How would I create a bash script that the user can use to sum any amount of command line arguments, but it will also ignore incompatible arguments? For example, say my script is called sum:
sum 10 -argument1 -argument2 -argument3 20
30
sum 7 -argument1 2 -argument2 1 -argument3
10
With awk, assuming the sum and each individual number argument can fit in your system's long type, and only considering sequences of decimal digits with an optional leading - sign:
#! /bin/sh -
awk 'BEGIN{
sum = 0
for (i = 1; i < ARGC; i++)
if (ARGV[i] ~ /^-?[0123456789]+$/)
sum += ARGV[i]
print sum}' "$@"
For arbitrary precision, you can use bc:
#! /bin/sh -
awk 'BEGIN{
for (i = 1; i < ARGC; i++)
if (ARGV[i] ~ /^-?[0123456789]+$/)
print "s+="ARGV[i]
print "s"}' "$@" | bc
Example:
$ ./sum1 999999999999999999999999 1
999999999999999983222784
$ ./sum2 999999999999999999999999 1
1000000000000000000000000
One (nonoptimal) example to do the sum:
echo "$@" |awk '{for (i = 1; i <= NF; i++) j+=$i} END {print j}'
(assuming number is everything awk accept as such)
Example:
./ff 1 2 3.1 0x10 0100
86.1
And warning: any word which start with number will be interpreted as number
awk, in that context, 1WHATEVER2 would be understood as 1.
$@ might as well be $*
"$@" so the parameter "*" does not expand to a list of files.
Here's another possibility
printf "%s\n" "$@" | awk '$1 ~ /^[[:digit:]]+$/ {s+=$1} END {print s}'
It takes each argument from the command line and prints it, one per line. The awk script accepts this data and adds up non-zero positive values, printing the result when it's received all the data
Example
#!/bin/sh
printf "%s\n" "$@" | awk '$1 ~ /^[[:digit:]]+$/ {s+=$1} END {print s}'
./sum 1 -argument2 2 -argument4 3
6
If you could guarantee that your numeric arguments contained only digits you could simplify the check expression from $1 ~ /^[[:digit:]]+$/ to $1+0 > 0.
$'foo\n1234'.