[ruby-core:75059] [Ruby trunk Feature#12306] Implement String #blank? #present? and improve #strip and family to handle unicode

Issue #12306 has been updated by Sam Saffron.


Nobuyoshi Nakada wrote:
> I think that `String#blank?` equals to `/\P{space}/ !~ self`.
> Is it slow, especially than `strip`?

Yes, in Rails apps this blank very quickly becomes a bottleneck see: http://tmm1.net/ruby21-profiling/ for a real world example. I have seen Rails applications spend 2-5% of the time in Active Support String#blank? it is used super aggressively due to the "style" of programming used. 

Recently Richard Schneems has been micro optimizing blank? due to this discovery see: https://github.com/rails/rails/pull/24658

The regex approach will never come close to a native implementation see bench here: https://github.com/SamSaffron/fast_blank 


@Matz as to why this tends to happen so much in Rails world, this is a super common scenario 

- Allow a textinput with User name
- Then from controller run

```
# one letter name is fine, especially for Chinese names, but blank is not
raise YouGotToAddAName if params[:name].blank?
```

In Rails programming it turns out that people are checking for "blankness" of strings in tons and tons of very very common scenario. The speed boost of having a fast native implementation would be very welcome. 

I am fine with not adding present? ( unless s.blank?) fills the need just fine 

Regarding timing, would a PR adding blank? and unicode support to strip be welcome for 2.4 timeframe? 

Regarding non-unicode strings I think strip should just keep working the way it does now. 



----------------------------------------
Feature #12306: Implement String #blank? #present? and improve #strip and family to handle unicode
https://bugs.ruby-lang.org/issues/12306#change-58180

* Author: Sam Saffron
* Status: Open
* Priority: Normal
* Assignee: 
----------------------------------------
Time and again there have been rejected feature requests to Ruby core to implement `blank` and `present` protocols across all objects as ActiveSupport does. I am fine with this call and think it is fair. 

However, for the narrow case of String having `#blank?` and `#present?` makes sense. 

- Provides a natural extension over `#strip`, `#lstrip` and `#rstrip`. `("   ".strip.length == 0) == "    ".blank?`

- Plays nicely with ActiveSupport, providing an efficient implementation in Ruby core: see: https://github.com/SamSaffron/fast_blank, implementing blank efficiently requires a c extension. 

However, if this work is to be done, `#strip` and should probably start dealing with unicode blanks, eg: 

```
irb(main):008:0> [0x3000].pack("U")
=> "　"
irb(main):009:0> [0x3000].pack("U").strip.length
=> 1
```

So there are 2 questions / feature requests here

1. Can we add blank? and present? to String? 
2. Can we amend strip and family to account for unicode per: https://github.com/SamSaffron/fast_blank/blob/master/ext/fast_blank/fast_blank.c#L43-L74



-- 
https://bugs.ruby-lang.org/

Unsubscribe: <mailto:[e-mail]?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>