Struggling with while loop with two conditions and a for loop - bash

Question

I'm trying to learn how to use arrays in bash. I'm writing a script that asks the user for three numbers and figures out what the biggest number is. I want the script to force the user to enter only numeric values. Furthermore, I want the three numbers should be different. I'm using an array to store user input. This is what I have so far:
## declare variables
order=("first" "second" "third")
answers=()
RED='\033[0;31m'
NC='\033[0m' # No Color
numbersonly() {
if [[ ! $1 =~ ^[0-9]+$ ]]; then
echo "${RED}$1 is not a valid number.${NC}"
else
answers+=("$input")
break
fi
}
inarray(){
for e in ${answers[#]}; do
if [[ $1 == $e ]]; then
echo "${RED}Warning.${NC}"
fi
done
}
readnumber(){
for i in {1..3}; do
j=$(awk "BEGIN { print $i-1 }")
while read -p "Enter the ${order[$j]} number: " input ; do
inarray $input
numbersonly $input
done
done
}
displayanswers(){
echo "Your numbers are: ${answers[#]}"
}
biggestnumber(){
if (( ${answers[0]} >= ${answers[1]} )); then
biggest=${answers[0]}
else
biggest=${answers[1]}
fi
if (( $biggest <= ${answers[2]} )); then
biggest=${answers[2]}
fi
echo "The biggest number is: $biggest"
}
main(){
readnumber
displayanswers
biggestnumber
}
main
Right now, I can get the script to display a warning when the user enters a number that was previously entered, but I can't seem to find the proper syntax to stay in the while loop if the user input has already been entered. Thoughts?

Answer 1

I found a way around it. My problem was twofold: 1) I didn't realize that if you have a for loop nested in a while loop, you'll need two break statements to exit the while loop; 2) having two functions within the same while loop made it hard to control what was happening. By merging inarray() and numbersonly() into a new function, I solved the double conditional issue. The new function looks like this:
testing(){
for item in ${answers[*]}
do
test "$item" == "$1" && { inlist="yes"; break; } || inlist="no"
done
if [[ $inlist == "yes" ]]; then
echo "${RED}$1 is already in list.${NC}"
else
if [[ ! $1 =~ ^[0-9]+$ ]]; then
echo "${RED}$1 is not a valid number.${NC}"
else
answers+=("$input")
break
fi
fi
}

Answer 2

Without much study here is what leapt off the screen to me follows. Beware I haven't actually tested it... debugging is an exercise left to the student.
Recommend using newer "function" definitions as you can declare local variables. () definitions do not allow localized variables.
function inarray
{
local e; #don't muck up any variable e in caller
...
}
To calculate values avoid extra awk and use j=$(( i - 1 ));
biggestnumber should likely use a loop.
Overall comment:
nummax=3; #maximum value defined in just one place
# loop this way... showing optional {} trick for marking larger loops
for (( n = 0; n < nummax; ++n )); do
{
nx=$(( 1 + n )); #1 based index
} done;
Hint: should stop input loop once all input present. Could also add:
if [ "" == "${input:-}" ]; then break;
for (( a = 0; a < ${#answer[*]}; ++a )); do
Note the extensive use of double quotes to avoid syntax errors if the variable value is empty or contains many shell metacharacters, like spaces. I can't tell you how many bug reports I've fixed by adding the quotes to existing code.
[[ ... ]] expressions use file name tests, not regular expressions. The closest you can get to [[ ! $1 =~ ^[0-9]+$ ]]; is using [[ "$1" != [0-9]* ]] && [[ "$1" != *[^0-9]* ]].
But I suspect ! expr >/dev/null "$i" : '[0-9][0-9]*$'; is more what you want as "expr" does use regular expressions. Don't enclose in []s. Used [0-9][0-9]* rather than [0-9]+ as "+" has given me mixed successes across all dialects of UNIX regular expressions.