sub bug_format_comment {
    my ($self, $args) = @_;

    # This replaces every occurrence of the word "foo" with the word
    # "bar"

    my $regexes = $args->{'regexes'};

    # And this links every occurrence of the word "bar" to example.com,
    # but it won't affect "foo"s that have already been turned into "bar"
    # above (because each regex is run in order, and later regexes don't modify
    # earlier matches, due to some cleverness in Bugzilla's internals).
    #
    # For example, the phrase "foo bar" would become:
    # bar <a href="http://example.com/bar">bar</a>
    my $commit_match = qr/\bcommit(:*)\s(:*)(\s*)([0-9a-z]{5,})\b/;
    push(@$regexes, { match => $commit_match, replace => \&_replace_commit });
    my $review_match = qr/^review(:*)\s(:*)(\s*)([0-9]{4,})$/;
    push(@$regexes, { match => $review_match, replace => \&_replace_review });
}

# Used by bug_format_comment--see its code for an explanation.
sub _replace_commit {
    my $args = shift;
    my @value = $args->{matches};
    # $match is the first parentheses match in the $bar_match regex 
    # in bug-format_comment.pl. We get up to 10 regex matches as 
    # arguments to this function.
    my $match = $args->{matches}->[0];
   # Remember, you have to HTML-escape any data that you are returning!
    $match = html_quote($match);
    return qq{<a href="http://cgit.scilab.org/scilab/commit/?id=">match $match</a>};
};