how to remove the first two columns in a file using shell (awk, sed, whatever)

You can do it with cut:

cut -d " " -f 3- input_filename > output_filename

Explanation:

• cut: invoke the cut command
• -d " ": use a single space as the delimiter (cut uses TAB by default)
• -f: specify fields to keep
• 3-: all the fields starting with field 3
• input_filename: use this file as the input
• > output_filename: write the output to this file.

Alternatively, you can do it with awk:

awk '{$1=""; $2=""; sub("  ", " "); print}' input_filename > output_filename

Explanation:

• awk: invoke the awk command
• $1=""; $2="";: set field 1 and 2 to the empty string
• sub(...);: clean up the output fields because fields 1 & 2 will still be delimited by " "
• print: print the modified line
• input_filename > output_filename: same as above.

Here's one way to do it with Awk that's relatively easy to understand:

awk '{print substr($0, index($0, $3))}'

This is a simple awk command with no pattern, so action inside {} is run for every input line.

The action is to simply prints the substring starting with the position of the 3rd field.

• $0: the whole input line
• $3: 3rd field
• index(in, find): returns the position of find in string in
• substr(string, start): return a substring starting at index start

If you want to use a different delimiter, such as comma, you can specify it with the -F option:

awk -F"," '{print substr($0, index($0, $3))}'

You can also operate this on a subset of the input lines by specifying a pattern before the action in {}. Only lines matching the pattern will have the action run.

awk 'pattern{print substr($0, index($0, $3))}'

Where pattern can be something such as:

• /abcdef/: use regular expression, operates on $0 by default.
• $1 ~ /abcdef/: operate on a specific field.
• $1 == blabla: use string comparison
• NR > 1: use record/line number
• NF > 0: use field/column number

Thanks for posting the question. I'd also like to add the script that helped me.

awk '{ $1=""; print $0 }' file

You can use sed:

sed 's/^[^ ][^ ]* [^ ][^ ]* //'

This looks for lines starting with one-or-more non-blanks, a blank, another set of one-or-more non-blanks and another blank, and deletes the matched material, aka the first two fields. The [^ ][^ ]* is marginally shorter than the equivalent but more explicit [^ ]\{1,\} notation, and the second might run into issues with GNU sed (though if you use --posix as an option, even GNU sed can't screw it up). OTOH, if the character class to be repeated was more complex, the numbered notation wins for brevity. It is easy to extend this to handle 'blank or tab' as separator, or 'multiple blanks' or 'multiple blanks or tabs'. It could also be modified to handle optional leading blanks (or tabs) before the first field, etc.

For awk and cut, see Sampson-Chen's answer. There are other ways to write the awk script, but they're not materially better than the answer given. Note that you might need to set the field separator explicitly (-F" ") in awk if you do not want tabs treated as separators, or you might have multiple blanks between fields. The POSIX standard cut does not support multiple separators between fields; GNU cut has the useful but non-standard -i option to allow for multiple separators between fields.

You can also do it in pure shell:

while read junk1 junk2 residue
do echo "$residue"
done < in-file > out-file