How to convert in specific column in csv in shell script linux-CodePudding

i have a file output.csv like this :

No | Timestamp | Table | agg | result | percentage | status | year | month | day
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|287.472.321|-4.83%|Normal|2022|202209|20220930
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|498.512.903|-4.05%|Normal|2022|202209|20220930
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|19.089.222|-19.4%|Normal|2022|202209|20220930
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|29.014.921|-2.47%|Normal|2022|202209|20220930
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|148.770.415|-0.14%|Normal|2022|202209|20220930

all i wanna do is converting "." to blankspace in 5th column without affect on 6th column like this :

2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|287472321|-4.83%|Normal|2022|202209|20220930
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|498512903|-4.05%|Normal|2022|202209|20220930
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|19089222|-19.4%|Normal|2022|202209|20220930
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|29014921|-2.47%|Normal|2022|202209|20220930
2022-09-30 00:00:00|2022-09-30 00:00:00|Source|count|148770415|-0.14%|Normal|2022|202209|20220930

i've tried using cat output.csv | grep -v -i "No " | sed 's/\.//g' > output3.csv But it will convert all columns that have "." .

So how to converting "." to blankspace in 5th column without affect on 6th column with still csv format output?

Thank you for the help!

CodePudding user response：

It is easier to do this in awk:

awk 'BEGIN{ FS=OFS="|" } { gsub("[.]","",$5) } 1' file

But if you insist on using sed:

sed '
    :a
    s/^\(\([^|]*|\)\{4\}[^|]*\)\./\1/
    ta
' file

CodePudding user response：

This might work for you (GNU sed and shell):

sed 'y/|/\n/;s/.*/echo "&"|sed "5s#\\.##g"/e;y/\n/|/' file

In essence, turn each line into a file and run a second invocation of sed where each address represents a column in the original line.

Translate |'s to newlines.

Write a mini sed invocation each time the pattern space is presented, that removes all .'s from the 5th line (each line in the invocation represents a column in the original line).

Translate newlines back to |'s.