Regex to delete specific spaces with sed

Question

I have a text file that looks like this:

HETATM11584
HETATM 2514
HETATM22222
HETATM 8585
HETATM 1213
HETATM62626
HETATM 6754

There are white spaces separating HETATM from a string of numbers. I want to delete theses white spaces and output a new file that looks like this:

HETATM11584
HETATM2514
HETATM22222
HETATM8585
HETATM1213
HETATM62626
HETATM6754

I tried to use sed

sed 's/.*HETATM [1-9]*.*/.*HETATM[1-9]*.*/' HETATM.txt > teste.txt

But the output looks like this:

HETATM11584
.*HETATM[1-9]*.*
HETATM22222
.*HETATM[1-9]*.*
.*HETATM[1-9]*.*
HETATM62626
.*HETATM[1-9]*.*

What is the proper sed synthax for this operation?

Answer 1

If the whole file is like your example, the easiest way would be removing all the spaces:

$ sed -r 's/ //g' file.txt
HETATM11584
HETATM2514
HETATM22222
HETATM8585
HETATM1213
HETATM62626
HETATM6754

If the file is composed by whatever followed by space(s) and a ending with a number, then this will do it:

$ sed -r 's/ +([0-9]+)$/\1/g' file.txt
HETATM11584
HETATM2514
HETATM22222
HETATM8585
HETATM1213
HETATM62626
HETATM6754

Answer 2

You should not use a pattern as the substitution string. You can remove the space from your string using backreference:

sed 's/\(HETATM\) \([0-9]\)/\1\2/'  HETATM.txt > teste.txt 

HETATM and digits are captured with parenthesis. Captured strings are output(without space) with \1 and \2.

Answer 3

Do you have to use sed?

$ tr -d ' ' < HETATM.txt > teste.txt

might serve your purpose as well.