Mark, Spacing Combining should be in \w

[chx]

The documentation says "\w any character that matches \p{L}, \p{N}, \p{Mn}, or \p{Pc}" but I believe Mc should be considered a word character as well.

[NWilson]

Thank you, I will consider this for some future release. Our primary goal would be to align with Perl, as they have a potentially larger and more active community to decide these details. I would consider differing if there is a consensus among other regex major engines, or a formal unicode recommendation with a different definition.

[NWilson]

Hmm. Unicode TR18 defines "word" as a compatibility property for use in regular expressions (https://unicode.org/reports/tr18/tr18-23.html#word).

It seems to correspond to:

• \p{alpha} (a "derived core property" consisting of Uppercase + Lowercase + Lt + Lm + Lo + Nl + Other_Alphabetic)
• \p{gc=Mark} (Mn + Mc + Me)
• \p{digit} (Nd)
• \p{gc=Connector_Punctuation} (Pc)
• \p{Join_Control} (this is its own non-derived property)

So you are right that Mc should ideally be included.

Perl manpage state that it follows the Unicode definition of "word". I haven't delved into its code to verify this however.

Nor have I delved into our own implementation to verify our documentation is up to date.

[chx]

As far as I can see the code aligns with the documentation https://3v4l.org/VU6v9 (sorry for using PHP here, I am more familiar)