Sed Regex Issue

Question

I'm attempting to transform:

 pid 2928's current affinity list: 0-3 pid 2928's new affinity list: 0

to

2928: 0-3 --> 0

At the moment I have the following transformations that work properly:

Regex: pid Output: 2928's current affinity list: 0-3 2928's new affinity list: 0

Regex: 's current affinity list Output: 2928: 0-3 2928's new affinity list: 0

And then the final regex, which is is not working properly:

Regex: (\d+)(.)(.)(.)((?:[a-z][a-z]+))(\s+)((?:[a-z][a-z]+))(\s+)((?:[a-z][a-z]+)) Output: 2928: 0-3 2928's new affinity list: 0

As you can see, it's not applying the regex correctly. If you test this regex in an online tester, you can see that it does pick out the right area for deletion.

Full command:

echo "pid 2928's current affinity list: 0-3 pid 2928's new affinity list: 0" | sed -e "s/pid //g" -e "s/'s current affinity list//g" -e "s/(\d+)(.)(.)(.)((?:[a-z][a-z]+))(\s+)((?:[a-z][a-z]+))(\s+)((?:[a-z][a-z]+))/-->/gim"

I must not be executing the regex correctly in the third -e for sed. Does this look wrong in any way?

Answer 1

The problem was that you were using perl regular expression features (\d, \s) not recognized by sed:

sed 's/pid \([0-9]\+\)'\''s current affinity list: \([0-9]\+-[0-9]\+\) pid [0-9]\+'\''s new affinity list: /\1: \2 --> /'

The awk answer is clearly easier to maintain.

Answer 2

Using awk is more simple here:

awk -F "[ ']" '{ print $2 ": " $7 " --> " $14}'

all the line:

echo "pid 2928's current affinity list: 0-3 pid 2928's new affinity list: 0" | awk -F "[ ']" '{ print $2 ": " $7 " --> " $14}'

Answer 3

Here's a quick and dirty solution to your task, though it may not answer your underlying question:

awk '{print $2" "$6" --> "$12}' | sed 's/.s/:/'

Note that this depends on the field 2, 6, and 12 (as separated by whitespace) being predictable.

Answer 4

In case you are interested in a non-pure-sed solution, but a simpler one, here is a hybrid sed and awk solution:

echo  "pid 2928's current affinity list: 0-3 pid 2928's new affinity list: 0"|sed "s/[a-z']//g"|awk '{print $1 ": " $3 " --> " $6}'