如何获取bash变量在awk管道上工作

Question

I can't get variable from my bash script to work on awk piping.

SERVER_PORT="27018"
...
HOLDER=`netstat -tlpn | awk -v sp=$SERVER_PORT '/:sp */ {split($NF,a,"/"); print a[1]}'`

After executing command for this variable next equation is TRUE!

...
if [ "$HOLDER" == "" ]; then
...

If I run

netstat -tlpn | awk '/:27018 */ {split($NF,a,"/"); print a[1]}'

on my shell window everything is fine and it gives me the PID of the process.

So my question is: how to pass variable to awk correctly or something has broken? Running CentOS 7

Answer 1

Inside a regular expression constant awk doesn't expand variables. So /:sp */ is literally /:sp */ .

You need to make that match a string: $0 ~ ":"sp" *"

awk scripts are made up of pattern {action} pairs. /:27018 */ is a pattern (using a regexp constant). As I said at the beginning awk doesn't expand variables inside a regexp constant so /:sp */ is matching those literal characters.

There are other patterns possible beyond a regexp constant. Even static strings can be patterns (they are always true but they work).

~ is a regular expression match operator in awk. You can use that to manually perform the same sort of match that the regexp constant pattern does. ~ takes two string arguments.

Putting that together we can match ( ~ ) the whole line ( $0 ) against a string constant regular expression (rather than a regexp constant) of ":"sp" *" (the string ":" , our variable sp , and the string " *" ).

Which, when put together, gets you (as I wrote above) $0 ~ ":"sp" *" as the pattern instead of the regexp constant.