bash script to sort data by field

lothario · 08-25-2009, 02:39 AM

I have some very large data files that look like this:

"blue","right","mhduei573","0","0"
"pink","left","bvop4967","0","0"
"yellow,"top","zpbm2368","0","0"
"green","right","cdw432lmjt","0","0"
"purple,"left","plmnhy3zd21","0","0"
"white","top","yg36c0wxqa","0","0"
"blue","left","bvop4967","0","0"
"pink","top","yg36c0wxqa","0","0"
"gray,"right","bvop4967","0","0"

Each field has quotes around it and there is a comma between fields.

I would like to sort this data by the 3rd field then the 1st field, using a bash script.
How do I do that?

colucix · 08-25-2009, 02:44 AM

No need for bash script, just a one-liner with the proper options:

Code:

sort -t, -k3 -k1 file

lothario · 08-25-2009, 01:09 PM

sort -t, -k3 -k1 ./a1.txt
works perfectly on this file:

"blue","right","mhduei573","0","0"
"pink","left","bvop4967","0","0"
"yellow,"top","zpbm2368","0","0"
"green","right","cdw432lmjt","0","0"
"purple,"left","plmnhy3zd21","0","0"
"white","top","yg36c0wxqa","0","0"
"blue","left","bvop4967","0","0"
"pink","top","yg36c0wxqa","0","0"
"gray,"right","bvop4967","0","0"

It results in:

"blue","left","bvop4967","0","0"
"gray,"right","bvop4967","0","0"
"pink","left","bvop4967","0","0"
"green","right","cdw432lmjt","0","0"
"blue","right","mhduei573","0","0"
"purple,"left","plmnhy3zd21","0","0"
"pink","top","yg36c0wxqa","0","0"
"white","top","yg36c0wxqa","0","0"
"yellow,"top","zpbm2368","0","0"

Exactly what I needed.

To take it a step further, I need to group the data by the 3rd field.
So that it adds a blank line every time there is a change in the 3rd field.
How do you make it look like this?

"blue","left","bvop4967","0","0"
"gray,"right","bvop4967","0","0"
"pink","left","bvop4967","0","0"

"green","right","cdw432lmjt","0","0"

"blue","right","mhduei573","0","0"

"purple,"left","plmnhy3zd21","0","0"

"pink","top","yg36c0wxqa","0","0"
"white","top","yg36c0wxqa","0","0"

"yellow,"top","zpbm2368","0","0"

Tinkster · 08-25-2009, 01:13 PM

Code:

sort -t, -k3 -k1 file |awk -F, '{if($3 != hold){print ""}; hold=$3;print $0}'

I'm too lazy to strip the first empty line ;}

Cheers,
Tink

colucix · 08-26-2009, 02:23 AM

Quote:

Originally Posted by Tinkster

I'm too lazy to strip the first empty line ;}

Just a slight modification: NR > 1

Code:

sort -t, -k3 -k1 file | awk -F, 'NR>1{if($3 != hold)print ""}{hold=$3; print}'