Finding gaps in a sorted sequence


#1

Hi Folks,

Given a sorted enumeration I need to find the first gap in a sequence.

e.g.
3 == find_gap [1,2,4,5]
nil == find_gap [1,2,3,4]

Here’s the best I can come up with

def first_gap( seq )
seq.each_cons(2) do |l,r|
_next = l.next
return _next if r!= _next
end
nil
end

but it seems rather ugly. Anyone have a more elegant implementation?

Cheers,
Pete


#2

On Thu, Nov 6, 2008 at 12:39 PM, Pete H. removed_email_address@domain.invalid
wrote:

Given a sorted enumeration I need to find the first gap in a sequence.

e.g.
3 == find_gap [1,2,4,5]
nil == find_gap [1,2,3,4]

$ cat bar.rb
def find_gap array
f = array.first
l = array.last
s = l - f + 1

return nil if s == array.size

if array.size == 2
return f + 1
end

c = array.size / 2
if array[c] != f + c
a = find_gap(array[0…c])
else
b = find_gap(array[c…-1])
end

a ? a : b
end

p find_gap([1,2,4,5])
p find_gap([1,2,3,4])
p find_gap([1,2,3,4,5,6,7,11,12])
p find_gap([2,3,4,5,6,7,11,12])

$ ruby bar.rb
3
nil
8
8


#3

Pete H. wrote:

Hi Folks,

Given a sorted enumeration I need to find the first gap in a sequence.

e.g.
3 == find_gap [1,2,4,5]
nil == find_gap [1,2,3,4]

Here’s the best I can come up with

def first_gap( seq )
seq.each_cons(2) do |l,r|
_next = l.next
return _next if r!= _next
end
nil
end

but it seems rather ugly. Anyone have a more elegant implementation?

Cheers,
Pete

The following does not quite meet your specs (it returns an array with
all gaps, empty if there are no gaps). Anyway, here is my try:

def find_gaps( ar )
(ar.first … ar.last).to_a - ar
end

p find_gaps( [1,2,4,5] )
p find_gaps( [“a”, “b”, “d”, “e”, “h”] )

require ‘date’
d1 = Date.new(2008,11,6)
d2 = Date.new(2008,11,7)
d3 = Date.new(2008,11,9)
puts find_gaps( [d1,d2,d3] )

=> [3]
=> [“c”, “f”, “g”]
=> 2008-11-08

hth,

Siep


#4

PH = Pete H.
SK = Siep K.

PH> Given a sorted enumeration I need to find the first gap in a
sequence.
PH> e.g.
PH> 3 == find_gap [1,2,4,5]
PH> nil == find_gap [1,2,3,4]

SK> The following does not quite meet your specs (it returns an array
with
SK> all gaps, empty if there are no gaps). Anyway, here is my try:

def find_gaps( ar )
(ar.first … ar.last).to_a - ar
end

Given the above definition, it’s easy to get Pete’s behavior exactly:

def find_gap( ar )
find_gaps(ar).first
end

And then you have a choice of methods to use, depending on whether you
need the full list or just the first gap.


#5

On Thu, Nov 6, 2008 at 9:39 AM, Pete H. removed_email_address@domain.invalid wrote:

Hi Folks,

Given a sorted enumeration I need to find the first gap in a sequence.

e.g.
3 == find_gap [1,2,4,5]
nil == find_gap [1,2,3,4]

irb(main):001:0> g = [1,2,3,5,6,8,9,10]
=> [1, 2, 3, 5, 6, 8, 9, 10]

irb(main):002:0> gap = g.inject {|a, e| e == a.next ? e : (break
a.next)}
=> 4

martin


#6

unsubscribe


#7

Pete H. removed_email_address@domain.invalid writes:

def first_gap( seq )
seq.each_cons(2) do |l,r|
_next = l.next
return _next if r!= _next
end
nil
end

but it seems rather ugly. Anyone have a more elegant implementation?

I think your version is pretty readable. However, it appears you do
pay a 60% speed penalty with the Enumerator version. So I guess it
depends on whether conciseness and readability are more important than
efficiency, or not.

— snip —
require ‘enumerator’
require ‘benchmark’
include Benchmark

def first_gap list
list.each_cons(2) do |l,r|
_next = l.next
return _next unless r == _next
end
nil
end

def other_gap list
prev = nil
list.each do |e|
if prev && (n = prev.next) != e
return n
end
prev = e
end
nil
end

t = (1…100).to_a

[100, 250, 500].each do |n|
t = (1…n).to_a
bm(5) do |x|
x.report(‘first’) { 1_000.times { first_gap(t) } }
x.report(‘other’) { 1_000.times { other_gap(t) } }
end
end
— snip —

~/temp$ ruby first_gap.rb
user system total real
first 0.700000 0.250000 0.950000 ( 0.957255)
other 0.430000 0.170000 0.600000 ( 0.604145)
user system total real
first 1.720000 0.630000 2.350000 ( 2.354915)
other 1.070000 0.420000 1.490000 ( 1.490321)
user system total real
first 3.470000 1.250000 4.720000 ( 4.838116)
other 2.120000 0.840000 2.960000 ( 3.018840)


#8

On Fri, Nov 7, 2008 at 9:28 AM, Brian A. removed_email_address@domain.invalid
wrote:

Martin DeMello removed_email_address@domain.invalid writes:

I still think I like it best though because it seems to be the most
natural Ruby solution to the original problem.
Well not quite, it does not return nil if there is no gap, one has to
check if the return value is the last element of the array
and replace it with nil in that case.
One should know that inject is quite slow, but no reason not to use it
unless performance is a problem.

Cheers
Robert

C’est véritablement utile puisque c’est joli.

Antoine de Saint Exupéry


#9

Martin DeMello removed_email_address@domain.invalid writes:

=> [1, 2, 3, 5, 6, 8, 9, 10]

irb(main):002:0> gap = g.inject {|a, e| e == a.next ? e : (break a.next)}
=> 4

Beautiful. Excellent use of inject :slight_smile: I would’ve thought this would be
better performing also, but it lags the ‘each’ version by quite a bit:

       user     system      total        real

OP 3.490000 1.240000 4.730000 ( 4.755358)
Brian 2.170000 0.830000 3.000000 ( 3.000508)
Martin 3.320000 1.240000 4.560000 ( 4.582286)

I still think I like it best though because it seems to be the most
natural Ruby solution to the original problem.


#10

Robert D. removed_email_address@domain.invalid writes:

On Fri, Nov 7, 2008 at 9:28 AM, Brian A. removed_email_address@domain.invalid wrote:

Martin DeMello removed_email_address@domain.invalid writes:

I still think I like it best though because it seems to be the most
natural Ruby solution to the original problem.
Well not quite, it does not return nil if there is no gap,

D’oh! I missed that. Ok, I guess I like my version best again then :slight_smile: