Question

I would like to replace extra spaces (instances of consecutive whitespace characters) with one space, as long as those extra spaces are not in double or single quotes (or any other enclosures I may want to include).

I saw some similar questions, but I could not find a direct response to my needs above. Thank you!

Answer 1

Hope you’re still looking, or come back to check! This seems to work for me:

'/\s+((["\']).*?(?=\2)\2)|\s\s+/'

…and replace with  $1

EDIT

Also, if you need to allow for escaped quotes like \" or \', you could use this expression:

 '/\s+((["\'])(\\\\\2|(?!\2).)*?(?=\2)\2)|\s\s+/'

It gets a bit stickier if you want to add support for “balanced” quotes like brackets (e.g. () or {})

END EDIT

Let me know if you find problems or would like some explanation!

---

HOPEFULLY FINAL EDIT AND WARNINGS

• Potential problem: If a quoted string starts at the beginning of the string variable (or file), it will either not count as a quoted string (and have any whitespace reduced) or it will throw off the whole thing, making anything NOT in quotes get treated as though it was in quotes and vice versa –
  • A potential change that might remedy this is to use the following match expression
  • /(?:^|\s+)((["\'])(\\\\\2|(?!\2).)*?(?=\2)\2)|\s\s+/
  • this replaces \s+ with (?:^|\s+) at the beginning of the expression
  • this will add a space at the beginning of the variable if the string starts with a quote – just trim() or remove that whitespace to continue
• I seem to have used the “line by line” approach (like sed, if I’m not mistaken) to reach my original results – if you use the “whole file” or “whole string” setting or approach, carriage-return-line-feed seems to count as two whitespace characters (can’t imagine why…), thus turning any newlines into single spaces (unless they are inside quotes and “dot-matches-newline” is used, of course)
  • this could be resolved by replacing the . and \s shorthand character classes with the specific characters you want to match, like the following:
  • /(?:^|[ \t]+)((["\'])(\\\\\2|(?!\2)[\s\S])*?(?=\2)\2)|[ \t]{2,}/
  • this does not require the dot-matches-newline switch and only replaces multiple spaces or tabs – not newlines – with a single space (and of course, only if they are not quoted)

---

EXAMPLE

This link shows an example of the first expression and last expression in use on sample text on http://codepad.viper-7.com