Linux SED command help

[Witlof]

Hello,

Hopefully someone can help me. I'm having troubles with splitting files. I got lots of logfiles with names like YYYYMMDD.log with the following content:

2009-03-05 07:35:11,<IP-ADDRESS>,<EMAIL-ADDRESS>,26/8-22-5-19-23,36692479174,1,1,91010507
2009-03-05 07:35:11,<IP-ADDRESS>,<EMAIL-ADDRESS>,26/8-10-9-26-5,108531796850,1,1,91010507
2009-03-05 07:35:11,<IP-ADDRESS>,<EMAIL-ADDRESS>,26/8-10-9-26-13,78524645570,1,1,91010507
2009-03-05 07:36:31,<IP-ADDRESS>,<EMAIL-ADDRESS>,27/4-23-70-9-10,82631681819,1,1,91030509
2009-03-05 07:46:32,<IP-ADDRESS>,<EMAIL-ADDRESS>,27/4-23-58-16-15,134007486202,46,17,91030509
2009-03-05 07:46:32,<IP-ADDRESS>,<EMAIL-ADDRESS>,27/4-23-70-10-9,75059663011,1,1,91030509

I need to split these logfiles into 2 logfiles. 1 with the lines containing ",26/" and the other one with containing ",27/".

For those who wants to know why I need this, It's a Distributed.net personal proxy logfile. We want to split project-stats, but it's to much work to do this by hand. So hopefully someone can help me. I know it's possible with the SED command, but don't know how to use it.

[Witlof]

Almost solved it, but how can I do this automatic?

sed -n '/,26\//p' <DATE>.log > 26-<DATE>.log
sed -n '/,27\//p' <DATE>.log > 27-<DATE>.log

Sorry for not using edit....

[Witlof]

After a long day searching on the web, I almost got it. But it won't work on multiple files.

This command works for one file:
sed -n '/,26\//p' 20090303.log > ogr26-20090303.log

But this isn't working:
sed -n '/,26\//p' *.log > ogr26-*.log

I'm getting one big file named ogr26-*.log instead of date seperated. What am I doing wrong?

[sideways]

sed is the wrong tool, use grep

grep ",26/" [file list] > 26.log

grep ",27/" [file list] > 27.log

[Witlof]

Thanks for your reply, but grep does the same thing. I don't need the results in 1 file, but like the original seperated by day. So:

20090401.log -> lines with ",26/" -> 26-20090401.log
-> lines with ",27/" -> 27-20090401.log

20090402.log -> lines with ",26/" -> 26-20090402.log
-> lines with ",27/" -> 27-20090402.log

[Witlof]

I did it \o/

for i in $( ls ); do
  sed -n '/,26\//p' $i > ogr26/$i
done

Problem solved

[sideways]

it's still the wrong tool, and you have a Useless use of ls

for i in *; do grep ",26/" $i > ogr26/$i; done

[creeping death]

what sideways gave should give you two files...(notice the stdout re-direction to 27.log and 26.log)

sed is a stream editor, while grep is for pattern matching...grep seems more appropriate here...but anywho...

good that you solved the problem on your own and we nice enough to share your solution.

[sideways]

btw, don't run those commands in a directory containing other files, it will empty them. To be safe you should make the ls command useful with something like

for i in $(ls ogr26*); do ...

(which also qualifies for a "Useless use of ls" award, simpler is: 'for i in ogr26*; do ...' )