How to sed the first/last ocurrence of a row in a CSV?

Question

Hello I have a CSV file with this random structure

,,,,,VALUE1,,,,,VALUE2,,,, ,,VALUE3,,,VALUE4,,,VALUE5, ,,,,,,,,,,,,,,,,,,, ,,,,,,,VALUE6,,,,VALUE7,, ,,,,,,,,VALUE8,,,,,,

And I want to output the first and last ocurrence of each line.

in this manner

For first ocurrence

VALUE1 VALUE3 (empty) VALUE6 VALUE8

for last ocurrence

VALUE2 VALUE5 (empty) VALUE7 VALUE8

I've searched for hours and didn't find the best way to achieve this.

Thank you very much.

EDIT: Thank you everybody, this ,+ as separator is the key.

It there data between the , like this a,b,s,VALUE1,d,e or is it just the , there. — Jotne
– Jotne, Commented Dec 28, 2014 at 18:17

glenn jackman · Accepted Answer · 2014-12-28 18:18:41Z

Using awk with custom field separator:

First non-empty field:

awk -F ',+' '{print $2}' file VALUE1 VALUE3 VALUE6 VALUE8

Last non-empty field:

awk -F ',+' '{print $(NF-1)}' file VALUE2 VALUE5 VALUE7 VALUE8

Regex pattern ',+' will make 1 or more commas as a field separator.

glenn jackman · Accepted Answer · 2014-12-28 18:22:44Z

Perl is good for this task

perl -F, -lane '@G = grep {$_ ne ""} @F; print $G[0]' file

The grep command filters out empty fields.
To print the last field, change $G[0] to $G[-1]

With sed:

 sed 's/^,*//; s/,.*//' file # first field sed 's/,*$//; s/.*,//' file # last field

Jotne · Accepted Answer · 2014-12-28 18:18:15Z

From this data:

cat file ,,,,,VALUE1,,,,,VALUE2,,,, ,,VALUE3,,,VALUE4,,,VALUE5, ,,,,,,,,,,,,,,,,,,, ,,,,,,,VALUE6,,,,VALUE7,, ,,,,,,,,VALUE8,,,,,,

First occurrence:

tr ',' ' ' <file | awk '{print $1}' VALUE1 VALUE3 VALUE6 VALUE8

Last:

tr ',' ' ' <file | awk '{print $NF}' VALUE2 VALUE5 VALUE7 VALUE8

Daniel · Accepted Answer · 2014-12-28 18:24:05Z

sed -e "s/,*\([^,]*\).*/\1/"

should find the first, and

sed -e "s/\(,*\([^,][^,]*\)\)*,*/\2/"

should find the last.

Jasen · Accepted Answer · 2014-12-28 18:29:47Z

if you wanr (empty) as part of the output.

first:

sed ' s/,*\([^,]\+\).*/\1/; s/^,*$/(empty)/; '

last:

sed ' s/\(^\|.*,\)\([^,]\+\),*/\2/; s/^,*$/(empty)/; '

5 Answers 5