There are quite a few non-printable ascii characters scattered throughout the log files which cause Windows programs to malfunction when processing or parsing the log files.  I write a program that parses the SpamFilter logfiles and compiles into a database where each row is a complete record on each email processed.  The non printable codes scattered throughout SpamFilter log files have been very problematic.  There are even carriage return line feeds that show up in the middle of log entries!

For those that need a quick reference on ascii codes:

http://www.cplusplus.com/doc/papers/ascii.html

Questions-

Are the SawMill users having problems with log processing due to non printable ascii characters too?  Typically, once a parser runs into certain codes it thinks it's finished so the symptom is that your entire log files don't get parsed.

Is this only a problem on Windows boxes?  Do SpamFilter logs also contain non-printable characters when running in *nix environments? 

DEC(26) aka HEX(1A) is frequent. Here is the text representation that you can copy and search by:



These chars come from returned values from MAPS and SPF queries.

I've seen just about every code DEC 1-32 scatter throughout, including CR/LF's in the middle of SPF returned values.

On Windows boxes the unprintable chars show up as little boxes like the above. I don't have any *nix boxes to see what the results are like on those.

Not causing any SpamFilter malfunctions.  

This is something to be concerned about if you are using a log file parser like SawMill or are parsing the log files in your own program, like I am.

The problems begin when you try to parse the log files.   If you saw any symptoms it would be in your parser.  You're using SawMill, right?  So, if SawMill is stripping the nonprintable characters already, you would see few problems. The only problem is when a carriage return/line feed is included, because I don't know you can strip those out of the log file before processing.  If SawMill is not stripping out those characters from the log file, then it's possible that you are not getting complete log file processing by SawMill.

I'm now stripping all non-printable characters out of the entire log file before processing using the program I mentioned where I parse into a SQL database that includes columns Server, Recipient Domain, Date, Time, Full Date & Time, DOW, Month, Thread ID, Source IP Address, Source Host Name, Country, Sender Email Address, Recipient Email Addres, Action Taken, Reason, Message, Completed Time, Keywords, SPF Checked for each individual email processed.