I have a procmail rule that works perfectly for me:

:0
* ^To:.*anonymized
* ^Subject:.*Report domain: anonymized Submitter: [^ ]* Report-ID:
{
LOG="1"
:0 cw
* ^To:\/.+
| /etc/procmail/process_dmarc_report.sh "$MATCH"
}

But the subject is sometimes encoded, which breaks the rule. Based on

I wanted to replace the test on the Subject header with a test on the SUBJECT variable containing the possibly decoded subject, as follows:

:0
* ^Subject:.*=\?utf-8\?B\?
SUBJECT=`formail -cXSubject: | perl -MMIME::Base64 -ne 's/^=\?utf-8\?B\?(.*?)\?=$/print decode_base64($1)/e'`
:0 E
SUBJECT=`formail -cXSubject:`
:0
* ^To:.*anonymized
* SUBJECT ?? Subject:.*Report domain: anonymized Submitter: [^ ]* Report-ID:.*

The subject is correctly decoded, but I'm doing it wrong because I have an error in the logs:

procmail: Skipped "[Preview] Report Domain: anonymized Submitter: enterprise.protection.outlook Report-ID: c136884d7adc4f5c82e2fdbbde75e2ab"

And the email is delivered to Folder: SUBJECT=Subject:.

I tried with -x instead of -X and not including Subject: in the test, and several other small adaptations, but without success. I can't find my mistake. It may be simple, but I can't seem to get it right.

If, as suggested, I do something like:

:0
* ^Subject:.*=\?utf-8\?B\?
SUBJECT=| formail -cXSubject: | perl -MMIME::Base64 -ne 's/^=\?utf-8\?B\?(.*?)\?=$/print decode_base64($1)/e'

:0 E
SUBJECT=| formail -cXSubject:

$SUBJECT is empty

My entire rule (not working) at the moment:

:0
* ^Subject:.*=\?utf-8\?B\?
SUBJECT=`formail -cXSubject: | perl -MMIME::Base64 -ne 's/^=\?utf-8\?B\?(.*?)\?=$/print decode_base64($1)/e'`
:0 E
SUBJECT=`formail -cXSubject:`
:0
* ^To:.*anonymized
* SUBJECT ?? Subject:.*Report domain: anonymized Submitter: [^ ]* Report-ID:.*
{
        LOG="1"
        :0 cw
        * ^To:\/.+
        | /etc/procmail/process_dmarc_report.sh "$MATCH"
}

I have a procmail rule that works perfectly for me:

:0
* ^To:.*anonymized
* ^Subject:.*Report domain: anonymized Submitter: [^ ]* Report-ID:
{
LOG="1"
:0 cw
* ^To:\/.+
| /etc/procmail/process_dmarc_report.sh "$MATCH"
}

But the subject is sometimes encoded, which breaks the rule. Based on https://stackoverflow/a/29715987/13463349

I wanted to replace the test on the Subject header with a test on the SUBJECT variable containing the possibly decoded subject, as follows:

:0
* ^Subject:.*=\?utf-8\?B\?
SUBJECT=`formail -cXSubject: | perl -MMIME::Base64 -ne 's/^=\?utf-8\?B\?(.*?)\?=$/print decode_base64($1)/e'`
:0 E
SUBJECT=`formail -cXSubject:`
:0
* ^To:.*anonymized
* SUBJECT ?? Subject:.*Report domain: anonymized Submitter: [^ ]* Report-ID:.*

The subject is correctly decoded, but I'm doing it wrong because I have an error in the logs:

procmail: Skipped "[Preview] Report Domain: anonymized Submitter: enterprise.protection.outlook Report-ID: c136884d7adc4f5c82e2fdbbde75e2ab"

And the email is delivered to Folder: SUBJECT=Subject:.

I tried with -x instead of -X and not including Subject: in the test, and several other small adaptations, but without success. I can't find my mistake. It may be simple, but I can't seem to get it right.

If, as suggested, I do something like:

:0
* ^Subject:.*=\?utf-8\?B\?
SUBJECT=| formail -cXSubject: | perl -MMIME::Base64 -ne 's/^=\?utf-8\?B\?(.*?)\?=$/print decode_base64($1)/e'

:0 E
SUBJECT=| formail -cXSubject:

$SUBJECT is empty

My entire rule (not working) at the moment:

:0
* ^Subject:.*=\?utf-8\?B\?
SUBJECT=`formail -cXSubject: | perl -MMIME::Base64 -ne 's/^=\?utf-8\?B\?(.*?)\?=$/print decode_base64($1)/e'`
:0 E
SUBJECT=`formail -cXSubject:`
:0
* ^To:.*anonymized
* SUBJECT ?? Subject:.*Report domain: anonymized Submitter: [^ ]* Report-ID:.*
{
        LOG="1"
        :0 cw
        * ^To:\/.+
        | /etc/procmail/process_dmarc_report.sh "$MATCH"
}

• procmail

2 Answers 2

It's of course possible to assign a value to a variable the way you did, however, not in a recipe action line. There the syntax requires a | for the assignment. From the manpage for "procmailrc":

|      Starts  the  specified program, possibly in $SHELL if any of the
       characters $SHELLMETAS are spotted.  You can optionally  prepend
       this  pipe symbol with variable=, which will cause stdout of the
       program to be captured in  the  environment  variable  (procmail
       will not terminate processing the rcfile at this point).  If you
       specify just this pipe symbol, without any program,  then  proc‐
       mail will pipe the mail to stdout.

So, in your case it should be

SUBJECT=| formail -cXSubject:

Same for the other "SUBJECT=" line, of course.

Regarding your edit saying that SUBJECT would still be empty:

Replace the Perl code with e.g.

perl -mEncode=from_to -pe 'from_to $_, "MIME-Header", "UTF-8"'

The problem was about the variable test:

* ^Subject:.*Report domain: anonymized Submitter: [^ ]* Report-ID:

won't work because of [^ ]*
but

* ^Subject:.*Report domain: anonymized Submitter: .* Report-ID:

works !

* variable ?? regex
Test the value of variable against regex.

The documentation says it's a regex, but it really isn't.

My final working rule is:

        :0
        * ^Subject: \/=\?utf-8\?B\?.+
        {
                SUBJECT=`echo $MATCH | perl -MMIME::Base64 -ne 's/^=\?utf-8\?B\?(.*?)\?=$/print decode_base64($1)/e'` 
        }

        :0E
        * ^Subject: \/.+
        {
                SUBJECT=$MATCH
        }


        :0
        * SUBJECT ?? Report domain: novazur\.fr Submitter: .* Report-ID:
        {
                :0 cw
                * ^To:\/.+
                | /etc/procmail/process_dmarc_report.sh "$MATCH"
        }

NB: Never remove curly braces that you might think are unnecessary around variable assignments. They are essential, even in:

:0
* ^Subject: \/=\?utf-8\?B\?.+
{
    SUBJECT=`echo $MATCH | perl -MMIME::Base64 -ne 's/^=\?utf-8\?B\?(.*?)\?=$/print decode_base64($1)/e'`
}
:0E
* ^Subject: \/.+
{
    SUBJECT=$MATCH
}