I'm learning bash and I'm trying identify a sequence elements (patch or subarray) position into a array.
For example:
array=(9 5 8 3 2 7 5 9 0 1 1 5 4 3 8 9 6 2 6 5 7 9 8);
patch=(0 1 1 5)
I would like to obtain a output equals to 8 (start position of my patch in relation to array) or 11 (final position).
bash doesn't really have any built-in facility to do this; you need to walk the array yourself:
for ((i=0; i<${#array[#]}; i++)); do
for ((j=0; j<${#patch[#]}; j++)); do
# Make sure the corresponding elements
# match, or give up. RHS is quoted to ensure
# actual string equality, rather than just pattern matching
[[ ${array[i+j] == "${patch[j]}" ]] || break
done
if [[ $j == ${#patch[#] ]]; then
# All the comparisons succeeded!
start=$i
finish=$((i+j-1))
break
fi
done
Hello I have been trying to get the numbers in the columns of a file for two days by reading a file via a bash script. Here is the file sample.txt
1 1 1 1 1
9 3 4 5 5
6 7 8 9 7
3 6 8 9 1
3 4 2 1 4
6 4 4 7 7
By column I mean i.e the first column is
1
9
6
3
3
6
I need to have the column elements each be in a given array col1 or col2 etc so that I can manipulate the values further.
Here's what I have done so far using while loop I have read the contents of the file assigning them each line to an array.
If I set IFS=$'\n'
while read -a line
do
IFS=$'\n'
#I can get the whole column 1 with this
echo ${line[0]}
#for column 2 I can get it by this an the others too
echo ${line[1]}
done < sample.txt
Now that may seem good as i thought but since I want to calculate averages of the columns putting in another loop like a for loop becomes impossible since ${line[0]} has all the elements in column 1 but they are all as a single string (i have tried to observe) that cannot be acted upon.
What would be the best way to get those elements be members of a given array and then compute the averages on them. help appreciated .
In bash I'd write
declare -A cols
n=0
while read -ra fields; do
for ((i=0; i<${#fields[#]}; i++)); do
cols[$i,$n]=${fields[i]}
((n[i]++))
done
done < sample.txt
read -a reads the fields of the line into the named array.
I'm using cols as an associative array to fake a multi-dimensional array. That's way easier to deal with than using a dynamic variable name:
eval "column${i}[$n]=\${fields[$i]}"
#!/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
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