Hi I am trying to use a Unix FIFO to communicate between a Python script and a shell script. The intention is for the shell script to capture all output of the python script. So far I have the following:
#!/bin/bash
# Test IPC using FIFOs.
## Create a named pipe (FIFO).
mkfifo ./myfifo
## Launch the Python script asynchronously and re-direct its output to the FIFO.
python3 helloworld.py > ./myfifo &
PID_PY=$!
echo "Python script (PID=$PID_PY) launched."
## Read from the FIFO using cat asynchronously.
## Note that running asynchronously using & runs it the program (in this case `cat`)
## in a child shell "subshell", so I will collect the output in a file.
echo "Reading FIFO."
>output.log cat ./myfifo &
PID_CAT=$!
## Sleep for 10 seconds.
sleep 10
## Kill the Python script.
kill -15 $PID_PY && echo "Python script (PID=$PID_PY) killed."
## Kill the cat!
kill -15 $PID_CAT
## Remove the pipe when done.
rm -fv ./myfifo
## Check for the existence of the output log file and print it.
[[ -f output.log ]] && cat output.log || echo "No logfile found!." 1>&2
However when I open the log file output.log, it is empty which is why the last command returns empty. Is there something I am doing wrong. I understand the above might be easily accomplished using an anonymous pipe like so: python3 helloworld.py | cat >output.log (or even python3 helloworld.py > output.log for that matter) but my intention is to understand the use of named pipes in Unix/Linux.
The python script just prints something to stdout every 1 second:
if __name__ == "__main__":
import time
try:
while True:
print("Hello, World")
time.sleep(1)
except KeyboardInterrupt:
print('Exiting.')
finally:
pass