Installs pre-commit hooks for OCaml projects that run dune fmt automatically
precommit: expand AI attribution detection and rewriting
- Fix sed quoting bug (use double quotes for msg-filter, single for sed pattern)
- Add detection and rewriting of commits authored by Claude/Claude Code
- Rewrite AI authors to current git user via --env-filter
+109
-24
+109
-24
lib/precommit.ml
+109
-24
lib/precommit.ml
···
243
243
244
244
type ai_commit = { hash : string; subject : string }
245
245
246
+
(* Patterns that indicate AI-generated content in commit messages *)
247
+
let ai_message_patterns =
248
+
[
249
+
"Co-Authored-By.*[Cc]laude";
250
+
"https://claude\\.ai/";
251
+
"Generated with.*[Cc]laude";
252
+
]
253
+
254
+
(* Author names that indicate AI-authored commits *)
255
+
let ai_author_patterns =
256
+
[ "Claude"; "Claude Code"; "Claude Opus"; "Claude Sonnet" ]
257
+
258
+
let parse_commit_line line =
259
+
if String.length line > 8 then
260
+
let hash = String.sub line 0 7 in
261
+
let subject = String.sub line 8 (String.length line - 8) in
262
+
Some { hash; subject }
263
+
else None
264
+
246
265
let check_ai_attribution ~process_mgr ~fs dir =
247
266
if not (file_exists ~fs (Filename.concat dir ".git")) then []
248
267
else
249
-
(* Get commits by the configured user that contain AI attribution patterns.
250
-
Use run_in_dir_opt to handle repos with no commits gracefully. *)
251
-
let cmd =
252
-
"git log --format='%h %s' --grep='Co-Authored-By.*[Cc]laude' \
253
-
--author=\"$(git config user.name)\" 2>/dev/null"
268
+
(* Find commits with AI attribution in message body *)
269
+
let grep_args =
270
+
ai_message_patterns
271
+
|> List.map (fun p -> "--grep='" ^ p ^ "'")
272
+
|> String.concat " "
273
+
in
274
+
let msg_cmd =
275
+
Printf.sprintf
276
+
"git log --format='%%h %%s' %s --author=\"$(git config user.name)\" \
277
+
2>/dev/null"
278
+
grep_args
279
+
in
280
+
let msg_commits =
281
+
match run_in_dir_opt ~process_mgr ~fs dir msg_cmd with
282
+
| Error _ -> []
283
+
| Ok lines -> List.filter_map parse_commit_line lines
284
+
in
285
+
(* Find commits authored by AI *)
286
+
let author_commits =
287
+
ai_author_patterns
288
+
|> List.concat_map (fun author ->
289
+
let cmd =
290
+
Printf.sprintf
291
+
"git log --format='%%h %%s' --author='%s' 2>/dev/null" author
292
+
in
293
+
match run_in_dir_opt ~process_mgr ~fs dir cmd with
294
+
| Error _ -> []
295
+
| Ok lines -> List.filter_map parse_commit_line lines)
254
296
in
255
-
match run_in_dir_opt ~process_mgr ~fs dir cmd with
256
-
| Error _ -> [] (* No commits or command failed *)
257
-
| Ok lines ->
258
-
List.filter_map
259
-
(fun line ->
260
-
if String.length line > 8 then
261
-
let hash = String.sub line 0 7 in
262
-
let subject = String.sub line 8 (String.length line - 8) in
263
-
Some { hash; subject }
264
-
else None)
265
-
lines
297
+
(* Combine and deduplicate by hash *)
298
+
let all = msg_commits @ author_commits in
299
+
let seen = Hashtbl.create 16 in
300
+
List.filter
301
+
(fun c ->
302
+
if Hashtbl.mem seen c.hash then false
303
+
else (
304
+
Hashtbl.add seen c.hash ();
305
+
true))
306
+
all
266
307
267
308
let current_branch ~process_mgr ~fs dir =
268
309
match
···
295
336
if not (file_exists ~fs (Filename.concat dir ".git")) then
296
337
Error (Printf.sprintf "%s: No .git directory found" dir)
297
338
else
339
+
(* Build sed command to delete all AI attribution patterns from messages *)
340
+
let sed_args =
341
+
ai_message_patterns
342
+
|> List.map (fun p -> "-e '/" ^ p ^ "/d'")
343
+
|> String.concat " "
344
+
in
345
+
(* Build env-filter to replace AI authors with current user *)
346
+
let author_conditions =
347
+
ai_author_patterns
348
+
|> List.map (fun a -> Printf.sprintf "\"$GIT_AUTHOR_NAME\" = '%s'" a)
349
+
|> String.concat " -o "
350
+
in
351
+
let env_filter =
352
+
Printf.sprintf
353
+
"if [ %s ]; then export GIT_AUTHOR_NAME=\"$(git config user.name)\"; \
354
+
export GIT_AUTHOR_EMAIL=\"$(git config user.email)\"; export \
355
+
GIT_COMMITTER_NAME=\"$(git config user.name)\"; export \
356
+
GIT_COMMITTER_EMAIL=\"$(git config user.email)\"; fi"
357
+
author_conditions
358
+
in
298
359
let cmd =
299
-
"FILTER_BRANCH_SQUELCH_WARNING=1 git filter-branch -f --msg-filter \"sed \
300
-
'/[Cc]o-[Aa]uthored-[Bb]y:.*[Cc]laude/d'\" -- HEAD 2>&1"
360
+
Printf.sprintf
361
+
"FILTER_BRANCH_SQUELCH_WARNING=1 git filter-branch -f --env-filter \
362
+
'%s' --msg-filter \"sed %s\" -- HEAD 2>&1"
363
+
env_filter sed_args
301
364
in
302
365
match run_in_dir_opt ~process_mgr ~fs dir cmd with
303
366
| Error e -> Error (Printf.sprintf "%s: %s" dir e)
304
367
| Ok _lines ->
305
-
(* Count how many commits were actually rewritten *)
306
-
let count_cmd =
307
-
"git log --format='%H' HEAD --grep='Co-Authored-By.*[Cc]laude' \
308
-
2>/dev/null | wc -l"
368
+
(* Count commits still with AI attribution in messages *)
369
+
let grep_args =
370
+
ai_message_patterns
371
+
|> List.map (fun p -> "--grep='" ^ p ^ "'")
372
+
|> String.concat " "
373
+
in
374
+
let msg_count_cmd =
375
+
Printf.sprintf "git log --format='%%H' HEAD %s 2>/dev/null | wc -l"
376
+
grep_args
309
377
in
310
-
let remaining =
311
-
match run_in_dir_opt ~process_mgr ~fs dir count_cmd with
378
+
let msg_remaining =
379
+
match run_in_dir_opt ~process_mgr ~fs dir msg_count_cmd with
312
380
| Ok (n :: _) -> ( try int_of_string (String.trim n) with _ -> 0)
313
381
| _ -> 0
314
382
in
383
+
(* Count commits still authored by AI *)
384
+
let author_remaining =
385
+
ai_author_patterns
386
+
|> List.fold_left
387
+
(fun acc author ->
388
+
let cmd =
389
+
Printf.sprintf
390
+
"git log --format='%%H' --author='%s' 2>/dev/null | wc -l"
391
+
author
392
+
in
393
+
match run_in_dir_opt ~process_mgr ~fs dir cmd with
394
+
| Ok (n :: _) -> (
395
+
try acc + int_of_string (String.trim n) with _ -> acc)
396
+
| _ -> acc)
397
+
0
398
+
in
399
+
let remaining = msg_remaining + author_remaining in
315
400
(* Clean up refs/original *)
316
401
let _ =
317
402
run_in_dir_opt ~process_mgr ~fs dir