I want to push my input parameters through two sets of checks. Each of these checks is a function. So case would not work because in KSH.Case will "esac" after satisfying the 1st condition.
die () {
echo "ERROR: $*. Aborting." >&2
exit 1
}
var=$1
var1=$2
[ -z "$var1" ] && var1=$var
echo "var is $var . Var1 is $var1 "
# test $var1 != $var2 ||
dtc() {
# small function where I am checking if the input parameter is a valid date
}
vlc(){
# function where I am checking if the input parameters year is after 2012, because earlier years cannot exist
}
if dtc $var && dtc $var1
then
if vlc $var && vlc $var1
then
stuff
else
die "message"
fi
else
die "message"
fi
The nested if looks a bit clumsy. If there is a more elegant way convey this to shell.
Improving your indentation will go a long way toward readable code.
If your die message are different, then you have no choice. If they are the same, then you can combine the conditions:
If you just want to die if not all the commands are successful, you can write
dtc "$var" && dtc "$var1" &&
vlc "$var" && vlc "$var1" || die "message"
# OK, all passed
stuff
One style tip for if-else is to put the shorter block first
if ! { dtc "$var" && dtc "$var1"; }
then
die "message 1"
else
if ! { vlc "$var" && vlc "$var1"; }
then
die "message 2"
else
we will do
lots of stuff
here
fi
fi
and of course, functions to encapsulate code
stuff() {
we will do
lots of stuff
here
}
Related
I'm new to bash so assume that I don't understand everything in this simple script as I've been putting this together as of today with no prior experience with bash.
I get this error when I run test.sh:
command substitution: line 29: syntax error near unexpected token `$1,'
./f.sh: command substitution: line 29: `index_of($1, $urls))'
FILE: f.sh
#!/bin/bash
urls=( "example.com" "example2.com")
error_exit()
{
echo "$1" 1>&2
exit 1
}
index_of(){
needle=$1
haystack=$2
for i in "${!haystack[#]}"; do
if [[ "${haystack[$i]}" = "${needle}" ]]; then
echo "${i}"
fi
done
echo -1
}
validate_url_param(){
index=-2 #-2 as flag
if [ $# -eq 0 ]; then
error_exit "No url provided. Exiting"
else
index=$(index_of($1, $urls)) #ERROR POINTS TO THIS LINE
if [ $index -eq -1 ]; then
error_exit "Provided url not found in list. Exiting"
fi
fi
echo $index
}
FILE: test.sh
#!/bin/bash
. ./f.sh
index=$(validate_url_param "example.com")
echo $index
echo "${urls[0]}"
I've lost track of all of the tweaks I tried but google is failing me and I'm sure this is some basic stuff so... thanks in advance.
The immediate error, just like the error message tells you, is that shell functions (just like shell scripts) do not require or accept commas between their arguments or parentheses around the argument list. But there are several changes you could make to improve this code.
Here's a refactored version, with inlined comments.
#!/bin/bash
urls=("example.com" "example2.com")
error_exit()
{
# Include script name in error message; echo all parameters
echo "$0: $#" 1>&2
exit 1
}
# A function can't really accept an array. But it's easy to fix:
# make the first argument the needle, and the rest, the haystack.
# Also, mark variables as local
index_of(){
local needle=$1
shift
local i
for ((i=1; i<=$#; ++i)); do
if [[ "${!i}" = "${needle}" ]]; then
echo "${i}"
# Return when you found it
return 0
fi
done
# Don't echo anything on failure; just return false
return 1
}
validate_url_param(){
# global ${urls[#]} is still a bit of a wart
if [ $# -eq 0 ]; then
error_exit "No url provided. Exiting"
else
if ! index_of "$1" "${urls[#]}"; then
error_exit "Provided url not found in list. Exiting"
fi
fi
}
# Just run the function from within the script itself
validate_url_param "example.com"
echo "${urls[0]}"
Notice how the validate_url_param function doesn't capture the output from the function it is calling. index_of simply prints the result to standard output and that's fine, just let that happen and don't intervene. The exit code tells us whether it succeeded or not.
However, reading the URLs into memory is often not useful or necessary. Perhaps you are simply looking for
grep -Fx example.com urls.txt
getAnimalFolder() {
local animal=""
if [[ ${ANIMAL} == "lion" ]]; then
animal = "./animals/lion/"
elif [[ ${ANIMAL} == "tiger" ]]; then
animal = "./animals/tiger/"
elif [[ ${ANIMAL} == "cheetah" ]]; then
animal = "./animals/cheetah/"
else
echo "inavalid animal"
exit 1`enter code here`
fi
echo $animal
}
result=$(getAnimalFolder)
cd ../result/age/
If the animal is not lion, tiger or cheetah, the function returns invalid animal and hence gives an error 'No such file or directory', instead I need to do an exit with code = 1. Hence I went for the second option -
if [[ ${ANIMAL} != "lion" && ${ANIMAL} != "tiger" && ${ANIMAL} != "cheetah" ]]; then
echo "Invalid animal"
exit 1
fi
getAnimalFolder() {
local animal=""
if [[ ${ANIMAL} == "lion" ]]; then
animal = "./animals/lion/"
elif [[ ${ANIMAL} == "tiger" ]]; then
animal = "./animals/tiger/"
elif [[ ${ANIMAL} == "cheetah" ]]; then
animal = "./animals/cheetah/"
fi
echo $animal
}
result=$(getAnimalFolder)
cd ../result/age/
This looks like a fix to my problem but if in the future more animals are added, then I need to remember to make changes in 2 places for every new animal added. So is there a better way to do this?
There are a number of problems here; #1 and #3 are the ones that directly address your question.
When a function/command/whatever may need to print both regular output (e.g. the path to an animal directory) and error/status output (e.g. "inavalid animal"), it should send the regular output to standard output (aka stdout aka FD #1, the default), and error/status output to standard error (aka stderr aka FD #2), like this:
echo "Invalid animal" >&2 # This is sent to stderr
Generally, functions should return rather than exiting. If a function does exit, it exits the entire shell, but in this case the function is running in a subshell due to $( ), so it only exits that. Using return avoids this inconsistency.
When a function/command/whatever may fail, you should check its exit status; there are a number of ways to do this:
if result=$(getAnimalFolder); then
: # command succeeded!
else
echo "OMG it failed!" >&2
exit 1
fi
or
result=$(getAnimalFolder)
if [ $? -ne 0 ]; then # $? is the status of the last command
echo "OMG it failed!" >&2
exit 1
fi
or
result=$(getAnimalFolder) || {
echo "OMG it failed!" >&2
exit 1
}
I use the last form a lot, since there are a lot of steps in a script might fail, and having a simple & compact way to include the failure handing code makes the overall script more readable.
In general, functions should take their input as arguments rather than via global variables. So in the function you'd refer to $1 instead of $ANIMAL, and you'd run the function with something like:
result=$(getAnimalFolder "$ANIMAL")
There are also a number of basic syntax errors and bad scripting practices in the script: don't put spaces around the equal sign in assignments; do put double-quotes around variable references; don't use all-caps variable names (to avoid conflicts with the many all-caps variables that have special meanings); do check for errors on cd commands (if they fail, the rest of the script will run in the wrong place); and when comparing a single variable against a bunch of values, use case instead of a bunch of if elseif etc.
shellcheck.net is good at recognizing many of these common mistakes. Strongly recommended.
Here's what I get with all fixes in place:
#!/bin/bash
getAnimalFolder() {
local animalname=$1
local animaldir=""
case "$animalname" in
lion ) animaldir="./animals/lion/" ;;
tiger ) animaldir="./animals/tiger/" ;;
cheetah ) animaldir="./animals/cheetah/" ;;
* )
echo "Invalid animal: $animalname" >&2
return 1 ;;
esac
echo "$animaldir"
}
read -p "Give me an animal: " animalname
result=$(getAnimalFolder "$animalname") || {
exit 1 # Appropriate error message has already been printed
}
cd "../$result/age/" || {
echo "Error changing directory to ../$result/age/ -- aborting" >&2
exit 1
}
Put the animals in an array:
#!/bin/bash
animals=(lion tiger cheetah)
getAnimalFolder() {
local i
for i in "${animals[#]}"; do
if [ "$i" == "${1}" ] ; then
animaldir="./animals/${1}"
return 0
fi
done
exit 1
}
read -rp "Give me an animal: " animalname
getAnimalFolder "${animalname}"
echo "Animaldir=${animaldir}"
EDIT:
I did not use the construction result=$(getAnimalFolder), assuming the OP wants to use the new path once. When needed, the function can be changed into
echo "./animals/${1}"
When the function is called with result=$(getAnimalFolder), OP needs to look at the line
cd ../result/age/
Is resulta fixed path or does he want to use the path from the function:
cd ../${result}/age/
I am trying to check for duplicate records in my database using shell scripting.
For this, I have created a function named "check()" which echo's True or False and is stored in variable "result". But while evaluating using if statement it is always returning "True".
#redundancy check function
check() {
temp=$(grep -w -c "$1" database.dat)
echo $temp
if [ "$temp" != 0 ]
then
echo True
else
echo False
fi
}
insert() {
option="y"
while [ "$option" == "y" ]
do
echo "Rollno: \c"
read roll
result="$(check $roll)"
echo $result
if [ "$result" == "False" ]
then
echo Do something
else
echo "ERROR: Duplicate record found...\nEXITING...\n"
option="n"
fi
done
}
If you're using a shell that doesn't support the == extension to test, then your tests will always, unconditionally fail simply on account of invalid syntax. Use = for string comparisons to be portable to all POSIX-compliant implementations.
Moreover, there's no point to storing and then comparing the output from grep at all: Use the exit status of grep -q when your only goal is to check whether the number of matches is zero or more-than-zero; this allows grep to exit immediately when a match is seen, rather than needing to read the rest of the file.
# with -q, this emits no stdout, but exits w/ status 0 (matches exist) or 1 (otherwise)
check() { grep -q -w -e "$1" database.dat; }
insert() {
option=y
while [ "$option" = y ]; do
printf '%b\n' "Rollno: \c"
read -r roll
if check "$roll"; then
printf "ERROR: Duplicate record found...\nEXITING...\n"
option=n
else
echo "Check failed; do something"
fi
done
}
In perl, you can exit with an error msg with die "some msg". Is there an equivalent single command in bash? Right now, I'm achieving this using commands: echo "some msg" && exit 1
You can roll your own easily enough:
die() { echo "$*" 1>&2 ; exit 1; }
...
die "Kaboom"
Here's what I'm using. It's too small to put in a library so I must have typed it hundreds of times ...
warn () {
echo "$0:" "$#" >&2
}
die () {
rc=$1
shift
warn "$#"
exit $rc
}
Usage: die 127 "Syntax error"
This is a very close function to perl's "die" (but with function name):
function die
{
local message=$1
[ -z "$message" ] && message="Died"
echo "$message at ${BASH_SOURCE[1]}:${FUNCNAME[1]} line ${BASH_LINENO[0]}." >&2
exit 1
}
And bash way of dying if built-in function is failed (with function name)
function die
{
local message=$1
[ -z "$message" ] && message="Died"
echo "${BASH_SOURCE[1]}: line ${BASH_LINENO[0]}: ${FUNCNAME[1]}: $message." >&2
exit 1
}
So, Bash is keeping all needed info in several environment variables:
LINENO - current executed line number
FUNCNAME - call stack of functions, first element (index 0) is current function, second (index 1) is function that called current function
BASH_LINENO - call stack of line numbers, where corresponding FUNCNAME was called
BASH_SOURCE - array of source file, where corresponfing FUNCNAME is stored
Yep, that's pretty much how you do it.
You might use a semicolon or newline instead of &&, since you want to exit whether or not echo succeeds (though I'm not sure what would make it fail).
Programming in a shell means using lots of little commands (some built-in commands, some tiny programs) that do one thing well and connecting them with file redirection, exit code logic and other glue.
It may seem weird if you're used to languages where everything is done using functions or methods, but you get used to it.
# echo pass params and print them to a log file
wlog(){
# check terminal if exists echo
test -t 1 && echo "`date +%Y.%m.%d-%H:%M:%S` [$$] $*"
# check LogFile and
test -z $LogFile || {
echo "`date +%Y.%m.%d-%H:%M:%S` [$$] $*" >> $LogFile
} #eof test
}
# eof function wlog
# exit with passed status and message
Exit(){
ExitStatus=0
case $1 in
[0-9]) ExitStatus="$1"; shift 1;;
esac
Msg="$*"
test "$ExitStatus" = "0" || Msg=" ERROR: $Msg : $#"
wlog " $Msg"
exit $ExitStatus
}
#eof function Exit
I am working on a bash script where I need to conditionally execute some things if a particular file exists. This is happening multiple times, so I abstracted the following function:
function conditional-do {
if [ -f $1 ]
then
echo "Doing stuff"
$2
else
echo "File doesn't exist!"
end
}
Now, when I want to execute this, I do something like:
function exec-stuff {
echo "do some command"
echo "do another command"
}
conditional-do /path/to/file exec-stuff
The problem is, I am bothered that I am defining 2 things: the function of a group of commands to execute, and then invoking my first function.
I would like to pass this block of commands (often 2 or more) directly to "conditional-do" in a clean manner, but I have no idea how this is doable (or if it is even possible)... does anyone have any ideas?
Note, I need it to be a readable solution... otherwise I would rather stick with what I have.
This should be readable to most C programmers:
function file_exists {
if ( [ -e $1 ] ) then
echo "Doing stuff"
else
echo "File $1 doesn't exist"
false
fi
}
file_exists filename && (
echo "Do your stuff..."
)
or the one-liner
file_exists filename && echo "Do your stuff..."
Now, if you really want the code to be run from the function, this is how you can do that:
function file_exists {
if ( [ -e $1 ] ) then
echo "Doing stuff"
shift
$*
else
echo "File $1 doesn't exist"
false
fi
}
file_exists filename echo "Do your stuff..."
I don't like that solution though, because you will eventually end up doing escaping of the command string.
EDIT: Changed "eval $*" to $ *. Eval is not required, actually. As is common with bash scripts, it was written when I had had a couple of beers ;-)
One (possibly-hack) solution is to store the separate functions as separate scripts altogether.
The cannonical answer:
[ -f $filename ] && echo "it has worked!"
or you can wrap it up if you really want to:
function file-exists {
[ "$1" ] && [ -f $1 ]
}
file-exists $filename && echo "It has worked"