I am trying to add the elements of an array that is defined by user input from the read -a command. How can I do that?
read -a array
tot=0
for i in ${array[#]}; do
let tot+=$i
done
echo "Total: $tot"
Given an array (of integers), here's a funny way to add its elements (in bash):
sum=$(IFS=+; echo "$((${array[*]}))")
echo "Sum=$sum"
e.g.,
$ array=( 1337 -13 -666 -208 -408 )
$ sum=$(IFS=+; echo "$((${array[*]}))")
$ echo "$sum"
42
Pro: No loop, no subshell!
Con: Only works with integers
Edit (2012/12/26).
As this post got bumped up, I wanted to share with you another funny way, using dc, which is then not restricted to just integers:
$ dc <<< '[+]sa[z2!>az2!>b]sb1 2 3 4 5 6 6 5 4 3 2 1lbxp'
42
This wonderful line adds all the numbers. Neat, eh?
If your numbers are in an array array:
$ array=( 1 2 3 4 5 6 6 5 4 3 2 1 )
$ dc <<< '[+]sa[z2!>az2!>b]sb'"${array[*]}lbxp"
42
In fact there's a catch with negative numbers. The number '-42' should be given to dc as _42, so:
$ array=( -1.75 -2.75 -3.75 -4.75 -5.75 -6.75 -7.75 -8.75 )
$ dc <<< '[+]sa[z2!>az2!>b]sb'"${array[*]//-/_}lbxp"
-42.00
will do.
Pro: Works with floating points.
Con: Uses an external process (but there's no choice if you want to do non-integer arithmetic — but dc is probably the lightest for this task).
My code (which I actually utilize) is inspired by answer of gniourf_gniourf. I personally consider this more clear to read/comprehend, and to modify. Accepts also floating points, not just integers.
Sum values in array:
arr=( 1 2 3 4 5 6 7 8 9 10 )
IFS='+' sum=$(echo "scale=1;${arr[*]}"|bc)
echo $sum # 55
With small change, you can get the average of values:
arr=( 1 2 3 4 5 6 7 8 9 10 )
IFS='+' avg=$(echo "scale=1;(${arr[*]})/${#arr[#]}"|bc)
echo $avg # 5.5
gniourf_gniourf's answer is excellent since it doesn't require a loop or bc. For anyone interested in a real-world example, here's a function that totals all of the CPU cores reading from /proc/cpuinfo without messing with IFS:
# Insert each processor core count integer into array
cpuarray=($(grep cores /proc/cpuinfo | awk '{print $4}'))
# Read from the array and replace the delimiter with "+"
# also insert 0 on the end of the array so the syntax is correct and not ending on a "+"
read <<< "${cpuarray[#]/%/+}0"
# Add the integers together and assign output to $corecount variable
corecount="$((REPLY))"
# Echo total core count
echo "Total cores: $corecount"
I also found the arithmetic expansion works properly when calling the array from inside the double parentheses, removing the need for the read command:
cpuarray=($(grep cores /proc/cpuinfo | awk '{print $4}'))
corecount="$((${cpuarray[#]/%/+}0))"
echo "Total cores: $corecount"
Generic:
array=( 1 2 3 4 5 )
sum="$((${array[#]/%/+}0))"
echo "Total: $sum"
I'm a fan of brevity, so this is what I tend to use:
IFS="+";bc<<<"${array[*]}"
It essentially just lists the data of the array and passes it into BC which evaluates it. The "IFS" is the internal field separate, it essentially specifies how to separate arrays, and we said to separate them with plus signs, that means when we pass it into BC, it receives a list of numbers separated by plus signs, so naturally it adds them together.
Another dc & bash method:
arr=(1 3.88 7.1 -1)
dc -e "0 ${arr[*]/-/_} ${arr[*]/*/+} p"
Output:
10.98
The above runs the expression 0 1 3.88 7.1 _1 + + + + p with dc. Note the dummy value 0 because there's one too many +s, and also note the usual negative number prefix - must be changed to _ in dc.
arr=(1 2 3) //or use `read` to fill the array
echo Sum of array elements: $(( ${arr[#]/%/ +} 0))
Sum of array elements: 6
Explanation:
"${arr[#]/%/ +}" will return 1 + 2 + 3 +
By adding additional zero at the end we will get 1 + 2 + 3 + 0
By wrapping this string with BASH's math operation like this$(( "${arr[#]/%/ +} 0")), it will return the sum instead
This could be used for other math operations.
For subtracting just use - instead
For multiplication use * and 1 instead of 0
Can be used with logic operators too.
BOOL AND EXAMPLE - check if all items are true (1)
arr=(1 0 1)
if [[ $((${arr[#]/%/ &} 1)) -eq 1 ]]; then echo "yes"; else echo "no"; fi
This will print: no
BOOL OR EXAMPLE - check if any item is true (1)
arr=(1 0 0)
if [[ $((${arr[#]/%/ |} 0)) -eq 1 ]]; then echo "yes"; else echo "no"; fi
This will print: yes
A simple way
function arraySum
{
sum=0
for i in ${a[#]};
do
sum=`expr $sum + $i`
done
echo $sum
}
a=(7 2 3 9)
echo -n "Sum is = "
arraySum ${a[#]}
I find this very simple using an increasing variable:
result2=0
for i in ${lineCoffset[#]};
do
result2=$((result2+i))
done
echo $result2
Related
I have an array trf. Would like to compute the sum of the second element in each array entry.
Example of array contents
trf=( "2 13 144" "3 21 256" "5 34 389" )
Here is the current implementation, but I do not find it robust enough. For instance, it fails with arbitrary number of elements (but considered constant from one array element to another) in each array entry.
cnt=0
m=${#trf[#]}
while (( cnt < m )); do
while read -r one two three
do
sum+="$two"+
done <<< $(echo ${array[$count]})
let count=$count+1
done
sum+=0
result=`echo "$sum" | /usr/bin/bc -l`
You're making it way too complicated. Something like
#!/usr/bin/env bash
trf=( "2 13 144" "3 21 256" "5 34 389" )
declare -i sum=0 # Integer attribute; arithmetic evaluation happens when assigned
for (( n = 0; n < ${#trf[#]}; n++)); do
read -r _ val _ <<<"${trf[n]}"
sum+=$val
done
printf "%d\n" "$sum"
in pure bash, or just use awk (This is handy if you have floating point numbers in your real data):
printf "%s\n" "${trf[#]}" | awk '{ sum += $2 } END { print sum }'
You can use printf to print the entire array, one entry per line. On such an input, one loop (while read) would be sufficient. You can even skip the loop entirely using cut and tr to build the bc command. The echo 0 is there so that bc can handle empty arrays and the trailing + inserted by tr.
{ printf %s\\n "${trf[#]}" | cut -d' ' -f2 | tr \\n +; echo 0; } | bc -l
For your examples this generates prints 68 (= 13+21+34+0).
Try this printf + awk combo:
$ printf '%s\n' "${trf[#]}" | awk '{print $2}{a+=$2}END{print "sum:", a}'
13
21
34
sum: 68
Oh, it's already suggested by Shawn. Then with loop:
$ for item in "${trf[#]}"; do
echo $item
done | awk '{print $2}{a+=$2}END{print "sum:", a}'
13
21
34
sum: 68
For relatively small arrays a for/while double loop should be ok re: performance; placing the final sum in the $result variable (as in OP's code):
result=0
for element in "${trf[#]}"
do
while read -r a b c
do
((result+=b))
done <<< "${element}"
done
echo "${result}"
This generates:
68
For larger data sets I'd probably opt for one of the awk-only solutions (for performance reasons).
I have an array with different values, and if the value is 3 (integer) this value will be exchanged for 999 (integer). But I have a syntax problem in the 'if' block of the array.
The correct return would be: 1 999 5 7 9 999
vetor=(1 3 5 7 9 3)
for i in ${vetor[*]}
do
if [[ ${vetor[i]} = 3 ]]; then
${vetor[i]} = 999
fi
echo $i
done
This produces the correct output in Bash:
vetor=(1 3 5 7 9 3);
for i in ${!vetor[*]};
do
if [[ ${vetor[i]} -eq 3 ]]; then
vetor[i]=999;
fi;
echo ${vetor[i]};
done
I added ! in the for loop expression to get the indices of vetor instead of the values, and I removed the ${} around the assignment in the if condition (this was giving "if 3 is not a typo" warnings). Also changed the echo to get the value at vetor[i] instead of printing the index.
So I'm learning bash and need to do a simple script to multiply array elements by calling a function.
My code so far is this, but it ain't working at all. I believe there is a much simpler way than this (incrementing the pos variable so as to move to the next array element feels simply wrong).
array=(1 2 3 4 5 100)
sum=0
pos=1
function multiplicate {
for i in ${array[*]};do
sum=$(($i * $array[pos]))
let pos++
done
}
multiplicate
echo $sum
I did my best to google the solution but was unable to find any relevant information, I found how to sum by using bc, but it simply wouldn't work by replacing + with *.
Use this script:
#!/bin/bash
array=(1 2 3 4 5 100)
function multiplicate {
local mul=1
for i in "${array[#]}"; do
((mul *= i))
done
echo "$mul"
}
multiplicate
$ ./script
12000
Or better yet:
#!/bin/bash
multiplicate() {
local mul=1
for i
do ((mul *= i))
done
echo "$mul"
}
multiplicate 1 2 3 4 5 100
And if you like to play with string variables use this:
multiplicate() { local IFS=* ; echo $(( $* )); }
multiplicate 1 2 3 4 5 100
Here is a method using bc:
multiply ()
{
printf '%s\n' "$#" | paste -s -d '*' | bc
}
Used as follows:
$ multiply 1 2 3 4 5 100
12000
The first command in the pipeline prints each array elements on a separate line:
$ printf '%s\n' 1 2 3 4 5 100
1
2
3
4
5
100
The paste -s ("serial") then turns the output into a single line again, but the elements now separated by *:
$ printf '%s\n' 1 2 3 4 5 100 | paste -s -d '*'
1*2*3*4*5*100
And bc finally evaluates the expression.
Alternatively, we can save a subshell and skip bc:
multiply () {
echo $(( $(printf '%s\n' "$#" | paste -s -d '*') ))
}
This uses an arithmetic expression to evaluate the output of printf and paste (which now is in a command substitution), but readability suffers a bit.
Alternatively, in pure Bash (hat tip sorontar):
multiply () {
local IFS='*'
echo "$(( $* ))"
}
This sets the field separator IFS to * so the arguments, $*, expand to a string separated by *, which is then evaluated in the arithmetic expression $(()).
#!/bin/bash
array=( 2 4 5 8 15 )
a_2=( 2 4 8 10 )
a_4=( 2 4 8 10 )
a_5=( 10 12 )
a_8=( 8 12 )
a_15=( 2 4 )
numberOfTests=5
while [ $i -lt ${#array[#]} ]; do
j=0
currentArray =${array[$i]}
*while [ $j -lt ${#a_$currentArray [#]} ]; do #### this line i get ->>>> bad substitution*
./test1.sh "${array[$i]}" -c "${a_"$currentArray "[$j]}" &
let j=j+1
done
let i=i+1
done
so Im trying this code, loop over an array(called array), The array should point out
the array number we are now looping(a_X). And every time to point out the current place and value.
can anybody help me how im using the $currentArray to work properly so I can know the length of the array and the value?
I get in the line I marked an error.
Thank you guys!
The simplest solution is to store the full names of the arrays, not just the numerical suffix, in array. Then you can use indirect parameter expansion while iterating directly over the values, not the indices, of the arrays.
# Omitting numberOfTests has it does not seem to be used
array=(a_2 a_4 a_5 a_8 a_15)
a_2=( 2 4 8 10 )
a_4=( 2 4 8 10 )
a_5=( 10 12 )
a_8=( 8 12 )
a_15=( 2 4 )
for arr in "${array[#]}"; do
currentArray=$arr[#]
for value in "${!currentArray}"; do
./test1.h "${arr#a_}" -c "$value" &
done
done
I have a test file that looks like this:
1 1 1 1 1
9 3 4 5 5
6 7 8 9 7
3 6 8 9 1
3 4 2 1 4
6 4 4 7 7
Each row is supposed to represent a students grades. So the user puts in either an 'r' or a 'c' into the command line to choose to sort by rows or columns, followed by the file name. Sorting by rows would represent getting a students average and sorting my columns would represent a particular assignments average.
I am not doing anything with the choice variable yet because I need to get the array sorted first so I can take the averages and then get the median for each column and row.
So im not sure how I can choose to sort by those specific options. Here is what I have so far:
#!/bin/bash
choice="$1"
filename="$2"
c=0
if [ -e "$filename" ]
then
while read line
do
myArray[$c]=$line
c=$(expr $c + 1)
done < "$filename"
else
echo "File does not exist"
fi
printf -- '%s\n' "${myArray[#]}"
FS=$'\n' sorted=($(sort -n -k 1,1<<<"${myArray[*]}"))
echo " "
printf '%s\n' "${sorted[#]}"
This is only sorting the first column though and im not sure why its even doing that. Any push in the right direction would be appreciated. Examples would help a ton, thanks!
UPDATE:
With the changes that were suggested I have this so far:
#!/bin/sh
IFS=$'\n';
choice="$1"
filename="$2"
if [ -e "$filename" ]
then
while read line
do
myArray[$c]=$line
c=$(expr $c + 1)
done < "$filename"
else
echo "File does not exist."
fi
printf -- '%s\n' "${myArray[#]}"
width=${myArray[0]// /}
width=${#width}
height=${#myArray[#]}
bar=()
for w in $(seq 0 1 $((${width}-1)))
do
tmp=($(sort -n <<<"${myArray[*]}"))
for h in $(seq 0 1 $((${height}-1)))
do
myArray[h]=${myArray[h]#* }
bar[h]="${bar[h]} ${tmp[h]%% *}"
bar[h]="${bar[h]# }"
done
done
printf -- '%s\n' "${bar[*]}"
But now I am getting some really strange output of way more numbers than i started with and in a seemingly random order.
actually it is sorting $line(s) which are strings. you need to initialize the column to sort correctly, so that it is an array
UPDATE:
the following code is really straight forward. no performance aspects are regarded. so for large datasets this will take a while to sort column wise. your datasets have to contain lines of numbers seperated by single spaces to make this work.
#!/bin/bash
IFS=$'\n';
# here you can place your read line function
ar[0]="5 3 2 8"
ar[1]="1 1 1 1"
ar[2]="3 2 4 5"
printf -- '%s\n' "${ar[*]}" # print the column wise unsorted ar
echo
# sorting
width=${ar[0]// /}
width=${#width}
height=${#ar[#]}
bar=()
for w in $(seq 0 1 $((${width}-1))); do # for each column
#sort -n <<<"${ar[*]}" # debug, see first column removal
tmp=($(sort -n <<<"${ar[*]}")) # this just sorts lexigraphically by "first column"
# rows are strings, see initial definition of ar
#echo
for h in $(seq 0 1 $((${height}-1))); do # update first column
ar[h]=${ar[h]#* } # strip first column
bar[h]="${bar[h]} ${tmp[h]%% *}" # add sorted column to new array
bar[h]="${bar[h]# }" # trim leading space
done
#printf -- '%s\n' "${bar[*]}" # debug, see growing bar
#echo "---"
done
printf -- '%s\n' "${bar[*]}" # print the column wise sorted ar
prints out the unsorted and sorted array
5 3 2 8
1 1 1 1
3 2 4 5
1 1 1 1
3 2 2 5
5 3 4 8