I am familiar with comparing numbers in a row using bash, but what if I wanted to compare columns? Like if i had a file with
4 2 5 7 6 1 3 8 And I want to find the largest out of each column
6 2 5 8 How can I do this using awk?
- Is the number of columns fixed?unxnut– unxnut2014-03-23 19:44:48 +00:00Commented Mar 23, 2014 at 19:44
- There's surely a bash/awk solution for your problem. That being said, why would you tackle such a problem in bash in the first place?! It's just the wrong language for anything but simple 10-line scripts. Sure, you can write more complex code (I did), but it's so much harder than using Python, Perl, TCL, Lua, etc.Christian Aichinger– Christian Aichinger2014-03-23 19:52:08 +00:00Commented Mar 23, 2014 at 19:52
- @unxnut no, the number of columns is not fixedTree55Topz– Tree55Topz2014-03-23 19:56:41 +00:00Commented Mar 23, 2014 at 19:56
- Is it the same number of fields in each record?unxnut– unxnut2014-03-23 19:58:01 +00:00Commented Mar 23, 2014 at 19:58
Add a comment |
2 Answers
Assuming that your data is in a file called "data":
$ awk '{for(j=1;j<=NF;++j){max[j]=(max[j]>$j)?max[j]:$j};mNF=mNF>NF?mNF:NF} END{for(j=1;j<=mNF;++j)printf " " max[j]; print "" }' data 6 2 5 8 Or, with different output formatting:
$ awk '{for(j=1;j<=NF;++j){max[j]=(max[j]>$j)?max[j]:$j};mNF=mNF>NF?mNF:NF} END{for(j=1;j<=mNF;++j)print "max of column " j "=" max[j]}' data max of column 1=6 max of column 2=2 max of column 3=5 max of column 4=8 The body of the above awk program consists of:
{ for(j=1;j<=NF;++j) {max[j]=(max[j]>$j)?max[j]:$j};mNF=mNF>NF?mNF:NF } This loop is run for every line in the input file. The loop runs through every column (field) in that input line from 1 to the number of fields (NF). For each column, it checks to see if the current value is greater than the previous maximum. If it is, it updates the value in the max array.
5 Comments
jaypal singh
+1: OP can use
END{for(j=1;j<=NF;++j)printf "%s%s",max[j],(j==NF?"\n":FS)}' to print according to the desired output.
glenn jackman
@jaypal, or even
END {for (j=1;j<=NF;j++) printf "%s ",max[j]; print ""} -- a bit less trickyjaypal singh
@glennjackman Absolutely .. I am just more inclined more towards job security!
;).
BMW
you need fix it with no-fix columns.
awk '{for(j=1;j<=NF;++j)max[j]=(max[j]>$j)?max[j]:$j;mNF=mNF>NF?mNF:NF} END{for(j=1;j<=mNF;++j) printf max[j] FS; print RS }' file
John1024
@BMW Thanks! That is a nice generalization. I added it to the answer.
Bash solution using arrays:
max=(); while read -a line ; do for ((i=0; i<${#line[@]} ; i++)) ; do ((line[i]>max[i])) && max[i]=${line[i]} done done < input echo "${max[@]}" 