sort command is used to sort a file, arranging the records in a particular order. By default, the sort command sorts file assuming the contents are ascii. Using options in sort command, it can also be used to sort numerically. Let us discuss it with some examples:
File with Ascii data:
Let us consider a file with the following contents:
$ cat file Unix Linux Solaris AIX Linux HPUX
1. sort simply sorts the file in alphabetical order:
$ sort file AIX HPUX Linux Linux Solaris UnixAll records are sorted alphabetically.
2. sort removes the duplicates using the -u option:
$ sort -u file AIX HPUX Linux Solaris UnixThe duplicate 'Linux' record got removed. '-u' option removes all the duplicate records in the file. Even if the file have had 10 'Linux' records, with -u option, only the first record is retained.
File with numbers:
Let us consider a file with numbers:
$ cat file 20 19 5 49 200
3. The default sort 'might' give incorrect result on a file containing numbers:
$ sort file 19 20 200 49 5In the above result, 200 got placed immediately below 20, not at the end which is incorrect. This is because the sort did ASCII sort. If the file had not contained '200', the default sort would have given proper result. However, it is incorrect to sort a numerical file in this way since the sorting logic is incorrect.
4. To sort a file numericallly:
$ sort -n file 5 19 20 49 200-n option can sort the decimal numbers as well.
5. sort file numerically in reverse order:
$ sort -nr file 200 49 20 19 5'r' option does a reverse sort.
Let us consider examples with multiple files, say file1 and file2, containing numbers:
$ cat file1 20 19 5 49 200
$ cat file2 25 18 5 48 200
6. sort can sort multiple files as well.
$ sort -n file1 file2 5 5 18 19 20 25 48 49 200 200The result of sort with multiple files will be a sorted and merged output of the multiple files.
7. Sort, merge and remove duplicates:
$ sort -nu file1 file2 5 18 19 20 25 48 49 200-u option becomes more handy in case of multiple files. With this, the output is now sorted, merged and without duplicate records.
Files with multiple fields and delimiter:
Let us consider a file with multiple fields:
$ cat file Linux,20 Unix,30 AIX,25 Linux,25 Solaris,10 HPUX,100
8. sorting a file containing multiple fields:
$ sort file AIX,25 HPUX,100 Linux,20 Linux,25 Solaris,10 Unix,30As shown above, the file got sorted on the 1st field, by default.
9. sort file on the basis of 1st field:
$ sort -t"," -k1,1 file AIX,25 HPUX,100 Linux,20 Linux,25 Solaris,10 Unix,30This is being more explicit. '-t' option is used to provide the delimiter in case of files with delimiter. '-k' is used to specify the keys on the basis of which the sorting has to be done. The format of '-k' is : '-km,n' where m is the starting key and n is the ending key. In other words, sort can be used to sort on a range of fields just like how the group by in sql does. In our case, since the sorting is on the 1st field alone, we speciy '1,1'. Similarly, if the sorting is to be done on the basis of first 3 fields, it will be: '-k 1,3'.
Note: For a file which has fields delimited by a space or a tab, there is no need to specify the "-t" option since the white space is the delimiter by default in sort.
10. sorting file on the basis of the 2nd field:
$ sort -t"," -k2,2 file Solaris,10 HPUX,100 Linux,20 AIX,25 Linux,25 Unix,30
11. sorting file on the basis of 2nd field , numerically:
$ sort -t"," -k2n,2 file Solaris,10 Linux,20 AIX,25 Linux,25 Unix,30 HPUX,100
12. Remove duplicates from the file based on 1st field:
$ sort -t"," -k1,1 -u file AIX,25 HPUX,100 Linux,20 Solaris,10 Unix,30The duplicate Linux record got removed. Keep in mind, the command "sort -u file" would not have worked here becuase both the 'Linux' records are not same, the values were different. However, in the above, sort is told to remove the duplicates based on the 1st key, and hence the duplicate 'Linux' record got removed. According to sort, in case of a group of similar records, except the first one, the rest are considered duplicate.
13. Sort the file numerically on the 2nd field in reverse order:
$ sort -t"," -k2nr,2 file HPUX,100 Unix,30 AIX,25 Linux,25 Linux,20 Solaris,10
14. sort the file alphabetically on the 1st field, numerically on the 2nd field:
$ sort -t"," -k1,1 -k2n,2 file AIX,25 HPUX,100 Linux,20 Linux,25 Solaris,10 Unix,30
15. sort a file based on the 1st and 2nd field, and numerically on 3rd field on a file containing 5 columns:
$ sort -t"," -k1,2 -k3n,3 file