checkpatch: Emit a warning on embedded filenames
diff mbox series

Message ID 1fd5f9188a14acdca703ca00301ee323de672a8d.camel@perches.com
State In Next
Commit 7d7f8a50ab353d4099deb01ea7eaa182b0060a9f
Headers show
Series
  • checkpatch: Emit a warning on embedded filenames
Related show

Commit Message

Joe Perches Oct. 1, 2020, 6:28 p.m. UTC
Embedding the complete filename path inside the file
isn't particularly useful as often the path is moved
around and becomes incorrect.

Emit a warning when the source contains the filename.

Signed-off-by: Joe Perches <joe@perches.com>
---
 scripts/checkpatch.pl | 6 ++++++
 1 file changed, 6 insertions(+)

Comments

Joe Perches Oct. 1, 2020, 6:47 p.m. UTC | #1
It's rather unnecessary for files to contain their
path/filename in source code comments.

Here's a trivial little script that can remove
embedded filenames in c90 style comments from files.

This requires git.

It does the following types of removals:

remove individual lines like /* filename */ completely
remove filename from /* filename -- comment */, leave /* comment */
remove filename and any trailing ' *\n' from /* filename, leave /*
remove filename from /* filename, leave /*
remove filename from continuation ' * filename -- comment' leave ' * comment'
remove filename and any trailing ' *\n' from continuation ' * filename\n *\n'

It seems to work well enough.

It does not handle c99 comments.
No // filename variants are removed.

Running it on today's -next gives:

$ perl remove_embedded_filenames.pl
$ git diff --shortstat
 2310 files changed, 354 insertions(+), 4239 deletions(-)

It's also possible to give any filename or path
as an argument to the script

For instance:

$ perl remove_embedded_filenames.pl drivers/net
Bhaskar Chowdhury Oct. 2, 2020, 2:49 p.m. UTC | #2
On 11:47 Thu 01 Oct 2020, Joe Perches wrote:
>It's rather unnecessary for files to contain their
>path/filename in source code comments.
>
>Here's a trivial little script that can remove
>embedded filenames in c90 style comments from files.
>
>This requires git.
>
>It does the following types of removals:
>
>remove individual lines like /* filename */ completely
>remove filename from /* filename -- comment */, leave /* comment */
>remove filename and any trailing ' *\n' from /* filename, leave /*
>remove filename from /* filename, leave /*
>remove filename from continuation ' * filename -- comment' leave ' * comment'
>remove filename and any trailing ' *\n' from continuation ' * filename\n *\n'
>
>It seems to work well enough.
>
>It does not handle c99 comments.
>No // filename variants are removed.
>
>Running it on today's -next gives:
>
>$ perl remove_embedded_filenames.pl
>$ git diff --shortstat
> 2310 files changed, 354 insertions(+), 4239 deletions(-)
>
>It's also possible to give any filename or path
>as an argument to the script
>
>For instance:
>
>$ perl remove_embedded_filenames.pl drivers/net
>

>#!/usr/bin/perl -w
>
># script to remove * <filename> comments;
># use: perl remove_embedded_filenames.pl <paths|files>
># e.g.: perl remove_embedded_filenames.pl drivers/net/ethernet/intel
>
>use strict;
>
>my $P = $0;
>my $modified = 0;
>my $quiet = 0;
>
>sub expand_tabs {
>    my ($str) = @_;
>
>    my $res = '';
>    my $n = 0;
>    for my $c (split(//, $str)) {
>	if ($c eq "\t") {
>	    $res .= ' ';
>	    $n++;
>	    for (; ($n % 8) != 0; $n++) {
>		$res .= ' ';
>	    }
>	    next;
>	}
>	$res .= $c;
>	$n++;
>    }
>
>    return $res;
>}
>
>my $args = join(" ", @ARGV);
>my $output = `git ls-files -- $args`;
>my @files = split("\n", $output);
>
>foreach my $file (@files) {
>    my $f;
>    my $cvt = 0;
>    my $text;
>
># read the file
>
>    next if ((-d $file));
>
>    open($f, '<', $file)
>	or die "$P: Can't open $file for read\n";
>    $text = do { local($/) ; <$f> };
>    close($f);
>
>    next if ($text eq "");
>
># Remove the embedded filenames
>
>    # remove individual lines like /* filename */ completely
>    $cvt += $text =~ s@/\*[ \t]+(?:linux\/)?\Q$file\E[ \t]*\*/[ \t]*\n@@g;
>    pos($text) = 0;
>    # remove filenamee from /* filename -- comment */, leave /* comment */
>    $cvt += $text =~ s@/\*([ \t]+)(?:linux\/)?\Q$file\E[ \t]*[:-]+[ \t]*@/*$1@g;
>    pos($text) = 0;
>    # remove filename and any trailing ' *\n' from /* filename, leave /*
>    $cvt += $text =~ s@/\*([ \t]+)(?:linux\/)?\Q$file\E[ \t]*\n([ \t]*\*[ \t]*\n)*(?:[ \t]*\*)?@/*@g;
>    pos($text) = 0;
>    # remove filename from /* filename, leave /*
>    $cvt += $text =~ s@/\*([ \t]+)(?:linux\/)?\Q$file\E[ \t]*\n@/*@g;
>    pos($text) = 0;
>    # remove filename from continuation ' * filename -- comment'
>    # leave ' * comment'
>    $cvt += $text =~ s/([ \t]+)\*([ \t]*)(?:linux\/)?\Q$file\E[ \t]*[:-]+[ \t]*/$1*$2/g;
>    pos($text) = 0;
>    # remove filename and any trailing ' *\n' from
>    # continuation ' * filename\n *\n'
>    $cvt += $text =~ s/([ \t]*)\*([ \t]*)(?:linux\/)?\Q$file\E[ \t]*\n([ \t]*\*[ \t]*\n)*//g;
>    pos($text) = 0;
>
># write the file if something was changed
>
>    if ($cvt > 0) {
>	$modified = 1;
>	print("$file\n");
>	open($f, '>', $file)
>	    or die "$P: Can't open $file for write\n";
>	print $f $text;
>	close($f);
>    }
>}
>
>if ($modified && !$quiet) {
>    print <<EOT;
>
>Warning: these changes may not be correct.
>
>These changes should be carefully reviewed manually and not combined with
>any functional changes.
>
>Compile, build and test your changes.
>
>You should understand and be responsible for all object changes.
>
>Make sure you read Documentation/SubmittingPatches before sending
>any changes to reviewers, maintainers or mailing lists.
>EOT
>}
  Joe,

  Suggestion.... please take those damn EOT lines out of it ..absolutely not
  required...or did you put for your own purpose?? As I believe it not the final
  product. Anyway, it would be good if those not there.

  Yup, I do like the "individual option" stuff ...so, you can only mess around
  single thing than the whole lot.

  ~Bhaskar
Joe Perches Oct. 2, 2020, 3:13 p.m. UTC | #3
( * removing mentors, I originally meant to send that to mentees )

On Fri, 2020-10-02 at 20:19 +0530, Bhaskar Chowdhury wrote:
> On 11:47 Thu 01 Oct 2020, Joe Perches wrote:
> > It's rather unnecessary for files to contain their
> > path/filename in source code comments.
> > 
> > Here's a trivial little script that can remove
> > embedded filenames in c90 style comments from files.
> > 
> > This requires git.
[]
> > Running it on today's -next gives:
> > 
> > $ perl remove_embedded_filenames.pl
> > $ git diff --shortstat
> > 2310 files changed, 354 insertions(+), 4239 deletions(-)
> > 
> > It's also possible to give any filename or path
> > as an argument to the script
> > 
> > For instance:
> > 
> > $ perl remove_embedded_filenames.pl drivers/net

The below was an attachment, it's odd that your
mailer quoted it.

> > #!/usr/bin/perl -w
[]
> > if ($modified && !$quiet) {
> >    print <<EOT;
> > Warning: these changes may not be correct.
> > 
> > These changes should be carefully reviewed manually and not combined with
> > any functional changes.
> > 
> > Compile, build and test your changes.
> > 
> > You should understand and be responsible for all object changes.
> > 
> > Make sure you read Documentation/SubmittingPatches before sending
> > any changes to reviewers, maintainers or mailing lists.
> > EOT
> > }
> 
> Suggestion.... please take those damn EOT lines out of it

No.  What's your actual problem with it?

It's a tool and it may not be perfect.

It merely emits a single message if it removes
filenames from files.
Andrew Morton Oct. 2, 2020, 10:18 p.m. UTC | #4
On Thu, 01 Oct 2020 11:28:10 -0700 Joe Perches <joe@perches.com> wrote:

> Embedding the complete filename path inside the file
> isn't particularly useful as often the path is moved
> around and becomes incorrect.
> 
> Emit a warning when the source contains the filename.
> 
> ...
>
> --- a/scripts/checkpatch.pl
> +++ b/scripts/checkpatch.pl
> @@ -3273,6 +3273,12 @@ sub process {
>  			}
>  		}
>  
> +# check for embedded filenames
> +		if ($rawline =~ /^\+.*\Q$realfile\E/) { di
> +			WARN("EMBEDDED_FILENAME",
> +			     "It's generally not useful to have the filename in the file\n" . $herecurr);
> +		}
> +

I removed that " di".  Please check that I merged the correct version
of this!
Joe Perches Oct. 2, 2020, 11:35 p.m. UTC | #5
On Fri, 2020-10-02 at 15:18 -0700, Andrew Morton wrote:
> On Thu, 01 Oct 2020 11:28:10 -0700 Joe Perches <joe@perches.com> wrote:
> 
> > Embedding the complete filename path inside the file
> > isn't particularly useful as often the path is moved
> > around and becomes incorrect.
> > 
> > Emit a warning when the source contains the filename.
> > 
> > ...
> > 
> > --- a/scripts/checkpatch.pl
> > +++ b/scripts/checkpatch.pl
> > @@ -3273,6 +3273,12 @@ sub process {
> >  			}
> >  		}
> >  
> > +# check for embedded filenames
> > +		if ($rawline =~ /^\+.*\Q$realfile\E/) { di
> > +			WARN("EMBEDDED_FILENAME",
> > +			     "It's generally not useful to have the filename in the file\n" . $herecurr);
> > +		}
> > +
> 
> I removed that " di".  Please check that I merged the correct version
> of this!

Thanks, it must have been added accidentally in my email client.

Combined, the patches are correct.

Patch
diff mbox series

diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
index a213cdb82ab0..84d2ee118c55 100755
--- a/scripts/checkpatch.pl
+++ b/scripts/checkpatch.pl
@@ -3273,6 +3273,12 @@  sub process {
 			}
 		}
 
+# check for embedded filenames
+		if ($rawline =~ /^\+.*\Q$realfile\E/) { di
+			WARN("EMBEDDED_FILENAME",
+			     "It's generally not useful to have the filename in the file\n" . $herecurr);
+		}
+
 # check we are in a valid source file if not then ignore this hunk
 		next if ($realfile !~ /\.(h|c|s|S|sh|dtsi|dts)$/);