course: memoized version of `get_unicode'

To avoid creating buffers to decode an HTML entity, this create a
memoized version of `get_unicode' in hope to operate a bit faster and
reduce memory allocations.

Signed-off-by: Alban Gruin <alban at pa1ch dot fr>

1 files changed, 13 insertions, 1 deletions

diff --git a/src/course.ml b/src/course.ml
index 4d02df3..461c99c 100644
--- a/src/course.ml
+++ b/src/course.ml
@@ -19,18 +19,30 @@ open CalendarLib
 
 module J = Json_encoding
 
+let memoize f =
+  let hashtbl = Hashtbl.create ~random:true 100 in
+  fun v ->
+  match Hashtbl.find_opt hashtbl v with
+  | Some r -> r
+  | None ->
+     let r = f v in
+     Hashtbl.add hashtbl v r;
+     r
+
 let get_unicode v =
   let b = Buffer.create 1 in
   Buffer.add_utf_8_uchar b (Uchar.of_int v);
   Buffer.contents b
 
+let memoized_get_unicode = memoize get_unicode
+
 let html_entities_regex = Re.Perl.compile_pat "&#(\\d+);"
 
 let replace_entities str =
   Re.Pcre.full_split ~rex:html_entities_regex str
   |> List.filter_map (function
          | Re.Pcre.Group (_, v) ->
-            Some ("&#" ^ v ^ ";", get_unicode @@ int_of_string v)
+            Some ("&#" ^ v ^ ";", memoized_get_unicode @@ int_of_string v)
          | _ -> None)
   |> Stringext.replace_all_assoc str