Create tab delimited file with header

SuKHI · 10-29-2015, 03:54 AM

Hi,

I have requirement to create tab delimited file with values coming from variables.
1. File will contain only two columns separated by tab.
2. Header will be added once.
3. Values will be keep adding upon the script run.
4. If values already exists then values will be replaced.

I have done so far below work:

Code:

#!/bin/bash

NDIR="/opt/nodes/"
DIRS=`ls -l $NDIR | egrep '^d' | awk '{print $9}'`

HEADER="Environment\tNode"

printf "1\ni\n$HEADER \n.\n\nw\nq\n"|ed -s /usr/node.txt

for DIR in $DIRS
do
echo  ${DIR}
NFILE="/opt/nodes/$DIR/config"

ENV=$(sed -n '/<user-props>/,\|</user-props>| {
   s/ *<kv-pair name="environment.name" value="\([^"]*\)"\/>/\1/p
   }' $NFILE/node.xml)
   
NODE=$(sed -n '/<sys-props>/,\|</sys-props>| {
   s/ *<kv-pair name="node.name" value="\([^"]*\)"\/>/\1/p
   }' $NFILE/node.xml)

echo -e "$ENV\t$NODE" >> /usr/node.txt
done

echo "###### END OF SCRIPT ######"

Above piece of code:
Will read node.xml file. Look for props. assign variables as below:
e.g.,
ENV = SystemEnv
NODE = SystemNode
There are multiple Nodes in nodes folder so for loop to take care.

Currently I'am able to write values into output file node.txt

Problms:
First time script will run ok. File will be perfect.
But Next time on run Header will be added again and if the folder contains same node names those will be added.
If I use only ">" method with echo then only last node will be added, ignoring the others.
Header method I used to add always print
?
?
upon run. I couldn't figure out why.

1. Is there any other method to add header in tab delimited style.
2. Shall use sed or awk instead of echo to make output file?
3. Replace the existing entries.

help needed to make it working as per needs.
Thanks,

chrism01 · 10-29-2015, 09:02 PM

If its ok to replace the entire file on each run, then either

Code:

rm /usr/node.txt

# OR

>/usr/node.txt

at the start of the run.

IF it needs to be created only once, then 'updated' on each run, I personally would use eg Perl instead of bash etc.
One of Perl's basic var types is a hash, which is ideal for ensuring uniqueness of records.

syg00 · 10-29-2015, 09:13 PM

Totally agree. But if it has to be bash ...
The reason you get multiple headers is because you tell "ed" to insert the header at line 1 - every time. Use "echo -e" like you do later - bash has several test that would help; maybe a simple "if [ !-e /usr/node.txt ]" would suffice.
As for multiple entries, that of course will happen if you do nothing to interrogate the /usr/node/txt - your script simple appends everything it finds.