Hi all,

I have a requirement where I have to substitute a | with nothing in a file which has records of the type

This is the regular format of my file

aaa|.|bbb|.|ccc|.|ddd
eee|.|fff|.|ggg|.|hhh
iii|.|jjj|.|kkk|.|lll

Sometimes I get a | in these fieds like below (check the g|gg)

aaa|.|bbb|.|ccc|.|ddd
eee|.|fff|.|g|gg|.|hhh
iii|.|jjj|.|kkk|.|lll

Due to this my ETL fails.
Can anybody suggest how I can remove only the extra pipes in the file.
Any suggestions greatly appreciated.
Please get back if you need more clarification.

Thanks
Narasimha

Using this data and placing it into a file called data:

aaa|.|bbb|.|ccc|.|ddd
eee|.|fff|.|g|gg|.|hhh
iii|.|jjj|.|kkk|.|lll

sed s/\|//g < data

gives me this output:

aaa.bbb.ccc.ddd
eee.fff.ggg.hhh
iii.jjj.kkk.lll

If I'm understanding you correctly, your field seperator is |.| that is, a pipe symbol, a period, a pipe symbol. A stray pipe symbol messes up your data.

One way of doing this is to choose a character that is not used in the data file for a temporary separator. Then, replace |.| with the temporary seperator, delete any extraneous pipe symbols, and, finally, replace the temporary seperator with the true one:

sed 's/\|\.\|/#/g' < data.file|sed 's/\|//g' |sed 's/#/\|\.\|/g'

Maybe somebody with better regex skills than I have, can come up with a better solution.

nails,

You have understood the scenario correctly.Currently I am using the same appraoch you have suggested.
Is there any other way ???

I realize the solution I provided is inefficient, but I still don't have any other solution.

nails,

Thats the best I have got till now, Thanks!