sed script to delete all characters up to & including the 2nd comma on a line

Question

Can anyone explain how to use sed to delete all characters up to & including the 2nd comma on a line in a CSV file?

The beginning of a typical line might look like

1234567890,ABC/DEF, and the number of digits in the first column varies i.e. there might be 9 or 10 or 11 separate digits in random order, and the letters in the second column could also be random. This randomness and varying length makes it impossible to use any explicit pattern searching.

Answer 1

You could do it with sed like this

sed -e 's/^\([^,]*,\)\{2\}//'

not 100% sure on the syntax, I tried it, and it seems to work though. It'll delete zero-or-more of anything-but-a-comma followed by a comma, and all that is matched twice in succession.

But even easier would be to use cut, like this

cut -d, -f3-

which will use comma as a delimiter, and print fields 3 and up.

EDIT:
Just for the record, both sed and cut can work with a file as a parameter, just append it at the end like so

cut -d, -f3- myfile.txt

or you can pipe the output of your program through them

./myprogram | cut -d, -f3-

Answer 2

sed is not the "right" choice of tool (although it can be done). since you have structured data, you can use fields/delimiter method instead of creating complicated regex.

you can use cut

$ cut -f3- -d"," file

or gawk

$ gawk -F"," '{$1=$2=""}1'  file
$ gawk -F"," '{for(i=3;i<NF;i++) printf "%s,",$i; print $NF}'  file

Answer 3

Thanks for all replies - with the help provided I have written the simple executable script below which does what I want.

!/bin/bash

cut -d, -f3- ~/Documents/forex_convert/input.csv |

sed -e '1d' \

-e 's/-/,/g' \

-e 's/ /,/g' \

-e 's/:/,/g' \

-e 's/,D//g' > ~/Documents/forex_convert/converted_input

exit