does emacs regular expression support (?!expression)

[土星五号]

does emacs regular expression support (?!expression)  ?

[Tassilo Horn]

土星五号 <laf...@gmail.com> writes:

> does emacs regular expression support (?!expression) ?

It would be more helpful if you'd tell us what (?!expression) would
match, preferably with some examples. Syntactically, emacs regexp's shy
groups look similar. The regular expression

"\\(:?foo\\|bar\\)\\([0-9]+\\)"

matches "foo19" or "bar23", but doesn't capture foo or bar, so that
(match-string 1) is the number.

Bye,
Tassilo

[Tim Landscheidt]

Tassilo Horn <tas...@member.fsf.org> wrote:

>> does emacs regular expression support (?!expression) ?

> It would be more helpful if you'd tell us what (?!expression) would
> match, preferably with some examples. Syntactically, emacs regexp's shy
> groups look similar. The regular expression

> "\\(:?foo\\|bar\\)\\([0-9]+\\)"

> matches "foo19" or "bar23", but doesn't capture foo or bar, so that
> (match-string 1) is the number.

In Perl, "(?!pattern)" is a zero-width negative look-ahead
assertion. Emacs does not support these AFAIK.

Tim

[Tassilo Horn]

Tim Landscheidt <t...@tim-landscheidt.de> writes:

Hi Tim,

>> It would be more helpful if you'd tell us what (?!expression) would
>> match, preferably with some examples. Syntactically, emacs regexp's shy
>> groups look similar. The regular expression
>
>> "\\(:?foo\\|bar\\)\\([0-9]+\\)"

BTW, that should have been (?:...).

>> matches "foo19" or "bar23", but doesn't capture foo or bar, so that
>> (match-string 1) is the number.
>
> In Perl, "(?!pattern)" is a zero-width negative look-ahead
> assertion. Emacs does not support these AFAIK.

I see. So when you do /foo(?!bar)/ in Perl, you'd need to do
"foo\\(?:[^b][^a][^r]\\)" in elisp.

Bye,
Tassilo

[Glenn Morris]

Tim Landscheidt wrote:

> In Perl, "(?!pattern)" is a zero-width negative look-ahead
> assertion. Emacs does not support these AFAIK.

http://debbugs.gnu.org/cgi/bugreport.cgi?bug=5393
(to date, not applied)

See also the linked
http://lists.gnu.org/archive/html/emacs-devel/2012-01/msg00732.html

[Tim Landscheidt]

Tassilo Horn <tas...@member.fsf.org> wrote:

> [...]

>>> matches "foo19" or "bar23", but doesn't capture foo or bar, so that
>>> (match-string 1) is the number.

>> In Perl, "(?!pattern)" is a zero-width negative look-ahead
>> assertion. Emacs does not support these AFAIK.

> I see. So when you do /foo(?!bar)/ in Perl, you'd need to do
> "foo\\(?:[^b][^a][^r]\\)" in elisp.

You have to account for possible end-of-buffer as well so
usually it's "easier" to use two matches ('(while (and
(search-forward-regexp "foo") (not (looking-at "bar"))))'
(untested)) - that's why I (still :-)) think these features
should be implemented in Emacs as well.

Tim

[Tim Landscheidt]

I wrote:

>> [...]

>>>> matches "foo19" or "bar23", but doesn't capture foo or bar, so that
>>>> (match-string 1) is the number.

>>> In Perl, "(?!pattern)" is a zero-width negative look-ahead
>>> assertion. Emacs does not support these AFAIK.

>> I see. So when you do /foo(?!bar)/ in Perl, you'd need to do
>> "foo\\(?:[^b][^a][^r]\\)" in elisp.

> You have to account for possible end-of-buffer as well so
> usually it's "easier" to use two matches ('(while (and
> (search-forward-regexp "foo") (not (looking-at "bar"))))'

> (untested)) - [...]

Eh, yes, there is an "(untested)", but still the logic is
obviously plain wrong. But you get the idea.

Tim

[Kevin Rodgers]

On 2/2/12 2:45 PM, Tassilo Horn wrote:

> Tim Landscheidt<t...@tim-landscheidt.de> writes:

>> In Perl, "(?!pattern)" is a zero-width negative look-ahead
>> assertion. Emacs does not support these AFAIK.
>
> I see. So when you do /foo(?!bar)/ in Perl, you'd need to do
> "foo\\(?:[^b][^a][^r]\\)" in elisp.

IIUC, the Perl regex would successfully match "foo" if it were followed by
"far", but the Emacs regexp would not.

Maybe \(?:[^b][^a][^r]\) should be \(?:[^b]\|b[^a]\|ba[^r]\)

--
Kevin Rodgers
Denver, Colorado, USA

[土星五号]

I would like match Err in any words(e.g. LastError), but not ErrorMode.
(defvar txt-mode-font-lock-keywords
  `(
    ;; 文件名
    ("\\\\\\(\\w+\\.exe\\)" 1 font-lock-keyword-face)
    ;; IP和版本
    ("[^0-9]\\([0-9]\\{1,3\\}\\.[0-9]\\{1,3\\}\\.[0-9]\\{1,3\\}\\.[0-9]\\{1,3\\}\\)" 1 font-lock-keyword-face)
    ;; Error
    ("\\<\\([Dd][Bb][Ee][Rr][Rr][Oo][Rr]\\)\\>" 1 font-lock-warning-face)
    ("\\<\\([Dd][Bb][Ee][Rr][Rr]\\)\\>" 1 font-lock-warning-face)
    ("\\<\\([Ee][Rr][Rr][Oo][Rr]\\)\\>" 1 font-lock-warning-face)
    ("\\([Ee][Rr][Rr]\\)\\(?!orMode\\)" 1 font-lock-warning-face)
    ("\\(错误\\)" 1 font-lock-warning-face)
    ("\\(失败\\)" 1 font-lock-warning-face)
    ("\\(严重\\)" 1 font-lock-warning-face)
    ("\\<\\([Ff][Aa][Ii][Ll][Ee][Dd]\\)\\>" 1 font-lock-warning-face)
    ("\\<\\([Ff][Aa][Ii][Ll]\\)\\>" 1 font-lock-warning-face)
    ("\\<\\([Cc][Aa][Nn]\\s+[Nn][Oo][Tt]\\)\\>" 1 font-lock-warning-face)
    ("\\<\\([Cc][Aa][Nn]'[Tt]\\)\\>" 1 font-lock-warning-face)
    ("\\<\\([Ww][Rr][Oo][Nn][Gg]\\)\\>" 1 font-lock-warning-face)
    ;; Warning
    ("\\<\\([Ww][Aa][Rr][Nn][Ii][Nn][Gg]\\)\\>" 1 font-lock-warning-face)
    ("\\<\\([Ww][Aa][Rr][Nn]\\)\\>" 1 font-lock-warning-face)
    ("\\(警告\\)" 1 font-lock-warning-face)
    ;; Exception
    ("\\<\\([Ee][Xx][Cc][Ee][Pp][Tt][Ii][Oo][Nn]\\)\\>" 1 font-lock-warning-face)
    ("\\<\\([Ee][Xx][Cc][Ee][Pp][Tt]\\)\\>" 1 font-lock-warning-face)
    ("\\(异常\\)" 1 font-lock-warning-face)
    )
  "Basic font lock keywords for txt mode.  Highlights keywords.")

2012/2/3 Kevin Rodgers <kevin.d...@gmail.com>

On 2/2/12 2:45 PM, Tassilo Horn wrote:

Tim Landscheidt<tim@tim-landscheidt.de>  writes:

[Thien-Thi Nguyen]

() 土星五号 <laf...@gmail.com>
() Fri, 3 Feb 2012 10:12:27 +0800

I would like match Err in any words(e.g. LastError), but not ErrorMode.

("\\([Ee][Rr][Rr]\\)\\(?!orMode\\)" 1 font-lock-warning-face)

You can add a spec for ‘ErrorMode’, with default face, prior to this one.

[Tim Landscheidt]

Thien-Thi Nguyen <t...@gnuvola.org> wrote:

> I would like match Err in any words(e.g. LastError), but not ErrorMode.

> ("\\([Ee][Rr][Rr]\\)\\(?!orMode\\)" 1 font-lock-warning-face)

> You can add a spec for ‘ErrorMode’, with default face, prior to this one.

Which Emacs version does support that?

Tim

[Thien-Thi Nguyen]

() Tim Landscheidt <t...@tim-landscheidt.de>
() Fri, 03 Feb 2012 11:17:25 +0000

> You can add a spec for ‘ErrorMode’, with default face, prior to this one.

Which Emacs version does support that?

Any that processes these specs in the given order, with
"masking semantics", i suppose. (I haven't tried.)

[Tim Landscheidt]

Thien-Thi Nguyen <t...@gnuvola.org> wrote:

> > You can add a spec for ‘ErrorMode’, with default face, prior to this one.

> Which Emacs version does support that?

> Any that processes these specs in the given order, with
> "masking semantics", i suppose. (I haven't tried.)

What does:

| (search-forward-regexp "\\([Ee][Rr][Rr]\\)\\(?!orMode\\)")

give on your Emacs?

Tim

[土星五号]

No, i put it in txt-mode-load.el.

 .emacs:
(require 'txt-mode-load)

2012/2/4 Tim Landscheidt <t...@tim-landscheidt.de>

[Tim Landscheidt]

Thien-Thi Nguyen <t...@gnuvola.org> wrote:

> [...]
> It throws an error "Invalid regular expression".
> Sorry for being unclear. [...]

Thanks.

Tim

[David Combs]

In article <mailman.3005.132822070...@gnu.org>,

And, by the above, that's not a "regexp" (emacs), but a bunch of
elisp code. Meaning that (as far as I know) you can't just
type it in, in response to a prompt-for-a-regexp, as you'd get
with a M-C s, query-replace-regexp, etc, but would have to --
well, what *would* you have to do for a M-C % -- you would have
to write a whole paragraph of elisp to duplicate what query-replace-regexp
does, somewhere inserting the above elisp that emulates the perl
regexp?

Easier to just get out of emacs for a bit, and do that perl-like
edit in perl itself?

David