Re: [PATCH] fix double ;;s in code

[Michael Ellerman <mpe@ellerman.id.au>]

Daniel Vetter <daniel@ffwll.ch> writes:
> On Sun, Feb 18, 2018 at 11:00:56AM +0100, Christophe LEROY wrote:
>> Le 17/02/2018 =C3=A0 22:19, Pavel Machek a =C3=A9crit=C2=A0:
>> >=20
>> > Fix double ;;'s in code.
>> >=20
>> > Signed-off-by: Pavel Machek <pavel@ucw.cz>
>>=20
>> A summary of the files modified on top of the patch would help understand
>> the impact.
>>=20
>> A maybe there should be one patch by area, eg one for each arch specific
>> modif and one for drivers/ and one for block/ ?
>
> Yeah, pls split this into one patch per area, with a suitable patch
> subject prefix. Look at git log of each file to get a feeling for what's
> the standard in each area.

This part is actually pretty annoying.

I hacked up a script (below) which seems to do a reasonable job in most
cases.

For this patch it gives:

  $ for f in $(git ls-files -m); do ./guess-prefix.py $f; done
  ARC:=20
  ARC:=20
  ARM:=20
  arm64: ptrace:=20
  KVM: PPC:=20
  powerpc/powernv:
  x86/efi:
  block/sed-opal:
  clocksource: mips-gic:=20
  clocksource/drivers/sun5i:
  drm/amd/display:
  drm/amd/powerplay:
  drm/msm/mdp5:
  drm:=20
  iommu/vt-d:
  md:=20
  soc: imx: gpc:=20

I think those are correct except for:
 - "drm:" for "drivers/gpu/drm/scheduler" which has only a single commit.
 - "md:" for "drivers/md/raid1.c" which is tricked by inconsistent
    usage of "md: raid1:" and "md/raid1:".

But that seems like a reasonable hit rate.

Another approach would be to have a file that defines for each subsystem
what the preferred style is, but that is likely to be a PITA to
maintain.

cheers


#!/usr/bin/python3

import sys
import re
from subprocess import check_output
from collections import Counter

if len(sys.argv) !=3D 2:
    print('Usage: %s <path>' % sys.argv[0], file=3Dsys.stderr)
    sys.exit(1)

fname =3D sys.argv[1]

cmd =3D ['git', 'log', '--format=3D%s', '-n', '100', fname]
output =3D check_output(cmd).decode('utf-8')

ignore =3D ['License', 'Merge']

# Ordered list of patterns
patterns =3D [
    # Common style "foo/bar/baz: Fix the foo"
    re.compile('^([\w\-_]+: )+'),
    # Less common "foo bar baz: Fix the foo"
    re.compile('^([\w\-_]+:? )+: ')
]

words =3D []
for line in output.splitlines():
    prefix =3D line.split()[0]
    for patt in patterns:
        match =3D patt.search(line)
        if match:
            prefix =3D match.group(0)
            break

    if prefix in ignore:
        continue

    words.append(prefix)

# Warn if we didn't find many examples
if len(words) < 5:
    print("Warning: only found %d previous commits to guess from for" % len=
(words),
          fname, file=3Dsys.stderr)

counts =3D Counter(words)
print(counts.most_common(1)[0][0])