May 12, 2022    Share on: Twitter | Facebook | HackerNews | Reddit


Validate CSV number of fields:

awk 'BEGIN{FS=",";x=1}{x++}NF!=3{print "wrong number of fields:" NF " in line:" x; exit}' valid.csv

test data:

echo "a,b,c\n1,2,3\n11,12,13,14\n21,22,23" > invalid.csv
echo "a,b,c\n1,2,3\n11,12,13\n21,22,23" > valid.csv

Other scripts (inspiration)

awk 'BEGIN{FS=OFS=","} NF!=20{print "not enough fields"; exit}!($1~/^[0-9]$/) {print "1st field invalid"; exit}' file.csv
awk 'BEGIN{FS=","}!n{n=NF}n!=NF{failed=1;exit}END{exit !failed}'

Skip the first row, print comma separated

awk -F "^" 'NR>1 {print $1", "$2}' xaa

display 80 columns from the first two rows

gawk -F',' 'NR<10 {for (i=1;i<=80;i++){printf "%d ", $i}print""}' file.txt


sum column

awk -F',' '{sum+=$57;} END{print sum;}' file.txt

sum along rows

awk -F ',' '{sum=0; for (i=1; i<=NF; i++) { sum+= $i } print sum}' file.txt

NOTE: When the processing speed is critical, in many cases it is possible to replace AWK with a faster alternative MAWK - as in the famous example: Command-line Tools can be 235x Faster than your Hadoop Cluster - Adam Drake However, there is a warning that MAWK contains some bugs (not sure if fixed at the moment of writing).