带可选组的正则表达式将多个组包装起来，对于未采用的分支返回未定义

Question

I'm trying to write a regular expression in JavaScript that returns the first quoted or non-quoted word in a string without the quotes (if present). For example:

'"quoted phrase" followed by text' => 'quoted phrase'
'phrase without quotes followed by text' => 'phrase'

My regular expression currently is this: (?:"([^"]*)"|([^"\\s]+))

However, what I'm noticing is that the output always includes two match groups, one that's always undefined, presumably from the branch that wasn't taken (ie it's the first match if the first word is not quoted, second otherwise).

What kind of changes can I make to avoid getting the undefined match group and still get the quote-stripping behavior?

NOTE: The words are NOT strictly "word-only" (eg alphanumeric) characters. They can include non-word characters, just not the " character.

Answer 1

You need to use ^ (Start anchor) to match the first word and simply use \\w+ to match the word also i think you don't need the main group :

"([^"]*)"|(^\w+)

Demo

Answer 2

You are getting extra matches because of the nested groupings you have defined inside your regular expression. The corrected expression should be (?:"[^"]*"|[^"\\s]+) which would produce the following for your inputs (without string quotes)

'"quoted phrase" followed by text' => "quoted phrase"
'phrase without quotes followed by text' => phrase

Answer 3

You can't do what you want using just the regex. Other regex flavors have power features like the Branch Reset Group (which causes capturing groups in each branch to start with the same number):

(?|"([^"]*)"|([^"\s]+))

...or they let you use the same name for more than one group:

(?:"(?<token>[^"]*)"|(?<token>[^"\s]+))

...but JavaScript has nothing. Of all the regex flavors associated with programming languages (Perl, Python, Java, etc.), JavaScript is the most lacking in useful features. You just have to go through all the groups and find the one that's not undefined .