Custom sorting in Unix

Question

I want to sort the following data in a particular order. I tried sort in different way but couldn't find any solution. please help. I am a newbie in Unix.:(

Data:-

method1:entry:2013.09.18.19.18.30
method1:exit:2013.09.18.19.18.30
method2:entry:2013.09.18.19.18.30
method2:exit:2013.09.18.19.18.30
method3:entry:2013.09.18.19.18.30
method4:entry:2013.09.18.19.18.30
method4:exit:2013.09.18.19.18.30
method1:entry:2013.09.18.19.18.30
method1:exit:2013.09.18.19.18.30
method3:exit:2013.09.18.19.18.30
method3:entry:2013.09.18.19.18.30
method5:entry:2013.09.18.19.18.30
method5:exit:2013.09.18.19.18.30
method3:exit:2013.09.18.19.18.30

Desired output:-

method1:entry:2013.09.18.19.18.30
method1:exit:2013.09.18.19.18.30
method1:entry:2013.09.18.19.18.30
method1:exit:2013.09.18.19.18.30
method2:entry:2013.09.18.19.18.30
method2:exit:2013.09.18.19.18.30
method3:entry:2013.09.18.19.18.30
method3:exit:2013.09.18.19.18.30
method3:entry:2013.09.18.19.18.30
method3:exit:2013.09.18.19.18.30
method4:entry:2013.09.18.19.18.30
method4:exit:2013.09.18.19.18.30
method5:entry:2013.09.18.19.18.30
method5:exit:2013.09.18.19.18.30

The sorting should be based on method name and 'entry-exit' occurrence.

How do you know which method1:exit should be placed after a method1:entry? Both method1:exits are identical. The same is true for method1:entry itself. You just want to match pairs of entry and exits which have same method name? — zaerymoghaddam
– zaerymoghaddam, Commented Oct 3, 2013 at 7:39
@moghaddam: Yes, I want to just pair entry and exit with same method name. — goosebump
– goosebump, Commented Oct 3, 2013 at 7:44
I Think there is no way to do this directly in shell using commands and utilities (like sort, set or awk). You have to write some python or perl scripts to do this — zaerymoghaddam
– zaerymoghaddam, Commented Oct 3, 2013 at 15:51

chepner · Accepted Answer · 2013-10-03 12:08:27Z

2

It appears you simply want to sort by method name, which is the first colon-delimited field.

sort -t: -s -k1,1 file.txt

The -s flag (stable sort) prevents sort from modifying the relative order of lines with the same first field.

answered Oct 3, 2013 at 12:08

chepner

538k77 gold badges594 silver badges746 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

tripleee · Accepted Answer · 2022-01-13 05:57:29Z

1

Try this :

sed -e 's/:/ /g' file.txt | sort |
awk 'BEGIN { var_entry="entry"; var_exit="exit"; flag="entry" }
    { if (flag == $2 && var_entry ==$2 ){
        i = 0; flag=var_exit; }
      else if (flag == $2 && var_exit == $2 ){
        i = 0; flag=var_entry; };
      i++ ; print i, $0 }' |
sort -t" " -k 2,2 -k 1,1 | sed 's/^[0-9]* //g'

The logic behind this is :

sed -e 's/:/ /g replaces : with a space so the delimiters are consistent.
sort simply sorts on method1 column.
awk step appends another column so that we can sort on that column so that we have a pattern like entry exit for matching method1, output is :

1 method1 entry 2013.09.18.19.18.30
2 method1 entry 2013.09.18.19.18.30
1 method1 exit 2013.09.18.19.18.30
2 method1 exit 2013.09.18.19.18.30
1 method2 entry 2013.09.18.19.18.30
1 method2 exit 2013.09.18.19.18.30
1 method3 entry 2013.09.18.19.18.30
2 method3 entry 2013.09.18.19.18.30
1 method3 exit 2013.09.18.19.18.30
2 method3 exit 2013.09.18.19.18.30
1 method4 entry 2013.09.18.19.18.30
1 method4 exit 2013.09.18.19.18.30
1 method5 entry 2013.09.18.19.18.30
1 method5 exit 2013.09.18.19.18.30
sort -t" " -k 2,2 -k 1,1 : then we sort on method1 column (2nd column) and if there are conflicts we resolve them on the newly added column viz. 1st column. output is :

1 method1 entry 2013.09.18.19.18.30
1 method1 exit 2013.09.18.19.18.30
2 method1 entry 2013.09.18.19.18.30
2 method1 exit 2013.09.18.19.18.30
1 method2 entry 2013.09.18.19.18.30
1 method2 exit 2013.09.18.19.18.30
1 method3 entry 2013.09.18.19.18.30
1 method3 exit 2013.09.18.19.18.30
2 method3 entry 2013.09.18.19.18.30
2 method3 exit 2013.09.18.19.18.30
1 method4 entry 2013.09.18.19.18.30
1 method4 exit 2013.09.18.19.18.30
1 method5 entry 2013.09.18.19.18.30
1 method5 exit 2013.09.18.19.18.30
sed 's/^[0-9]* //g' : we remove the extra column which was created.

edited Jan 13, 2022 at 5:57

tripleee

192k37 gold badges318 silver badges367 bronze badges

answered Oct 3, 2013 at 8:33

Ashish Gaur

2,0403 gold badges20 silver badges34 bronze badges

4 Comments

goosebump Over a year ago

@snyder.The output is coming as a single line in your code. Is there any way can I get the output in different lines? wc -l filename is returning 1.

Ashish Gaur Over a year ago

@anz try this cat file.txt | sed 's/method/\nmetthod/g'

goosebump Over a year ago

method name won't be constant, it can be some other name also.

Ashish Gaur Over a year ago

@anz I have tested the above command on my shell, it gives newline. I am using bash, which shell are you using ?

Collectives™ on Stack Overflow

Custom sorting in Unix

2 Answers 2

Comments

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related