(These will be marked as "Failing" until you have all the tests passing.)
First and Foremost: Please read the entire lab before starting. It's quite possible that some questions that arise while reading the lab are answered later in the lab. That said... let's get to to it!
This provides a basic introduction to shell programming. If you
use Linux much at all, you'll at least occasionally find yourself needing
to use the shell/command line (i.e., what you get when you open the terminal program). Having experience with the shell is extremely useful, as you often
end up needing to, e.g., ssh
into a remote, cloud-based system where you won't
have access to the nice GUI tools. This lab provides an introduction to a
variety of important shell tools and how programming/scripting is done
using shell commands.
:warning: Remember to complete the Command line introduction pre-lab reading and preparation before this lab begins.
You will need to write a few different scripts for this lab:
tar
archive, and then
compile and run the C program it containsNone of these will be very long, but most or all of them will require
you to learn new shell commands or tools. We’ll give you hints and
pointers as to what commands/tools to be using, but you’ll need to
do some digging in the man
pages and/or searching on-line to find
the details. Don’t bang your head against any piece of this for too
long. If you’ve spent more than 10 minutes on a single part or command,
you probably need to take a break and ask someone (like your instructor)
for some help. On the other side of the coin, however, don’t immediately
give up and ask at every step. Learning how to find and use this sort
of information is an enormously valuable skill, and will be useful
far longer and more often than the details that you’re actually looking
up. So make a bit of an effort, but know when to stop.
This is all structured around the bash
shell and "standard"
command-line tools like find
and grep
. There are a variety of
other shells (e.g., fish
or
zsh
) and tools
(e.g., fd
and
rg
),
but those won't necessarily be available on the the random system you
find yourself working on, so we're sticking to the "old standbys".
Before you start writing scripts, you’ll need get a copy of this repository to work on. This is a two step process:
:warning: IMPORTANT :warning: Because of a bug in GitHub's handling of templates, your copy of the repository will not work as it is. You'll need to run the following two commands to import the submodule files correctly:
git submodule init
git submodule update
If you're working in pairs or larger groups only one of you needs to create
your group's copy in GitHub Classroom, but everyone else will need to join
that team in GitHub Classroom so they have access to their team's project.
Also note that if Pat checks out the project on the first day of lab, and then
later Chris is logged in when you sit down to work on it again, Chris will
need to check out the project. Similarly, if Pat and Chris are working
on different computers, then both of them will need to clone
a local copy.
It's also crucial that everyone commit and push their
work (perhaps to a branch) at the end of each work session so that it will
accessible to everyone in the team.
You’ll “turn in” your work simply by having it committed to the repository and pushed to GitHub. We’ll check it out from there to run and grade it. We'll obviously need to be able to find your repository to grade it, so make sure to submit the URL of your repository using whatever technique is indicated by your instructor.
Be certain to commit often, and frequently trade places as driver and navigator. At a minimum you should probably trade every time you solve a specific problem that comes out of the test script. You should probably consider committing that often as well.
Part of this lab's rubric is readability, and shell scripts are notoriously difficult to read. So remember all the nice habits that you've learned, like using good variable names and commenting non-obvious commands. It's worth noting that almost everything in a shell script is non-obvious when you first see it. While we definitely would not recommend commenting every line of a program in a language like Java or Python, commenting every (or nearly every) line in a shell script isn't a bad idea. Students almost never comment too much on these, so when in doubt comment more rather than less.
You should make sure you run the shellcheck
command on your shell
scripts, e.g.,
/snap/bin/shellcheck big_clean.sh
and heed (or at least ask questions about) any warnings that it generates.
Some resources on shell style:
examples
directory
that has some nice examples, such as the steam.sh
example_
) at the front
variable names don't seem all that "standard", so follow your heart on
those issues.shellcheck
looks for. It would probably make sense to at least skim this since it's quite concise, and we're using shellcheck
so the payoff will be fairly direct. They don't actually explain why
these things are bad, though, so feel free to ask questions
if you're uncertain about something (esp. if it comes up when
you run shellcheck
).You should complete the following exercises for this lab, each of which has tests and relevant files in the indicated sub-directory:
compiling
)cleaning
)The tests and any relevant files for each part are in the appropriate sub-directory in this repository:
compiling
cleaning
In each case there are tests written using the bats
testing tool for bash
scripts in a file called bats_tests.sh
.
You should be able to run the tests with bats bats_tests.sh
, and use the testing
results as a guide to the development of your scripts. If you ever find that you
don't understand what the tests are "telling you", definitely ask; they are
there to help you, and if they aren't communicating effectively then they're
not doing their job.
You should get all the tests to pass before you "turn in" your work. Having the tests pass doesn't guarantee that your scripts are 100% correct, but it's a strong initial indicator.
The tests and data for this problem are in the compiling
directory of this project, and the discussion of this problem will all assume that you've cd
ed into that directory. Your goal is complete the desired
script. This includes at a minimum getting the tests in bats_tests.sh
(in the compiling
directory) to pass.
For this you should write a bash script called extract_and_compile.sh
that:
NthPrime.tgz
into the
current directory.
gz
, for gzip
, in the file extension),
so you’ll need to uncompress and then extract.gunzip
and then extract with tar
. Even
better, though, is to use tar
(with the right flags) to do both
things in one step.man
pages
for tar
useful.NthPrime
in your current
directory; that NthPrime
directory should contain several *.c
and
*.h
files that can be compiled to create an executable.NthPrime
directory that the tar
extraction created.NthPrime
in the same directory.
gcc
compiler.NthPrime
). NthPrime
requires a single
number as a command line argument; you should pass it the command line
argument your script received.As an example, imagine you are in the directory /home/chris/lab0
and
your script is called using:
./extract_and_compile.sh 17
Then it should
NthPrime.tgz
into
the current directory, creating a directory /home/chris/lab0/NthPrime
/home/chris/lab0/NthPrime
to generate the
binary /home/chris/lab0/NthPrime/NthPrime
.17
(the first argument in
this example); this should generate the output Prime 17 = 59.
:exclamation: When you run the program you compiled (NthPrime
) you need give
NthPrime
a single command line argument. The value you should pass it is
the number your script received as its command line argument.
The final file structure in the example above (as
displayed by the tree
program) should be:
$ tree .
.
└── NthPrime
├── NthPrime
├── main.c
├── nth_prime.c
└── nth_prime.h
1 directory, 4 files
Please note that if you are using bash on windows (say, perhaps, via the Linux for Windows Subsystem) then you may need to install gcc
and/or bats
. Talk to your instructor about how to do this if the situation arises (the exact installation command depends upon which distribution of linux is installed)
Let the provided Bats tests drive your solution. Run bats bats_tests.sh
and look at the first failure. What's the simplest thing you can do to
get that test to pass? Look at the source for the tests (in bats_tests.sh
)
for hints on what to do if a given test fails.
The tests require that the .tgz
version of the tar archive will still
be in the specified directory
when you’re done. This means that if you first gunzip
and then, in a
separate step, untar, the test is likely to fail since you’ll end up
with a .tar
file instead of a .tgz
file. So you should use the appropriate tar
flags that uncompress and untar in a single step.
The tests also assume that your script generates no
"extraneous" output. If, for example, you use the -v
flag with tar
, you'll generate a bunch of output that
will cause some of the tests to fail. You may want to have
"extra" output as a debugging tool while you're working
on the script, but you'll need to remove all that to get
the tests to pass. This is consistent with standard
practice in Unix shell programming, where most commands
provide little to no output if things went fine, making it
much easier for you to chain them together into more
complex behaviors.
Remember that you can call your script "by hand" as a debugging aid so you can see exactly what it's doing and where. So you could do something like
./extract_and_compile.sh 8
and then go look around and see what your script did. You'll want to clean
up after each test like that (e.g., rm -rf NthPrime
) to make sure that
your script successfully re-creates NthPrime
and doesn't "succeed" just
because NthPrime
was left over from an earlier run. If you lose
NthPrime.tgz
the command git restore NthPrime.tgz
will bring it back,
assuming you having committed the deletion.
The C compiler in the lab is the Gnu C Compiler: gcc
.
There are two .c
files in this program, both of which will need to be
compiled and linked for form an executable. You can do this in a single line (handing gcc both
.c
files) or you can compile them separately and then link them.
You can tell gcc
what you want the executable called, or you can take
the default output and rename it.
Most of you have never compiled a C program before, so this might be a
good time to ask me to say a little about how that works. Alternately,
you might see what you can figure out with man gcc
.
Your goal here to build a script that removes files that have been marked
for deletion; at a minimum, you want to get the tests in bats_tests.sh
to pass and make sure shellcheck
is happy. For this you should write a
bash script named big_clean.sh
that:
tar
archive (.tgz
file
that contains the files you'll process.mktemp
for this.tar
archive in their directory.tar
archive into the newly created
scratch directory.
tar
that allows you specify where
the extracted files should go, which is a lot cleaner than extracting
them "where they stand" and then moving them to the target directory.
It also ensures that there won't be a conflict with any existing files.big_dir.tgz
since that has
over 1000 files in it.tar
archive) containing the line “DELETE ME!”, while
leaving all the others alone.
grep
family of tools is probably the easiest way to
see if a file has the “DELETE ME!” line. You could then use xargs
to remove the ones that have the line.grep
and a shell
loop to loop through all the files and remove the ones that have the
line.while/read
approach
described on this page about properly
unquoting variables.tar
archive that contains the files in the scratch directory after you've removed the "DELETE ME!" files. The files in the archive should not have the path to the scratch directory in their filenames. The new tar file should have the name cleaned_...
where the ellipsis is replaced by the name of the original file, e.g., if your original file is little_dir.tgz
then the newly created file should be called cleaned_little_dir.tgz
. The commands basename
and dirname
might be useful in constructing the desired file name.
tar
archive or you'll end up with the path to the scratch directory in all the file names.cd $SCRATCH
or pushd $SCRATCH
to get to the scratch directory to run the tar -zcf...
command, but then how do you know where you came from, so you can put the new tar file in the right place? The pwd
command returns your current working directory, so something like here=$(pwd)
will capture your current directory in a shell variable called here
so you can use $here
later to refer to where you had been.If we assume that your scratch directory is, for example,
/tmp/tmp.eMvVweqb
, then after the first step
(uncompressing) the sample tar file little_dir.tgz
you should end up with:
$ tree /tmp/tmp.eMvVweqb/
/tmp/tmp.eMvVweqb/
└── little_dir
├── file_0
├── file_1
├── file_10
├── file_11
├── file_12
├── file_13
├── file_14
├── file_15
├── file_16
├── file_17
├── file_18
├── file_19
├── file_2
├── file_3
├── file_4
├── file_5
├── file_6
├── file_7
├── file_8
└── file_9
1 directory, 20 files
Then after deleting the appropriate files, you should have:
$ tree /tmp/tmp.eMvVweqb/
/tmp/tmp.eMvVweqb/
└── little_dir
├── file_1
├── file_10
├── file_11
├── file_12
├── file_15
├── file_16
├── file_17
├── file_18
├── file_19
├── file_2
├── file_3
├── file_4
├── file_5
├── file_6
├── file_8
└── file_9
1 directory, 16 files
Finally, after creating the new cleaned tar file
(cleaned_little_dir.tgz
in this case) your project
directory should look like:
$ tree cleaning/
cleaning/
├── bats_tests.sh
├── big_dir.tgz
├── cleaned_little_dir.tgz
└── little_dir.tgz
0 directories, 4 files
The tests assume that the .tgz
version of the tar archive will be in the specified directory
when you’re done. This means that if you first gunzip
and then, in a
separate step, untar, the test is likely to fail since you’ll end up
with a .tar
file instead of a .tgz
file. So you should use the appropriate tar
flags that uncompress and untar in a single step.
You should also assume that if you untar frogs.tgz
that will result in a directory called frogs
that
contains the files you need to process. (There's nothing
magic about tar
that requires this to be true – see
"Notes on archive structures" for more).
You can assume that the first argument
has the form frogs.tgz
and not some alternative like
frogs.tar.gz
.
The tests assume that your script generates no
"extraneous" output. If, for example, you use the -v
flag with tar
, you'll generate a bunch of output that
will cause some of the tests to fail. You may want to have
"extra" output as a debugging tool while you're working
on the script, but you'll need to remove all that to get
the tests to pass. This is consistent with standard
practice in Unix shell programming, where most commands
provide little to no output if things went fine, making it
much easier for you to chain them together into more
complex behaviors.
Make sure that all your code passes the appropriate tests. Passing the test will make up the majority of your grade. There will also be a portion of your grade that will take into account how clean your code is. Also when we said that you should commit often –- we meant it. Also be professional and informative about your commit messages; we'll be looking at them in the grading. Finally, it is easy to overlook important details. If the test isn’t being passed go back and re-read the directions carefully.
You'll "turn this in" by committing your work to your GitHub Classroom copy of the project. You should also submit the URL of your repository in whatever way indicated by your instructor. Remember to make sure you've completed each of the assigned tasks:
compiling/extract_and_compile.sh
cleaning/big_clean.sh