替换匹配的正则表达式地址的内容而不重复正则表达式

Question

Is there a way in (GNU) sed to perform a s substitution on the exact string matched by the regexp-address?

Something that avoids repeating the regexp already used to select the line?

Example of an hard-to-read sed invocation where the same regexp is used to match the line where the s command will operate as well as to determine what will be substituted in that line.

sed -i -e '/verycomplexregexp/s/verycomplexregexp/something/'

Answer 1

Using an empty regexp // in the s command avoids having to repeat the regexp used to select the line:

sed -i -e '/verycomplexregexp/s//something/'

The empty regexp // refers the previously matched text in the pattern space. When a regexp-address is used, the empty regexp points to the string that made the line match the address.