First you should comment out everything what prints to terminal eg "exit" at exit, printf's for debugging etc
You also need to make sure Readline is writting to stderr instead of stdout (the same way bash does).
You can change this by setting
rl_output = stderr
in your main.
Clone it to the root of your minishell
cd to the testers folder
bash tester.sh [m vm ne d b a] {b, builtins, pa, parsing, r, redirections, pi, pipelines, c, cmds, v, variables, co, corrections, path, s, syntax}
m # Mandatory tests
vm # Mandatory tests with valgrind
ne # Mandatory tests without environment
d # Mandatory pipe segfault test (BRUTAL)
b # Bonus tests
a # All the tests
For the mandatory and valgrind mandatory tests, additionnal flags can be specified :
b, builtins # builtins-specific tests
pa, parsing # parsing-specific tests
r, redirections # redirections-specific tests
pi, pipelines # pipelines-specific tests
c, cmds # full-commands tests
v, variables # environment variables tests
co, corrections # corrections' subject tests
path # path resolving tests
s, syntax # syntax errors tests
To install the script, copy and run following command:
bash -c "$(curl -fsSL https://raw.githubusercontent.com/zstenger93/42_minishell_tester/master/install.sh)"
The tester will be installed in the $HOME/42_minishell_tester
directory.
After installation an alias mstest
will be automaticly added in .zshrc
or .bashrc
So that you can run the program in any directory (where your minishell is) by calling
mstest
Tests without environment now are updated, separated and can be run with:
bash tester.sh ne
Now they should be working better, still don't trust it 100% and do test yourself as well to understand.
It's a bit tricky to do test's well in this case because if you run env -i bash
it disables only partially.
It will still have most things, but if you do unset PATH
afterwards, will see the difference.
Also this part is pretty much what you aren't required to handle.
The main point is to not to crash/segfault when you launch without environment.
DO NOT FAIL SOMEONE BECAUSE THEY AREN'T PASSING ALL TESTS!
NEITHER LET THEM PASS JUST BECAUSE THEY DO, CHECK THE CODE WELL!
DO YOUR OWN TESTING. TRY TO BREAK IT! ^^
HAVE FUN WITH YOUR BEAUTIFUL MINISHELL
Don't trust 100% the leak check, try it yourself as well and the linked tester below Try to write your own test first and don't just run a tester mindlessly You don't have to pass all the cases in this tester If you want to check leaks outside of your manual checking:
This is also a good one to check valgrind A bit more time to set it up, but worth it The first time if you run the tester above and expect a lot of errors Then redirect each of the output from stdin and strerror to a file otherwise you won't be able see all of the errors
Even though the required changes have been made to your proram, it might still going to throw you only KO STD_OUT. This is because readline version. (then you probably have the older version where it isn't checking where does the input coming from(the tester or you))
If a test just hanging in infinite loop, you can use the link to go there and comment it out in the test file until you fix it.
Upgraded by: Zsolt
Parsing hell and mini_death by: KΔrlis
Extra bonus tests by: Mouad
and
My minishell pain
Later on I might make changes so you can run test for each part like cd, echo, pipes, redirs etc.. as well Or if someone takes on the challenge, you are welcome to do so.
Feel free to ask on slack if you have a question Or open a pull request if you would like to add more tests Looking for people who would like to add more tests to the bonus part, because we haven't done it