anil.recoil.org
Detect which human language a document uses from OCaml, from the Nu Html validator
languages
unicode
ocaml
fix
+6
-5
lib/langdetect.ml
+6
-5
lib/langdetect.ml
···
146
146
let prob = init_prob n_langs in
147
147
let alpha = t.config.alpha +. (random_gaussian () *. alpha_width) in
148
148
let converged = ref false in
149
-
let i = ref 0 in
150
-
while (not !converged) && !i < iteration_limit do
149
+
let iter_count = ref 0 in
150
+
while (not !converged) && !iter_count < iteration_limit do
151
151
let r = random_int (Array.length ngrams) in
152
152
let (_ : bool) = update_lang_prob prob ngrams.(r) t.word_lang_prob alpha in
153
-
if !i mod 5 = 0 then
153
+
if !iter_count mod 5 = 0 then begin
154
154
let max_p = normalize_prob prob in
155
-
if max_p > t.config.conv_threshold then converged := true;
156
-
incr i
155
+
if max_p > t.config.conv_threshold then converged := true
156
+
end;
157
+
incr iter_count
157
158
done;
158
159
for j = 0 to n_langs - 1 do
159
160
lang_prob.(j) <- lang_prob.(j) +. (prob.(j) /. float_of_int t.config.n_trial)
+36
test/test_langdetect.ml
+36
test/test_langdetect.ml
···
538
538
with exn ->
539
539
Alcotest.fail (Printf.sprintf "Exception testing profiles: %s" (Printexc.to_string exn))
540
540
541
+
(* Regression test: ensure detection loop completes in reasonable time.
542
+
This catches bugs like the iter_count variable name mismatch that caused infinite loops. *)
543
+
let test_detection_completes_quickly () =
544
+
let d = get_detector () in
545
+
let start_time = Unix.gettimeofday () in
546
+
(* Run detection on several texts to ensure it completes *)
547
+
List.iter (fun (_, _, text) ->
548
+
let _ = Langdetect.detect_best d text in ()
549
+
) all_test_corpus;
550
+
let elapsed = Unix.gettimeofday () -. start_time in
551
+
(* All detections should complete within 5 seconds total *)
552
+
if elapsed > 5.0 then
553
+
Alcotest.fail (Printf.sprintf "Detection took too long: %.2f seconds (expected < 5s)" elapsed)
554
+
else
555
+
Printf.printf "Detection completed in %.2f seconds\n" elapsed
556
+
557
+
(* Regression test: verify iteration_limit is respected in detect_block *)
558
+
let test_iteration_limit_respected () =
559
+
let d = get_detector () in
560
+
(* Use a text that might not converge quickly *)
561
+
let mixed_text = String.concat " " (List.init 100 (fun i ->
562
+
if i mod 3 = 0 then "hello"
563
+
else if i mod 3 = 1 then "bonjour"
564
+
else "hallo"
565
+
)) in
566
+
let start_time = Unix.gettimeofday () in
567
+
let _ = Langdetect.detect d mixed_text in
568
+
let elapsed = Unix.gettimeofday () -. start_time in
569
+
(* Single detection should complete within 1 second *)
570
+
if elapsed > 1.0 then
571
+
Alcotest.fail (Printf.sprintf "Single detection took too long: %.2f seconds (expected < 1s)" elapsed)
572
+
541
573
(* Main test suite *)
542
574
let () =
543
575
Alcotest.run "Langdetect" [
···
583
615
Alcotest.test_case "very long text" `Quick test_very_long_text;
584
616
Alcotest.test_case "repeated detection consistency" `Quick test_repeated_detection_consistency;
585
617
Alcotest.test_case "all profiles functional" `Quick test_all_profiles_functional;
618
+
]);
619
+
("Regression tests", [
620
+
Alcotest.test_case "detection completes quickly" `Quick test_detection_completes_quickly;
621
+
Alcotest.test_case "iteration limit respected" `Quick test_iteration_limit_respected;
586
622
]);
587
623
]