parkerm at pobox
Sep 24, 2009, 12:16 AM
Post #1 of 1
(202 views)
Permalink
|
|
Fwd: split_into_array_of_short_paragraphs vs short_lines
|
|
Anyone have any thoughts on this? the BLANK_LINES rules appear to be
broken.
Michael
Begin forwarded message:
> From: Michael Parker <parkerm [at] pobox>
> Date: September 21, 2009 10:22:26 AM CDT
> To: SpamAssassin Dev <dev [at] spamassassin>
> Subject: split_into_array_of_short_paragraphs vs short_lines
>
> Howdy,
>
> I was looking at why our old 3.1 instance of SA was hitting a few of
> the BLANK_LINES_NN_NN rules where as 3.3 stuff wasn't hitting at
> all. I narrowed it down to what get_decoded_body_text_array returns.
>
> For instance, I have a short mail that is 4 lines long. In 3.1
> get_decoded_body_text_array would return an array with 4 elements
> (ie lines), however in 3.3 that call now returns a single element
> array with a "paragraph."
>
> This of course breaks this code in check_blank_line_ratio:
>
> if (scalar @{$fulltext} >= $minlines) {
> foreach my $line (@{$fulltext}) {
> next if ($line =~ /\S/);
> $blank++;
> }
> $pms->{blank_line_ratio}->{$minlines} = 100 * $blank / scalar
> @{$fulltext};
> }
> else {
> $pms->{blank_line_ratio}->{$minlines} = -1; # don't report if
> it's a blank message ...
> }
>
> Because its looking at array elements and not actual lines.
>
> I suspect there may be other similar eval rules doing that same thing.
>
> Thoughts?
>
> Michael
|
