Regex to replace multiple spaces to single space excluding leading spaces

Question

I'm trying to write a regex which will capture two or more whitespaces excluding leading whitespaces. Let's take the bellow example

One OS to rule them     all,
    One  OS  to  find    them.
    One     OS to call them    all,
    And  in  salvation    bind         them.
    In  the  bright  land  of  Linux,
    Where the     hackers play.

I want it to become

One OS to rule them all,
    One OS to find them.
    One OS to call them all,
    And in salvation bind them.
    In the bright land of Linux,
    Where the hackers play.

By using this regex ([ ]* ){2,} I can capture two or more whitespaces. The problem with this is that it also captures the leading whitespaces on lines 2 - 5.

Note: I want to use this regex inside Intellij IDEA.

You probably want negative lookbehind for the beginning of the line: `(?<!^)` from http://stackoverflow.com/q/15669557/2988730 — Mad Physicist, Jul 05 '16 at 14:49
@rock321987 I'll use that regex in Intellij to fix a messy file :) — mackatozis, Jul 05 '16 at 14:51

Federico Piazza · Accepted Answer · 2016-07-05T16:12:09.640

3

You can use a regex like this:

\b\s+\b

With a space _ substitution

Working demo

Update for IntelliJ: seems the lookarounds aren't working on IntelliJ, so you can try this other workaround:

(\w+ )\s+

With replacement string: $1

Working demo

Of course, above regex will narrow the scenarios but you can try with that.

edited Jul 05 '16 at 16:12

answered Jul 05 '16 at 14:53

Federico Piazza

30,085
15
87
123

1

Note that using a trailing `\b` will avoid matching multiple spaces at the end of the line – Aaron Jul 05 '16 at 14:55
@Aaron good catch. Will take a look, anyway will leave the answer incase OP find it useful. – Federico Piazza Jul 05 '16 at 14:56
@FedericoPiazza this works fine on SublimeText, but not on Intellij. (It also catches the leading spaces) – mackatozis Jul 05 '16 at 15:03
1

@efthialex don't forget to use the flags I used `gm` – Federico Piazza Jul 05 '16 at 15:04
1

@efthialex on IntelliJ you can try `(?<=\w)\s+` but it narrows the amount of scenarios – Federico Piazza Jul 05 '16 at 15:16
@FedericoPiazza both `\b\s+\b` and `(?<=\w)\s+` match the leading spaces as well (on Intellij only) – mackatozis Jul 05 '16 at 15:17
@efthialex I don't have IntelliJ to test, however it seems IntelliJ doesn't a good regex engine since `(?<=\w)\s+` matches only spaces after a word character. – Federico Piazza Jul 05 '16 at 15:25
@efthialex, I've updated the answer with a trick you can try on IntelliJ. – Federico Piazza Jul 05 '16 at 15:30
@FedericoPiazza `(\w+ )\s+` ignores the leading whitespaces but it also captures the word before the multiple spaces. When I try to substitute with `\1` it also removes the word before the space. – mackatozis Jul 05 '16 at 15:47
@efthialex yes, correct, that's why you have to use the replacement string `\1` or `$1`. The parentheses mean capturing groups and then you have to use the group index to reference the group content, so the replacement will put the word back plus a single space. This works on Eclipse, so should work on IntelliJ too – Federico Piazza Jul 05 '16 at 15:53
@FedericoPiazza it workes only with `$1`. Thanks again for your time. Please update your post. – mackatozis Jul 05 '16 at 15:54
@efthialex glad to help – Federico Piazza Jul 05 '16 at 16:12

Aaron · Answer 2 · 2016-07-05T15:39:28.877

2

In your example, you could use the word-boundary meta-character :

\b\s{2,}

That will match any number of spaces greater than 2 that follow the end of a word (or the beginning, but a word can't start with spaces).

However, it would fail in a more general case where you could have multiples spaces following a special character, which won't be considered part of a word.

If your language supports unbounded-width lookbehind, you can match the following :

(?<!^\s*)\s{2,}

edited Jul 05 '16 at 15:39

answered Jul 05 '16 at 14:52

Aaron

24,009
2
33
57

1

What regex engine do you use to use quantifiers on lookbehind? I though none supports that – Federico Piazza Jul 05 '16 at 15:00
1

probably it should be `{2,}` – Dmitry Bychenko Jul 05 '16 at 15:02
@Aaron I'm not trying to bug your answer, I always wanted to use quantifiers on lookbehind even on PCRE and never work. Have you tried your regex? Using your 2nd expression you can see it is not compiling https://regex101.com/r/tJ8zO3/2 – Federico Piazza Jul 05 '16 at 15:07
infinite lookbehinds are not supported on all PCRE regex engine..`.NET` is one which supports it – rock321987 Jul 05 '16 at 15:08
`regex` module of python supports infinite lookbehind..[source](http://stackoverflow.com/a/24987519/1996394) – rock321987 Jul 05 '16 at 15:15
@FedericoPiazza ok, I can't read... "each alternative still has to be fixed-length". It's variable length sure, but not unbounded length. Hopefully rock321987 was here to answer your question, so at least `python`'s `regex` module and `.NET` support unbounded-width lookbehinds. – Aaron Jul 05 '16 at 15:15

Jan · Answer 3 · 2016-07-05T15:50:33.997

2

With a support for (*SKIP)(*FAIL) you could also come up with:

^[ ]+(*SKIP)(*FAIL)  # match spaces at the beginning of a line
                     # these shall fail
|                    # OR
[ ]{2,}              # at least two spaces

See a demo on regex101.com (mind the modifiers!).

edited Jul 05 '16 at 15:50

answered Jul 05 '16 at 15:25

Jan

42,290
8
54
79

Regex to replace multiple spaces to single space excluding leading spaces

3 Answers3

Linked