问题
Is there a way to get awk to return the number of fields that met a field-separator criteria? Say, for instance, my file contains
a b c d
so, awk --field-separator=" " | <something> should return 4
回答1:
The NF variable is set to the total number of fields in the input record. So:
echo "a b c d" | awk --field-separator=" " "{ print NF }"
will display
4
Note, however, that:
echo -e "a b c d\na b" | awk --field-separator=" " "{ print NF }"
will display:
4 2
Hope this helps, and happy awking
回答2:
NF gives the number of fields for a given record:
[]$ echo "a b c d" | gawk '{print NF}'
4
回答3:
If you would like to know the set of all the numbers of fields in a multiline content you can run:
X | awk '{print NF}' | sort -n | uniq
being X a command that outputs content in the standard output: cat, echo, etc. Example:
With file.txt:
a b
b c
c d
e t a
e u
The command cat file.txt | awk '{print NF}' | sort -n | uniq will print:
2
3
And with file2.txt:
a b
b c
c d
e u
The command cat file2.txt | awk '{print NF}' | sort -n | uniq will print:
2
回答4:
awk(1) on FreeBSD does not recognize --field-separator. Use -v instead:
echo "a b c d" | awk -v FS=" " "{ print NF }"
It is a portable, POSIX way to define the field separator.
来源：https://stackoverflow.com/questions/5582405/number-of-fields-returned-by-awk