sv_spamchecker: New subcommand 'usercheck'

1 files changed, 91 insertions, 59 deletions

diff --git a/backend/misc/sv_spamchecker b/backend/misc/sv_spamchecker
index e33b4a3..10242c8 100755
--- a/backend/misc/sv_spamchecker
+++ b/backend/misc/sv_spamchecker
@@ -46,6 +46,17 @@ my $verbose;
 my %stats;
 my $spamc;
 
+sub item_tag {
+    my ($tracker, $item_id, $comment_id) = @_;
+    if (defined($item_id)) {
+	$tracker .= '#' . $item_id;
+	if ($comment_id) {
+	    $tracker .= 'c' . $comment_id;
+	}
+    }
+    return $tracker;
+}
+
 sub parse_time_interval {
     my $input = shift;
 
@@ -144,9 +155,9 @@ sub adjust_spamscore {
 sub check_tracker_comment {
     my ($tracker, $item_id, $comment_id, %opts) = @_;
 
-    debug("checking entry: "
+    debug("checking "
 	  . (exists($opts{queue_id}) ? "queue_id=$opts{queue_id}, " : '')
-	  . "tracker=$tracker, item_id=$item_id, comment_id=$comment_id");
+	  . item_tag($tracker, $item_id, $comment_id));
 
     if (exists($opts{queue_id})) {
 	db_delete(table => "trackers_spamcheck_queue",
@@ -188,9 +199,12 @@ sub check_tracker_comment {
 			  'bug_id=? LIMIT 1', 
 			  $item_id);
     }
-	
-    my $tag = "$tracker#$item_id";
-    $tag .= "c$comment_id" if $comment_id;
+
+    if (exists($opts{spamscore})) {
+	$spamscore = $opts{spamscore};
+    }
+    
+    my $tag = item_tag($tracker, $item_id, $comment_id);
 	
     if ($spamscore < 5) {
 	# The item was unflagged
@@ -309,9 +323,9 @@ sub learn_from_item {
 				       $ref->{text});
 
     my $learntype = ($ref->{spamscore} > 4) ? TYPE_SPAM : TYPE_HAM;
+    my $tag = item_tag($tracker, $ref->{item_id}, $ref->{comment_id});
+
     if (debug()) {
-	my $tag = "$tracker#$ref->{item_id}";
-	$tag .= ", comment \#$ref->{comment_id}" if $ref->{comment_id};
 	logit("debug", "$tag: $ref->{uid} $ref->{spamscore} $isspam{$learntype}");
     }
 
@@ -324,7 +338,8 @@ sub learn_from_item {
     return if dry_run();
 
     my $res = $spamc->learn($ref->{text}, $learntype);
-    logit('err', "Error learning from $tracker:$ref->{item_id}" . $spamc->{resp_msg}) unless defined $res;
+    logit('err', "Error learning from $tag: " . $spamc->{resp_msg})
+	unless defined $res;
     db_modify(qq{
 INSERT INTO trackers_spamcheck_cache (artifact, item_id, comment_id, isspam)
 VALUES (?, ?, ?, ?)
@@ -435,45 +450,22 @@ WHERE $cond},
     return @userlist;
 }
 
-sub com_mark {
+sub userproc {
+    my $cb = shift;
+    
     my @ulist = userlist_collect(@_);
     foreach my $user (@ulist) {
 	if ($user->{resume} ne '') {
 	    debug("$user->{user_name}: resume");
-	    unless (dry_run()) {
-		db_insert({
-		        table => 'trackers_spamcheck_queue',
-			values => {
-			    artifact => 'resume',
-			    item_id => $user->{user_id},
-			    comment_id => 0,
-			    priority => 3
-			},
-			update => {
-			    priority => 3,
-		        }
-		      });
-		db_modify(q{UPDATE user SET resume_spamscore=? WHERE user_id=?},
-			  8, $user->{user_id});
-	    }
+	    &{$cb}($user, 'resume', $user->{user_id}, 0);
 	}
 	
 	foreach my $t (TrackerNames()) {
 	    db_foreach(
 		sub {
 		    my $ref = shift;
-		    debug("$user->{user_name}: $t#$ref->{bug_id}");
-		    db_insert({ table => 'trackers_spamcheck_queue',
-				values => {
-				    artifact => $t,
-				    item_id => $ref->{bug_id},
-				    comment_id => 0,
-				    priority => 3
-				},
-				update => {
-				    priority => 3,
-                                }					    
-			      }) unless dry_run();
+		    debug("$user->{user_name}: ". item_tag($t, $ref->{bug_id}));
+		    &{$cb}($user, $t, $ref->{bug_id}, 0);
 		},
 		qq{
 SELECT tracker.bug_id AS bug_id
@@ -488,18 +480,9 @@ ORDER BY tracker.date
 	    db_foreach(
 		sub {
 		    my $ref = shift;
-		    debug("$user->{user_name}: $t#$ref->{bug_id}c$ref->{comment_id}");
-		    db_insert({ table => 'trackers_spamcheck_queue',
-				values => {
-				    artifact => $t,
-				    item_id => $ref->{bug_id},
-				    comment_id => $ref->{comment_id},
-				    priority => 3
-				},
-				update => {
-				    priority => 3,
-			        }
-			      }) unless dry_run();
+		    debug("$user->{user_name}: "
+			  . item_tag($t, $ref->{bug_id}, $ref->{comment_id}));
+		    &{$cb}($user, $t, $ref->{bug_id}, $ref->{comment_id});
 		},
 		qq{
 SELECT history.bug_history_id AS comment_id,
@@ -512,23 +495,65 @@ LEFT JOIN user_group grp
 WHERE history.field_name='details' AND history.mod_by=? AND grp.user_group_id IS NULL
 ORDER BY history.date},
 		$user->{user_id}, $user->{user_id});
-
-	    unless (dry_run()) {
-		db_modify(qq{UPDATE $t SET spamscore=8 WHERE submitted_by=?},
-			  $user->{user_id});
-		db_modify(qq{UPDATE ${t}_history SET spamscore=8 WHERE field_name=? AND mod_by=?},
-			  'details', $user->{user_id});
-	    }
 	}
     }
 }
 
+sub com_mark {
+    userproc(
+	sub {
+	    return if dry_run();
+
+	    my ($user, $tracker, $bug_id, $comment_id) = @_;
+	    
+	    db_insert({ table => 'trackers_spamcheck_queue',
+			values => {
+			    artifact => $tracker,
+			    item_id => $bug_id,
+			    comment_id => $comment_id,
+			    priority => 3
+			},
+			update => {
+			    priority => 3,
+		        }
+		      });
+	    
+	    if ($tracker eq 'resume') {
+		db_modify(q{UPDATE user SET resume_spamscore=? WHERE user_id=?},
+			  8, $user->{user_id});
+	    } elsif ($comment_id == 0) {
+		db_modify(qq{UPDATE $tracker SET spamscore=? WHERE bug_id=?},
+			  8, $bug_id);
+	    } else {
+		db_modify(qq{UPDATE ${tracker}_history SET spamscore=? WHERE bug_history_id=?},
+			  8, $comment_id);
+	    }
+	},
+	@_);
+}
+
+sub com_usercheck {
+    userproc(
+	sub {
+	    my ($user, $tracker, $bug_id, $comment_id) = @_;
+	    check_tracker_comment($tracker,
+				  $bug_id,
+				  $comment_id,
+				  spamscore => 8,
+				  notify => 0);
+	},
+	@_);
+    update_user_spamscores(U_IDS, [@_]);
+}
+
 my %command = (
     queue => { code => \&com_check },
     learn => { code => \&com_learn, prereq => 'queue' },
     # Recompute user spamscore
     userscore => { code => \&com_userscore, wantargs => 1 },
     mark => { code => \&com_mark, wantargs => 1 },
+    usercheck => { code => \&com_usercheck, wantargs => 1 },
+    usrck => { alias => 'usercheck' }
 );
 
 
@@ -637,6 +662,8 @@ B<sv_spamchecker>
 [B<queue>]
 [B<learn>]
 [B<userscore>]
+[B<usercheck>]
+[B<usrck>]    
 [B<mark>] [I<USER>...]
 
 B<sv_spamchecker> [B<-h>] [B<--help>]
@@ -669,11 +696,16 @@ By default, both stages are run in the order they are described above.
 To instruct B<sv_spamchecker> to run only a particular stage, supply
 an additional argument: B<queue> or B<learn>.    
 
-Two special modes are available, which take optional list of user names
+Three special modes are available, which take optional list of user names
 in the command line.  When given the B<userscore> argument, B<sv_spamchecker>
 recomputes total spam scores for the listed users.  When given the B<mark>
-argument, it flags the listed users for subsequent spam checking.  If no
-user names are given, both commands apply to all Savane users (use with care).
+argument, it flags the listed users for subsequent spam checking.  Finally,
+the B<usercheck> (or B<usrck>) argument recomputes the spam score for the
+given users, which is (equivalent to running B<sv_spamchecker mark I<USERS>;
+sv_spamchecker queue>)    
+    
+If no user names are given, these three commands apply to all Savane users
+(use with care). 
     
 =head1 OPTIONS