[ruby-core:98540] [Ruby master Feature#15771] Add `String#split` option to set `split_type string` with a single space separator

Issue #15771 has been updated by sawa (Tsuyoshi Sawada).


Dan0042 (Daniel DeLorme) wrote in #note-10:
> That optimization is nice to have, but I think the point of this ticket is that it's currently not possible to have an arbitrary string separator.

I agree.

Dan0042 (Daniel DeLorme) wrote in #note-10:
> sawa (Tsuyoshi Sawada) wrote in #note-7:
> > My guess is that, perhaps, even now, it is very rare to use a single space string argument with the expectation to match single or multiple spaces.

> 
> It looks like using a single space string argument is not so rare:
> https://pastebin.com/pPyEf2GA

I didn't write that using a single space string argument is rare, I wrote that using a single space string argument **with the expectation to match single or multiple spaces** is rare.

In fact, my guess is that using a single space string argument is frequent, and that most of them expect to match only single space**. I have not confirmed this. If it turns out to be correct, then that would constitute the use cases that asked for.


----------------------------------------
Feature #15771: Add `String#split` option to set `split_type string` with a single space separator
https://bugs.ruby-lang.org/issues/15771#change-85831

* Author: 284km (kazuma furuhashi)
* Status: Feedback
* Priority: Normal
----------------------------------------
When `String#split`'s separator is a single space character, it executes under `split_type: awk`.

When you want to split literally by a single space `" "`, and not a sequence of space characters, you need to take special care. For example, the CSV library detours this behavior like [this](https://github.com/ruby/csv/blob/7ff57a50e81c368029fa9b664700bec4a456b81b/lib/csv/parser.rb#L508-L512):

```ruby
if @column_separator == " ".encode(@encoding)
  @split_column_separator = Regexp.new(@escaped_column_separator)
else
  @split_column_separator = @column_separator
end
```

Unfortunately, using a regexp here makes it slower than using a string. [The following result](https://github.com/284km/benchmarks_no_yatu#stringsplitstring-or-regexp) shows it is about nine times slower.

```sh
$ be benchmark-driver string_split_string-regexp.yml --rbenv '2.6.2'
Comparison:
              string:   3161117.6 i/s
              regexp:    344448.0 i/s - 9.18x  slower
```

I want to add a `:literal` option to execute the method under `split_type: string` as follows:

```ruby
" a  b   c    ".split(" ")                    # => ["a", "b", "c"]
" a  b   c    ".split(" ", literal: true)     # => ["", "a", "", "b", "", "", "c"]
" a  b   c    ".split(" ", -1)                # => ["a", "b", "c", ""]
" a  b   c    ".split(" ", -1, literal: true) # => ["", "a", "", "b", "", "", "c", "", "", "", ""]
```

### Implementation

- https://github.com/284km/ruby/tree/split_space
    - test code: https://github.com/284km/ruby/blob/split_space/test/ruby/test_string.rb#L1708-L1713




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

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