Sorting in Linux

Question

I have a string "abc" and I am greping it in 15 files in my directory in shell. When I grep "abc" in my 15 files than it returns me the whole line of the files in which its present like this:

abc 0.2  9.0
abc 0.01 8.0
abc 0.06 9.4
abc 0.02 8.7

Now I want this output to be sorted in ascending order according to the second column.

So I wrote the command like this:

grep "abc" *.txt | sort -nr -t_ -k2

but the above command is not working and I don't know why.

Answer 1

Your command is not working because you don't have underscores separating the columns; further, you wanted the data ascending but you told it to sort in reverse (descending) order. Use:

grep "abc" *.txt | sort -n -k 2

Or:

grep "abc" *.txt | sort -k 2n

Note that if there are multiple files, your grep output will be prefixed with a file name. You will have to decide whether that matters. It only screws things up if there are spaces in any of the file names. The -h option to grep suppresses the file names.

Answer 2

I suggest to delete -t_ parameter, because as I see you use spaces as separator, not underscore. After that it works for me:

$ cat t | sort -n -k2
abc 0.01 8.0
abc 0.02 8.7
abc 0.06 9.4
abc 0.2  9.0

Updated: and yes, as @jonathan-leffler said you also should omit -r option for sorting in ascending order.

Answer 3

You can replace your entire script, including the call to grep , with one call to awk

awk '/abc/{a[$2,i++]=$0}END{l=asorti(a,b);for(i=1;i<=l;i++)print a[b[i]]}' *.txt

Example

$ ls *.txt
four.txt  one.txt  three.txt  two.txt

$ cat *.txt
abc 0.02 8.3
foo
abc 0.06 9.4
bar
blah blah
abc 0.2  9.0
blah
abc 0.01 8.0
blah
abc 0.02 8.7
blah blah

$ awk '/abc/{a[$2,i++]=$0}END{l=asorti(a,b);for(i=1;i<=l;i++)print a[b[i]]}' *.txt
abc 0.01 8.0
abc 0.02 8.3
abc 0.02 8.7
abc 0.06 9.4
abc 0.2  9.0

Answer 4

First problem: you've used -t_ to specify the field separator, but the output doesn't contain _ characters.

Next problem: the -k option has two arguments, start field and end field. By omitting the end field, you get end_of_line by default.

I'd suggest writing it like this:

grep "abc" *.txt | sort -nr -k2,2