如何在循环中使用os.pipe（）（复制多个管道）？

Question

I am trying to replicate 'cat < hello.txt | cat | cat > hello2.txt' 'cat < hello.txt | cat | cat > hello2.txt' 'cat < hello.txt | cat | cat > hello2.txt' in python by only using os.pipe() . I am not trying to make something fancy, it is just for education(uni).

I am really confused on how to do this. But I am thinking of doing it as follows:

so i made a list of all commands between piping is to be performed -

pipe_commands = [['/bin/cat < hello.txt'] , ['/bin/cat'], ['/bin/cat > hello2']]

for i in pipe_command:
    r, w = os.pipe()
    if '<' in i:
        pid =  os.fork() 
        if pid == 0:
            os.close(r)
            w = os.fdopen(w, 'w')   # Start Writting on write end of pipe 
            os.dup2(w.fileno(), sys.stdout.fileno())

            f= open('hello.txt', 'r')   # start reading from hello.txt 
            os.dup2(f.fileno(), sys.stdin.fileno())

            os.execl('/bin/echo', 'echo')
    else:
        os.wait()
        os.close(w)
        r = os.fdopen(r, 'r')
        os.dup2(r.fileno(), sys.stdin.fileno()) # read from read end of pipe 
        # OUTPUT TO BE GIVEN TO NEXT COMMAND

    elif '>' in i:

        pid =  os.fork() 

        if pid == 0:

            os.close(w)         # Start reading from previous commands pipe output 
            r = os.fdopen(r, 'r')
            os.dup2(r.fileno(), sys.stdin.fileno())

            f = open('hello2.txt', 'w')     # write to hello2.txt
            os.dup2(f.fileno(), sys.stdout.fileno())

            os.execl('/bin/echo', 'echo')

        else:
            os.wait()

    else:
        pid =  os.fork() 
        if pid == 0:
            os.close(r)
            w = os.fdopen(w, 'w')   # Start Writting on write end of pipe 
            os.dup2(w.fileno(), sys.stdout.fileno())

            os.execl('/bin/echo', 'echo')   #input from read end of the previous command  
        else:
            os.wait()
            os.close(w)
            r = os.fdopen(r, 'r')
            os.dup2(r.fileno(), sys.stdin.fileno()) # read from read end of pipe 
        # OUTPUT TO BE GIVEN TO NEXT COMMAND

I am confused after this, who do i make it next command which is '/bin/cat' read from the output to perform the first piping ( cat < hello.txt | cat ) ?

Also any hints on how to put it on loop so it is automated? (I know the for loop i have used is wrong)

I know it is not the most optimal solution to perform piping but we are only taught os.pipe() and are refrained from using os.system and os.subprocess .

Thanks in Advance

Answer 1

You look like you're going the right way with this.

The basic idea is that stdin, stdout and/or stderr need to be replaced with a pipe for the processes to communicate with each other. Since you'll be doing that several times, it's good to have a function for it.

eg

def start_process(command, newin, newout, newerr):
    pid = os.fork()
    if pid == 0:
        # Child process
        os.dup2(newin, 0)  # Replace stdin
        os.dup2(newout, 1)  # Replace stdout
        os.dup2(newerr, 2)  # Replace stderr
        os.execv(command, (command,))
    else:
        # Parent process
        return pid

Once you have the function, you can start to think about how you want to run the commands. I'll leave putting the loop to you, but it's something like the following.

# cat < hello.txt
# stdin=hello.txt, stdout=pipe1, stderr=unchanged
hello = open('hello.txt', 'r')
pipe1_read, pipe1_write = os.pipe()
proc1 = start_process('/bin/cat', hello.fileno(), pipe1_write, sys.stderr.fileno())
os.close(pipe1_write)  # only proc1 should have this

# /bin/cat
# stdin=pipe1 stdout=pipe2, stderr=unchanged
pipe2_read, pipe2_write = os.pipe()
proc2 = start_process('/bin/cat', pipe1_read, pipe2_write, sys.stderr.fileno())
os.close(pipe2_write)  # only proc2 should have this

# etc...

Once you've started all the processes, you just need to wait for them to finish.

os.waitpid(proc1, 0)
os.waitpid(proc2, 0)
os.waitpid(proc3, 0)