Lintian checking ARB packages

[Niels Thykier]

On 2012-12-31 14:19, Bhavani Shankar R wrote:
> On Thu, Dec 27, 2012 at 4:05 PM, Niels Thykier <niels at thykier.net> wrote:
>> [...]
> 
>>  * test_has_maintainer_scripts
>>    - checks/debhelper might be useful as an example.
> 
> Yes checks/debhelper has the required functionality I believe when I
> checked for maintainer scripts. and Indeed it has some dh_ checks.
> Thanks Niels!
> 
>>  * test_is_lightweight_app
>>    - checks/cruft might be useful as an example (e.g. File::Find
>>      and $info->unpacked)
> 
> Looked at $info->unpacked, but as ARB, I think we define lightweight
> app in terms of code if m not wrong, (If you have a look at the
> snippet below:
> 
>  @requires('source_files')
>     def test_is_lightweight_app(self):
>         source_file_extensions = [ ".py", ".rb", ".vala", ".c", ".h", ".cpp",
>                 ".cc", ".hh", ".pm", ".java", ".js", ".pl" ]
>         number_of_max_source_files = 30
>         number_of_max_lines_of_code = 15000
>         source_code_files = filter(lambda a: os.path.splitext(a)[1] in
> source_file_extensions,
>                                    self.source_files)
>         loc = sum(map(lambda a: len(open(a).readlines()), source_code_files))
>         if len(source_code_files) >= number_of_max_source_files or \
>            loc >= number_of_max_lines_of_code:
> 
> indicates the same)
> 
> Wanted your inputs on whether this type of check is feasible in lintian.
> 

I believe it is feasible to do.  In its basics it would look something like:

  use Lintian::Util qw(fail);

  [...]

  my $no_files = 0;
  my $no_lines = 0;
  foreach my $filename ($info->sorted_index) {
      next unless $info->index ($filename)->is_file;
      next unless $filename =~ m/\.(?:py|rb|vala|...)$/;
      my $fpath = $info->unpacked ($filename);
      $no_files++;
      last if $no_files > 30;
      open my $fd, '<', $fpath or fail "open $fpath: $!";
      foreach my $line (<$fd>) {
          $no_files++;
      }
      close $fd;
      last if $no_lines > 15000;
  }
  if ($no_files > 30) {
      tag 'too-many-files', $no_files, '> 30';
  }

  if ($no_lines > 30) {
      tag 'too-many-lines-of-source', $no_lines, '> 15 000';
  }

Of course, it is probably possible to replace some of the loop with
stuff like grep and map (though I am not sure it makes it more readable
or concise for that matter).

>> [...]
> 
> 
> Thanks Niels for your everlasting patience and reviews :)
> 
> Regards,

You are more than welcome.  On a related note, I have devised a short
POD tutorial in commit 0b7b5d86a502c27ce61c5b1157d43fa7de4b8b19.  It
will show up in the API docs (run "debian/rules api-doc").
  I am not sure it teaches you anything atm, but feel free to send
suggestions or patches for it.  :)

~Niels