Pattern engine extension for \< and \> support?

Pattern engine extension for \< and \> support?

[Sebastian Gniazdowski]

[-- Attachment #1: Type: text/plain, Size: 917 bytes --]

Hi,
in regexps, there are \<, \> and also \b sequences, that match zero-length
string on the word boundary. They are particularly useful for e.g.:
refactoring substitutions, like e.g.: with sed:

sed -r -i -e 's/\<myvar\>/other/'

This automatically protects pre/suffixed strings like str_myvar and
myvar_str. In mcedit, I often use \b that's available there (\< and \> do
not work), to obtain the same effect with the regex:

\bmyvat\b

I'm not sure if distinction of left and right boundary offered by \<, \>
actually makes a difference.

Zsh has similar sequences already - (#s) and (#e). They do match 0=length
strings on begin or end of the matched string. So maybe the topic is known
to any implementing person? Is it hard to add such \b or \<,\> sequences?
Could it be also possible to make it backward-safe, i.e. the sequence be
ignored by proceeding Zsh versions?




-- 
Best regards,
Sebastian Gniazdowski

[-- Attachment #2: Type: text/html, Size: 2272 bytes --]

Re: Pattern engine extension for \< and \> support?

[Stephane Chazelas]

2023-01-27 10:01:15 +0000, Sebastian Gniazdowski:
> Hi,
> in regexps, there are \<, \> and also \b sequences, that match zero-length
> string on the word boundary. They are particularly useful for e.g.:
> refactoring substitutions, like e.g.: with sed:
> 
> sed -r -i -e 's/\<myvar\>/other/'
> 
> This automatically protects pre/suffixed strings like str_myvar and
> myvar_str. In mcedit, I often use \b that's available there (\< and \> do
> not work), to obtain the same effect with the regex:
[...]

Some regexp engines like on BSDs also have [[:<:]] / [[:>:]] for
that which would possibly be easy to add to zsh (I'd think
anything using backslash would be a no-no).

zsh has [[:WORD:]] already as similar to \w. Its [[:<:]],
[[:>:]] would need to honour $WORDCHARS as well for consistency.

\< / \> come from ex/vi in the 70s AFAIK, \b from perl in the 80s.
Perl also has \B and:

     \b{}   Match at Unicode boundary of specified type
     \B{}   Match where corresponding \b{} doesn't match
     \b     Match a \w\W or \W\w boundary
     \B     Match except at a \w\W or \W\w boundary

-- 
Stephane

Re: Pattern engine extension for \< and \> support?

[Daniel Shahaf]

Sebastian Gniazdowski wrote on Fri, Jan 27, 2023 at 10:01:15 +0000:
> Hi,
> in regexps, there are \<, \> and also \b sequences, that match zero-length
> string on the word boundary. They are particularly useful for e.g.:
> refactoring substitutions, like e.g.: with sed:
> 
> sed -r -i -e 's/\<myvar\>/other/'
> 
> This automatically protects pre/suffixed strings like str_myvar and
> myvar_str. In mcedit, I often use \b that's available there (\< and \> do
> not work), to obtain the same effect with the regex:
> 
> \bmyvat\b
> 
> I'm not sure if distinction of left and right boundary offered by \<, \>
> actually makes a difference.
> 
> Zsh has similar sequences already - (#s) and (#e). They do match 0=length
> strings on begin or end of the matched string. So maybe the topic is known
> to any implementing person? Is it hard to add such \b or \<,\> sequences?
> Could it be also possible to make it backward-safe, i.e. the sequence be
> ignored by proceeding Zsh versions?

This happens to be covered here:

https://github.com/zsh-users/zsh-syntax-highlighting/blob/master/docs/highlighters/regexp.md

tl;dr For portability use RE_MATCH_PCRE.