Find and replace {} in sed
The problem
Replacing all instances of vcr: {} with vcr: { record: :new_episodes }, in all files below spec/features.
Normally this find -exec can come to the rescue, so a reasonable approach might be:
$ find spec/features/ -type f -exec sed -i 's#vcr: {}#vcr: { record: :new_episodes }#' {} \;
Pro-tip: change the delimiter to #
Unfortunately we run in to trouble here: It just so happens that {} is the sequence find’s -exec switch uses to substitute in filenames.
Well, surely we can just escape the {} through find, right?
$ find spec/features/ -type f -exec sed -i 's#vcr: \{\}#vcr: { record: :new_episodes }#' {} \;`
sed: -e expression #1, char 45: Invalid content of \{\}
Not quite. Now the \{\} are being passed straight through to sed, including the \s, and unfortunately when sed sees escaped {}s it assumes them to be a consecutive occurrence matcher.
The solution
Sometimes it’s best to just ditch -exec altogether, and use the ever versatile xargs:
$ find spec/features/ -type f -print0 | xargs -0 sed -i 's#vcr: {}#vcr: { record: :new_episodes }#'