regex in bash expression

Question

I have 2 questions about regex in bash expression.

1.non-greedy mode

local temp_input='"a1b", "d" , "45"'
if [[ $temp_input =~ \".*?\" ]]
then
    echo ${BASH_REMATCH[0]}
fi

The result is

"a1b", "d" , "45"

In java

String str = "\"a1b\", \"d\" , \"45\"";
Matcher m = Pattern.compile("\".*?\"").matcher(str);
while (m.find()) {
    System.out.println(m.group());
}

I can get the result below.

"a1b"
"d"
"45"

But how can I use non-greedy mode in bash?
I can understand why the \"[^\"]\" works.
But I don't understand why does the \".?\" do not work.

2.global matches

local temp_input='abcba'
if [[ $temp_input =~ b ]]
then
    #I wanna echo 2 b here. 
    #How can I set the global flag?
fi

How can I get all the matches?
ps:I only wanna use regex.

For the second question, sorry for the confusing.
I want to echo "b" and "b", not count "b".

Help!

bash regular expressions do not have a non-greedy mode. See man 7 regex for what is supported. — chepner
– chepner, Commented Sep 11, 2013 at 13:33
@chepner Thanks for your help. I also find nothing about non-greedy mode in the specific. I think you're right. — android_su
– android_su, Commented Sep 12, 2013 at 1:53

konsolebox · Accepted Answer · 2013-09-11 11:41:13Z

3

For your first question, an alternative is this:

[[ $temp_input =~ \"[^\"]*\" ]]

For your second question, you can do this:

temp_input=abcba
t=${temp_input//b}
echo "$(( (${#temp_input} - ${#t}) / 1 )) b"

Or for convenience place it on a function:

function count_matches {
    local -i c1=${#1} c2=${#2}
    if [[ c2 -gt 0 && c1 -ge c2 ]]; then
        local t=${1//"$2"}
        echo "$(( (c1 - ${#t}) / c2 )) $2"
    else
        echo "0 $2"
    fi
}

count_matches abcba b

Both produces output:

2 b

Update:

If you want to see the matches you can use a function like this. You can also try other regular expressions not just literals.

function find_matches {
    MATCHES=() 
    local STR=$1 RE="($2)(.*)"
    while [[ -n $STR && $STR =~ $RE ]]; do
        MATCHES+=("${BASH_REMATCH[1]}")
        STR=${BASH_REMATCH[2]}
    done
}

Example:

> find_matches abcba b
> echo "${MATCHES[@]}"
b b

> find_matches abcbaaccbad 'a.'
> echo "${MATCHES[@]}"
ab aa ad

edited Sep 11, 2013 at 11:41

answered Sep 11, 2013 at 10:36

konsolebox

76.3k13 gold badges110 silver badges114 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

android_su Over a year ago

Thanks for you help. Sorry for confusing you. For the second question, I want b and b, not count "b".

android_su Over a year ago

I can understand why the \"[^\"]*\" works. But I don't understand why does the \".*?\" do not work.

konsolebox Over a year ago

@android_su Probably because bash does not implement it, or interprets it a different way. In a strict sense of interpreting \".*?\" I believe bash is not wrong anyway. The feature is just available in other programs. I made an update by the way.

android_su Over a year ago

Thanks for your great help! But I still want to know how to use non-greedy mode in bash expression.

nickie · Accepted Answer · 2013-09-11 10:38:15Z

2

Your regular expression matches the string starting with the first quotation mark (before ab) and ending with the last quotation mark (after ef). This is greedy, even though your intention was to use a non-greedy match (*?). It seems that bash uses POSIX.2 regular expression (check your man 7 regex), which does not support a non-greedy Kleene star.

If you want just "ab", I'd suggest a different regular expression:
```
if [[ $temp_input =~ \"[^\"]*\" ]]
```
which explicitly says that you don't want quotation marks inside your strings.
I don't understand what you mean. If you want to find all matches (and there are two occurrences of b here), I think you cannot do it with a single ~= match.

answered Sep 11, 2013 at 10:38

nickie

5,8282 gold badges28 silver badges40 bronze badges

1 Comment

android_su Over a year ago

Thanks for your help. For the first question, I can understand your regex string, but I wanna know why does mine do not work. For the second question, could you give me a good way to get the two "b"s please? I am a beginner for the bash.

user251764 · Accepted Answer · 2016-02-11 09:05:42Z

1

This is my first post, and I am very amateur at bash, so apologies if I haven't understood the question, but I wrote a function for non-greedy regex using entirely bash:

regex_non_greedy () {
    local string="$1"
    local regex="$2"
    local replace="$3"

    while [[ $string =~ $regex ]]; do
        local search=${BASH_REMATCH}
        string=${string/$search/$replace}
    done

    printf "%s" "$string"
}

Example invocation:

regex_non_greedy "all cats are grey and green" "gre+." "white"

Which returns:

all cats are white and white

answered Feb 11, 2016 at 9:05

user251764

112 bronze badges

Collectives™ on Stack Overflow

regex in bash expression

3 Answers 3

4 Comments

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

4 Comments

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related