Thread: Overeager spam fighting for link?

Overeager spam fighting for link?

From
Greg Sabino Mullane
Date:
I came across this page today:

Subject: Re: redacting password in SQL statement in server log


The spam filtering[1] that occurs for the header information appears to have been applied to the bottom link as well - making it 404. Not sure where that is getting applied, but doing a "raw source" and checking my original inbox seems to indicate the link at the bottom of the email was sent unobscured.

[1] In which alice@gmail.com becomes alice(at)gmail(dot)com

Cheers,
Greg

--
Enterprise Postgres Software Products & Tech Support

Re: Overeager spam fighting for link?

From
Julien Rouhaud
Date:
On Tue, Feb 18, 2025 at 09:27:32AM -0500, Greg Sabino Mullane wrote:
> I came across this page today:
>
> Subject: Re: redacting password in SQL statement in server log
>
> https://www.postgresql.org/message-id/20220724114449.qqjbwgmpcgn66yhy%40jrouhaud
>
> The spam filtering[1] that occurs for the header information appears to
> have been applied to the bottom link as well - making it 404. Not sure
> where that is getting applied, but doing a "raw source" and checking my
> original inbox seems to indicate the link at the bottom of the email was
> sent unobscured.

FWIW I confirm that I did send a valid / unobscured link.

I may be wrong but I think that this issue happened (and was solved) a few
years ago.



Re: Overeager spam fighting for link?

From
Tom Lane
Date:
Julien Rouhaud <rjuju123@gmail.com> writes:
>> The spam filtering[1] that occurs for the header information appears to
>> have been applied to the bottom link as well - making it 404. Not sure
>> where that is getting applied, but doing a "raw source" and checking my
>> original inbox seems to indicate the link at the bottom of the email was
>> sent unobscured.

> FWIW I confirm that I did send a valid / unobscured link.

IME the trick to keeping this from happening is to manually convert
any "@" characters in the URL you send to "%40".  Usually if you
copy-and-paste from your browser address bar that happens for free.

            regards, tom lane



Re: Overeager spam fighting for link?

From
Julien Rouhaud
Date:
On Tue, Feb 18, 2025 at 07:51:01PM -0500, Tom Lane wrote:
> Julien Rouhaud <rjuju123@gmail.com> writes:
> >> The spam filtering[1] that occurs for the header information appears to
> >> have been applied to the bottom link as well - making it 404. Not sure
> >> where that is getting applied, but doing a "raw source" and checking my
> >> original inbox seems to indicate the link at the bottom of the email was
> >> sent unobscured.
>
> > FWIW I confirm that I did send a valid / unobscured link.
>
> IME the trick to keeping this from happening is to manually convert
> any "@" characters in the URL you send to "%40".  Usually if you
> copy-and-paste from your browser address bar that happens for free.

That's what I usually do.  I guess that time I just copied the link from some
email header.



Re: Overeager spam fighting for link?

From
Greg Sabino Mullane
Date:
Yes, but surely our script can be taught that \bhttps://\S* should be excluded from the munging?

Secondarily, maybe make the web server smart enough to undo and redirect the (and) (at) stuff that would otherwise be 404?

Cheers,
Greg

Re: Overeager spam fighting for link?

From
Magnus Hagander
Date:
On Tue, Feb 18, 2025 at 3:28 PM Greg Sabino Mullane <htamfids@gmail.com> wrote:
I came across this page today:

Subject: Re: redacting password in SQL statement in server log


The spam filtering[1] that occurs for the header information appears to have been applied to the bottom link as well - making it 404. Not sure where that is getting applied, but doing a "raw source" and checking my original inbox seems to indicate the link at the bottom of the email was sent unobscured.

[1] In which alice@gmail.com becomes alice(at)gmail(dot)com


The regexp used is this:
_re_mail = re.compile(r'(/m(essage-id)?/)?[^()<>@,;:\/\s"\'&|]+@[^()<>@,;:\/\s"\'&|]+')


The comment indicates it's trying to special case these links already, but clearly it's failing. I think maybe because it's not considering the /flat/ part of the URL. I haven't got time to test it out properly right this moment, but I think that might be it.

--