How do I skip lines using sed?

Question

I want to perform a sed substitution globally in a file. However I want to skip certain lines: - those where the substitution would occur after a ! - those containing a comma (,) - those containing the word "parameter" (lower or upper case)

How do I do this?

Do it have to be `sed`. Other program may do it better? And -1 since you did not post what you have tried. — Jotne, Jul 11 '14 at 07:57
Possible duplicate of [How to skip lines matching a string](https://stackoverflow.com/questions/6684857/how-to-skip-lines-matching-a-string) — Evan Carroll, Jul 07 '18 at 17:05

tripleee · Answer 1 · 2014-07-11T08:05:45.867

Your first exception is not trivial to implement, but maybe you can rephrase it as part of the substitution? So to replace "foo" with "bar" except on those lines, I'd go with

sed '/parameter\|,/!s/^\([^!]*\)foo/\1bar/'

(Some sed dialects will want more or fewer backslashes.)

/parameter\|,/! will only select lines which do not match this regular expression.

s/^\([^!]*\)foo/\1bar/ will substitute "foo" with "bar" provided that "foo" is not proceded by an exclamation mark. The leading text is captured into a back reference \1 and basically replaced with itself.

The case insensitive matching requires an ugly regex in sed -- [Pp][Aa][Rr][Aa][Mm][Ee][Tt][Ee][Rr]. For anything more complex, maybe switch to Awk or Perl, which provide better facilities for complex matching as well as much more readable code.

How do I skip lines using sed?

1 Answers1