Sort command

djfog · 07-03-2008, 01:31 AM

Hi to all.My question is how I can use sort with two references.Example I want to sort a file named Mail.txt by surname and name,if I have more than a time the same surname.Thanks in advance..

Mr. C. · 07-03-2008, 01:36 AM

Show examples of what the text input looks like. Sounds like you want two sort keys, a primary based on surname, and secondary based on first name.

jschiwal · 07-03-2008, 01:39 AM

The "info sort" manual has examples using two or more keys.

djfog · 07-03-2008, 01:51 AM

This is the text:

Code:

George Pappas 12136 Peristeri 5757675
Nick Nikolaoy 12232 Aigaleo 5314555
John Ioannoy 13222 Athens 3245890
Nick Pappas 11223 Aigaleo 5324123
George Georgioy 11132 Athens 3245678
Helen Georgioy 12136 Peristeri 5748456
Nick Pappas 11223 Aigaleo 5324123
John Ioannoy 13222 Athens 3245890
Helen Thanoy 11132 N.Smyrni 9718345

where the 2nd column is the surname and the first the name.I'm from Greece as you see.

Mr. C. · 07-03-2008, 01:54 AM

Use the -k option to indicate which field contains the key. Since your data is space-separated, which is the default field for sort, the sort is trivial:

sort -k2 -k1

Χαίρετε

djfog · 07-03-2008, 02:05 AM

I did not quite understand.I used sort -2 -1 Mail.txt but I get an error message saying invalid option --2.Could you be more specific..?

Έλληνας..?

jschiwal · 07-03-2008, 02:14 AM

Notice the letter "k" in the example.

Mr. C. · 07-03-2008, 02:14 AM

You need the -k to preceed the key number:

Code:

$ cat in3
George Pappas 12136 Peristeri 5757675
Nick Nikolaoy 12232 Aigaleo 5314555
John Ioannoy 13222 Athens 3245890
Nick Pappas 11223 Aigaleo 5324123
George Georgioy 11132 Athens 3245678
Helen Georgioy 12136 Peristeri 5748456
Nick Pappas 11223 Aigaleo 5324123
John Ioannoy 13222 Athens 3245890
Helen Thanoy 11132 N.Smyrni 9718345

$ sort -k2 -k1 in3
George Georgioy 11132 Athens 3245678
Helen Georgioy 12136 Peristeri 5748456
John Ioannoy 13222 Athens 3245890
John Ioannoy 13222 Athens 3245890
Nick Nikolaoy 12232 Aigaleo 5314555
Nick Pappas 11223 Aigaleo 5324123
Nick Pappas 11223 Aigaleo 5324123
George Pappas 12136 Peristeri 5757675
Helen Thanoy 11132 N.Smyrni 9718345

Αμερικάνος, ανόητος.

djfog · 07-03-2008, 02:16 AM

Thanks.It really works...

Mr. C. · 07-03-2008, 02:17 AM

By the way, that was me calling myself a silly american!

djfog · 07-03-2008, 02:22 AM

And how can I erase the double names..??

Mr. C. · 07-03-2008, 02:28 AM

pipe everything to uniq

sort ... | uniq

that's all there is to it.

djfog · 07-03-2008, 02:32 AM

Thanks for your help.I can now suppose that if I want to sort surname and the 5 digit mail code the sort command would be like this:
sort -k3 -k2 Mail.txt
Am I right..?
Thanks again..?

Γιατί ανόητος Αμερικάνος??

Mr. C. · 07-03-2008, 02:35 AM

Well, numeric sorts are different than character sorts. You add options to the end of the sort key to change the type of sort. For example, just sorting numerically on the last field:

sort -k5n

Note the addition of "n" to the -k5 key field.

έξοχος ερώτημα !

djfog · 07-03-2008, 02:40 AM

I just noticed something...the command

Quote:

sort -k2 -k1

does not seem to work properly as George Pappas is under Nick Pappas and I want the opposite.Any ideas..?