Hi, i neeed to sort a file which has lines like 310494550615007O ptt 310494550615008P ptt 310494550615007P ptt 310494550615008Q ptt 310494550615007P ptt 310494550615007Q ptt 310494550615008O ptt the need to sort from 1.1 to 1.15 and then after that sort the with the char(16i.e., O|P|Q)but in reverse alphabetic order...the o/p should be like, 310494550615007Q ptt 310494550615007P ptt 310494550615007P ptt 310494550615007O ptt 310494550615008Q ptt 310494550615008P ptt 310494550615008O ptt can anybody suggest how to do that... can i use sort with loops ... how to sort among a set of lines in the same file? is there any such option Please suggest its very urgent thanks, Vijay

Gosh, I was thinking that the reverse flag could be specified on a key by key basis. Until you find something better, following is an awk solution that inputs a sorted file, and it just reverses each group of lines.  It stores a group into array slots 1 thru n, then pulls them out of the array in reverse order. sort file.in | awk '\ BEGIN { getline holdkey=substr($0,1,15) lbuf[1]=$0 top=1 } function flush_array() { #print "flushing group: " holdkey for (k=top;k>0;k--)    print lbuf[k] top=0 } { key=substr($0,1,15) if (key!=holdkey)    {flush_array()     holdkey=key} top++ lbuf[top]=$0 } END {flush_array()}' > file.out

perl -e 'print map{$_->[0]} sort{$a->[1] <=> $b->[1] || $b->[2] cmp $a->[2]} map{[$_, /(\d+)/, uc($_)]} <>;' file.txt

It might be easier to understand if it's split across multiple lines. perl -e '\ print map{ $_->[0] } sort{ $a->[1] <=> $b->[1] || $b->[2] cmp $a->[2] } map{[$_, /(\d+)/, uc($_)]} <>;' file.txt