Please don't bump.

-F tells awk what to use as a separator -- where fields stop and start. It's a regular expression. Any single , . ? \ ! will stop the current record and start the next.

The first for-loop:

Code:

for (i=1;i<=NF;i++) {arr[$i]++}

loops through i=1, i=2, i=3, ..., i=NF, incrementing arr[$i] each time.

Note the red $, it's important: if i is 1, $i becomes field 1.

So if you had fields A B C D E, it'd do arr["A"]++, then arr["B"]++, then arr["C"]++, and so forth.

NF is a special awk variable for "number of fields". Fields go from 1 to NF inclusive. You can use $NF to mean "the contents of the last field".

Another useful special variable is NR, the 'number of records'. Usually this'd be the line number, but awk can be used to operate on larger blocks than lines, so they consider each thing a 'record' instead.

Already explained NF.

I'll explain about awk codeblocks in general.

The usual {} global code block gets run once for every line, with updated values of NF, NR, $1,$2,... and so forth each time.

You're allowed to have more than one of them.

Code:

echo asdf | awk '{ print }
{ print }'

will print 'asdf' twice.

You can put statements before the block which control when they are run, and statements after the blocks which control when they print. The entire line is printed when you print that way. 'echo asdf | awk '{} 1' will print 'asdf' since the expression '1' is considered always true. You could put 'NR=5' in it instead so it'd only print the fifth line.

END {} tells a code block to run once, after all input has been processed. Useful for printing out things that've been stored but not printed earlier.

BEGIN {} tells a code block to run before any input has been processed. Useful for setting up variables.

/regex/ {} tells a code block to only run when the current line matches the given regular expression.


You can mix and match them however you like, i.e.

Code:

# Only run the code block if NR is 1.
# Only print the line if it matches /sd/.
echo asdf | awk 'NR==1 {} /sd/'