Bug 703788: Improve performance of diff_arrays() with large arrays

r/a=mkanat

1 files changed, 29 insertions, 26 deletions

diff --git a/Bugzilla/Util.pm b/Bugzilla/Util.pm
index 1f698f80e..b75c5c340 100644
--- a/Bugzilla/Util.pm
+++ b/Bugzilla/Util.pm
@@ -305,36 +305,39 @@ sub use_attachbase {
 
 sub diff_arrays {
     my ($old_ref, $new_ref, $attrib) = @_;
-
-    my @old = @$old_ref;
-    my @new = @$new_ref;
     $attrib ||= 'name';
 
-    # For each pair of (old, new) entries:
-    # If object arrays were passed then an attribute should be defined;
-    # If they're equal, set them to empty. When done, @old contains entries
-    # that were removed; @new contains ones that got added.
-    foreach my $oldv (@old) {
-        foreach my $newv (@new) {
-            next if ($newv eq '' or $oldv eq '');
-            if (blessed($oldv) and blessed($newv)) {
-                if ($oldv->$attrib eq $newv->$attrib) {
-                    $newv = $oldv = '';
-                }
-            }
-            else {
-                if ($oldv eq $newv) {
-                    $newv = $oldv = ''
-                }
-            }
-        }
+    my (%counts, %pos);
+    # We are going to alter the old array.
+    my @old = @$old_ref;
+    my $i = 0;
+
+    # $counts{foo}-- means old, $counts{foo}++ means new.
+    # If $counts{foo} becomes positive, then we are adding new items,
+    # else we simply cancel one old existing item. Remaining items
+    # in the old list have been removed.
+    foreach (@old) {
+        next unless defined $_;
+        my $value = blessed($_) ? $_->$attrib : $_;
+        $counts{$value}--;
+        push @{$pos{$value}}, $i++;
     }
-
-    my @removed;
     my @added;
-    @removed = grep { $_ ne '' } @old;
-    @added   = grep { $_ ne '' } @new;
-
+    foreach (@$new_ref) {
+        next unless defined $_;
+        my $value = blessed($_) ? $_->$attrib : $_;
+        if (++$counts{$value} > 0) {
+            # Ignore empty strings, but objects having an empty string
+            # as attribute are fine.
+            push(@added, $_) unless ($value eq '' && !blessed($_));
+        }
+        else {
+            my $old_pos = shift @{$pos{$value}};
+            $old[$old_pos] = undef;
+        }
+    }
+    # Ignore cancelled items as well as empty strings.
+    my @removed = grep { defined $_ && $_ ne '' } @old;
     return (\@removed, \@added);
 }