Operating systems 1 -- 2008-2009 -- info.uvt.ro/Laboratory 4

Quick links: front; laboratories agenda, 1, 2, 3, 4, 5, 6, projects, evaluation, tools, references.

Exit status

each process has an exit status;
if a command succeeds its exit status is 0:

true
echo "${?}"
cat /etc/passwd
echo "${?}"

if a command fails its exit status is different that 0:

false
echo "${?}"
cat /tmp/non-existing-file
echo "${?}"

to exit from a script with a certain status:

exit <status>

references:

Command chaining

syntax:

<command> [argument] ... { && <command> [argument] ... } ...
<command> [argument] ... { || <command> [argument] ... } ...

execute something if a command has succeeded:

true && echo 'command succeeded.'
false && echo 'command succeeded.'
test -f /etc/passwd && echo 'file exists.'

execute something if a command has failed:

true || echo 'command failed!'
false || echo 'command failed!'
test -f /tmp/non-existing-file || echo 'file does not exist!'

references:
- Advanced Bash-Scripting Guide:
  - List Constructs;

'[' or test command

syntax: please consult the man page;
testing if an argument is empty:

test -z "${1}" || echo 'empty.'
test ! -z "${2}" && echo 'non-empty.'
test -n "${1}" || echo 'empty.'
test -n "${1}" && echo 'non-empty.'

testing if a file exists (otherwise creating it):

test -f /tmp/some-file-1 || nano /tmp/some-file-1
test ! -f /tmp/some-file-2 && nano /tmp/some-file-2

testing if a directory exists (otherwise creating it):

test -d /tmp/some-directory || mkdir /tmp/some-directory

testing if two arguments are equal:

test "${1}" == "${2}" || echo 'non equal.'
test "${1}" != "${2}" && echo 'non equal.'

testing if two arguments are less one than another (integer comparison):

test "${1}" -lt "${2}" && echo 'lesser.'
test "${2}" -lt "${2}" || echo 'greater.'

important operators:
- !;
- string: -z, -n, =, !=;
- numbers: -eq, -ne, -gt, -lt, -le, -ge;
- files: -e, -f, -d, -x, -w;
references:
- Advanced Bash-Scripting Guide:
- test.1;

Command grouping and sub-shells

block syntax:

{ <command> [arguments] ... ; <command> [arguments] ... ; ... ; }
{
    <command> [arguments] ...
    <command> [arguments] ...
    ...
}

simple example:

{ echo a ; echo b ; }

do not forget the last ;:

{ echo a ; echo b }

if we want to take more actions if a file does not exist:

cat /tmp/expected-file-1 || { echo 'file does not exist; creating...' ; sleep 5 ; nano /tmp/expected-file-1 ; }

or:

cat /tmp/expected-file-2 || {
    echo 'file does not exist; creating...'
    sleep 5
    nano /tmp/expected-file-2
}

sub-shell syntax: just like the block syntax, but with () instead of {};
if we issue exit from a sub-shell only it terminates, not the entire shell:

( exit 1 ; ) || echo 'sub-shell failed!'

in contrast if we issue exit' from a block, the current shell exits:

{ exit 1 ; } || echo 'block failed!'

references:
- Advanced Bash-Scripting Guide:

Loops and branches

if

syntax:

if <command> <argument> ... ; then { <command> <argument> ... ; } ... else { <command> <argument> ... ; } fi
if <command> <argument> ...
then
    <command> <argument> ...
    ...
else
    <command> <argument> ...
    ...
fi

testing if a file exists, and if not creating an empty one:

if test ! -f /tmp/expected-file-3 ; then touch /tmp/expected-file-3 ; fi

testing if a file exists, and if not editing it:

if test ! -f /tmp/expected-file-4
then
    echo 'file does not exit; creating...'
    touch /tmp/expected-file-4
    sleep 1
    nano /tmp/expected-file-4
else
    echo 'file exists.'
fi

references:
- Advanced Bash-Scripting Guide:

case

syntax:

case <string> in
    ( <pattern> { | <pattern> } ... )
        <command> <argument> ...
        ...
    ;;
    ...
esac

example script for controling a daemon:

case "${1}" in
    ( start )
        echo 'starting daemon...'
        daemon &
    ;;
    ( stop )
        echo 'stopping daemon...'
        killall daemon
    ;;
    ( * )
        echo "unknown action ${1}!"
    ;;
esac

references:
- Advanced Bash-Scripting Guide:
  - Testing and Branching;

for

syntax:

for <variable> in <string> ... ; do { <command> <argument> ... ; } ... done
for <variable> in <string> ...
do
    <command> <argument> ...
    ...
done

simple for command:

for a in 1 2 3 4 ; do echo "${a}" ; done

or:

for a in 1 2 3 4
do
    echo "${a}"
done

editing all .bash<something> files in the current folder:

for file in ./.bash* ; do nano "${file}" ; done

or:

for file in ./.bash*
do
    echo "editing file ${file}..."
    sleep 5
    nano "${file}"
done

references:
- Advanced Bash-Scripting Guide:
  - Loops;
  - Loop Control;
  - Nested Loops;

while

syntax -- using a test command:

while <command> <argument> ... ; do { <command> <argument> ... ; } ... done
while <command> <argument> ...
do
    <command> <argument> ...
    ...
done

syntax -- reading lines:

while read variable ; do { <command> <argument> ... ; } ... done
while read variable
do
    <command> <argument> ...
    ...
done

waiting for a file to be created:

while test ! -f /tmp/expected-file-5 ; do sleep 1 ; done

or:

while test ! -f /tmp/expected-file-5
do
    echo 'file does not exit; waiting to be created...'
    sleep 1
done

doing something with all files (recursively in a folder):

find /etc -type f | while read file
do
    echo "found file ${file}."
done

references:
- Advanced Bash-Scripting Guide:
  - Loops;
  - Loop Control;
  - Nested Loops;

Exercises

write a script that takes as arguments a file and a folder, and tries to copy the file into the specified folder:
1. write the script without any checks;
2. make it check if the source exists, and if not refuse to do anything (exit);
3. make it check if the destination exists, and if so refuse to do anything (exit);
4. make it check if the destination folder exists, and if not tries to create it;
5. make it check if the source is a folder (and if so copy it recursively);

write a script that takes as argument a file name, and writes to the standard output if it represent a file, folder, something else, or if it doesn't exist;

write a script like the previous one, but which takes any number of arguments that represent file names; (hint: use for;)

write a script that finds all the users defined in the /etc/passwd file (passwd.5); (hint: use twice grep with -o, or sed;)

write a script that uses the previous one, and checks for which users there is a homedir (/home/<username>); (hint: use ./find-users.sh | while read user ...;)

Ciprian Dorin Crăciun, ccraciun@info.uvt.ro