Dumb oneliner challenge

probably_the_wrong_thing_todo.rb

# So to explain this contrived example a bit. Say you have an array of input that you wanted to match with
# multiple regexes how do you in ~1 line match the lot of things.
#
# So say for a given input of %w/ab abc abcde bcde bc efgh etc/
# You have multiple regexes to match against, in this case its just an array of strings that we build regexes off of.
#
# This is my kitbash first try.
#
# Anyone have any better ideas? I know you could just union them both if === works as your comparison:
# %w/ab abc bc bcde efgh/ & %w/ab bc/
# ["ab", "bc"]

all = %w/ab abcd abcdef bcde bc ladkfjlkadsjfja lajflkajfkaf/
matches = %w/ab bc/
p all.dup.keep_if {|l| matches.collect {|r| l !~ /#{r}/}.include?(false) }
p all.dup.keep_if {|l| matches.collect{|r| not l !~ /#{r}/}.include?(true)}

# Output (order is unimportant)
# ["ab", "abcd", "abcdef", "bcde", "bc"]
# ["ab", "abcd", "abcdef", "bcde", "bc"]

This also works.
p all.dup.delete_if {|l| matches.collect{|r| l !~ /#{r}/}.inject(:&)}

["ab", "abcd", "abcdef", "bcde", "bc"]

Not sure where this Regexp#& method is coming from (my ri lists it, but irb throws NoMethodError). But in any case, here's a better way:

Regexp.union(*matches).tap {|r| break all.select {|l| l =~ r } }

Notably:

1. Uses #select
2. Avoids iterating over matches (twice!) for each element of all
3. Sets up the regular expression once, reuses the same compiled instance for each iteration