Custom sort array of hashes based on multiple key/value pairs

Question

I have an array of hashes which I need to sort based on two different key value pairs.

Here is the array I am trying to sort:

array_group =  [
 {operator: OR, name: "some string", status: false},
 {operator: AND, name: "other string", status: false},
 {operator: _NOT_PRESENT, name: "another string", status: true},
 {operator: AND, name: "just string", status: true}
]

I want to sort array_group so I have items with status: true first, followed by status: false, followed by the items with operator: _NOT_PRESENT and finally sort it based on name, resulting in something like:

array_group = [
 {operator: AND, name: "just string", status: true},
 {operator: AND, name: "other string", status: false},
 {operator: OR, name: "some string", status: false},
 {operator: _NOT_PRESENT, name: "another string", status: true},
]

Is there a way that I can get this done without creating sub-arrays and sorting them and concatenating them back?

The statement of your question suggests that status can have a value other than True or False. Is that correct? — Cary Swoveland
– Cary Swoveland, Commented Nov 6, 2015 at 21:52
yes status is not actually a boolean its a string which I am later wrapping to boolean values if it has true or false in some other part of code — sujay
– sujay, Commented Nov 6, 2015 at 21:54
How did "other string"'s status switch from true to false? Why is "another string" not first in the sorted array, consider that it ties with "other string" on the value of :status and is the only one of the two to have _NOT_PRESENT as the value of :operator? — Cary Swoveland
– Cary Swoveland, Commented Nov 6, 2015 at 22:12
In that case you need to edit the question and correct the statement of the desired ordering. You should leave the current statement and then add something like "Edit: this statement is not correct ect. It should read:...". If you merely replace the current statement it would render comments and answers meaningless and puzzling. Also correct your desired result (array). — Cary Swoveland
– Cary Swoveland, Commented Nov 7, 2015 at 23:20
And while those things might have been done, we'll waste our time suggesting things you've already tried and rejected unless you tell us. Help us help you by reducing the possible things to try. — the Tin Man
– the Tin Man, Commented Nov 7, 2015 at 23:23

sschmeck · Accepted Answer · 2015-11-10 07:41:10Z

6

You can also use Enumerable#sort_by. The example builds an array which is compared element by element when sorting.

array_group.sort_by { |e| [e[:operator] == "_NOT_PRESENT" ? 1 : 0, 
                           e[:status] ? 0 : 1, 
                           e[:name]] }

The example above orders records with operator: "_NOT_PRESENT" also by :status. The following snippet precisely performs the ordering from the question.

def priority(h)
  case
  when h[:operator] == "_NOT_PRESENT" then 3
  when h[:status] == false            then 2
  # h[:status] == true
  else 1
  end
end

array_group.sort_by { |e| [priority(e), e[:name]] }

edited Nov 10, 2015 at 7:41

answered Nov 6, 2015 at 21:15

sschmeck

7,7936 gold badges45 silver badges78 bronze badges

Sign up to request clarification or add additional context in comments.

10 Comments

Cary Swoveland Over a year ago

If embedded in a method this will not work because _NOT_PRESENT is a local variable. If you change _NOT_PRESENT to NOT_PRESENT, both here and in array_group, after having defined the constant NOT_PRESENT, you get {:operator=>"and", :name=>"just string", :status=>true} as the first element of the sorted array, which is incorrect.

sschmeck Over a year ago

@CarySwoveland The first element you mention is the first element of the desired result. What's wrong with it?

Cary Swoveland Over a year ago

Your first line puts "another string" after all other hashes (regardless of the value of :status, which is the first filter). Suppose :status=true for "another string" (as it is now), but :status=false for all other hashes. Then "another string" should be sorted first, not last.

sschmeck Over a year ago

@CarySwoveland I understand the question in another way. 1st place ordering is [status: true, status: false, operator: _NOT_DEFINED], 2nd is sorted by :name. That's what the desired result looks like.

sujay Over a year ago

@sschmeck you are right about the order of array items i.e., status: true comes first (all status true items are further sorted by name) then comes status false (all status false items further sorted by name) and then finally status _NOT_PRESENT will be present :) this answer works for me but I am trying few more scenarios before I accept it ,,, thank you for your answer

|

Dmitry Sokurenko · Accepted Answer · 2015-11-06 22:16:40Z

1

You can use the Array.sort method. It accepts a block with two arguments (x, y), when x is larger than y it should return 1, otherwise -1, and 0 if they are equal.

The code:

OR = "OR"
AND = "AND"
_NOT_PRESENT = "_NOT_PRESENT"

array_group =  [
 {operator: OR, name: "some string", status: false},
 {operator: AND, name: "other string", status: true},
 {operator: _NOT_PRESENT, name: "another string", status: true},
 {operator: AND, name: "just string", status: true}
]

results = array_group.sort do |x, y|
  next x[:operator] == _NOT_PRESENT ? 1 : -1 if x[:operator] == _NOT_PRESENT || y[:operator] == _NOT_PRESENT
  next x[:status] ? -1 : 1 if x[:status] != y[:status]
  next x[:name] <=> y[:name]
end

And btw, you your input and output data doesn't match each other — the hash with OR is false in the input, but true in the output.

I believe your output should look like:

[{:operator=>"AND", :name=>"just string", :status=>true},
 {:operator=>"AND", :name=>"other string", :status=>true},
 {:operator=>"OR", :name=>"some string", :status=>false},
 {:operator=>"_NOT_PRESENT", :name=>"another string", :status=>true}]

That output will actually match your sorting logic.

edited Nov 6, 2015 at 22:16

answered Nov 6, 2015 at 21:11

Dmitry Sokurenko

6,1324 gold badges34 silver badges54 bronze badges

2 Comments

sujay Over a year ago

thank you for the answer,,, sorry for presenting wrong info,,, it was a typo regarding one entry but my final output should always have _NOT_PRESENT at the last..

Dmitry Sokurenko Over a year ago

I've added another version, so _NOT_PRESENTs are always last, otherwise status == true goes before status == false, when statuses are equal then names are compared alphabetically.

Doguita · Accepted Answer · 2015-11-06 22:44:34Z

You can use Enumerable#sort_by using an array to keep the groups you want and sorting by name without lose those groups. For example:

[[0,0,1], [1,1,1], [0,1,0], [1,0,0], [0,1,1]].sort_by &:itself
# => [[0, 0, 1], [0, 1, 0], [0, 1, 1], [1, 0, 0], [1, 1, 1]]

Using this, you can set an array with length 2. Using index 0 to "group" by the elements you want and index 1 to sort by name.

OR = "OR"
AND = "AND"
_NOT_PRESENT = "_NOT_PRESENT"

array_group =  [
 { operator: OR, name: "z string", status: false },
 { operator: AND, name: "a string", status: false },
 { operator: AND, name: "z string", status: true },
 { operator: OR, name: "a string", status: true },
 { operator: _NOT_PRESENT, name: "d string", status: true },
 { operator: _NOT_PRESENT, name: "b string", status: true },
 { operator: _NOT_PRESENT, name: "c string", status: false },
 { operator: _NOT_PRESENT, name: "a string", status: false }
]

# Types of "groups" you want to keep. Greater values will be at the end
BY_NOT_PRESENT = 2
BY_STATUS_FALSE = 1
BY_STATUS_TRUE = 0

array_group.sort_by do |a|
  group = if a[:operator] == _NOT_PRESENT
            BY_NOT_PRESENT
          else
            a[:status] ? BY_STATUS_TRUE : BY_STATUS_FALSE
          end
  [group, a[:name]]
end
#=> [{:operator=>"OR", :name=>"a string", :status=>true},
#     {:operator=>"AND", :name=>"z string", :status=>true},
#     {:operator=>"AND", :name=>"a string", :status=>false}, 
#     {:operator=>"OR", :name=>"z string", :status=>false}, 
#     {:operator=>"_NOT_PRESENT", :name=>"a string", :status=>false}, 
#     {:operator=>"_NOT_PRESENT", :name=>"b string", :status=>true}, 
#     {:operator=>"_NOT_PRESENT", :name=>"c string", :status=>false}, 
#     {:operator=>"_NOT_PRESENT", :name=>"d string", :status=>true}]

I am doing this way, because if just sort_by an array like [status, not_present, name] then status will have precedence on sort, overriding the name sort, resulting this:

#=> [{:operator=>"OR", :name=>"a string", :status=>true}, 
#    {:operator=>"AND", :name=>"z string", :status=>true}, 
#    {:operator=>"AND", :name=>"a string", :status=>false}, 
#    {:operator=>"OR", :name=>"z string", :status=>false}, 
#    {:operator=>"_NOT_PRESENT", :name=>"b string", :status=>true}, 
#    {:operator=>"_NOT_PRESENT", :name=>"d string", :status=>true}, 
#    {:operator=>"_NOT_PRESENT", :name=>"a string", :status=>false}, 
#    {:operator=>"_NOT_PRESENT", :name=>"c string", :status=>false}]

Cary Swoveland · Accepted Answer · 2015-11-07 03:15:28Z

I assume the unspecified constants OR, AND and NOT_PRESENT are defined as follows (for example):

OR = "or"
AND = "and"
NOT_PRESENT = "not present"

(Constants must begin with a capital letter. _NOT_PRESENT is a local variable.)

When using Enumerable#sort_by, you need to sort on an array containing three elements, corresponding to :status, :operator and :name, in that order. The first of those three elements is smallest (say 0) if the value of :status is true, next smallest (say 1) if the value of :status is false and largest (say 2) if it is anything else. We have:

def rearrange(arr)
  arr.sort_by do |h|
    [
      h[:status]==true ? 0 : (h[:status]==false ? 1 : 2),
      (h[:operator]==NOT_PRESENT) ? 0 : 1, 
      h[:name]
    ]
  end
end

array_group = [
 {operator: AND, name: "just string", status: true},
 {operator: OR, name: "some string", status: false},
 {operator: AND, name: "other string", status: false},
 {operator: NOT_PRESENT, name: "another string", status: true},
]

rearrange(array_group)
  #=> [{:operator=>"not present", :name=>"another string", :status=>true},
  #    {:operator=>"and", :name=>"just string", :status=>true},
  #    {:operator=>"and", :name=>"other string", :status=>false}, 
  #    {:operator=>"or", :name=>"some string", :status=>false}]

This is not the result desired, but it is consistent with my understanding of the question.

Collectives™ on Stack Overflow

Custom sort array of hashes based on multiple key/value pairs

4 Answers 4

10 Comments

2 Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

10 Comments

2 Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related