Re: [PATCH v3] gitweb: redacted e-mail addresses feature.
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Mar 21, 2021, 18:48 UTC
- Message-ID
- <xmqqv99k9wc9.fsf@gitster.g>
- In-Reply-To
- <87r1k8qs73.fsf@evledraar.gmail.com>
Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:
Show 33 quoted lines
> But:
>
>> sub parse_tag {
>> my $tag_id = shift;
>> my %tag;
>> @@ -3471,6 +3493,10 @@ sub parse_tag {
>> if ($tag{'author'} =~ m/^([^<]+) <([^>]*)>/) {
>> $tag{'author_name'} = $1;
>> $tag{'author_email'} = $2;
>> + if (gitweb_check_feature('email_privacy')) {
>> + $tag{'author_email'} = "private";
>> + $tag{'author'} = hide_mailaddr($tag{'author'});
>> + }
>
> This code seems quite awkward, we've already done the regex match, but
> this code:
>
>> [snip]
>> +sub hide_mailaddr_if_private {
>> + my $line = shift;
>> + return $line unless (gitweb_check_feature('email_privacy') &&
>> + $line =~ m/^([^<]+) <([^>]*)>/);
>> + return hide_mailaddr($line)
>> +}
>> +
>> +sub hide_mailaddr {
>> + my $mailaddr = shift;
>> + $mailaddr =~ s/<([^>]*)>/<private>/;
>> + return $mailaddr;
>> +}
>
> Is going to do it again incrementally, and then just act on a
> search-replacement if we've got the feature enabled.I think you misread the patch the same way as I did initially. What is called in parse_tag is not the "if-private" version---the caller knows that $tag{'author'} MUST BE an address so unconditionally redact when the feature is set by bypassing the _if_private variant.
Having said that, I do think
Show 6 quoted lines
> sub maybe_hide_email {
> my ($email, $ref) = shift;
> return $email unless gitweb_check_feature('email_privacy');
> $$ref = "private" if $ref;
> return hide_email($email);
> }this helper is how you would simplify these numerous callers.
Show 19 quoted lines
>> } elsif ($format eq 'patch') {
>> - local $/ = undef;
>> - print <$fd>;
>> + while (my $line = <$fd>) {
>> + print hide_mailaddr_if_private($line);
>> + }
>
> Urm, have you tested this? How does a while loop over a <$fd> make sense
> when $/ is undef, the readline() operator will always return just one
> record, so having a while loop doesn't make sense.
>
> I'm not sure of the input here, but given that if you're expecting to
> replace all e-mail addresses on all lines with this function that's not
> how it'll work, the s/// doesn't have a /g, so it'll stop at the first
> replacement.
>
>> close $fd
>> or print "Reading git-format-patch failed\n";
>> }All true. This one is reading from "format-patch --stdout" output, but for the reasons I mentioned in my review, I do not think it should touch the 'patch' output at all.