Starting with a 109,000 row LibreOffice Calc file containing many groups of
identical strings separated by blank rows,
how can I re-order this file with the largest groups at the top, still
separated by blank rows ?
Here's how that original file was created:
First, sort the original 159,000 row file on the third (hostname) column:
>> sort -k3 HNs.bst.lt.visitors03.txt > HNs.bst.lt.sorted03.txt
The original file had two fields separated by a colon(:) that I converted to
tabs with Leafpad in ~30 minutes.
It now has the Current Visitor filename in the first column, the occurrence
count in the second column, and
the master file's hostnames in the third column, the one on which this file
has to be sorted.
Second, gather the duplicates into bunches separated by blank lines,
retaining the original general sorted sequence:
>> time awk 'NR >= 2 { print $1, $2, $3 }' 'HNs.bst.lt.sorted03.txt' | uniq
-f 2 --all-repeated=separate | awk '{ print $1 "\t" $2 "\t" $3 }' >
HNs.bst.lt.Duplicates03.txt
This makes a very pretty .ods file that's slightly larger than the original
file, but there's so much there that it
would be tedious to rank the most-often-repeated Recent Visitors' hostnames
by hand.
How can I re-sort the .ods file so that the longest identical-hostname lists
are at the top ? I want to retain the associations
with the original domains' Recent Visitor data. This is a task which has to
be repeated many times ...
BTW, the longest identical-hostname list that I found visually in the .ods
file was 42 out of 45 original Recent Visitor lists.
All the original data is freely available on the Internet.
George Langford
