How to express variables efficiently for SED?

Question

There are many ways to express arithmetic things: new and old ways. However, I would like to write this line efficiently and well

sed -n '$START,$ENDp;$ENDq' < data.tex

which output

sed: 1: "$START,$ENDp;$ENDq": invalid command code S

Total Code

#!/bin/bash
function getStart {
        local START="$(awk '/begin\{document\}/{ print NR; exit }' data.tex)"
        echo $START
}

function getEnd {
        local END="$(awk '/end\{document\}/{ print NR; exit }' data.tex)"
        echo $END
}

START=$(getStart)
(( START-- ))
echo $START

END=$(getEnd)
(( END++ ))
echo $END

sed -n '$START,$ENDp;$ENDq' < data.tex

where the SED is getting the linenumbers from local variables of functions. How can you express the first line by expressing the variables efficiently for the SED?

@don_crissti No, just me keeping up my Unix skills by processing my notes, but hitting my head sometimes. Feel free to say a better way about processing the thing once. — Léo Léopold Hertz 준영, Sep 11 '15 at 13:28
sed '/begin{document}/,/end{document}/!d;/end{document}/q' data.tex — don_crissti, Sep 11 '15 at 13:37
Yours is superior! I thought the thing absolutely too complicated. Thank you! — Léo Léopold Hertz 준영, Sep 11 '15 at 13:38
I want open boundaries. I opened a new thread about it here http://unix.stackexchange.com/q/229060/16920 — Léo Léopold Hertz 준영, Sep 11 '15 at 13:45

score 3 · Accepted Answer · edited Sep 11 '15 at 23:03

Variables aren't expanded in single quotes. Use double quotes:

sed -n "$START,${END}p;${END}q" < data.tex

Also, since you want the value of END followed by the character p, and not the value of ENDp, you need to indicate where the name of the variable ends. You can use the ${END} syntax, which makes it explicit where the variable name ends.

Note that if $START or $END are not numbers but regular expressions, you might need to escape some characters.

How to express variables efficiently for SED?

Total Code

1 Answers1

Linked