdlib.Printf.eprintf "edit: %5.3f \n" (t2 -. t1);
+  Stdlib.Printf.eprintf "outline: %5.3f\n" (t3 -. t2);
   ()

language-server/tests/rd_tests.ml

@@ -0,0 +1,37 @@
+open Base
+
+type raw = { lines : int array }
+
+let rec position_of_loc_bisect raw loc i m n len =
+  (* Printf.eprintf "%i | %i(%d) <= %i(%d) < %i(%d)\n" loc m raw.lines.(m) i raw.lines.(i) n raw.lines.(n); *)
+  if i = len || i = n || raw.lines.(i) <= loc && loc < raw.lines.(i+1) then i
+  else if loc < raw.lines.(i) then position_of_loc_bisect raw loc (m + (max 1 ((i - m) / 2))) m i len
+  else position_of_loc_bisect raw loc (i + (max 1 ((n - i) / 2))) i n len
+
+let go raw loc =
+  let nlines = Array.length raw.lines in
+  let line = if loc = 0 then 0 else position_of_loc_bisect raw loc (nlines/2) 0 (nlines-1) (nlines-1) in
+  line
+
+let raw = { lines = Array.init 5 ~f:(fun i -> i * 5) }
+
+let%test_unit "bisect hole" =
+  [%test_eq: int] (go raw 0) 0;
+  let last = ref 0 in
+  for i = 0 to 41 do
+    let p = go raw i in
+    [%test_pred: int] ((<=) !last) p;
+    last := p;
+  done
+
+let raw = { lines = Array.init 5 ~f:(fun i -> i) }
+
+let%test_unit "bisect hole" =
+  [%test_eq: int] (go raw 0) 0;
+  let last = ref 0 in
+  for i = 0 to 41 do
+    let p = go raw i in
+    [%test_pred: int] ((<=) !last) p;
+    last := p;
+  done
+