]> Sergey Matveev's repositories - public-inbox.git/blobdiff - public-inbox-learn
-learn: nuke HTML portions when training as ham
[public-inbox.git] / public-inbox-learn
index 13b75b763cfc9d3280011b29a237065d013de987..db0a1bb35e5ec8115e007c4dcd8cd12fdee82a70 100755 (executable)
@@ -24,9 +24,16 @@ foreach my $h (qw(Cc To)) {
        }
 }
 
-my $in = $mime->as_string;
-$mime->body_set('');
+my ($name, $email, $date);
+
+if ($train eq "ham") {
+       require PublicInbox::MDA;
+       require PublicInbox::Filter;
+       PublicInbox::Filter->run($mime);
+       ($name, $email, $date) = PublicInbox::MDA->author_info($mime);
+}
 
+my $in = $mime->as_string;
 my $err = 0;
 my @output = qw(> /dev/null > /dev/null);
 
@@ -50,16 +57,10 @@ foreach my $recipient (keys %dests) {
                        }
                }
        } else { # $train eq "ham"
-               require PublicInbox::MDA;
-               require PublicInbox::Filter;
-
-               # no checking for errors here, we assume the message has
+               # no checking for spam here, we assume the message has
                # been reviewed by a human at this point:
-               PublicInbox::Filter->run($mime);
                PublicInbox::MDA->set_list_headers($mime, $dst);
 
-               my ($name, $email, $date) =
-                               PublicInbox::MDA->author_info($mime);
                local $ENV{GIT_AUTHOR_NAME} = $name;
                local $ENV{GIT_AUTHOR_EMAIL} = $email;
                local $ENV{GIT_AUTHOR_DATE} = $date;