Feature #9826: Enumerable#slice_between

• Status: Closed
• Assignee: akr (Akira Tanaka)
• Target version: -

[akr (Akira Tanaka)]

I'd like to add a new method, Enumerable#slice_between.

It is similar to Enumerable#slice_before but it can use
not only the element after the slice position
but also the element before the slice position.

enum.slice_between(pattern_before, pattern_after=nil) -> an_enumerator
enum.slice_between {|elt_before, elt_after| bool }    -> an_enumerator

I found several people try to use Enumerable#slice_before for
compacting sequence of integers using hyphens:
1,2,4,9,10,11,12,15,16,19,20,21 to 1,2,4,9-12,15,16,19-21.

• ruby-talk:370132 Dave Thomas and James Edward Gray II
• http://d.hatena.ne.jp/keyesberry/20120107/p1 (in Japanese)

slice_before needs state management to do it.
slice_between can be used more easily for this situation:

a = [1,2,4,9,10,11,12,15,16,19,20,21]
p a.slice_between {|i, j| i+1 != j }.map {|a| a.length < 3 ? a : "#{a.first}-#{a.last}" }.join(",")

Or more verbosely as:

a = [1,2,4,9,10,11,12,15,16,19,20,21]
b = a.slice_between {|i, j| i+1 != j }
p b.to_a #=> [[1, 2], [4], [9, 10, 11, 12], [15, 16], [19, 20, 21]]
c = b.map {|a| a.length < 3 ? a : "#{a.first}-#{a.last}" }
p c #=> [[1, 2], [4], "9-12", [15, 16], "19-21"]
d = c.join(",")
p d #=> "1,2,4,9-12,15,16,19-21"

Also, I found several usages for Enumerable#slice_between.

• ruby-talk:359255 split logs where interval is 30s or more.
• http://stackoverflow.com/questions/6258971/how-do-i-return-a-group-of-sequential-numbers-that-might-exist-in-an-array
• ruby-talk:415057 collects same elements. (Enumerable#chunk can be used, though.)

Any idea?

[akr (Akira Tanaka)]

• File slice_between2.patch slice_between2.patch added

I updated the patch to simplify argument handling.

[matz (Yukihiro Matsumoto)]

• Status changed from Open to Feedback

I understand the use-case, and I'd like to add the feature.
But slice_between does not describe what it does.
Try another name.

Matz.

[akr (Akira Tanaka)]

• File slice_between3.patch slice_between3.patch added

I updated the patch to be applied cleanly for HEAD after slice_after merge.

[akr (Akira Tanaka)]

Yukihiro Matsumoto wrote:

But slice_between does not describe what it does.
Try another name.

I searched "between" with wordnet.

% wordnet between -over

Overview of adv between

The adv between has 2 senses (first 2 from tagged texts)

1. (1) between, betwixt -- (in the interval; "dancing all the dances with little rest between")
2. (1) between, 'tween -- (in between; "two houses with a tree between")

I agree that enum.slice_between(X,Y) doesn't behave as "in the interval".
There is no interval between X and Y.

However I feel the 2nd meaning is appropriate.
So, slice_between itself may be possible.
slice_in_between may be more clear.

Some idea:

• slice_between
• slice_in_between
• slice_boundary_between
• slice_boundary
• slice_at

I'd like to ask others (especially English native speakers).

Any opinion?

[sawa (Tsuyoshi Sawada)]

There can be another method that works the opposite to the proposed method with respect to the truthfulness of the block, and I think the two methods should come in a pair. And I have a feeling that cons should somehow be used in the name of these methods. For example, the method implemented by Akira Tanaka may be called slice_cons, and the opposite one may be called chunk_cons, and the following could be equivalent:

[1,2,4,9,10,11,12,15,16,19,20,21].slice_cons{|i, j| i+1 != j}
[1,2,4,9,10,11,12,15,16,19,20,21].chunk_cons{|i, j| i+1 == j}

But by analogy from the method each_cons, the cons in the names may imply that the arity of the block need not be two. I have no clear idea how this feature can be extended to take a block of other arity, and if the implication is a problem, then the word pair might work better:

[1,2,4,9,10,11,12,15,16,19,20,21].slice_pair{|i, j| i+1 != j}
[1,2,4,9,10,11,12,15,16,19,20,21].chunk_pair{|i, j| i+1 == j}

[mrkn (Kenta Murata)]

[1,2,4,9,10,11,12,15,16,19,20,21].slice_when {|i, j| i+1 != j}

[matz (Yukihiro Matsumoto)]

I prefer #slice_when. Besides that, could you explain the behavior when no block is given?
#slice_when might not suitable for that calling pattern. But maybe we don't need that.

Matz.

[akr (Akira Tanaka)]

I'm grad to see an acceptable name.

Non-block form can be used to split paragraphs (sequence of non-empty lines
with trailing empty lines), for example.

% ruby -e '
lines = ["foo\n", "bar\n", "\n", "baz\n", "\n", "\n", "qux\n", "quux\n"]
lines.slice_when(/\A\s*\z/, /\S/).each {|para| p para }'
["foo\n", "bar\n", "\n"]
["baz\n", "\n", "\n"]
["qux\n", "quux\n"]

If the name, slice_when, is not appropriate to this form, I can drop
the form.

The block form can same task as follows. (It is bit longer, of course.)

% ruby -e '
lines = ["foo\n", "bar\n", "\n", "baz\n", "\n", "\n", "qux\n", "quux\n"]
lines.slice_when {|l1, l2| /\A\s*\z/ =~ l1 && /\S/ =~ l2 }.each {|para| p para }' 
["foo\n", "bar\n", "\n"]
["baz\n", "\n", "\n"]
["qux\n", "quux\n"]

[akr (Akira Tanaka)]

• File slice_when.patch slice_when.patch added

I updated the patch at today's lunch break: slice_when.patch
It defines Enumerable#slice_when and it doesn't support non-block form.

[matz (Yukihiro Matsumoto)]

• Status changed from Feedback to Open
• Assignee set to akr (Akira Tanaka)

Agreed with slice_when. Gi ahead.

Matz.

[akr (Akira Tanaka)]

• Status changed from Open to Closed
• % Done changed from 0 to 100

Applied in changeset r47652.

---

• enum.c (enum_slice_when): New method: Enumerable#slice_when.
  (slicewhen_i): New function.
  (slicewhen_ii): New function.

• enumerator.c (InitVM_Enumerator): New method:
  Enumerator::Lazy#slice_when.

  [ruby-core:62499] [Feature #9826]