Substitute in multiple isolated parts of a string

Question

I have this string:

b'32^"Strings" c32 d32 c32 b32 r8. \break

I want to replace [a-z] (except r) with r and remove ' everywhere outside "..". I want to ignore \break.

Final output should be:

r32^"Strings" r32 r32 r32 r32 r8. \break

The tricky part is sometimes the ".." and/or \break may not be present.

For example:

b'32 c32 d32 c32 b32 a32 g32 fis32 d'32 \break

Should be converted to:

r32 r32 r32 r32 r32 r32 r32 r32 r32 \break

A single command would be desirable, but multiple commands are acceptable too.

I tried:

s/^\(.\{-}\)\ze".*$/\=substitute(submatch(1), "[a-z']", 'r', 'g')/

and

s/^.*"\zs\(.*\)/\=substitute(submatch(1), "[a-z]", "r", "g")/

But I am not sure how to handle optional ".." and \break.

Answer 1

I would propose you:

s/\v(<[a-z]'?|"[^"]*"|\\\w+)/\=submatch(1)[0]=='"'||submatch(1)[0]=='\'?submatch(1):'r'/g

The idea is to match by token either:

• starting with a letter and an optional quote: [a-z]'?
• a quoted string: "[^"]*"
• a back-slashed word: \\\w+

Then to decide to make the replacement by r only on the first case.

i.e.: if the token is not a string or a back-slashed word (submatch(1)[0]=='"'||submatch(1)[0]=='\')

Remark: the leading \v allow to clarify the expression avoiding to have to escape the special characters: <, (, ), |, ?, + ([, ] and * don't need to be escaped).