sed to ignore a pattern as well as match a pattern in same line

May 22, 2022

Input file

<a href="perl.html">perl</a>     <a href="http://zoidberg.sourceforge.net/out.html">http://zoidberg.sourceforge.net</a>
<a href="zoiduser.html">zoiduser</a>    <a href="perl.html">perl</a>     <a href="http://zoidberg.sourceforge.net/sample.html">http://zoidberg.sourceforge.net</a>

I need to only remove .HTML extension from below URL from above file:

<a href="perl.html">perl</a>
<a href="zoiduser.html">zoiduser</a>

So that the final output should look like:

<a href="perl">perl</a>     <a href="http://zoidberg.sourceforge.net/out.html">http://zoidberg.sourceforge.net</a>
<a href="zoiduser">zoiduser</a>    <a href="perl.html">perl</a>     <a href="http://zoidberg.sourceforge.net/sample.html">http://zoidberg.sourceforge.net</a>

This is what I am doing:

sed '/"http\|"www\|"mailto/ ! s|\(.html\)||g' file

But it ignores the line as soon as it matches the first pattern i.e. avoid URLs that start with "http|"www|"mailto.

>Solution :

You can use

sed -E 's/("(http|www|mailto)[^"]*")|\.html/\1/g' file

Details:

-E – enables POSIX ERE syntax
("(http|www|mailto)[^"]*") – Group 1 (\1): " and then either http, www, or mailto and then zero or more chars other than " and then a "
| – or
\.html – .html string.

The replacement is Group 1 values.

See the online demo:

#!/bin/bash
s='<a href="perl.html">perl</a>     <a href="http://zoidberg.sourceforge.net/out.html">http://zoidberg.sourceforge.net</a>
<a href="zoiduser.html">zoiduser</a>    <a href="perl.html">perl</a>     <a href="http://zoidberg.sourceforge.net/sample.html">http://zoidberg.sourceforge.net</a>'
sed -E 's/("(http|www|mailto)[^"]*")|\.html/\1/g' <<< "$s"

Output:

<a href="perl">perl</a>     <a href="http://zoidberg.sourceforge.net/out.html">http://zoidberg.sourceforge.net</a>
<a href="zoiduser">zoiduser</a>    <a href="perl">perl</a>     <a href="http://zoidberg.sourceforge.net/sample.html">http://zoidberg.sourceforge.net</a>