Difference between revisions of "Awk"

From Noah.org
Jump to: navigation, search
 
Line 31: Line 31:
 
<pre>
 
<pre>
 
history | awk {'print $2'} | sort | uniq -c | sort -k1 -n
 
history | awk {'print $2'} | sort | uniq -c | sort -k1 -n
 +
</pre>
 +
 +
If you have $HISTTIMEFORMAT set then use this instead:
 +
 +
<pre>
 +
history | awk {'print $3'} | sort | uniq -c | sort -k1 -n
 
</pre>
 
</pre>
  

Latest revision as of 09:11, 1 October 2008

About the most I ever use awk for is to print fields I want from a delimited file. For example, to print field #2 from a file (assuming white space between columns):

 awk '{print $2;}' file

If the fields have other delimiters such as '/' then use the -F option to set the delimiter. For example to print the first directory from a list of file paths use this:

 awk -F '/' '{print $1;}' file

So if I had a list of filenames such as:

 subversion-1.3.2/README
 subversion-1.3.2/configure.in
 subversion-1.3.2/TRANSLATING
 subversion-1.3.2/subversion
 subversion-1.3.2/subversion/libsvn_fs_base
 subversion-1.3.2/subversion/libsvn_fs_base/bdb

Running that list through the example would produce

 subversion-1.3.2
 subversion-1.3.2
 subversion-1.3.2
 subversion-1.3.2
 subversion-1.3.2
 subversion-1.3.2

Once you get much more complicated than this then you probably want a real scripting language. Average of column 3:

 awk '{total+=$3;count++;}END{print total/count;}' file

show CLI usage statistics

This is probably useless, but interesting. This counts and sorts the commands as seen by the Bash CLI history.

history | awk {'print $2'} | sort | uniq -c | sort -k1 -n

If you have $HISTTIMEFORMAT set then use this instead:

history | awk {'print $3'} | sort | uniq -c | sort -k1 -n

Gives something like this:

...
     65 chmod
     71 svn
     84 df
    116 rm
    132 sudo
    390 ls
    396 cd
    631 vim
   1202 exit
   1359 mutt

Kill All by grep

This is similar to `killall` except that it matches a process by pattern -- this can be more dangerous.

alias ka='kill -9 `ps auxww | grep -i $1 | sed \$d`'