Word truncation

[oss92]

Patched the truncate strings method to include truncation by word

Examples:

'Once upon a time'.truncate(4, word: true)
=> "Once upon a time"

'Once upon a time in a world far far away'.truncate(4, word: true)
=> "Once upon a time..."

'Once<br>upon<br>a<br>time<br>in<br>a<br>world'.truncate(5, word: true, separator: '<br>')
 => "Once<br>upon<br>a<br>time<br>in..."

'Once<br>upon<br>a<br>time<br>in<br>a<br>world'.truncate(5, word: true, separator: '<br>', omission: '[...]')
 => "Once<br>upon<br>a<br>time<br>in[...]"

[rafaelfranca]

Can we implement another method instead of adding a new option?

[oss92]

yes we can, I will implement it an create a new pull request

[matthewd]

You can just update this pull request; no need for a new one.

If we're going to add such a method, hopefully we can make it a bit more performant, too... this sounds like it would be allocating a lot of objects.

[oss92]

okay will do. Thanks for the advice regarding the pull request

[oss92]

Done and tested, less objects allocated and everything is working fine.

Also added rtruncate to get the rest of a truncated string. This is useful in web apps when truncating strings and having a 'read more...' link, the rest of the string is now easily gotten.

Example
'Once upon a time in a world far far away'.truncate(27, omission: '') + 'Once upon a time in a world far far away'.rtruncate(27, omission: '')
=> "Once upon a time in a world far far away"

[matthewd]

👎

[matthewd]

I don't think it's sufficiently useful.

(Also, the '(Read more) ...' example seems at best misleading... if you're passing a different omission than you gave to truncate, things are surely going to get confusing?)

[oss92]

The read more part is just an example, yes it may be a bit misleading for some developers but the method is useful for, as I said, getting the rest of the truncated string without having to sub! two strings from each other.

[matthewd]

Yeah, I understand the use case... the problem is just that adding a new method to AS requires a pretty high standard of justification, and (personally) I don't think this clears it.

[oss92]

Okay, after all it is off-point from the 'word truncation' so it can be removed

[oss92]

The documentation is identical to that of truncate but that's for consistency. Shall I change these lines or do you think the rtruncate method is not at all useful?

[matthewd]

The split here is still allocating a separate string for every word... what about something like:

def truncate_words(words_count, options={})
  sep = options[:separator] || /(?<=\w)\b/
  sep = Regexp.escape(sep.to_s) unless Regexp === sep
  if self =~ /^((?:.+?#{sep}){#{words_count - 1}}.+?)#{sep}.*/
    $1 + (options[:omission] || '...')
  else
    dup
  end
end

[oss92]

Yes something like this is much better actually

It just has this problem of a character sticking to the word at the end
For one of the test for my method against yours
Expected: "Hello Big World!"
Actual: "Hello Big World..."

But overall it is much better, it should be the one used because of its efficiency.

[matthewd]

Ah, depending on what you want, try /\s+/ or /\b(?=\w)/.

Or, after building $1 + omission, it could compare the length of the result to decide whether it should still just return dup after all.

[oss92]

Yes it works now if
sep = options[:separator] || /\s+/

For me this is better than comparing the length at the end. All tests are fine now. Now, shall I commit this on the same branch or will you create your own pull request with that?

[oss92]

Pardon me I still have limited experience with "open-source"

[matthewd]

Yep, go ahead and update with that... and then squash down to a single commit, and add a CHANGELOG entry.

Then we can see whether anyone else has a 👍/👎 opinion on adding the new method. 😄

[oss92]

Ready to merge 😄

[matthewd]

Thanks!

I moved your changelog entry to the top, and fixed an issue with handling of multi-line strings in my suggested regexp.

After I pushed, I noticed that your git committer name is "root" 😟. You can submit a PR to https://github.com/fxn/rails-contributors/blob/master/app/models/names_manager.rb to fix up your entry on http://contributors.rubyonrails.org/, if you like. 😅

[oss92]

I shall add my commit SHA and my name to this method, correct me if I am mistaken please. Thanks

self.authors_of_special_cased_commits(commit)

[matthewd]

👍