Bug#795614: [PATCH] fix false-positive spelling-error-in-binary affecting many Go binaries

[Michael Stapelberg <stapelberg@debian.org>]

On Sun, Aug 16, 2015 at 10:11 AM, Niels Thykier <niels@thykier.net> wrote:
> On 2015-08-15 21:59, Michael Stapelberg wrote:
>> Package: lintian
>> Version: 2.5.33
>> Severity: normal
>> Tags: patch
>>
>>>From the patch description:
>>  [PATCH] spelling-error-in-binary: ignore ang/and
>>
>>  The Go stdlib html/ package contains the string "ang;" (for the &ang;
>>  entity). Every program which directly or indirectly imports the html/
>>  package will hence end up with an unuseful spelling-error-in-binary
>>  lintian warning.
>>
>> [...]
>
> Hi Michael,
>
> Thanks for the patch.
>
> I was wondering, have you tried something like?
>
> """
> $ git diff
> diff --git a/lib/Lintian/Check.pm b/lib/Lintian/Check.pm
> index 4b08fae..0b11477 100644
> --- a/lib/Lintian/Check.pm
> +++ b/lib/Lintian/Check.pm
> @@ -301,6 +301,7 @@ sub check_spelling {
>      $text =~ s/(\w-)\s*\n\s*/$1/;
>      $text =~ tr/\r\n \t/ /s;
>      $text =~ s/\s++/ /g;
> +    $text =~ s/\&[a-zA-Z0-9_-]+\;//g; # Ignore XML entities
>      strip($text);
>
>      for my $word (split(' ', $text)) {
> """
>
> AFAICT, it should strip out all XML entities (untested though), which
> should be a more generic solution.  Does this work for you?

I think your patch does indeed strip out XML entities, but it doesn’t
quite work for our use-case: note that the string that’s found in the
binary is “ang;”, not “&ang;”.

-- 
Best regards,
Michael