Delete n lines following a pattern (and the line matching the pattern)

Question

How can I delete a line containing a matching pattern and the following n lines using a tool supporting regular expressions?

Said differently, how can I write a regular expression matching a line containing a matching pattern and the following n lines, so that I can replace them with nothing?

For example, if I have a matching pattern bbbb and I want to delete also the 5 lines that follows it, for the input file:

aldjflajdkl
aaaabbbbaaaa
1l;adfjl
2aldfjl
3adlflkdas
4aldfjd
5aldfkld
6dlafjlkdas

The output would be:

aldjflajdkl
6dlafjlkdas

It probably simplify things that in my specific case, it cannot be that the matching pattern (bbbb) is contained in the following 5 lines.

A solution already exists for sed, but it relies only partially on regular expressions, and uses custom replacement commands which are not portable.

Answer 1

A possible solution is:

.*<matching pattern>(.*\r?\n){<N+1>}

where N is the number of lines I want to remove after the line containing the pattern.

For the example given, this translates in:

.*bbbb(.*\r?\n){6}

That's how it looks in grepWin:
Side notes:

• In the tab "The regex search string matches" also the 5aldfkld line is marked to be matched, indeed a scroll bar is visible on the right
• (grepWin specific) Because of a small bug, when applying this search on files, you'll see the count of Matches increasing by 7 for each match! That's probably because the match counter counts how many lines are matched, and in this case the pattern covers 7 lines: the matched line, the following 5 lines and the line reached with the last line feed
• (sed specific) This regex does not work for sed, which does not fully support regex and has no easy way to match/replace new lines.

---

The following explains how I got to the solution.

I started from:

.*bbbb.*\n.*\n.*\n.*\n.*\n.*\n

which would not work in my system. But the following would work:

.*bbbb.*\r\n.*\r\n.*\r\n.*\r\n.*\r\n.*\r\n

So, I am working in a CRLF system. However this doesn't look very pretty nor portable.

I can make it a little bit more portable (and uglier :-) ) by doing:

.*bbbb.*\r?\n.*\r?\n.*\r?\n.*\r?\n.*\r?\n.*\r?\n

(The carriage return becomes optional). It still looks ugly, but I can collect the repetitive term:

.*bbbb(.*\r?\n){6}

This guide was very handy.

Answer 2

An awk solution:

awk '/bbbb/ {i=5; next} {if (i>0) i--; else print}'

When it detects the pattern you're looking for, it sets i (which is a countdown counter) to 5, and skips the rest of the processing (i.e., skips to the next line of input).  In particular, it does not print the line.  (Saying /bbbb/ {i=5+1} for the first part would be equivalent; choose one based on your style preference.)  Then, if the counter is positive, decrement it (subtract 1) so as to count the lines that are being deleted (skipped), and do not print; otherwise, print the line.