sed / awk: Delete all lines matching a word not followed by another word

Question

I have the following lines in my file.

<stuff>
test1.*test2
test1
test1.*test2
test1
<other stuff>

I want to delete all lines containing test1, but not followed by test2 -- which means I should end up with:

<stuff>
test1.*test2
test1.*test2
<other stuff>

I've tried a lot of regular expressions, but can't seem to crack either sed or awk. This was my latest attempt, after testing with vim regexp.

sed -i .bk '/^.*test1\(\(test2\)\@!.\)*$/d' file

(This is part of a bash script on Mac OS X)

Answer 1

Use two expressions. The first one skips the cases where test1 is followed by test2 , the second one removes test1 - it can only be reached if test2 wasn't there.

sed -e '/test1.*test2/b' -e '/test1/d'

Answer 2

Try something like:

awk '/test1/&&!/test2/{next}1' file

We tell awk to:

• Look for lines with test1 .
• On the same line test2 should not be present
• If such line is found we skip
• If such line is not found we print

$ cat file
<stuff>
test1.*test2
test1
test1.*test2
test1
<other stuff>

$ awk '/test1/&&!/test2/{next}1' file
<stuff>
test1.*test2
test1.*test2
<other stuff>

Answer 3

带有perl正则表达式的GNU grep：

grep -vP 'test1(?!.*test2)' file

Answer 4

$ awk '!/test1/||/test2/' file
<stuff>
test1.*test2
test1.*test2
<other stuff>