Linux Script: Command Line Argument - shell

Question

Trying to write a script that accepts exactly one command line argument and must be a positive number. I have the one argument down, but can't figure out the positive number.
I have the following:
#!/bin/ksh
NUMBER=$1
# Accepts only one argument or does not execute.
if [ "$#" != 1 ]
then
echo "error: program must be executed with 1 argument."
exit
else
if [ "$#" < 1 ] # **NEED HELP HERE**. Check for negative number.
then
echo "error: argument must be a positive number."
fi
fi
do
echo -n $NUMBER
if [ $NUMBER -gt 1 ]
then
echo -n ", "
fi
NUMBER=$(($NUMBER - 1))
done
echo
...Need help with identifying if command line argument is a negative number.
Thanks

Answer 1

Verify it is a number using by checking the whether it is only digits:
#!/bin/ksh
NUMBER=$1
# Accepts only one argument or does not execute.
if [ "$#" != 1 ]
then
echo "error: program must be executed with 1 argument."
exit
else
case $NUMBER in
''|*[!0-9]*)
echo "error: argument must be a positive number."
exit
;;
esac
fi
while ! [ $NUMBER -lt 1 ]
do
echo -n $NUMBER
if [ $NUMBER -gt 1 ]
then
echo -n ", "
fi
NUMBER=$(($NUMBER - 1))
done
echo

Answer 2

Following will work:
test "$NUMBER" -ge 0 && printf "NOT NEGATIVE" || printf "NEGATIVE"

Answer 3

Assuming we're only talking about positive (base-10) integers ... and excluding commas, exponential notation and leading plus sign (+) ...
# test for ${NUMBER} containing anything other than digits 0-9
if [[ "${NUMBER}" = #(*[!0-9]*) ]]
then
echo "error: argument is not a positive number."
else
echo "success: argument is a positive number"
fi
Obviously it becomes a bit more interesting as you strip away the assumptions on what's defined as a *valid* positive number.
Expanding on the solution proposed by sqrt163 ... a simple function should allow us to handle exponential notation, float/real and numbers with a leading plus sign (+), while allowing the calling process to filter out messages about syntax errors ...
$cat test.sh
#!/bin/ksh
NUMBER=$1
isnumber()
{
# following should generate a non-zero for negative and non-numerics,
# while allowing the parent process to filter out syntax errors
[[ "${1}" -gt 0 ]]
}
# direct isnumber() output to /dev/null since we're only interested
# in the return code, and we want to mask any syntax issues if
# isnumber() is called with a non-numeric argument; unfortunately,
# for a valid numeric that contains a comma (eg, NUMBER = 1,000)
# our isnumber() call will fail
if isnumber ${NUMBER} >/dev/null 2>&1
then
echo "${NUMBER} is a positive number"
else
echo "${NUMBER} is not a positive number"
fi
And some tests ...
$ for s in 1 1.33 1e5 1e-4 +1 +1.65 +1e4 1,000 -1 -1.234 -1e5 -1,000 a b a1b ab1 1ab "a b c"
do
test.sh "$s"
done
1 is a positive number
1.33 is a positive number
1e5 is a positive number
1e-4 is a positive number
+1 is a positive number
+1.65 is a positive number
+1e4 is a positive number
1,000 is not a positive number
-1 is not a positive number
-1.234 is not a positive number
-1e5 is not a positive number
-1,000 is not a positive number
a is not a positive number
b is not a positive number
a1b is not a positive number
ab1 is not a positive number
1ab is not a positive number
a b c is not a positive number
This leaves us (I believe) with just the issue of an input containing 1 or more commas, which could be handled by stripping out the comma(s) using whatever method fits your fancy.