SED Regex rewriting between two patterns using backreferences - regex

Question

I have text in many html files that is in this format:
<!-- BEGIN FOOTER -->
<div id="footer">
<p align="right"> Slogan<br />
5555 Street East <br />
City, State 99999 <br />
Call Us [phone] <br />
</p>
<div align="center">
<a class="footer" href="http://www.example.com" title="Site">Site</a>
</div>
<br>
</div>
<!--END FOOTER-->
I am using this:
sed -E -i 's/(<!-- BEGIN FOOTER -->)(.|\n)*(<!--END FOOTER-->)/\1 <br>REPLACE<br> \3 /m' file.html
but not working to capture & backreference them:
\1 <!-- BEGIN FOOTER -->
\3 <!--END FOOTER-->
And insert this in between them:
REPLACE
So trying to end up with this:
<!-- BEGIN FOOTER -->
<br>REPLACE<br>
<!--END FOOTER-->

Answer 1

This might work for you (GNU sed):
sed '/<!-- BEGIN FOOTER -->/{:a;N;/<!--END FOOTER-->/!ba;s/\n.*\n/\n<br>REPLACE<br>\n/}' file
This gathers up lines between footers and replaces the lines between with the required string.
An alternative (similar to revo);
sed '/<!-- BEGIN FOOTER -->/,/<!--END FOOTER-->/!b;/<!-- BEGIN FOOTER -->/b;/<!--END FOOTER-->/!d;i\<br>REPLACE<br>' file
Another way:
sed '/<!-- BEGIN FOOTER -->/,/<!--END FOOTER-->/!b;/<!--END FOOTER-->/p | sed '/<!-- BEGIN FOOTER -->/p;/<!-- BEGIN FOOTER -->/,/<!--END FOOTER-->/c\<br>REPLACE<br>'

Answer 2

sed reads from input file one line at a time and directs the result to standard output. It means no where in input line there is \n character to be matched except in some cases that a few special commands are being used. You could use address ranges along with insert i command to achieve desired output:
sed '/BEGIN FOOTER/,/END FOOTER/{ /END FOOTER/{i\<br>REPLACE<br>
b}; /BEGIN FOOTER/b; d;}' file
Watch the linebreak after first line. i inserts data before current line in pattern space. b jumps over remaining commands causing a new iteration to be dobe. /BEGIN FOOTER/,/END FOOTER/ implies a range in which the other commands should execute.

Answer 3

You can first save the lines with BEGIN FOOTER and END FOOTER in beg and end variables:
beg=$(grep -n "BEGIN FOOTER" inputfile | cut -d: -f1)
end=$(grep -n "END FOOTER" inputfile| cut -d: -f1)
Then use sed's c\ command:
sed -i.bak "$((beg+1)),$((end-1))c\<br>REPLACE<br>" inputfile
Result:
<!-- BEGIN FOOTER -->
<br>REPLACE<br>
<!--END FOOTER-->
Alternatively use the single command:
sed "/BEGIN FOO/,/END FOO/c\<!--BEGIN FOOTER-->\n<br>REPLACE<br>\n<!--END FOOTER-->" inputfile