Understanding "\_W*" in vim regex

Question

I have been going through some answers to a vimgolf, and stumbled upon this regex:

:s/\_W*/-/g|x<CR>

Which converts

abcdefghijklm

to

-a-b-c-d-e-f-g-h-i-j-k-l-m-

The regex (without unrelated keystrokes) is /\_W*/-/. The manual indicates that: \W stands for non-word character: [^0-9A-Za-z_]

\_x Where "x" is any of the characters above: The character class with end-of-line added

So \_W is non-word characters with end-of-line added.

How does that put the dashes between the characters? Is there a hidden metacharacter between each character in vim text?

Answer 1

This is not a question of hidden character, the trick comes from the * combined with the flag g:

Specifying * after a character classe will match "zero or more" items (which includes an empty string): between a and b there is zero non-word character, thus \W* will match this "not existing space".

For example you could also have used \_s* or \_d* and still get the same result since you match zero items of the class.

Answer 2

To add to what statox mentioned above, if you want to avoid matching against the "not existing" space, use \+ instead of * to match one or more instances of the non-word character:

:s/\_W\+/-/g