CODEWITHSUNDEEP
bashawk
Sparhawk
Sparhawk

Reputation: 1677

Why can I use FS=x to define awk's field separator only in some situations

There are a few different ways of setting awk's field separator. These include

However, I often use the following: awk '{…}' FS='\t' a.txt (I like it because it's quick to type, and also allows for OFS='\t' in a similar format.)

I've noticed that the field separator won't be set if the last two items are reversed. i.e. awk '{…}' a.txt FS='\t'

What does this syntax mean exactly? Hence, why is the order of the last two items so important?

Upvotes: 1

Views: 235

Answers (2)

William Pursell
William Pursell

Reputation: 212198

awk processes its arguments in order. Consider:

awk 'BEGIN{ print "BEGIN:", a } 
  { print FILENAME":", a }
  END { print "END:", a }
' file1 a=foo file2 a=bar

In this example, awk first parses the script and executes the BEGIN blocks before it even reads its arguments. It then processes file1, then makes the variable assignment a=foo, then process file2, then makes the variable assignment a=bar, and finally processes the END blocks.

The first argument (loosely defining "argument" as elements of the command line that are not flags) passed to awk (assuming the absence of a -f flag) is the script to be executed. The arguments following the script that contain an equals sign are variable assignments that are made after any BEGIN blocks and before input lines are processed for files listed after the assignment. Arguments that do not contain an equal sign are files to be processed. So:

awk '{...}' a.txt FS='\t'

does not work because the argument FS='\t' is a variable assignment that is made after a.txt is processed. Since no files are listed after the assignment, it is never relevant to the processing.

Upvotes: 2

Jotne
Jotne

Reputation: 41446

Last input on awk need to be the file. That is why:

awk '{…}' a.txt FS='\t'  # this does not work



awk '{…}' FS='\t' a.txt  # this is OK



cat a.txt | awk '{…}' FS='\t'  # this is OK too



awk '{…}' FS='\t' a.txt  b.txt c.txt   # multiple file input OK

Upvotes: 0

Related Questions