sort / cut question

Shell Programming and Scripting
1

Hi All,

I have a small problem, hope you can help me out here.

I have a file that contains the same format of lines in 99% of the cases.
906516 XYZ.NNN V 0000 20070711164648 userID1 userID2 hostname 20070711164641

There are unfortunately several lines with these format:
906516 XYZ V 0000 20070711164647 userID1 userID2 hostname 20070711164641

The difference is in the 2nd coloumn. Sometimes its length is 3 , sometimes 7 digits long. Thus the number of delimiters between 2nd and 3rd column are also varied. Imagine the two lines as "V" in coloumn 3 would be in the same line...

My scenario is:
cut out the 1st(the number) and 5th (timestamp) coloumn and sort the result in reverse order using the timestamp column as key.

My command is:
cat filename |cut -f 1,15 -d " " |sort -r -k 2,2

Obviously this cannot work properly because sometimes the 5th column does not start at the 15th " " delimiter due the above mentioned reasons.

Could you guys let me know a solution I can carry out this scenario? I checked man for cut but could not find an option of using sophisticated delimiters, etc...

Thanks a lot!
BearCheese

2 
awk '{ print $5, $1}' filename | sort -rn > newfile
3

Thank you!
Now I have a file with the content like this:
20070711172841 905994
20070711172822 905994
20070711171431 906608
20070711171412 906608
20070711171142 905994
20070711171140 905994
20070711171139 905994
20070711164813 905994

I would like to keep only one line where the second column is the same. But that line should be the most lastest entry (timestamp in 1st coloumn).
I tried command:
sort -u -k 2,2
but this keeps the oldest timestamp :frowning:

I tought that -u suppress everything but the first match. The first match (that should be kept) should be :
20070711172841 905994

but my command magically keeps the oldest line:
20070711164813 905994

Could you please advise on why -u does not work as anticipated?

Thank you!
BearCheese

4 
mLastValue=''
mLastDate=''
sort -k2 -k1 input_file | \
while read mDate mValue
do
  if [ "$mValue" != "$mLastValue" ]; then
    if [ "$mLastValue" != "" ]; then
      echo $mLastDate" "$mLastValue
    fi
  fi
  mLastValue=$mValue
  mLastDate=$mDate
done
if [ "$mLastValue" != "" ]; then
  echo $mLastDate" "$mLastValue
fi
5 
awk '{ print $5, $1}' filename | sort -rn | \
     awk '
     { arr[$2]=$0 }      
     END { for ( item in arr) { print arr[item]}
     } ' filename > newfile
6 
awk '!arr[$2] { arr[$2] = $0 } 
       END { for ( item in arr ) { print arr[item] }
       }' file > newfile