[SOLVED] bash suggestions to convert horizontal columns to vertical columns

Guyverix · 01-21-2013, 08:44 PM

I am attempting what I ~thought~ was kind of straight forward. I am pulling column data from a Windows host via bash, and the results are good but the format is not what I am trying for. Here is what I get:

AvgDiskBytesPerRead|AvgDiskBytesPerTransfer|AvgDiskBytesPerWrite|AvgDiskQueueLength|AvgDiskReadQueue Length|AvgDisksecPerRead|AvgDisksecPerTransfer|AvgDisksecPerWrite|AvgDiskWriteQueueLength|(null)|Cur rentDiskQueueLength|Description|DiskBytesPersec|DiskReadBytesPersec|DiskReadsPersec|DiskTransfersPer sec|DiskWriteBytesPersec|DiskWritesPersec|FreeMegabytes|Frequency_Object|Frequency_PerfTime|Frequenc y_Sys100NS|Name|PercentDiskReadTime|PercentDiskTime|PercentDiskWriteTime|PercentFreeSpace|PercentIdl eTime|SplitIOPerSec|Timestamp_Object|Timestamp_PerfTime|Timestamp_Sys100NS
0|0|0|0|0|0|0|0|0|(null)|0|(null)|0|0|0|0|0|0|18196|0|0|0|C:|0|0|0|44|100|0|0|0|0
0|0|0|0|0|0|0|0|0|(null)|0|(null)|0|0|0|0|0|0|132411|0|0|0|D:|0|0|0|80|100|0|0|0|0
0|0|0|0|0|0|0|0|0|(null)|0|(null)|0|0|0|0|0|0|150607|0|0|0|_Total|0|0|0|73|100|0|0|0|0

What I am attempting is this:

AvgDiskBytesPerRead 0 0 0
AvgDiskBytesPerTransfer 0 0 0
... etc ...

As you can tell from the output, at minimum it will have two lines of return, and max will be variable based on hard drives.

I have messed with using tr and awk to break things up in a for loop, and I can make each line display vertically, however when adding the additional column output I just cannot make it work.

Any suggestions or direction would be greatly appreciated.

danielbmartin · 01-21-2013, 10:25 PM

This is not elegant but it works...

Code:

# Initialize the Output file
sed q $InFile |sed 's/|/\n/g' > $OutFile

# Lines = number of lines in the input file
Lines=$(cat $InFile |wc -l)  

for (( c=2; c<=$Lines; c++ ))
do
   sed -n $c'p' $InFile \
  |sed 's/|/\n/g'       \
  |paste $OutFile - > $Work1
  cat $Work1 > $OutFile
done

Daniel B. Martin

pan64 · 01-22-2013, 12:35 AM

awk is ok, can you show us what have you tried?

Theoretically you would need two loops: the first one to read all the lines into arrays the second is to display the result

danielbmartin · 01-22-2013, 07:48 AM

This awk solution is more concise than the one given in post #2 of this thread and does not use a temporary work file.

Code:

awk -F "|" '{for (i=1; i<=NF; i++) arr[i]=arr[i]$i"\t"} 
  END {i=1; while (i in arr) {print arr[i]; i++;}}'  \
  $InFile > $OutFile

Daniel B. Martin

Guyverix · 01-22-2013, 10:07 AM

Thank you very much! I went with answer #3, as I was able to run it inline with the WQL query itself and datafill a base varaible. With this, I can really make something useful here! Thank you again!

pan64 · 01-22-2013, 10:42 AM

glad to help you
Just in case you really want to say thanks press YES (bottom right corner)

konsolebox · 01-22-2013, 08:31 PM

Using same concept you could do that in Bash as well:

Code:

#!/bin/bash

while IFS=\| read -a LINE; do
    for I in "${!LINE[@]}"; do
        if [[ -n ${ARR[I]} ]]; then
            ARR[I]=${ARR[I]}$'\t'${LINE[I]}
        else
            ARR[I]=${LINE[I]}
        fi
    done
done

IFS=$'\n' A="${ARR[*]}"
echo "$A"

Code:

bash script.sh < INPUT > OUTPUT

grail · 01-23-2013, 02:00 AM

Just for another fun alternative:

Code:

ruby -ne 'BEGIN{a=[]};a << $_.chomp.split("|");END{a.transpose.each{|x| puts x.join(" ")}}' file

danielbmartin · 01-23-2013, 07:19 AM

Quote:

Originally Posted by grail

Just for another fun alternative ...

I tested your ruby solution and got this error:

Code:

-e:1: undefined local variable or method `a' for main:Object (NameError)

Please advise.

Daniel B. Martin

grail · 01-23-2013, 09:24 AM

Hmmm ... initially I would have said you missed the BEGIN ... which version of ruby are you running? Unfortunately there are a few quirks which catch me out between versions

Code:

grail@pilgrim:~$ ruby -v
ruby 1.9.3p362 (2012-12-25 revision 38607) [x86_64-linux]

schneidz · 01-23-2013, 09:32 AM

some of the above solutions are probably more elegant but heres an idea:

separate each row into a file with a single column:

Code:

sed -n 1p guy.txt | tr '\|' '\n' > guy-1.txt

then paste all the separated columns together:

Code:

paste guy-*.txt > guy-columns.txt

i am too lazy to wrap this up in a for loop at the moment.

danielbmartin · 01-23-2013, 09:39 AM

Quote:

Originally Posted by schneidz

i am too lazy to wrap this up in a for loop at the moment.

Take a look at post #2 in this thread. It uses the same idea, wrapped up in a loop.

Daniel B. Martin

konsolebox · 01-23-2013, 07:41 PM

grail's solution worked for me.

Code:

ruby 1.9.3p194 (2012-04-20 revision 35410) [x86_64-linux]

danielbmartin · 01-23-2013, 07:54 PM

Quote:

Originally Posted by konsolebox

grail's solution worked for me.

Code:

ruby 1.9.3p194 (2012-04-20 revision 35410) [x86_64-linux]

I've got

Code:

ruby 1.8.7 (2010-01-10 patchlevel 249) [i486-linux]

Perhaps that explains the error message cited previously.

Daniel B. Martin

grail · 01-24-2013, 11:03 AM

I believe the below excerpt 'may' identify the problem:

Quote:

Ruby 1.9 introduces many significant changes over the 1.8 series.[21] Examples:
Block local variables (variables that are local to the block in which they are declared)

Unfortunately I started with 1.9 so am unsure on what is required to be changed to make it work in the older version