TO: vs CC:

[David Relson]

I just noticed an interesting anomaly....

My wife is a Girl Scout troop leader.  This morning she received a
message about a meeting.  The message had 31 people in its "To:" field
of a message.  Bogofilter tagged the messages as ham and auto-updated
the wordlist, just like it should have.

Due to weather the meeting has been cancelled and she got to use "Reply
All" to send out the message.  Bogofilter tagged the cancellation notice
as "unsure".  As I thought this was peculiar, I used "-vvv" to see
tokens and scores.  Here are a few lines:

"head:Leonor"                        1  0.000000  0.000017  0.992115 +
"head:Betsy"                         2  0.000000  0.000034  0.996023 +
"head:Giles"                         2  0.000000  0.000034  0.996023 +
"head:Wendy"                         4  0.000000  0.000068  0.998003 +
"head:Cook"                          5  0.000000  0.000085  0.998401 +
"head:Sonja"                         5  0.000000  0.000085  0.998401 +
"head:jennifer"                      6  0.000000  0.000102  0.998666 +

What's peculiar about these tokens is that they're all purely spam --
all have ham counts of zero.  This shouldn't be the case since all these
names were in the To: section of a previous "ham" message and should all
have been entered into the wordlist by auto-update, i.e. "-u".

Thinking further, I realized that the first time around, the names were
processed as "to:Leonor", "to:Betsy", etc.  In the second message, they
were in a CC: section which was processed like a regular header line and
didn't receive the special tagging of a TO: line, i.e. they were
processed as "head:Leonor", "head:Betsy", etc.

It seems that bogofilter should tag CC: lines the same way it
processes TO: lines.

### Here's a test of the two messages using 0.16.1 ###

[relson at osage src]$ bogofilter -C -v -B msg.pc.0114*
msg.pc.0114.1114 X-Bogosity: No, tests=bogofilter, spamicity=0.000000,
version=0.16.1
msg.pc.0114.1312 X-Bogosity: No, tests=bogofilter, spamicity=0.500000,
version=0.16.1

### Here's a test of the two messages using 0.16.1.cvs, which tags CC:
lines like TO: lines ###

[relson at osage src]$ bogofilter -C -v -B msg.pc.0114*
msg.pc.0114.1114 X-Bogosity: No, tests=bogofilter, spamicity=0.000000,
version=0.16.1.cvs
msg.pc.0114.1312 X-Bogosity: No, tests=bogofilter, spamicity=0.000000,
version=0.16.1.cvs

Looks good to me!

Anybody object to changing bogofilter to tag CC: like TO: ???

David

-- 
David Relson                   Osage Software Systems, Inc.
relson at osagesoftware.com       Ann Arbor, MI 48103
www.osagesoftware.com          tel:  734.821.8800