Difference between \\w and [:word:]?

Difference between \\w and [:word:]?

[Eric Abrahamsen]

Can someone explain to me what's happening here? What's the difference
between \\w and [:word:], inside and outside of a bracket construction?

(setq case-fold-search nil)
(string-match-p "[:word:]" "P") => nil
(string-match-p "[[:word:]]" "P") => 0
(string-match-p "\\w" "P") => 0
(string-match-p "[\\w]" "P") = nil

Re: Difference between \\w and [:word:]?

[Robert Pluim]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> Can someone explain to me what's happening here? What's the difference
> between \\w and [:word:], inside and outside of a bracket construction?
>
> (setq case-fold-search nil)
> (string-match-p "[:word:]" "P") => nil
> (string-match-p "[[:word:]]" "P") => 0
> (string-match-p "\\w" "P") => 0
> (string-match-p "[\\w]" "P") = nil

Consider

(string-match-p "[\\w]" "w") => ?

Robert

Re: Difference between \\w and [:word:]?

[Eric Abrahamsen]

Robert Pluim <rpluim@gmail.com> writes:

> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Can someone explain to me what's happening here? What's the difference
>> between \\w and [:word:], inside and outside of a bracket construction?
>>
>> (setq case-fold-search nil)
>> (string-match-p "[:word:]" "P") => nil
>> (string-match-p "[[:word:]]" "P") => 0
>> (string-match-p "\\w" "P") => 0
>> (string-match-p "[\\w]" "P") = nil
>
> Consider
>
> (string-match-p "[\\w]" "w") => ?

Oh, ha.

(string-match-p "[\\w]" "\\") => 0

And my other confusion (about character classes) was cleared up by
reading the manual more carefully.

Thanks,
Eric

Re: Difference between \\w and [:word:]?

[Ben Bacarisse]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> Can someone explain to me what's happening here? What's the difference
> between \\w and [:word:], inside and outside of a bracket construction?
>
> (setq case-fold-search nil)
> (string-match-p "[:word:]" "P") => nil

This is the same as (string-match-p "[wrdo:]" "P").  It's just a
collection of characters to be matched.

> (string-match-p "[[:word:]]" "P") => 0

This looks for a member of the character class [:word:] in the string.
It finds it (case-insensitively) at position 0.  Technically, the
current character class table determines what is a word-constituent.

> (string-match-p "\\w" "P") => 0

This does the same.

> (string-match-p "[\\w]" "P") = nil

And this is just a normal character class match looking for either a w
or a \.

I can never quite decide if it's good or bad that the named character
classes use the same [...] syntax.  [:digit:] is just a plain character
set with a duplicated : and i in it.  It only becomes a named class
inside a character set.

-- 
Ben.

Re: Difference between \\w and [:word:]?

[Eric Abrahamsen]

Ben Bacarisse <ben.usenet@bsb.me.uk> writes:

> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Can someone explain to me what's happening here? What's the difference
>> between \\w and [:word:], inside and outside of a bracket construction?
>>
>> (setq case-fold-search nil)
>> (string-match-p "[:word:]" "P") => nil
>
> This is the same as (string-match-p "[wrdo:]" "P").  It's just a
> collection of characters to be matched.

Yeah, I did figure that out once I read the manual properly.

>> (string-match-p "[[:word:]]" "P") => 0
>
> This looks for a member of the character class [:word:] in the string.
> It finds it (case-insensitively) at position 0.  Technically, the
> current character class table determines what is a word-constituent.
>
>> (string-match-p "\\w" "P") => 0
>
> This does the same.
>
>> (string-match-p "[\\w]" "P") = nil
>
> And this is just a normal character class match looking for either a w
> or a \.
>
> I can never quite decide if it's good or bad that the named character
> classes use the same [...] syntax.  [:digit:] is just a plain character
> set with a duplicated : and i in it.  It only becomes a named class
> inside a character set.

It definitely felt a little weird writing something like
"\\w[[:word:]-']+", where we're required to use two different notations
for the same match.

Thanks,
Eric

Re: Difference between \\w and [:word:]?

[Robert Pluim]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> It definitely felt a little weird writing something like
> "\\w[[:word:]-']+", where we're required to use two different notations
> for the same match.
>

Nothing prevents you from writing

[[:word:]][[:word:]-']+

Robert

Re: Difference between \\w and [:word:]?

[Yuri Khan]

On Fri, Jun 8, 2018 at 4:17 PM Robert Pluim <rpluim@gmail.com> wrote:

> > It definitely felt a little weird writing something like
> > "\\w[[:word:]-']+", where we're required to use two different notations
> > for the same match.
> >
>
> Nothing prevents you from writing
>
> [[:word:]][[:word:]-']+

Or even

(rx wordchar (any wordchar ?- ?'))

Re: Difference between \\w and [:word:]?

[Eric Abrahamsen]

Yuri Khan <yurivkhan@gmail.com> writes:

> On Fri, Jun 8, 2018 at 4:17 PM Robert Pluim <rpluim@gmail.com> wrote:
>
>> > It definitely felt a little weird writing something like
>> > "\\w[[:word:]-']+", where we're required to use two different notations
>> > for the same match.
>> >
>>
>> Nothing prevents you from writing
>>
>> [[:word:]][[:word:]-']+
>
> Or even
>
> (rx wordchar (any wordchar ?- ?'))

I know, I know -- there are many ways to skin this cat, and for some
reason I went straight for one that didn't work. Thanks to all for the
other suggestions.

Eric

Re: Difference between \\w and [:word:]?

[Stefan Monnier]

> I know, I know -- there are many ways to skin this cat, and for some
> reason I went straight for one that didn't work. Thanks to all for the
> other suggestions.

If you need help to find other ways that don't work, we'd be happy to
provide many more.


        Stefan

Re: Difference between \\w and [:word:]?

[Eric Abrahamsen]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> I know, I know -- there are many ways to skin this cat, and for some
>> reason I went straight for one that didn't work. Thanks to all for the
>> other suggestions.
>
> If you need help to find other ways that don't work, we'd be happy to
> provide many more.

If I wanted that, I'd post to emacs-tangent, emacs-rabbit-hole, and
emacs-garden-path!