mjhjmtu

I've been given some challenge: there is an application that provides me some postfix expression and expects to get a solution within 5 seconds.
If there is no solution, it quits, otherwise, it provides additional expression.

The interface is like follow:

<some text>
<expression>
<expecting my result>
<next expression>
<expecting my result>

etc.

If there was only single expression I've thought to use tail -1 command to get the last line to get the expression and use the dc command to evaluate it.
However, as there are unknown number of expressions (so far), I can't use it and have to do some loop till the string which is read is not starting with a number (i.e expression).

Therefore I've though to use 2 scripts:

1. start the application always send text output to file1 and execute it in the background (script 1)


2. read the last row of a script the file1 (script 2), evaluate a solution and sent it to file2 (script 2)


3. read solution from file2, delete file2 (for next solutions), bring the application to foreground, send a solution and then send the application to background (script 1)

However, it seems to be little bit complicate for me.

Any suggestions how to simplify the procedure to solve it?

I suggest looking into named pipes as you shouldn't need a file for any of this.

However, something like:

mkfifo mypipe
script1 <mypipe | grep --line-buffered '^[0-9]' | dc >mypipe

Should be close to what you want. This assumes that the input has the necessary p commands to make dc actually print the result of the expressions, though.