Logfile analysing with pyparsing

Nick Vatamaniuc vatamane at gmail.com
Tue Sep 26 05:06:14 EDT 2006 

You can parse it just once, you just have to setup your data structure
(the structure of your XML schema) and fill it up as you parse.

For example, you can represent you data structure as a dictionaries in
Python:

message={
     MID : {
                   ' timestamp' : TIMESTAMP,
                   'from':FROM,
                   'tolist':[TO1, TO2, ... ],
                   'qids': [QID1, QID2, ...],
                   'spam_score':S_SCORE,
                   'spam_filtered':Y/N,
                   'spam_sentdo':S_SENDTO,
              }
}


Note: I inferred this from your XML.

Then as you parse each record(line?)  identify what token is what and
fill in the corresponding data structure. So if you pass by a line that
has the message_id, queue_id and a send_to fields, check if the
corresponding message_id dictionary has been created (if not do so) and
fill in what you just found out. Later on you might see another record
that will help you fill in other information regarding this particular
message_id (such as a 'timestamp' or a 'from' field).

Then translating the data into XML should be fairly easy.

Note: I am not familiar with the syntax of the mail log so I presented
a general idea only. My assumptions about the syntax might have been
wrong.

Hope this helps,

Nick Vatamaniuc




Andi Clemens wrote:
> Hi,
>
> we had some problems in the last weeks with our mailserver.
> Some messages were not delivered and we wanted to know why.
> But looking through the logfile is a time consuming process.
> So I wanted to write a parser to analyse the logs and parse them as XML.
>
> But I have never written a parser before and know I'm sitting in front
> of the logfile trying to write the grammar for pyparsing.
>
> First of all I need to know if it is possible to parse that kind of info
> into XML.
> Here is an excerpt of the logfile lines I'm interested in:
>
> Sep 18 04:15:22 mailrelay postfix/cleanup[12103]: 755387301:
> message-id=<200609180214.k8I2EuNo016264 at mforward2.dtag.de>
> Sep 18 04:15:22 mailrelay spamd[1364]: spamd: processing message
> <200609180214.k8I2EuNo016264 at mforward2.dtag.de> for nobody:65534
> Sep 18 04:15:25 mailrelay spamd[1364]: spamd: result: Y 15 -
> BAYES_99,DATE_IN_PAST_03_06,DNS_FROM_RFC_ABUSE,DNS_FROM_RFC_DSN,DNS_FROM_RFC_POST,DNS_FROM_RFC_WHOIS,FORGED_MUA_OUTLOOK,SPF_SOFTFAIL
> scantime=3.1,size=8086,user=nobody,uid=65534,required_score=5.0,rhost=localhost,raddr=127.0.0.1,rport=55277,mid=<200609180214.k8I2EuNo016264 at mforward2.dtag.de>,bayes=1,autolearn=no
>
> Sep 18 04:15:25 mailrelay postfix/cleanup[12074]: DA1431965E:
> message-id=<200609180214.k8I2EuNo016264 at mforward2.dtag.de>
> Sep 18 04:15:26 mailrelay postfix/cleanup[13057]: EF90720AD:
> message-id=<200609180214.k8I2EuNo016264 at mforward2.dtag.de>
> Sep 18 04:15:26 mailrelay postfix/smtp[10879]: EF90720AD:
> to=<SPAM-FOUND at OUR-MAILSERVER.mail.com>, relay=10.49.0.7[10.49.0.7],
> delay=1, status=sent (250 2.6.0
> <200609180214.k8I2EuNo016264 at mforward2.dtag.de> Queued mail for delivery)
>
> They are filtered by "message-id", so all these lines above have
> something to do with the message
> "200609180214.k8I2EuNo016264 at mforward2.dtag.de".
>
> The original logfile is about 25 MB big, so I can't post all of the
> lines of course ;-)
>
> Looking at these lines I realized that there are "Queue IDs":
> 755387301
> DA1431965E
> EF90720AD
>
> Filtering the log for these IDs results in the following lines:
>
> Sep 18 02:15:11 mailrelay postfix/smtpd[10841]: 755387301:
> client=unknown[194.25.242.123]
> Sep 18 04:15:22 mailrelay postfix/cleanup[12103]: 755387301:
> message-id=<200609180214.k8I2EuNo016264 at mforward2.dtag.de>
> Sep 18 04:15:22 mailrelay postfix/qmgr[11082]: 755387301:
> from=<sender at mail.net.mx>, size=8152, nrcpt=7 (queue active)
> Sep 18 04:15:25 mailrelay postfix/pipe[11659]: 755387301:
> to=<receiver1 at mail.com>, relay=procmail, delay=14, status=sent (filter)
> Sep 18 04:15:25 mailrelay postfix/pipe[11659]: 755387301:
> to=<receiver2 at mail.com>, relay=procmail, delay=14, status=sent (filter)
> Sep 18 04:15:25 mailrelay postfix/pipe[11659]: 755387301:
> to=<receiver3 at mail.com>, relay=procmail, delay=14, status=sent (filter)
> Sep 18 04:15:25 mailrelay postfix/pipe[11659]: 755387301:
> to=<receiver4 at mail.com>, relay=procmail, delay=14, status=sent (filter)
> Sep 18 04:15:25 mailrelay postfix/qmgr[11082]: 755387301: removed
>
> Sep 18 04:15:25 mailrelay postfix/pickup[13175]: DA1431965E: uid=65534
> from=<nobody>
> Sep 18 04:15:25 mailrelay postfix/cleanup[12074]: DA1431965E:
> message-id=<200609180214.k8I2EuNo016264 at mforward2.dtag.de>
> Sep 18 04:15:25 mailrelay postfix/qmgr[11082]: DA1431965E:
> from=<nobody at OUR-MAILSERVER.mail.com>, size=11074, nrcpt=1 (queue active)
> Sep 18 04:15:26 mailrelay postfix/smtp[11703]: DA1431965E:
> to=<SPAM-FOUND at OUR-MAILSERVER.mail.com>, relay=localhost[127.0.0.1],
> delay=1, status=sent (250 Ok: queued as EF90720AD)
> Sep 18 04:15:26 mailrelay postfix/qmgr[11082]: DA1431965E: removed
>
> Sep 18 04:15:25 mailrelay postfix/smtpd[11704]: EF90720AD:
> client=localhost[127.0.0.1]
> Sep 18 04:15:26 mailrelay postfix/cleanup[13057]: EF90720AD:
> message-id=<200609180214.k8I2EuNo016264 at mforward2.dtag.de>
> Sep 18 04:15:26 mailrelay postfix/smtp[11703]: DA1431965E:
> to=<SPAM-FOUND at OUR-MAILSERVER.mail.com>, relay=localhost[127.0.0.1],
> delay=1, status=sent (250 Ok: queued as EF90720AD)
> Sep 18 04:15:26 mailrelay postfix/qmgr[11082]: EF90720AD:
> from=<nobody at OUR-MAILSERVER.mail.com>, size=11263, nrcpt=1 (queue active)
> Sep 18 04:15:26 mailrelay postfix/smtp[10879]: EF90720AD:
> to=<SPAM-FOUND at OUR-MAILSERVER.mail.com>, relay=10.49.0.7[10.49.0.7],
> delay=1, status=sent (250 2.6.0
> <200609180214.k8I2EuNo016264 at mforward2.dtag.de> Queued mail for delivery)
> Sep 18 04:15:26 mailrelay postfix/qmgr[11082]: EF90720AD: removed
>
> All this work is done with command line and grep...
>
> Is it possible to parse this big logfile only ONCE and extract all this
> info into XML?
>
> Like this:
>
> <message id="200609180214.k8I2EuNo016264 at mforward2.dtag.de">
>    <timestamp>Sep 18 04:15:26</timestamp>
>    <from>sender at mail.net.mx</from>
>    <to>receiver1 at mail.com</to>
>    <to>receiver2 at mail.com</to>
>    <to>receiver3 at mail.com</to>
>    <to>receiver4 at mail.com</to>
>    <queueID>EF90720AD</queueID>
>    <queueID>DA1431965E</queueID>
>    <queueID>755387301</queueID>
>    <spamd>
> 	<score>15</score>
> 	<filtered>yes</filtered>
> 	<sendto>SPAM-FOUND at OUR-MAILSERVER.mail.com</sendto>
>    </spamd>
> </message>
>
> The goal of this is to provide a web interface were we can see if the
> messages were filtered as spam (or deleted by our virus scanner).
>
> Is it possible? Or do I have to scan / parse the file more than once?
> 
> Andi
> 
> -- 
> Mozilla Thunderbird 1.5.0.7
> Arch Linux

More information about the Python-list mailing list