Installs pre-commit hooks for OCaml projects that run dune fmt automatically
precommit: optimize fix command with commit range and parallelism
- Add find_oldest_ai_commit to locate the first commit needing rewrite
- Use commit range (oldest^..HEAD) instead of rewriting entire history
- Process multiple repos in parallel with Eio.Fiber.all
This dramatically speeds up `precommit fix` for repos where AI
attribution is only in recent commits.
+72
-22
+23
-20
bin/main.ml
+23
-20
bin/main.ml
···
356
356
info "Aborted";
357
357
exit 0
358
358
end;
359
-
let fixed = ref 0 in
360
-
let errors = ref 0 in
361
-
List.iter
362
-
(fun d ->
363
-
let backup = Precommit.backup_branch ~process_mgr ~fs d in
364
-
success "%s: backed up to %s" d backup;
365
-
match Precommit.rewrite_ai_attribution ~process_mgr ~fs d with
366
-
| Ok _ ->
367
-
incr fixed;
368
-
success "%s: attribution removed" d
369
-
| Error msg ->
370
-
incr errors;
371
-
error "%s" msg)
372
-
affected;
359
+
let fixed = Atomic.make 0 in
360
+
let errors = Atomic.make 0 in
361
+
Eio.Fiber.all
362
+
(List.map
363
+
(fun d () ->
364
+
let backup = Precommit.backup_branch ~process_mgr ~fs d in
365
+
success "%s: backed up to %s" d backup;
366
+
match Precommit.rewrite_ai_attribution ~process_mgr ~fs d with
367
+
| Ok _ ->
368
+
Atomic.incr fixed;
369
+
success "%s: attribution removed" d
370
+
| Error msg ->
371
+
Atomic.incr errors;
372
+
error "%s" msg)
373
+
affected);
373
374
Format.pp_print_newline Format.std_formatter ();
374
-
if !errors > 0 then begin
375
-
error "%d repo%s fixed, %d error%s" !fixed
376
-
(if !fixed = 1 then "" else "s")
377
-
!errors
378
-
(if !errors = 1 then "" else "s");
375
+
let n_fixed = Atomic.get fixed in
376
+
let n_errors = Atomic.get errors in
377
+
if n_errors > 0 then begin
378
+
error "%d repo%s fixed, %d error%s" n_fixed
379
+
(if n_fixed = 1 then "" else "s")
380
+
n_errors
381
+
(if n_errors = 1 then "" else "s");
379
382
exit 1
380
383
end
381
-
else success "%d repo%s fixed" !fixed (if !fixed = 1 then "" else "s")
384
+
else success "%d repo%s fixed" n_fixed (if n_fixed = 1 then "" else "s")
382
385
end
383
386
384
387
let yes =
+49
-2
lib/precommit.ml
+49
-2
lib/precommit.ml
···
332
332
in
333
333
backup_name
334
334
335
+
let find_oldest_ai_commit ~process_mgr ~fs dir =
336
+
let grep_args =
337
+
ai_message_patterns
338
+
|> List.map (fun p -> "--grep='" ^ p ^ "'")
339
+
|> String.concat " "
340
+
in
341
+
let msg_cmd =
342
+
Printf.sprintf "git log --format='%%H' --reverse %s 2>/dev/null | head -1"
343
+
grep_args
344
+
in
345
+
let author_cmd =
346
+
ai_author_patterns
347
+
|> List.map (fun a ->
348
+
Printf.sprintf
349
+
"git log --format='%%H' --reverse --author='%s' 2>/dev/null | head -1"
350
+
a)
351
+
|> String.concat "; "
352
+
in
353
+
let msg_oldest =
354
+
match run_in_dir_opt ~process_mgr ~fs dir msg_cmd with
355
+
| Ok (h :: _) when h <> "" -> Some h
356
+
| _ -> None
357
+
in
358
+
let author_oldest =
359
+
match run_in_dir_opt ~process_mgr ~fs dir author_cmd with
360
+
| Ok lines -> List.find_opt (fun h -> h <> "") lines
361
+
| _ -> None
362
+
in
363
+
match (msg_oldest, author_oldest) with
364
+
| Some m, Some a -> (
365
+
(* Find which is older by checking commit order *)
366
+
let cmd =
367
+
Printf.sprintf
368
+
"git merge-base --is-ancestor %s %s && echo %s || echo %s" m a m a
369
+
in
370
+
match run_in_dir_opt ~process_mgr ~fs dir cmd with
371
+
| Ok (h :: _) -> Some h
372
+
| _ -> Some m)
373
+
| Some h, None | None, Some h -> Some h
374
+
| None, None -> None
375
+
335
376
let rewrite_ai_attribution ~process_mgr ~fs dir =
336
377
if not (file_exists ~fs (Filename.concat dir ".git")) then
337
378
Error (Printf.sprintf "%s: No .git directory found" dir)
338
379
else
380
+
(* Find the oldest commit with AI attribution to limit rewrite scope *)
381
+
let commit_range =
382
+
match find_oldest_ai_commit ~process_mgr ~fs dir with
383
+
| Some oldest -> Printf.sprintf "%s^..HEAD" oldest
384
+
| None -> "HEAD"
385
+
in
339
386
(* Build sed command to delete all AI attribution patterns from messages.
340
387
Escape forward slashes in patterns for sed compatibility. *)
341
388
let escape_slashes s =
···
368
415
let cmd =
369
416
Printf.sprintf
370
417
"FILTER_BRANCH_SQUELCH_WARNING=1 git filter-branch -f --env-filter \
371
-
'%s' --msg-filter \"sed %s\" -- HEAD 2>&1"
372
-
env_filter sed_args
418
+
'%s' --msg-filter \"sed %s\" -- %s 2>&1"
419
+
env_filter sed_args commit_range
373
420
in
374
421
match run_in_dir_opt ~process_mgr ~fs dir cmd with
375
422
| Error e -> Error (Printf.sprintf "%s: %s" dir e)