How to grep netcat output

Answer is here: grep not matching in nc output

netcat outputs verbose logs to standard error, so we need to capture errors before we pipe to grep.

$ netcat -zv localhost 1-9000 2>&1 | grep succeeded

You could use the read command (bash builtin) to force characters to be read one by one :

netcat localhost 9090 | (
    cnt=0
    line=
    while read -N 1 c; do
        line="$line$c"
        if [ "$c" = "{" ]; then
            cnt=$((cnt+1))
        elif [ "$c" = "}" ]; then
            cnt=$((cnt-1))
            if [ $cnt -eq 0 ]; then
                printf "%s\n" "$line"
                line=
            fi
        fi
    done
) | grep sender

This script should print every full output with balancing {and }, but you can change the script to do whatever you want. This script would NOT do well on a benchmark compared to pretty much anything, but it's pretty simple and seems to work for me...

Note that your test sample didn't have matching {and }, so if this is the case of the real input, you might want to change the criteria to print the line.

I think the issue is the absence of newlines in the netcat output. I can see two workarounds:

1. Insert a newline every x seconds (with unfortunate consequences if the newline is inserted in the middle of source):

   ( while sleep 1; do echo; done & netcat ... ) | grep source
   

2. Use awk with an RS other than newline:

   netcat ... | awk -v RS='}' -v ORS='}' '/source/'