Question

I need to modify an existing script which uses SED to search and replace in a file.
My problem is that i need to use sed to do the same search and replace but only if 2 criteria is met.

I use find and sed like below :

find . -name 'search.do*' -exec sed -i "/href/\
s/href=\"[^\"]*sommerhus\/detail\.do[^\"]*unitId=\([^\"]*\)\"/href=\"sommerhus-\1.html\"/g" {} \;

but now i need to make sure that the replacement only takes place if the “href” AND another search criteria is met f.x “href” && “language” is present

I have tried with -e but without success.
Is it possible ??

Answer 1

You could use /href.*language\|language.*href/ (possibly with some extra constraints, like a space before the second word in each match).

You could also nest the commands in blocks:

sed '/href/{ /language/s/<match>/<replace>/ }'

This is more extensible: adding a third pattern only requires adding in one pair of braces and the pattern, while the alternation solution above requires 6 patterns in total (one for each possible order).

---

Also, to save having to backslash the / inside the pattern, sed allows you to use any character as a separator, e.g. @:

find . -name 'search.do*' -exec sed -i "/href/\
s@href=\"[^\"]*sommerhus/detail\.do[^\"]*unitId=\([^\"]*\)\"@href=\"sommerhus-\1.html\"@g" {} ;'

(A small gain in this case, but nice to know for other situations when handling file paths or URLs etc.)