I have the following JS:
"a a a a".replace(/(^|\s)a(\s|$)/g, '$1')
I expect the result to be '', but am instead getting 'a a'. Can anyone explain to me what I am doing wrong?
Clarification: What I am trying to do is remove all occurrences of 'a' that are surronded by whitespace (i.e. a whole token)
It's because this regex /(^|\s)a(\s|$)/g match the previous char and the next char to each a
in string "a a a a" the regex matches :
Edit: Little bit tricky but working (without regex):
var a = "a a a a";
// Handle beginning case 'a '
var startI = a.indexOf("a ");
if (startI === 0){
var off = a.charAt(startI + 2) !== "a" ? 2 : 1; // test if "a" come next to keep the space before
a = a.slice(startI + off);
}
// Handle middle case ' a '
var iOf = -1;
while ((iOf = a.indexOf(" a ")) > -1){
var off = a.charAt(iOf + 3) !== "a" ? 3 : 2; // same here
a = a.slice(0, iOf) + a.slice(iOf+off, a.length);
}
// Handle end case ' a'
var endI = a.indexOf(" a");
if (endI === a.length - 2){
a = a.slice(0, endI);
}
a; // ""
First "a " matches. Then it will try to match against "a a a", which will skip first a, and then match "a ". Then it will try to match against "a", which will not match.
The result will be "a a".
To get your desired result you can do this:
"a a a a".replace(/(?:\s+a(?=\s))+\s+|^a\s+(?=[^a]|$|a\S)|^a|\s*a$/g, '')
'ba a' to 'b'
console.log('"' + "ba a".replace(/(^|\s+)a(?=\s|$)/g, '') + '"'); and you'll see that it does, in fact output "ba".
As others have tried to point out, the issue is that the regex consumes the surrounding spaces as part of the match. Here's a [hopefully] more straight forward explanation of why that regex doesn't work as you expect:
First let's breakdown the regex, it says match the a space or start of string, followed by an 'a' followed by a space or the end of the string.
Now let's apply it to the string. I've added character indexes beneath the string to make things easier to talk about:
a a a a
0123456
The regex looks at the 0 index char, and finds an 'a' at that location, followed by a space at index 2. This is a match because it is the start of the string, followed by an a followed by a space. The length of our match is 2 (the 'a' and the space), so we consume two characters and start our next search at index 2.
Character 2 ('a') is neither a space nor the start of the string, and therefore it doesn't match the start of our regular expression, so we consume that character (without replacing it) and move on to the next.
Character 3 is a space, followed by an 'a' followed by another space, which is a match for our regex. We replace it with an empty string, consume the length of the match (3 characters - " a ") and move on to index 6.
Character 6 ('a') is neither a space nor the start of the string, and therefore it doesn't match the start of our regular expression, so we consume that character (without replacing it) and move on to the next.
Now we're at the end of the string, so we're done.
The reason why the regex @caeth suggested (/(^|\s+)a(?=\s|$)/g) works is because of the ?= quantifier. From the MDN Regexp Documentation:
Matches x only if x is followed by y. For example,
/Jack(?=Sprat)/matches "Jack" only if it is followed by "Sprat"./Jack(?=Sprat|Frost)/matches "Jack" only if it is followed by "Sprat" or "Frost". However, neither "Sprat" nor "Frost" is part of the match results.
So, in this case, the ?= quantifier checks to see if the following character is a space, without actually consuming that character.
/(?:^|\s)a(?=\s|$)/g, replace by ''Use this instead:
"a a a a".replace(/(^|\s*)a(\s|$)/g, '$1')
With "* this you replace all the "a" occurrences
Greetings
Or you can just split the string up, filter it and glue it back:
"a ba sl lf a df a a df r a".split(/\s+/).filter(function (x) { return x != "a" }).join(" ")
>>> "ba sl lf df df r"
"a a a a".split(/\s+/).filter(function (x) { return x != "a" }).join(" ")
>>> ""
Or in ECMAScript 6:
"a ba sl lf a df a a df r a".split(/\s+/).filter(x => x != "a").join(" ")
>>> "ba sl lf df df r"
"a a a a".split(/\s+/).filter(x => x != "a").join(" ")
>>> ""
I assume that there is no leading and trailing spaces. You can change the filter to x && x != 'a' if you want to remove the assumption.
'a a a a'.replace(/(?:^|\s)a(?=\s|$)/g, '');?