I need to count the number of occurrences of a char in a string using Bash.
In the following example, when the char is (for example) t, it echos the correct number of occurrences of t in var, but when the character is comma or semicolon, it prints out zero:
var = "text,text,text,text"
num = `expr match $var [,]`
echo "$num"
you can for example remove all other chars and count the whats remains, like:
var="text,text,text,text"
res="${var//[^,]}"
echo "$res"
echo "${#res}"
will print
,,,
3
or
tr -dc ',' <<<"$var" | awk '{ print length; }'
or
tr -dc ',' <<<"$var" | wc -c #works, but i don't like wc.. ;)
or
awk -F, '{print NF-1}' <<<"$var"
or
grep -o ',' <<<"$var" | grep -c .
or
perl -nle 'print s/,//g' <<<"$var"
y="${x//[^s|S]}"; echo "${#y}"
tr -dc ',' <<<"$var" | awk '{ print length; }' seems to be way faster than the first option.
wc incorrectly counts the number of lines, in the input. Of course, the wc -c is OK but because of the line counting problem I don't like it. example, the: printf "line1\nline2\n" | wc -l prints 2 but the printf "line1\nline2" | wc -l prints only 1.
aaa\nbbb\nccc into aa, a\nb, bb\ncc, c, count lines in all those parts and then just add them to get the result. This additive stability is important. You probably don't like cat too, since for files with data aaa\nbbb and ccc\nddd it joins the last "line" with the first one, creating bbbccc.I would use the following awk command:
string="text,text,text,text"
char=","
awk -F"${char}" '{print NF-1}' <<< "${string}"
I'm splitting the string by $char and print the number of resulting fields minus 1.
If your shell does not support the <<< operator, use echo:
echo "${string}" | awk -F"${char}" '{print NF-1}'
$(grep -o "$needle" < filename | wc -l)
wc -l, just use grep -c, it works on both bsd grep and linux grep.grep -c will only output the number of matching lines. It does not count multiple matches per line.
You can do it by combining tr and wc commands. For example, to count e in the string referee
echo "referee" | tr -cd 'e' | wc -c
output
4
Explanations: Command tr -cd 'e' removes all characters other than 'e', and Command wc -c counts the remaining characters.
Multiple lines of input are also good for this solution, like command cat mytext.txt | tr -cd 'e' | wc -c can counts e in the file mytext.txt, even thought the file may contain many lines.
*** Update ***
To solve the multiple spaces in from of the number (@tom10271), simply append a piped tr command:
tr -d ' '
For example:
echo "referee" | tr -cd 'e' | wc -c | tr -d ' '
awk is very cool, but why not keep it simple?
num=$(echo $var | grep -o "," | wc -l)
You could then re-use it as a function:
# usage: echo "1,2,3,4,5" | text.count # outputs 4
function text.count(){
grep -o "$1" | wc -l
}
Building on everyone's great answers and comments, this is the shortest and sweetest version:
grep -o "$needle" <<< "$haystack" | wc -l
awk works well if you your server has it
var="text,text,text,text"
num=$(echo "${var}" | awk -F, '{print NF-1}')
echo "${num}"
awk -F, looks for a ,. You can do the following: awk -F"${your_char}"I Would suggest the following:
var="any given string"
N=${#var}
G=${var//g/}
G=${#G}
(( G = N - G ))
echo "$G"
No call to any other program
also check this out, for example we wanna count t
echo "test" | awk -v RS='t' 'END{print NR-1}'
or in python
python -c 'print "this is for test".count("t")'
or even better, we can make our script dynamic with awk
echo 'test' | awk '{for (i=1 ; i<=NF ; i++) array[$i]++ } END{ for (char in array) print char,array[char]}' FS=""
in this case output is like this :
e 1
s 1
t 2
The awk solutions provided here so far all break if there's a line break in your text. E.g.:
text="one,two,thr
ee,four"
DELIM=','
count=$( awk -F"$DELIM" '{print NF-1}' <<<"${text}" )
echo $count
Result:
2
1
The solution that will also work correctly with line breaks is:
text="one,two,thr
ee,four"
DELIM=','
count=$( awk 'BEGIN{RS="'"$DELIM"'";FS=""}END{print NR-1}' <<<"${text}" )
echo $count
Result is 3.
export searchpattern=$(echo ",")
echo "text,text,text,text" | tr "," '\n' | sed 's/$/,/g' > filename
export count=$(grep -F $searchpattern filename | wc -l)
echo "$count-1" | bc
