Question

I run a minecraft server and like to periodically clean up my log files. I’ve developed a bunch of regex commands that I am able to use in gedit or notepad++, and they work perfectly, but I’d like to be able to automate the process.

The essential file-cleaning command would look like this:

cat server.log | sed -e 's/REGEXTODELETE//g' > server.log

but I’m having trouble getting the regex commands to translate properly to SED. I’m using sed on a CentOS6.3 box.

Here’s an example of a command that works in notepad++:

^[0-9]+-[0-9]+-[0-9]+ [0-9]+:[0-9]+:[0-9]+ \[INFO\] Connection reset$

But when I enter it into my bash script:

cat server.clean.log | sed -e 's/^[0-9]*-[0-9]*-[0-9]* [0-9]*:[0-9]*:[0-9]* \[INFO\] Connection reset$//g' > server.clean.log

it empties the whole file. I’ve looked all over, and I suspect I’m having an issue with whitespace, but after about an hour of searching I’m stuck. Any help is appreciated, and I can provide more examples.

Answer 1

The problem is you are overwriting the input file before it is being read.

cat server.clean.log | sed -e 's/^[0-9]*-[0-9]*-[0-9]* [0-9]*:[0-9]*:[0-9]* \[INFO\]  Connection reset$//g' > server.clean.log

Should be something like

cat server.log | sed -e 's/^[0-9]*-[0-9]*-[0-9]* [0-9]*:[0-9]*:[0-9]* \[INFO\]  Connection reset$//g' > server.clean.log

In general,

program <infile >infile

will not work (reliably) since infile will get truncated before program gets a chance to read it