Flatten a list

Flatten a list
You are encouraged to solve this task according to the task description, using any language you may know.

Task

Write a function to flatten the nesting in an arbitrary list of values.

Your program should work on the equivalent of this list:

  [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]

Where the correct result would be the list:

   [1, 2, 3, 4, 5, 6, 7, 8]

Related task

•   Tree traversal

8th

\ take a list (array) and flatten it:

: (flatten)  \ a -- a
	(
		\ is it a number?
		dup >kind ns:n n:= if
			\ yes.  so add to the list
			r> swap a:push >r
		else
			\ it is not, so flatten it
			(flatten)
		then
		drop
	) a:each drop ;
	
: flatten \ a -- a
	[] >r (flatten) r> ;

[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
dup . cr
flatten 
. cr 
bye

Output:

[[1],2,[[3,4],5],[[[]]],[[[6]]],7,8,[]]
[1,2,3,4,5,6,7,8]

ACL2

(defun flatten (tr)
   (cond ((null tr) nil)
         ((atom tr) (list tr))
         (t (append (flatten (first tr))
                    (flatten (rest tr))))))

ActionScript

function flatten(input:Array):Array {
	var output:Array = new Array();
	for (var i:uint = 0; i < input.length; i++) {
                //typeof returns "object" when applied to arrays. This line recursively evaluates nested arrays,
                // although it may break if the array contains objects that are not arrays.
		if (typeof input[i]=="object") {
			output=output.concat(flatten(input[i]));
		} else {
			output.push(input[i]);
		}
	}
	return output;
}

Ada

nestable_lists.ads:

generic
   type Element_Type is private;
   with function To_String (E : Element_Type) return String is <>;
package Nestable_Lists is

   type Node_Kind is (Data_Node, List_Node);
   
   type Node (Kind : Node_Kind);
   
   type List is access Node;
   
   type Node (Kind : Node_Kind) is record
      Next : List;
      case Kind is
         when Data_Node =>
            Data    : Element_Type;
         when List_Node =>
            Sublist : List;
      end case;
   end record;
   
   procedure Append (L : in out List; E : Element_Type);
   procedure Append (L : in out List; N : List);
   
   function Flatten (L : List) return List;

   function New_List (E : Element_Type) return List;
   function New_List (N : List) return List;
   
   function To_String (L : List) return String;
   
end Nestable_Lists;

nestable_lists.adb:

with Ada.Strings.Unbounded;

package body Nestable_Lists is

   procedure Append (L : in out List; E : Element_Type) is
   begin
      if L = null then
         L := new Node (Kind => Data_Node);
         L.Data := E;
      else
         Append (L.Next, E);
      end if;
   end Append;

   procedure Append (L : in out List; N : List) is
   begin
      if L = null then
         L := new Node (Kind => List_Node);
         L.Sublist := N;
      else
         Append (L.Next, N);
      end if;
   end Append;

   function Flatten (L : List) return List is
      Result  : List;
      Current : List := L;
      Temp    : List;
   begin
      while Current /= null loop
         case Current.Kind is
            when Data_Node =>
               Append (Result, Current.Data);
            when List_Node =>
               Temp := Flatten (Current.Sublist);
               while Temp /= null loop
                  Append (Result, Temp.Data);
                  Temp := Temp.Next;
               end loop;
         end case;
         Current := Current.Next;
      end loop;
      return Result;
   end Flatten;
   
   function New_List (E : Element_Type) return List is
   begin
      return  new Node'(Kind => Data_Node, Data => E, Next => null);
   end New_List;

   function New_List (N : List) return List is
   begin
      return new Node'(Kind => List_Node, Sublist => N, Next => null);
   end New_List;

   function To_String (L : List) return String is
      Current : List := L;
      Result  : Ada.Strings.Unbounded.Unbounded_String;
   begin
      Ada.Strings.Unbounded.Append (Result, "[");
      while Current /= null loop
         case Current.Kind is
            when Data_Node =>
               Ada.Strings.Unbounded.Append
                 (Result, To_String (Current.Data));
            when List_Node =>
               Ada.Strings.Unbounded.Append
                 (Result, To_String (Current.Sublist));
         end case;
         if Current.Next /= null then
            Ada.Strings.Unbounded.Append (Result, ", ");
         end if;
         Current := Current.Next;
      end loop;
      Ada.Strings.Unbounded.Append (Result, "]");
      return Ada.Strings.Unbounded.To_String (Result);
   end To_String;

end Nestable_Lists;

example usage:

with Ada.Text_IO;
with Nestable_Lists;

procedure Flatten_A_List is
   package Int_List is new Nestable_Lists
     (Element_Type => Integer,
      To_String    => Integer'Image);

   List : Int_List.List := null;
begin
   Int_List.Append (List, Int_List.New_List (1));
   Int_List.Append (List, 2);
   Int_List.Append (List, Int_List.New_List (Int_List.New_List (3)));
   Int_List.Append (List.Next.Next.Sublist.Sublist, 4);
   Int_List.Append (List.Next.Next.Sublist, 5);
   Int_List.Append (List, Int_List.New_List (Int_List.New_List (null)));
   Int_List.Append (List, Int_List.New_List (Int_List.New_List
                    (Int_List.New_List (6))));
   Int_List.Append (List, 7);
   Int_List.Append (List, 8);
   Int_List.Append (List, null);
   
   declare
      Flattened : constant Int_List.List := Int_List.Flatten (List);
   begin
      Ada.Text_IO.Put_Line (Int_List.To_String (List));
      Ada.Text_IO.Put_Line (Int_List.To_String (Flattened));
   end;
end Flatten_A_List;

Output:

[[ 1],  2, [[ 3,  4],  5], [[[]]], [[[ 6]]],  7,  8, []]
[ 1,  2,  3,  4,  5,  6,  7,  8]

Aikido

function flatten (list, result) {
    foreach item list {
        if (typeof(item) == "vector") {
            flatten (item, result)
        } else {
            result.append (item)
        }
    }
}

var l = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
var newl = []
flatten (l, newl)

// print out the nicely formatted result list
print ('[')
var comma = ""
foreach item newl {
    print (comma + item)
    comma = ", "
}
println("]")

Output:

 [1, 2, 3, 4, 5, 6, 7, 8]

Aime

void
show_list(list l)
{
    integer i, k;

    o_text("[");

    i = 0;
    while (i < ~l) {
        o_text(i ? ", " : "");
        if (l_j_integer(k, l, i)) {
            o_integer(k);
        } else {
            show_list(l[i]);
        }
        i += 1;
    }

    o_text("]");
}

list
flatten(list c, object o)
{
    if (__id(o) == INTEGER_ID) {
        c.append(o);
    } else {
        l_ucall(o, flatten, 1, c);
    }

    c;
}

integer
main(void)
{
    list l;

    l = list(list(1), 2, list(list(3, 4), 5),
             list(list(list())), list(list(list(6))), 7, 8, list());

    show_list(l);
    o_byte('\n');

    show_list(flatten(list(), l));
    o_byte('\n');

    return 0;
}

Output:

[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
[1, 2, 3, 4, 5, 6, 7, 8]

ALGOL 68

Works with: ALGOL 68 version Revision 1 - no extensions to language used

Works with: ALGOL 68G version Any - tested with release 1.18.0-9h.tiny

Works with: ELLA ALGOL 68 version Any (with appropriate job cards) - tested with release 1.8-8d

Flattening is built in to all of Algol68's transput routines. The following example also uses widening, where scalars are converted into arrays.

main:(
  [][][]INT list = ((1), 2, ((3,4), 5), ((())), (((6))), 7, 8, ());
  print((list, new line))
)

Output:

         +1         +2         +3         +4         +5         +6         +7         +8

APL

Dyalog APL

Flatten is a primitive in APL, named enlist

∊

Output:

      ∊((1) 2 ((3 4) 5) (((⍬))) (((6))) 7 8 (⍬))
1 2 3 4 5 6 7 8

AppleScript

my_flatten({{1}, 2, {{3, 4}, 5}, {{{}}}, {{{6}}}, 7, 8, {}})

on my_flatten(aList)
    if class of aList is not list then
        return {aList}
    else if length of aList is 0 then
        return aList
    else
        return my_flatten(first item of aList) & (my_flatten(rest of aList))
    end if
end my_flatten

Or, to make more productive use of the language (where "efficiency" is a function of the scripter's time, rather than the machine's) we can express this in terms of a generic concatMap:

Translation of: JavaScript

-- flatten :: Tree a -> [a]
on flatten(t)
    if class of t is list then
        concatMap(flatten, t)
    else
        t
    end if
end flatten

--------------------------- TEST ---------------------------
on run
    
    flatten([[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []])
    
    --> {1, 2, 3, 4, 5, 6, 7, 8}
end run


-------------------- GENERIC FUNCTIONS ---------------------

-- concatMap :: (a -> [b]) -> [a] -> [b]
on concatMap(f, xs)
    set lst to {}
    set lng to length of xs
    tell mReturn(f)
        repeat with i from 1 to lng
            set lst to (lst & |λ|(item i of xs, i, xs))
        end repeat
    end tell
    return lst
end concatMap

-- Lift 2nd class handler function into 1st class script wrapper 
-- mReturn :: Handler -> Script
on mReturn(f)
    if class of f is script then
        f
    else
        script
            property |λ| : f
        end script
    end if
end mReturn

Output:

{1, 2, 3, 4, 5, 6, 7, 8}

It can be more efficient to build just one list by appending items to it than to proliferate lists using concatenation:

on flatten(theList)
    script o
        property flatList : {}
        
        -- Recursive handler dealing with the current (sub)list.
        on flttn(thisList)
            script p
                property l : thisList
            end script
            
            repeat with i from 1 to (count thisList)
                set thisItem to item i of p's l
                if (thisItem's class is list) then
                    flttn(thisItem)
                else
                    set end of my flatList to thisItem
                end if
            end repeat
        end flttn
    end script
    
    if (theList's class is not list) then return theList
    o's flttn(theList)

    return o's flatList
end flatten

Arturo

print flatten [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]

Output:

1 2 3 4 5 6 7 8

AutoHotkey

Works with: AutoHotkey_L

AutoHotkey doesn't have built in list data type. This examples simulates a list in a tree type object and flattens that tree.

list := object(1, object(1, 1), 2, 2, 3, object(1, object(1, 3, 2, 4)
, 2, 5), 4, object(1, object(1, object(1, object()))), 5
, object(1, object(1, 6)), 6, 7, 7, 8, 9, object())
msgbox % objPrint(list) ; (( 1 ) 2 (( 3  4 ) 5 )(((())))(( 6 )) 7  8 ())
msgbox % objPrint(objFlatten(list)) ; ( 1  2  3  4  5  6  7  8 )
return

!r::reload
!q::exitapp

objPrint(ast, reserved=0)
{
  if !isobject(ast)
    return " " ast " "
  
  if !reserved
    reserved := object("seen" . &ast, 1)  ; to keep track of unique objects within top object
  
  enum := ast._newenum()
  while enum[key, value]
  {
    if reserved["seen" . &value]
      continue  ; don't copy repeat objects (circular references)
;   string .= key . ": " . objPrint(value, reserved)
    string .= objPrint(value, reserved)
  }
  return "(" string ")"
}


objFlatten(ast)
{
  if !isobject(ast)
    return ast
  
  flat := object() ; flat object
  
  enum := ast._newenum()
  while enum[key, value]
  {
    if !isobject(value)
      flat._Insert(value)
    else
    {
      next := objFlatten(value)
      loop % next._MaxIndex()
      flat._Insert(next[A_Index])
      
    }
  }
  return flat
}

BASIC

BaCon

BaCon has the concept of delimited strings, which may contain delimited strings within delimited strings etc. Such nested delimited strings must be surrounded by (escaped) double quotes in order to avoid their delimiter messing up operations on higher level delimited strings. However, from functional point of view, a delimited string is the same as a regular list. The special function FLATTEN$ can actually flatten out lists within lists. The last SORT$ in the program below makes sure no empty items remain in the list.

OPTION COLLAPSE TRUE

lst$ = "\"1\",2,\"\\\"3,4\\\",5\",\"\\\"\\\\\"\\\\\"\\\"\",\"\\\"\\\\\"6\\\\\"\\\"\",7,8,\"\""

PRINT lst$

REPEAT
    lst$ = FLATTEN$(lst$)
UNTIL AMOUNT(lst$, ",") = AMOUNT(FLATTEN$(lst$), ",")

PRINT SORT$(lst$, ",")

Output:

"1",2,"\"3,4\",5","\"\\"\\"\"","\"\\"6\\"\"",7,8,""
1,2,3,4,5,6,7,8

BASIC256

Translation of: FreeBASIC

sComma = "": sFlatter = ""
sString = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"

For siCount = 1 To Length(sString)
	If Instr("[] ,", Mid(sString, siCount, 1)) = 0 Then
		sFlatter = sFlatter & sComma & Mid(sString, siCount, 1)
		sComma = ", "
	End If
Next siCount

Print "["; sFlatter; "]"
End

Output:

Igual que la entrada de FreeBASIC.

Chipmunk Basic

Works with: Chipmunk Basic version 3.6.4

Works with: QBasic

10 cls
20 sstring$ = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"
30 for sicount = 1 to len(sstring$)
40   if instr("[] ,",mid$(sstring$,sicount,1)) = 0 then
50     sflatter$ = sflatter$+scomma$+mid$(sstring$,sicount,1)
60     scomma$ = ", "
70   endif
80 next sicount
90 print "[";sflatter$;"]"
100 end

FreeBASIC

Translation of: Gambas

Dim As String sComma, sString, sFlatter
Dim As Short siCount

sString = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"

For siCount = 1 To Len(sString)
    If Instr("[] ,", Mid(sString, siCount, 1)) = 0 Then 
        sFlatter += sComma + Mid(sString, siCount, 1)
        sComma = ", "
    End If
Next siCount

Print "["; sFlatter; "]"
Sleep

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

FutureBasic

Definitely old school.

local fn FlattenList( list as Str255 ) as Str255
  long   i
  Str255 flatStr, commaStr
  
  flatStr = ""
  for i = 1 to len$(list)
    if ( instr$( 0, "[] ,", mid$( list, i, 1 ) ) === 0 )
      flatStr += commaStr + mid$( list, i, 1 )
      commaStr = ", " 
    end if
  next
end fn = flatStr

window 1, @"Flatten a list", ( 0, 0, 350, 150 )

print "["; fn FlattenList( "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]" ); "]"

HandleEvents

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

Modern and a little outside the box.

void local fn FlattenAList
  CFStringRef listStr = @"[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]"
  CFArrayRef  listArr = fn StringComponentsSeparatedByCharactersInSet( listStr, fn CharacterSetWithCharactersInString( @"\"[ ]," ) )
  CFMutableArrayRef mutArr = fn MutableArrayWithArray( listArr )
  MutableArrayRemoveObject( mutArr, @"" )
  CFStringRef flatStr = fn ArrayComponentsJoinedByString( mutArr, @", " )
  printf @"[%@]", flatStr
end fn

fn FlattenAList

HandleEvents

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

Gambas

Click this link to run this code

'Code 'borrowed' from Run BASIC

Public Sub Main()
Dim sComma, sString, sFlatter As String
Dim siCount As Short

sString = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"
For siCount = 1 To Len(sString)
 If InStr("[] ,", Mid$(sString, siCount, 1)) = 0 Then 
  sFlatter = sFlatter & sComma & Mid(sString, siCount, 1)
  sComma = ","
 End If
Next
Print "["; sFlatter; "]"

End

Output:

[1,2,3,4,5,6,7,8]

GW-BASIC

Works with: Chipmunk Basic

Works with: QBasic

10 CLS
20 SSTRING$ = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"
30 FOR SICOUNT = 1 TO LEN(SSTRING$)
40   IF INSTR("[] ,",MID$(SSTRING$,SICOUNT,1)) = 0 THEN SFLATTER$ = SFLATTER$+SCOMMA$+MID$(SSTRING$,SICOUNT,1): SCOMMA$ = ", "
50 NEXT SICOUNT
60 PRINT "[";SFLATTER$;"]"
70 END

MSX Basic

Works with: QBasic

Works with: Chipmunk Basic

Works with: GW-BASIC

10 CLS
20 S$ = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"
30 FOR SICOUNT = 1 TO LEN(S$)
40   IF INSTR("[] ,",MID$(S$,SICOUNT,1)) = 0 THEN SFLATTER$ = SFLATTER$+SCOMMA$+MID$(S$,SICOUNT,1): SCOMMA$ = ", "
50 NEXT SICOUNT
60 PRINT "[";SFLATTER$;"]"
70 END

PureBasic

Structure RCList
  Value.i
  List A.RCList()
EndStructure

Procedure Flatten(List A.RCList())
  ResetList(A())
  While NextElement(A())
    With A()
      If \Value
        Continue
      Else
        ResetList(\A())
        While NextElement(\A())
          If \A()\Value: A()\Value=\A()\Value: EndIf
        Wend
      EndIf
      While ListSize(\A()): DeleteElement(\A()): Wend
      If Not \Value: DeleteElement(A()): EndIf
    EndWith
  Wend
EndProcedure

Set up the MD-List & test the Flattening procedure.

;- Set up two lists, one multi dimensional and one 1-D.
NewList A.RCList()

;- Create a deep list
With A()
  AddElement(A()):  AddElement(\A()): AddElement(\A()): \A()\Value=1
  AddElement(A()):                     A()\Value=2
  AddElement(A()):  AddElement(\A()): \A()\Value=3
  AddElement(\A()):                   \A()\Value=4
  AddElement(A()):  AddElement(\A()): \A()\Value=5
  AddElement(A()):  AddElement(\A()): AddElement(\A()): AddElement(\A())
  AddElement(A()):  AddElement(\A()): AddElement(\A()): \A()\Value=6
  AddElement(A()):                     A()\Value=7
  AddElement(A()):                     A()\Value=8
  AddElement(A()):  AddElement(\A()): AddElement(\A())
EndWith

Flatten(A())

;- Present the result
If OpenConsole()
  Print("Flatten: [")
  ForEach A()
    Print(Str(A()\Value))
    If ListIndex(A())<(ListSize(A())-1)
      Print(", ")
    Else
      PrintN("]")
    EndIf
  Next
  Print(#CRLF$+"Press ENTER to quit"): Input()
EndIf

Flatten: [1, 2, 4, 5, 6, 7, 8]

QBasic

Works with: QBasic version 1.1

Works with: QuickBasic version 4.5

sString$ = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"

FOR siCount = 1 TO LEN(sString$)
    IF INSTR("[] ,", MID$(sString$, siCount, 1)) = 0 THEN
        sFlatter$ = sFlatter$ + sComma$ + MID$(sString$, siCount, 1)
        sComma$ = ", "
    END IF
NEXT siCount

PRINT "["; sFlatter$; "]"
END

Run BASIC

This example is incorrect. Please fix the code and remove this message. Details: The task is not in string translation but in list translation.

n$ = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"
for i = 1 to len(n$)
 if instr("[] ,",mid$(n$,i,1)) = 0 then 
  flatten$ = flatten$ + c$ + mid$(n$,i,1)
  c$ = ","
 end if
next i
print "[";flatten$;"]"

Output:

[1,2,3,4,5,6,7,8]

TI-89 BASIC

There is no nesting of lists or other data structures in TI-89 BASIC, short of using variable names as pointers.

True BASIC

LET sstring$ = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"
FOR sicount = 1 TO LEN(sstring$)
    IF pos("[] ,",(sstring$)[sicount:sicount+1-1]) = 0 THEN
       LET sflatter$ = sflatter$ & scomma$ & (sstring$)[sicount:sicount+1-1]
       LET scomma$ = ", "
    END IF
NEXT sicount
PRINT "["; sflatter$; "]"
END

XBasic

Works with: Windows XBasic

PROGRAM	"Flatten a list"

DECLARE FUNCTION  Entry ()

FUNCTION  Entry ()
  n$ = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"
  FOR i = 1 TO LEN(n$)
    IF INSTR("[] ,",MID$(n$,i,1)) = 0 THEN
      flatten$ = flatten$ + c$ + MID$(n$,i,1)
      c$ = ", "
    END IF
  NEXT i
  PRINT "[";flatten$;"]"
END FUNCTION

END PROGRAM

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

Yabasic

Translation of: FreeBASIC

sString$ = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"

For siCount = 1 To Len(sString$)
	If Instr("[] ,", Mid$(sString$, siCount, 1)) = 0 Then
		sFlatter$ = sFlatter$ + sComma$ + Mid$(sString$, siCount, 1)
		sComma$ = ", "
	End If
Next siCount

Print "[", sFlatter$, "]"
End

Output:

Igual que la entrada de FreeBASIC.

ZX Spectrum Basic

This example is incorrect. Please fix the code and remove this message. Details: The task is not in string translation but in list translation.

10 LET f$="["
20 LET n$="[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8 []]"
30 FOR i=2 TO (LEN n$)-1
40 IF n$(i)>"/" AND n$(i)<":" THEN LET f$=f$+n$(i): GO TO 60
50 IF n$(i)="," AND f$(LEN f$)<>"," THEN LET f$=f$+","
60 NEXT i
70 LET f$=f$+"]": PRINT f$

BQN

Enlist ← {(∾𝕊¨)⍟(1<≡)⥊𝕩}

Output:

   Enlist ⟨⟨1⟩, 2, ⟨⟨3, 4⟩, 5⟩, ⟨⟨⟨⟩⟩⟩, ⟨⟨⟨6⟩⟩⟩, 7, 8, ⟨⟩⟩
⟨ 1 2 3 4 5 6 7 8 ⟩

Bracmat

A list is automatically flattened during evaluation if the items are separated by either commas, plusses, asterisks or white spaces.

On top of that, lists separated with white space, plusses or asterisks have 'nil'-elements removed when evaluated. (nil-elements are empty strings, 0 and 1 respectively.)

On top of that, lists separated with plusses or asterisks have their elements sorted and, if possible, combined when evaluated.

A list that should not be flattened upon evaluation can be separated with dots.

( (myList = ((1), 2, ((3,4), 5), ((())), (((6))), 7, 8, ()))
& put$("Unevaluated:")
& lst$myList
& !myList:?myList          { the expression !myList evaluates myList }
& put$("Flattened:")
& lst$myList
)

Brat

array.prototype.flatten = {
  true? my.empty?
    { [] }
    { true? my.first.array?
      { my.first.flatten + my.rest.flatten }
      { [my.first] + my.rest.flatten }
    }
}

list = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
p "List: #{list}"
p "Flattened: #{list.flatten}"

Burlesque

Usually flattening Blocks is done with the Concat command but it only removes one level of nesting therefore it is required to chain Concat calls until the Block does not contain Blocks anymore.

blsq ) {{1} 2 {{3 4} 5} {{{}}} {{{6}}} 7 8 {}}{\[}{)to{"Block"==}ay}w!
{1 2 3 4 5 6 7 8}

C

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

typedef struct list_t list_t, *list;
struct list_t{
	int is_list, ival; /* ival is either the integer value or list length */
	list *lst;
};

list new_list()
{
	list x = malloc(sizeof(list_t));
	x->ival = 0;
	x->is_list = 1;
	x->lst = 0;
	return x;
}

void append(list parent, list child)
{
	parent->lst = realloc(parent->lst, sizeof(list) * (parent->ival + 1));
	parent->lst[parent->ival++] = child;
}

list from_string(char *s, char **e, list parent)
{
	list ret = 0;
	if (!parent) parent = new_list();

	while (*s != '\0') {
		if (*s == ']') {
			if (e) *e = s + 1;
			return parent;
		}
		if (*s == '[') {
			ret = new_list();
			ret->is_list = 1;
			ret->ival = 0;
			append(parent, ret);
			from_string(s + 1, &s, ret);
			continue;
		}
		if (*s >= '0' && *s <= '9') {
			ret = new_list();
			ret->is_list = 0;
			ret->ival = strtol(s, &s, 10);
			append(parent, ret);
			continue;
		}
		s++;
	}

	if (e) *e = s;
	return parent;
}

void show_list(list l)
{
	int i;
	if (!l) return;
	if (!l->is_list) {
		printf("%d", l->ival);
		return;
	}

	printf("[");
	for (i = 0; i < l->ival; i++) {
		show_list(l->lst[i]);
		if (i < l->ival - 1) printf(", ");
	}
	printf("]");
}

list flatten(list from, list to)
{
	int i;
	list t;

	if (!to) to = new_list();
	if (!from->is_list) {
		t = new_list();
		*t = *from;
		append(to, t);
	} else
		for (i = 0; i < from->ival; i++)
			flatten(from->lst[i], to);
	return to;
}

void delete_list(list l)
{
	int i;
	if (!l) return;
	if (l->is_list && l->ival) {
		for (i = 0; i < l->ival; i++)
			delete_list(l->lst[i]);
		free(l->lst);
	}

	free(l);
}

int main()
{
	list l = from_string("[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []", 0, 0);

	printf("Nested: ");
	show_list(l);
	printf("\n");

	list flat = flatten(l, 0);
	printf("Flattened: ");
	show_list(flat);

	/* delete_list(l); delete_list(flat); */
	return 0;
}

Output:

Nested: [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
Flattened: [1, 2, 3, 4, 5, 6, 7, 8]

C#

Works with: C# version 3+

Actual Workhorse code

using System;
using System.Collections;
using System.Linq;

namespace RosettaCodeTasks
{
	static class FlattenList
	{
		public static ArrayList Flatten(this ArrayList List)
		{
			ArrayList NewList = new ArrayList ( );

			NewList.AddRange ( List );

			while ( NewList.OfType<ArrayList> ( ).Count ( ) > 0 )
			{
				int index = NewList.IndexOf ( NewList.OfType<ArrayList> ( ).ElementAt ( 0 ) );
				ArrayList Temp = (ArrayList)NewList[index];
				NewList.RemoveAt ( index );
				NewList.InsertRange ( index, Temp );
			}
			
			return NewList;
		}
	}
}

Method showing population of arraylist and usage of flatten method

using System;
using System.Collections;

namespace RosettaCodeTasks
{
	class Program
	{
		static void Main ( string[ ] args )
		{

			ArrayList Parent = new ArrayList ( );
			Parent.Add ( new ArrayList ( ) );
			((ArrayList)Parent[0]).Add ( 1 );
			Parent.Add ( 2 );
			Parent.Add ( new ArrayList ( ) );
			( (ArrayList)Parent[2] ).Add ( new ArrayList ( ) );
			( (ArrayList)( (ArrayList)Parent[2] )[0] ).Add ( 3 );
			( (ArrayList)( (ArrayList)Parent[2] )[0] ).Add ( 4 );
			( (ArrayList)Parent[2] ).Add ( 5 );
			Parent.Add ( new ArrayList ( ) );
			( (ArrayList)Parent[3] ).Add ( new ArrayList ( ) );
			( (ArrayList)( (ArrayList)Parent[3] )[0] ).Add ( new ArrayList ( ) );
			Parent.Add ( new ArrayList ( ) );
			( (ArrayList)Parent[4] ).Add ( new ArrayList ( ) );
			( (ArrayList)( (ArrayList)Parent[4] )[0] ).Add ( new ArrayList ( ) );

			( (ArrayList)( (ArrayList)( (ArrayList)( (ArrayList)Parent[4] )[0] )[0] ) ).Add ( 6 );
			Parent.Add ( 7 );
			Parent.Add ( 8 );
			Parent.Add ( new ArrayList ( ) );


			foreach ( Object o in Parent.Flatten ( ) )
			{
				Console.WriteLine ( o.ToString ( ) );
			}
		}

	}
}

Works with: C# version 4+

	public static class Ex {
		public static List<object> Flatten(this List<object> list) {

			var result = new List<object>();
			foreach (var item in list) {
				if (item is List<object>) {
					result.AddRange(Flatten(item as List<object>));
				} else {
					result.Add(item);
				}
			}
			return result;
		}
		public static string Join<T>(this List<T> list, string glue) {
			return string.Join(glue, list.Select(i => i.ToString()).ToArray());
		}
	}

	class Program {

		static void Main(string[] args) {
			var list = new List<object>{new List<object>{1}, 2, new List<object>{new List<object>{3,4}, 5}, new List<object>{new List<object>{new List<object>{}}}, new List<object>{new List<object>{new List<object>{6}}}, 7, 8, new List<object>{}};

			Console.WriteLine("[" + list.Flatten().Join(", ") + "]");
			Console.ReadLine();
		}
	}

C++

#include <list>
#include <boost/any.hpp>

typedef std::list<boost::any> anylist;

void flatten(std::list<boost::any>& list)
{
  typedef anylist::iterator iterator;

  iterator current = list.begin();
  while (current != list.end())
  {
    if (current->type() == typeid(anylist))
    {
      iterator next = current;
      ++next;
      list.splice(next, boost::any_cast<anylist&>(*current));
      current = list.erase(current);
    }
    else
      ++current;
  }
}

Use example:

Since C++ currently doesn't have nice syntax for initializing lists, this includes a simple parser to create lists of integers and sublists. Also, there's no standard way to output this type of list, so some output code is added as well.

#include <cctype>
#include <iostream>

// *******************
// * the list parser *
// *******************

void skipwhite(char const** s)
{
  while (**s && std::isspace((unsigned char)**s))
  {
    ++*s;
  }
}

anylist create_anylist_i(char const** s)
{
  anylist result;
  skipwhite(s);
  if (**s != '[')
    throw "Not a list";
  ++*s;
  while (true)
  {
    skipwhite(s);
    if (!**s)
      throw "Error";
    else if (**s == ']')
    {
      ++*s;
      return result;
    }
    else if (**s == '[')
      result.push_back(create_anylist_i(s));
    else if (std::isdigit((unsigned char)**s))
    {
      int i = 0;
      while (std::isdigit((unsigned char)**s))
      {
        i = 10*i + (**s - '0');
        ++*s;
      }
      result.push_back(i);
    }
    else
      throw "Error";

    skipwhite(s);
    if (**s != ',' && **s != ']')
      throw "Error";
    if (**s == ',')
      ++*s;
  }
}

anylist create_anylist(char const* i)
{
  return create_anylist_i(&i);
}

// *************************
// * printing nested lists *
// *************************

void print_list(anylist const& list);

void print_item(boost::any const& a)
{
  if (a.type() == typeid(int))
    std::cout << boost::any_cast<int>(a);
  else if (a.type() == typeid(anylist))
    print_list(boost::any_cast<anylist const&>(a));
  else
    std::cout << "???";
}

void print_list(anylist const& list)
{
  std::cout << '[';
  anylist::const_iterator iter = list.begin();
  while (iter != list.end())
  {
    print_item(*iter);
    ++iter;
    if (iter != list.end())
      std::cout << ", ";
  }
  std::cout << ']';
}

// ***************************
// * The actual test program *
// ***************************

int main()
{
  anylist list =
    create_anylist("[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]");
  print_list(list);
  std::cout << "\n";
  flatten(list);
  print_list(list);
  std::cout << "\n";
}

Output:

[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
[1, 2, 3, 4, 5, 6, 7, 8]

Ceylon

shared void run() {
    "Lazily flatten nested streams"
    {Anything*} flatten({Anything*} stream)
        =>  stream.flatMap((element)
            =>  switch (element)
                case (is {Anything*}) flatten(element)
                else [element]);

    value list = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []];
    
    print(list);
    print(flatten(list).sequence());
}

Output:

[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
[1, 2, 3, 4, 5, 6, 7, 8]

Clojure

The following returns a lazy sequence of the flattened data structure.

(defn flatten [coll]
  (lazy-seq
    (when-let [s  (seq coll)]
      (if (coll? (first s))
        (concat (flatten (first s)) (flatten (rest s)))
        (cons (first s) (flatten (rest s)))))))

The built-in flatten is implemented as:

(defn flatten [x]
  (filter (complement sequential?)
          (rest (tree-seq sequential? seq x))))

CoffeeScript

flatten = (arr) ->
  arr.reduce ((xs, el) ->
    if Array.isArray el
      xs.concat flatten el
    else
      xs.concat [el]), []

# test
list = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
console.log flatten list

Ouput:

> coffee foo.coffee 
[ 1, 2, 3, 4, 5, 6, 7, 8 ]

Common Lisp

(defun flatten (structure)
  (cond ((null structure) nil)
        ((atom structure) (list structure))
        (t (mapcan #'flatten structure))))

or, from Paul Graham's OnLisp,

(defun flatten (ls)
  (labels ((mklist (x) (if (listp x) x (list x))))
    (mapcan #'(lambda (x) (if (atom x) (mklist x) (flatten x))) ls)))

Note that since, in Common Lisp, the empty list, boolean false and nil are the same thing, a tree of nil values cannot be flattened; they will disappear.

A third version that is recursive, imperative, and reasonably fast.

(defun flatten (obj)
  (let (result)
    (labels ((grep (obj)
               (cond ((null obj) nil)
                     ((atom obj) (push obj result))
                     (t (grep (rest obj))
                        (grep (first obj))))))
      (grep obj)
      result)))

The following version is tail recursive and functional.

(defun flatten (x &optional stack out)
  (cond ((consp x) (flatten (rest x) (cons (first x) stack) out))
        (x         (flatten (first stack) (rest stack) (cons x out)))
        (stack     (flatten (first stack) (rest stack) out))
        (t out)))

The next version is imperative, iterative and does not make use of a stack. It is faster than the versions given above.

(defun flatten (obj)
  (do* ((result (list obj))
        (node result))
       ((null node) (delete nil result))
    (cond ((consp (car node))
           (when (cdar node) (push (cdar node) (cdr node)))
           (setf (car node) (caar node)))
          (t (setf node (cdr node))))))

The above implementations of flatten give the same output on nested proper lists.

Output:

CL-USER> (flatten '((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ()))
(1 2 3 4 5 6 7 8)

It should be noted that there are several choices that can be made when implementing flatten in Common Lisp:

-- should it work on dotted pairs?

-- should it work with non-nil atoms, presumably returning the atom or a copy of the atom?

-- when it comes to nil, should it be considered as an empty list and removed, or should it be considered as an atom and preserved?

So there are in fact several slightly different functions that correspond to flatten in common lisp. They may

-- collect all atoms, including nil,

-- collect all atoms in the car of the cons cells,

-- collect all atoms which are not in the cdr of a cell,

-- collect all non-nil atoms.

Which version is suitable for a given problem depends of course on the nature of the problem.

Crystal

[[1], 2, [[3, 4], 5], [[[] of Int32]], [[[6]]], 7, 8, [] of Int32].flatten()

[1, 2, 3, 4, 5, 6, 7, 8]

D

Instead of a Java-like class-based version, this version minimizes heap activity using a tagged union.

import std.stdio, std.algorithm, std.conv, std.range;

struct TreeList(T) {
    union { // A tagged union
        TreeList[] arr; // it's a node
        T data; // It's a leaf.
    }
    bool isArray = true; // = Contains an arr on default.

    static TreeList opCall(A...)(A items) pure nothrow {
        TreeList result;

        foreach (i, el; items)
            static if (is(A[i] == T)) {
                TreeList item;
                item.isArray = false;
                item.data = el;
                result.arr ~= item;
            } else
                result.arr ~= el;

        return result;
    }

    string toString() const pure {
        return isArray ? arr.text : data.text;
    }
}

T[] flatten(T)(in TreeList!T t) pure nothrow {
    if (t.isArray)
        return t.arr.map!flatten.join;
    else
        return [t.data];
}

void main() {
    alias TreeList!int L;
    static assert(L.sizeof == 12);
    auto l = L(L(1), 2, L(L(3,4), 5), L(L(L())), L(L(L(6))),7,8,L());
    l.writeln;
    l.flatten.writeln;
}

Output:

[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
[1, 2, 3, 4, 5, 6, 7, 8]

With an Algebraic Data Type

A shorter and more cryptic version.

import std.stdio, std.variant, std.range, std.algorithm;

alias T = Algebraic!(int, This[]);

int[] flatten(T t) {
    return t.peek!int ? [t.get!int] : t.get!(T[])().map!flatten.join;
}

void main() {
    T([T([ T(1) ]),
       T(2),
       T([ T([ T(3), T(4) ]), T(5) ]),
       T([ T([ T( T[].init ) ]) ]),
       T([ T([ T([ T(6) ]) ]) ]),
       T(7),
       T(8),
       T( T[].init )
      ]).flatten.writeln;
}

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

Déjà Vu

(flatten):
	for i in copy:
		i
		if = :list type dup:
			(flatten)

flatten l:
	[ (flatten) l ]


!. flatten [ [ 1 ] 2 [ [ 3 4 ] 5 ] [ [ [] ] ] [ [ [ 6 ] ] ] 7 8 [] ]

Output:

[ 1 2 3 4 5 6 7 8 ]

E

def flatten(nested) {
    def flat := [].diverge()
    def recur(x) {
        switch (x) {
            match list :List { for elem in list { recur(elem) } }
            match other      { flat.push(other) }
        }
    }
    recur(nested)
    return flat.snapshot()
}

? flatten([[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []])
# value: [1, 2, 3, 4, 5, 6, 7, 8]

EchoLisp

The built-in (flatten list) is defined as follows:

(define (fflatten l)
(cond
[(null? l) null]
[(not (list? l)) (list l)]
[else (append (fflatten (first l)) (fflatten (rest l)))]))

;;
(define L' [[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []])

(fflatten L) ;; use custom function
 → (1 2 3 4 5 6 7 8)
(flatten L) ;; use built-in
 → (1 2 3 4 5 6 7 8)

;; Remarks
;; null is the same as () - the empty list - 
(flatten '(null null null))
   → null
(flatten '[ () () () ])
  → null
(flatten null)
❗ error: flatten : expected list : null

;; The 'reverse' of flatten is group
(group '( 4 5 5 5 6 6 7 8 7 7 7 9))
    → ((4) (5 5 5) (6 6) (7) (8) (7 7 7) (9))

Ela

This implementation can flattern any given list:

xs =  [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
 
flat = flat' []
       where flat' n [] = n
             flat' n (x::xs) 
               | x is List = flat' (flat' n xs) x
               | else = x :: flat' n xs

flat xs

Output:

[1,2,3,4,5,6,7,8]

An alternative solution:

flat [] = [] 
flat (x::xs) 
  | x is List = flat x ++ flat xs
  | else = x :: flat xs

Elixir

defmodule RC do
  def flatten([]), do: []
  def flatten([h|t]), do: flatten(h) ++ flatten(t)
  def flatten(h), do: [h] 
end

list = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []] 

# Our own implementation
IO.inspect RC.flatten(list)
# Library function
IO.inspect List.flatten(list)

Output:

[1, 2, 3, 4, 5, 6, 7, 8]
[1, 2, 3, 4, 5, 6, 7, 8]

Elm

import Graphics.Element exposing (show)

type Tree a
  = Leaf a
  | Node (List (Tree a))

flatten : Tree a -> List a
flatten tree =
  case tree of
    Leaf a -> [a]
    Node list -> List.concatMap flatten list

-- [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
tree : Tree Int
tree = Node
  [ Node [Leaf 1]
  , Leaf 2
  , Node [Node [Leaf 3, Leaf 4], Leaf 5]
  , Node [Node [Node []]]
  , Node [Node [Node [Leaf 6]]]
  , Leaf 7
  , Leaf 8
  , Node []
  ]

main =
  show (flatten tree)

Emacs Lisp

(defun flatten (mylist)
  (cond
   ((null mylist) nil)
   ((atom mylist) (list mylist))
   (t
    (append (flatten (car mylist)) (flatten (cdr mylist))))))

The flatten-tree function was added in Emacs 27.1 or earlier.

(flatten-tree mylist)

Erlang

There's a standard function (lists:flatten/1) that does it more efficiently, but this is the cleanest implementation you could have;

flatten([]) -> [];
flatten([H|T]) -> flatten(H) ++ flatten(T);
flatten(H) -> [H].

Euphoria

Works with: Euphoria version 4.0.0

sequence a = {{1}, 2, {{3, 4}, 5}, {{{}}}, {{{6}}}, 7, 8, {}}

function flatten( object s )
	sequence res = {}
	if sequence( s ) then
		for i = 1 to length( s ) do
			sequence c = flatten( s[ i ] )
			if length( c ) > 0 then
				res &= c 
			end if
		end for
	else
		if length( s ) > 0 then 
			res = { s }  
		end if
	end if
	return res
end function

? a
? flatten(a)

Output:

{
  {1},
  2,
  {
    {3,4},
    5
  },
  {
    {{}}
  },
  {
    {
      {6}
    }
  },
  7,
  8,
  {}
}
{1,2,3,4,5,6,7,8}

F#

As with Haskell and OCaml we have to define our list as an algebraic data type, to be strongly typed:

type 'a ll =
    | I of 'a             // leaf Item
    | L of 'a ll list     // ' <- confine the syntax colouring confusion

let rec flatten = function
    | [] -> []
    | (I x)::y -> x :: (flatten y)
    | (L x)::y -> List.append (flatten x) (flatten y)

printfn "%A" (flatten [L([I(1)]); I(2); L([L([I(3);I(4)]); I(5)]); L([L([L([])])]); L([L([L([I(6)])])]); I(7); I(8); L([])])

// -> [1; 2; 3; 4; 5; 6; 7; 8]

An alternative approach with List.collect and the same data type. Note that flatten operates on all deepLists (ll) and atoms (I) are "flatened" to lists.

let rec flatten =
    function
    | I x -> [x]
    | L x -> List.collect flatten x

printfn "%A" (flatten (L [L([I(1)]); I(2); L([L([I(3);I(4)]); I(5)]); L([L([L([])])]); L([L([L([I(6)])])]); I(7); I(8); L([])]))

// -> [1; 2; 3; 4; 5; 6; 7; 8]

Factor

   USE: sequences.deep
   ( scratchpad ) { { 1 } 2 { { 3 4 } 5 } { { { } } } { { { 6 } } } 7 8 { } } flatten .
   { 1 2 3 4 5 6 7 8 }

Fantom

class Main
{ 
  // uses recursion to flatten a list
  static List myflatten (List items)
  {
    List result := [,]
    items.each |item|
    {
      if (item is List)
        result.addAll (myflatten(item))
      else
        result.add (item)
    }
    return result
  }
  
  public static Void main ()
  {
    List sample := [[1], 2, [[3,4], 5], [[[,]]], [[[6]]], 7, 8, [,]]
    // there is a built-in flatten method for lists
    echo ("Flattened list 1: " + sample.flatten)
    // or use the function 'myflatten'
    echo ("Flattened list 2: " + myflatten (sample))
  }
}

Forth

Works with: Forth

Works with any ANS Forth. Needs the FMS-SI (single inheritance) library code located here: http://soton.mpeforth.com/flag/fms/index.html

include FMS-SI.f
include FMS-SILib.f

: flatten {: list1 list2 --  :}
  list1 size: 0 ?do i list1 at: 
                  dup is-a object-list2
                  if list2 recurse else list2 add: then  loop ;

object-list2 list 
o{ o{ 1 } 2 o{ o{ 3 4 } 5 } o{ o{ o{ } } } o{ o{ o{ 6 } } } 7 8 o{ } } 
list flatten
list p: \ o{ 1 2 3 4 5 6 7 8 } ok

Fortran

! input   : [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
! flatten : [1, 2, 3, 4, 5, 6, 7, 8 ]

module flat
  implicit none

  type n
     integer                             :: a
     type(n), dimension(:), pointer      :: p => null()
     logical                             :: empty = .false.
  end type

contains

  recursive subroutine del(this)
  type(n), intent(inout) :: this
  integer                :: i
  if (associated(this%p)) then
    do i = 1, size(this%p)
       call del(this%p(i))
    end do
  end if
  end subroutine

  function join(xs) result (r)
  type(n), dimension(:), target :: xs
  type(n)                       :: r
  integer                       :: i
  if (size(xs)>0) then
    allocate(r%p(size(xs)), source=xs)
    do i = 1, size(xs)
      r%p(i) = xs(i)
    end do
  else
    r%empty = .true.
  end if
  end function

  recursive subroutine flatten1(x,r) 
  integer, dimension (:), allocatable, intent(inout) :: r
  type(n), intent(in)                                :: x
  integer, dimension (:), allocatable                :: tmp
  integer                                            :: i
  if (associated(x%p)) then
    do i = 1, size(x%p)
      call flatten1(x%p(i), r)
    end do
  elseif (.not. x%empty) then
    allocate(tmp(size(r)+1))
    tmp(1:size(r)) = r
    tmp(size(r)+1) = x%a
    call move_alloc(tmp, r)
  end if
  end subroutine

  function flatten(x) result (r)
  type(n), intent(in)                                :: x
  integer, dimension(:), allocatable                 :: r
  allocate(r(0))
  call flatten1(x,r)
  end function

  recursive subroutine show(x)
  type(n)   :: x
  integer   :: i
  if (x%empty) then 
    write (*, "(a)", advance="no") "[]"
  elseif (associated(x%p)) then
    write (*, "(a)", advance="no") "["
    do i = 1, size(x%p)
      call show(x%p(i))
      if (i<size(x%p)) then
        write (*, "(a)", advance="no") ", "
      end if
    end do
    write (*, "(a)", advance="no") "]"
  else
    write (*, "(g0)", advance="no") x%a
  end if
  end subroutine

  function fromString(line) result (r)
  character(len=*)                      :: line
  type (n)                              :: r
  type (n), dimension(:), allocatable   :: buffer, buffer1
  integer, dimension(:), allocatable    :: stack, stack1
  integer                               :: sp,i0,i,j, a, cur, start
  character                             :: c
 
  if (.not. allocated(buffer)) then
    allocate (buffer(5)) ! will be re-allocated if more is needed
  end if
  if (.not. allocated(stack)) then
    allocate (stack(5))
  end if

  sp = 1; cur = 1; i = 1
  do
    if ( i > len_trim(line) ) exit
    c = line(i:i)
    if (c=="[") then
      if (sp>size(stack)) then 
        allocate(stack1(2*size(stack)))
        stack1(1:size(stack)) = stack
        call move_alloc(stack1, stack)
      end if
      stack(sp) = cur;  sp = sp + 1; i = i+1
    elseif (c=="]") then
      sp = sp - 1; start = stack(sp)
      r = join(buffer(start:cur-1))
      do j = start, cur-1
        call del(buffer(j))
      end do
      buffer(start) = r; cur = start+1; i = i+1
    elseif (index(" ,",c)>0) then
      i = i + 1; continue
    elseif (index("-123456789",c)>0) then
      i0 = i
      do 
        if ((i>len_trim(line)).or. &
            index("1234567890",line(i:i))==0) then
          read(line(i0:i-1),*) a
          if (cur>size(buffer)) then
            allocate(buffer1(2*size(buffer)))
            buffer1(1:size(buffer)) = buffer
            call move_alloc(buffer1, buffer)
          end if
          buffer(cur) = n(a); cur = cur + 1; exit
        else
          i = i+1
        end if
      end do
    else
       stop "input corrupted"
    end if
  end do
  end function
end module

program main
  use flat
  type (n)  :: x
  x = fromString("[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]")
  write(*, "(a)", advance="no") "input   : "
  call show(x)
  print *
  write (*,"(a)", advance="no") "flatten : ["
  write (*, "(*(i0,:,:', '))", advance="no") flatten(x)
  print *, "]"
end program

Or, older style

Fortran does not offer strings, only CHARACTER variables of some fixed size. Functions can return such types, but, must specify a fixed size. Or, mess about with run-time allocation as above. Since in principle a list is arbitrarily long, the plan here is to crush its content in place, and thereby never have to worry about long-enough work areas. This works because the transformations in mind never replace something by something longer. A subroutine can receive an arbitrary-sized CHARACTER variable, and can change it. No attempt is made to detect improper lists.

      SUBROUTINE CRUSH(LIST)	!Changes LIST.
Crushes a list holding multi-level entries within [...] to a list of single-level entries. Null entries are purged.
Could escalate to recognising quoted strings as list entries (preserving spaces), not just strings of digits.
       CHARACTER*(*) LIST	!The text manifesting the list.
       INTEGER I,L		!Fingers.
       LOGICAL LIVE		!Scan state.
        L = 1		!Output finger. The starting [ is already in place.
        LIVE = .FALSE.	!A list element is not in progress.
        DO I = 2,LEN(LIST)	!Scan the characters of the list.
          SELECT CASE(LIST(I:I))	!Consider one.
           CASE("[","]",","," ")	!Punctuation or spacing?
            IF (LIVE) THEN		!Yes. If previously in an element,
              L = L + 1			!Advance the finger,
              LIST(L:L) = ","		!And place its terminating comma.
              LIVE = .FALSE.		!Thus the element is finished.
            END IF		!So much for punctuation and empty space.
           CASE DEFAULT		!Everything else is an element's content.
            LIVE = .TRUE.		!So we're in an element.
            L = L + 1			!Advance the finger.
            LIST(L:L) = LIST(I:I)	!And copy the content's character.
          END SELECT		!Either we're in an element, or, we're not.
        END DO			!On to the next character.
Completed the crush. At least one ] must have followed the last character of the last element.
        LIST(L:L) = "]"		!It had provoked a trailing comma. Now it is the ending ].
        LIST(L + 1:) = ""	!Scrub any tail end, just to be neat.
      END		!Trailing spaces are the caller's problem.

      CHARACTER*88 STUFF	!Work area.
      STUFF = "[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]"	!The example.
      WRITE (6,*) "Original: ",STUFF
      CALL CRUSH(STUFF)		!Can't be a constant, as it will be changed.
      WRITE (6,*) " Crushed: ",STUFF	!Behold!
      END

Output is

Original: [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
 Crushed: [1,2,3,4,5,6,7,8]

Note that if you insist on the rather flabby style of having spaces after commas, then there would be trouble. Instead of placing just a comma, a ", " would be required, which is two symbols going out when one symbol has come in: overwriting yet-to-be-scanned input is a bad idea. Either a more complex set of scan states would be required to squeeze in the extra or a separate work area would be needed to hold such output and the issue of "long enough" would arise.

All of this relies on the list being presented as a flat text, which text is then manipulated directly. If the list was manifested in a data structure of some kind with links and suchlike, then tree-traversal of that structure would be needed to reach the leaf entries.

Frink

a = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
println[flatten[a]]

GAP

Flat([[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]);

GNU APL

Using (monadic) enlist function ε. Sometimes called 'Super Ravel'.

      ⊢list←(2 3ρι6)(2 2ρ(7 8(2 2ρ9 10 11 12)13)) 'ABCD'
┏→━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃┏→━━━━┓ ┏→━━━━━━━━━┓ "ABCD"┃
┃↓1 2 3┃ ↓      7  8┃       ┃
┃┃4 5 6┃ ┃          ┃       ┃
┃┗━━━━━┛ ┃┏→━━━━┓ 13┃       ┃
┃        ┃↓ 9 10┃   ┃       ┃
┃        ┃┃11 12┃   ┃       ┃
┃        ┃┗━━━━━┛   ┃       ┃
┃        ┗∊━━━━━━━━━┛       ┃
┗∊∊━━━━━━━━━━━━━━━━━━━━━━━━━┛
      ∊list
┏→━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃1 2 3 4 5 6 7 8 9 10 11 12 13 'A''B''C''D'┃
┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

Go

package main

import "fmt"

func list(s ...interface{}) []interface{} {
    return s
}

func main() {
    s := list(list(1),
        2,
        list(list(3, 4), 5),
        list(list(list())),
        list(list(list(6))),
        7,
        8,
        list(),
    )
    fmt.Println(s)
    fmt.Println(flatten(s))
}

func flatten(s []interface{}) (r []int) {
    for _, e := range s {
        switch i := e.(type) {
        case int:
            r = append(r, i)
        case []interface{}:
            r = append(r, flatten(i)...)
        }
    }
    return
}

Output:

[[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []]
[1 2 3 4 5 6 7 8]

In the code above, flatten uses an easy-to-read type switch to extract ints and return an int slice. The version below is generalized to return a flattened slice of interface{} type, which can of course refer to objects of any type, and not just int. Also, just to show a variation in programming style, a type assertion is used rather than a type switch.

func flatten(s []interface{}) (r []interface{}) {
    for _, e := range s {
        if i, ok := e.([]interface{}); ok {
            r = append(r, flatten(i)...)
        } else {
            r = append(r, e)
        }
    }
    return
}

Groovy

List.flatten() is a Groovy built-in that returns a flattened copy of the source list:

assert [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []].flatten() == [1, 2, 3, 4, 5, 6, 7, 8]

Haskell

In Haskell we have to interpret this structure as an algebraic data type.

import Data.Tree (Tree(..), flatten)

-- [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
-- implemented as multiway tree:
-- Data.Tree represents trees where nodes have values too, unlike the trees in our problem.
-- so we use a list as that value, where a node will have an empty list value,
-- and a leaf will have a one-element list value and no subtrees
list :: Tree [Int]
list =
  Node
    []
    [ Node [] [Node [1] []]
    , Node [2] []
    , Node [] [Node [] [Node [3] [], Node [4] []], Node [5] []]
    , Node [] [Node [] [Node [] []]]
    , Node [] [Node [] [Node [6] []]]
    , Node [7] []
    , Node [8] []
    , Node [] []
    ]

flattenList :: Tree [a] -> [a]
flattenList = concat . flatten

main :: IO ()
main = print $ flattenList list

Output:

[1,2,3,4,5,6,7,8]

Alternately:

data Tree a
  = Leaf a
  | Node [Tree a]
 
flatten :: Tree a -> [a]
flatten (Leaf x) = [x]
flatten (Node xs) = xs >>= flatten
 
main :: IO ()
main =
  (print . flatten) $
  Node
    [ Node [Leaf 1]
    , Leaf 2
    , Node [Node [Leaf 3, Leaf 4], Leaf 5]
    , Node [Node [Node []]]
    , Node [Node [Node [Leaf 6]]]
    , Leaf 7
    , Leaf 8
    , Node []
    ]
    
-- [1,2,3,4,5,6,7,8]

Yet another choice, custom data structure, efficient lazy flattening:

(This is unnecessary; since Haskell is lazy, the previous solution will only do just as much work as necessary for each element that is requested from the resulting list.)

data NestedList a
  = NList [NestedList a]
  | Entry a

flatten :: NestedList a -> [a]
flatten nl = flatten_ nl []
  where
    flatten_ :: NestedList a -> [a] -> [a]
    flatten_ (Entry a) cont = a : cont
    flatten_ (NList entries) cont = foldr flatten_ cont entries

-- By passing through a list to which the results will be prepended,
-- we allow for efficient lazy evaluation
example :: NestedList Int
example =
  NList
    [ NList [Entry 1]
    , Entry 2
    , NList [NList [Entry 3, Entry 4], Entry 5]
    , NList [NList [NList []]]
    , NList [NList [NList [Entry 6]]]
    , Entry 7
    , Entry 8
    , NList []
    ]

main :: IO ()
main = print $ flatten example
-- output [1,2,3,4,5,6,7,8]

Hy

(defn flatten [lst]
  (sum (genexpr (if (isinstance x list)
                    (flatten x)
                    [x])
                [x lst])
       []))

(print (flatten [[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []]))
; [1, 2, 3, 4, 5, 6, 7, 8]

Icon and Unicon

The following procedure solves the task using a string representation of nested lists and cares not if the list is well formed or not.

link strings           # for compress,deletec,pretrim

procedure sflatten(s)  # uninteresting string solution
return pretrim(trim(compress(deletec(s,'[ ]'),',') ,','),',')       
end

Library: Icon Programming Library

The solution uses several procedures from strings in the IPL

This procedure is more in the spirit of the task handling actual lists rather than representations. It uses a recursive approach using some of the built-in list manipulation functions and operators.

procedure flatten(L)   # in the spirt of the problem  a structure
local l,x

l := []
every x := !L do
   if type(x) == "list" then l |||:= flatten(x)
   else put(l,x)
return l
end

Finally a demo routine to drive these and a helper to show how it works.

procedure main()
write(sflatten(" [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]"))
writelist(flatten( [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]))
end

procedure writelist(L)         
writes("[")
every writes(" ",image(!L))
write(" ]")
return
end

Insitux

Insitux has a built-in flatten function.

(flatten [[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []])

Output:

[1 2 3 4 5 6 7 8]

Ioke

iik> [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []] flatten
[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []] flatten
+> [1, 2, 3, 4, 5, 6, 7, 8]

Isabelle

theory Scratch
  imports Main
begin

datatype 'a tree = Leaf 'a ("<_>")
                 | Node "'a tree list" ("⟦ _ ⟧")

text‹The datatype introduces special pretty printing:›
lemma "Leaf a = <a>" by simp
lemma "Node [] = ⟦ [] ⟧" by simp

definition "example ≡ ⟦[ ⟦[<1>]⟧, <2>, ⟦[ ⟦[<3>, <4>]⟧, <5>]⟧, ⟦[⟦[⟦[]⟧]⟧]⟧, ⟦[⟦[⟦[<6>]⟧]⟧]⟧, <7>, <8>, ⟦[]⟧ ]⟧"

lemma "example = 
   Node [
     Node [Leaf 1],
     Leaf 2,
     Node [Node [Leaf 3, Leaf 4], Leaf 5],
     Node [Node [ Node []]],
     Node [Node [Node [Leaf 6]]],
     Leaf 7,
     Leaf 8,
     Node []
   ]"
  by(simp add: example_def)

fun flatten :: "'a tree ⇒ 'a list" where
  "flatten (Leaf a) = [a]"
| "flatten (Node xs) = concat (map flatten xs)"

lemma "flatten example = [1, 2, 3, 4, 5, 6, 7, 8]"
  by(simp add: example_def)

end

J

Solution:

flatten =: [: ; <S:0

Example:

   NB. create and display nested noun li
   ]li =.  (<1) ; 2; ((<3; 4); 5) ; ((<a:)) ; ((<(<6))) ; 7; 8; <a:
+---+-+-----------+----+-----+-+-+--+
|+-+|2|+-------+-+|+--+|+---+|7|8|++|
||1|| ||+-----+|5|||++|||+-+|| | ||||
|+-+| |||+-+-+|| |||||||||6||| | |++|
|   | ||||3|4||| |||++|||+-+|| | |  |
|   | |||+-+-+|| ||+--+|+---+| | |  |
|   | ||+-----+| ||    |     | | |  |
|   | |+-------+-+|    |     | | |  |
+---+-+-----------+----+-----+-+-+--+

  flatten li
1 2 3 4 5 6 7 8

Notes: The primitive ; removes one level of nesting.

<S:0 takes an arbitrarily nested list and puts everything one level deep.

[: is glue, here.

We do not use ; by itself because it requires that all of the contents be the same type and nested items have a different type from unnested items.

We do not use ]S:0 (which puts everything zero levels deep) because it assembles its results as items of a list, which means that short items will be padded to be equal to the largest items, and that is not what we would want here (we do not want the empty item to be padded with a fill element).

Alternative Solution:
The previous solution can be generalized to flatten the nesting and shape for a list of arbitrary values that include arrays of any rank:

flatten2 =: [: ; <@,S:0

Example:

   ]li2 =.  (<1) ; 2; ((<3;4); 5 + i.3 4) ; ((<a:)) ; ((<(<17))) ; 18; 19; <a:
+---+-+---------------------+----+------+--+--+--+
|+-+|2|+-------+-----------+|+--+|+----+|18|19|++|
||1|| ||+-----+| 5  6  7  8|||++|||+--+||  |  ||||
|+-+| |||+-+-+|| 9 10 11 12|||||||||17|||  |  |++|
|   | ||||3|4|||13 14 15 16|||++|||+--+||  |  |  |
|   | |||+-+-+||           ||+--+|+----+|  |  |  |
|   | ||+-----+|           ||    |      |  |  |  |
|   | |+-------+-----------+|    |      |  |  |  |
+---+-+---------------------+----+------+--+--+--+

   flatten2 li
1 2 3 4 5 6 7 8
   flatten2 li2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Here, we have replaced <S:0 with <@,S:0 so our leaves are flattened before the final step where their boxes are razed.

Java

Works with: Java version 1.5+

The flatten method was overloaded for better separation of concerns. On the first one you can pass any List and get it flat into a LinkedList implementation. On the other one you can pass any List implementation you like for both lists.

Note that both implementations can only put the result into type List<Object>. We cannot type-safely put the result into a generic type List<T> because there is no way to enforce that the original list contains elements of "type T or lists of elements which are T or further lists..."; there is no generic type parameter that will express that restriction. Since we must accept lists of any elements as an argument, we can only safely put them in a List<Object>.

Actual Workhorse code

import java.util.LinkedList;
import java.util.List;


public final class FlattenUtil {

	public static List<Object> flatten(List<?> list) {
		List<Object> retVal = new LinkedList<Object>();
		flatten(list, retVal);
		return retVal;
	}

	public static void flatten(List<?> fromTreeList, List<Object> toFlatList) {
		for (Object item : fromTreeList) {
			if (item instanceof List<?>) {
				flatten((List<?>) item, toFlatList);
			} else {
				toFlatList.add(item);
			}
		}
	}
}

Method showing population of the test List and usage of flatten method.

import static java.util.Arrays.asList;
import java.util.List;

public final class FlattenTestMain {

	public static void main(String[] args) {
		List<Object> treeList = a(a(1), 2, a(a(3, 4), 5), a(a(a())), a(a(a(6))), 7, 8, a());
		List<Object> flatList = FlattenUtil.flatten(treeList);
		System.out.println(treeList);
		System.out.println("flatten: " + flatList);
	}
	
	private static List<Object> a(Object... a) {
		return asList(a);
	}
}

Output:

[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
flatten: [1, 2, 3, 4, 5, 6, 7, 8]

Functional version

Works with: Java version 8+

import java.util.List;
import java.util.stream.Stream;
import java.util.stream.Collectors;

public final class FlattenUtil {

	public static Stream<Object> flattenToStream(List<?> list) {
		return list.stream().flatMap(item ->
			item instanceof List<?> ?
			flattenToStream((List<?>)item) :
			Stream.of(item));
	}

	public static List<Object> flatten(List<?> list) {
		return flattenToStream(list).collect(Collectors.toList());
	}
}

JavaScript

ES5

function flatten(list) {
  return list.reduce(function (acc, val) {
    return acc.concat(val.constructor === Array ? flatten(val) : val);
  }, []);
}

Or, expressed in terms of the more generic concatMap function:

(function () {
    'use strict';

    // flatten :: Tree a -> [a]
    function flatten(t) {
        return (t instanceof Array ? concatMap(flatten, t) : t);
    }

    // concatMap :: (a -> [b]) -> [a] -> [b]
    function concatMap(f, xs) {
        return [].concat.apply([], xs.map(f));
    }

    return flatten(
        [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
    );

})();

From fusion of flatten with concatMap we can then derive:

    // flatten :: Tree a -> [a]
    function flatten(a) {
        return a instanceof Array ? [].concat.apply([], a.map(flatten)) : a;
    }

For example:

(function () {
    'use strict';

    // flatten :: Tree a -> [a]
    function flatten(a) {
        return a instanceof Array ? [].concat.apply([], a.map(flatten)) : a;
    }

    return flatten(
        [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
    );

})();

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

ES6

Built-in

// flatten :: NestedList a -> [a]
const flatten = nest =>
    nest.flat(Infinity);

Recursive

// flatten :: NestedList a -> [a]
const flatten = t => {
    const go = x =>
        Array.isArray(x) ? (
            x.flatMap(go)
        ) : x;
    return go(t);
};

Iterative

function flatten(list) {
  for (let i = 0; i < list.length; i++) {
    while (true) {
      if (Array.isArray(list[i])) {
      	list.splice(i, 1, ...list[i]);
      } else {
      	break;
      }
    }
  }
  return list;
}

Or alternatively:

// flatten :: Nested List a -> a
const flatten = t => {
    let xs = t;

    while (xs.some(Array.isArray)) {
        xs = [].concat(...xs);
    }

    return xs;
};

Result is always:

[1, 2, 3, 4, 5, 6, 7, 8]

Joy

"seqlib" libload.

[[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []] treeflatten.

(* output: [1 2 3 4 5 6 7 8] *)

jq

Recent (1.4+) versions of jq include the following flatten filter:

def flatten:
   reduce .[] as $i
     ([];
     if $i | type == "array" then . + ($i | flatten)
     else . + [$i]
     end);

Example:

[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []] | flatten
[1,2,3,4,5,6,7,8]

Jsish

From Javascript entry, with change to test for typeof equal "array".

/* Flatten list, in Jsish */
function flatten(list) {
  return list.reduce(function (acc, val) {
    return acc.concat(typeof val === "array" ? flatten(val) : val);
  }, []);
}

if (Interp.conf('unitTest')) {
;   flatten([[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]);
}

/*
=!EXPECTSTART!=
flatten([[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]) ==> [ 1, 2, 3, 4, 5, 6, 7, 8 ]
=!EXPECTEND!=
*/

Output:

prompt$ jsish --U flatten.jsi
flatten([[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]) ==> [ 1, 2, 3, 4, 5, 6, 7, 8 ]

Julia

(Note that Julia versions prior to 0.5 automatically flattened nested arrays.)

The following version of flatten makes use of the higher order function mapreduce.

isflat(x) = isempty(x) || first(x) === x

function flat_mapreduce(arr)
    mapreduce(vcat, arr, init=[]) do x
        isflat(x) ? x : flat(x)
    end
end

An iterative recursive version that uses less memory but is slower:

function flat_recursion(arr)
    res = []
    function grep(v)
        for x in v
            if x isa Array 
                grep(x)
            else
                push!(res, x)
            end
        end
    end
    grep(arr)
    res
end

Using the Iterators library from the Julia base:

function flat_iterators(arr)
    while any(a -> a isa Array, arr)
        arr = collect(Iterators.flatten(arr))
    end
    arr
end

Benchmarking these three functions using the BenchmarkTools package yields the following results:

using BenchmarkTools

arr = [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]

@show flat_mapreduce(arr)
@show flat_recursion(arr)
@show flat_iterators(arr)

@btime flat_mapreduce($arr)
@btime flat_recursion($arr)
@btime flat_iterators($arr)

Output:

flat_mapreduce(arr) = Any[1, 2, 3, 4, 5, 6, 7, 8]
flat_recursion(arr) = Any[1, 2, 3, 4, 5, 6, 7, 8]
flat_iterators(arr) = [1, 2, 3, 4, 5, 6, 7, 8]
  14.163 μs (131 allocations: 4.27 KiB)
  500.824 ns (4 allocations: 176 bytes)
  28.223 μs (133 allocations: 4.33 KiB)

K

In K, join is: , and reduce/fold (called "over") is: /. With a monadic argument (as ,/ is), over repeats application until reaching a fixed-point.

So to flatten a list of arbitrary depth, you can join-over-over, or reduce a list with a function that reduces a list with a join function:

,//((1); 2; ((3;4); 5); ((())); (((6))); 7; 8; ())

Kotlin

// version 1.0.6

@Suppress("UNCHECKED_CAST")

fun flattenList(nestList: List<Any>, flatList: MutableList<Int>) {
    for (e in nestList)
        if (e is Int)
            flatList.add(e)
        else
            // using unchecked cast here as can't check for instance of 'erased' generic type
            flattenList(e as List<Any>, flatList) 
}
            
fun main(args: Array<String>) {
    val nestList : List<Any> = listOf(
        listOf(1),
        2,
        listOf(listOf(3, 4), 5),
        listOf(listOf(listOf<Int>())),
        listOf(listOf(listOf(6))),
        7,
        8,
        listOf<Int>()
    )
    println("Nested    : " + nestList)
    val flatList = mutableListOf<Int>()
    flattenList(nestList, flatList)
    println("Flattened : " + flatList)    
}

Or, using a more functional approach:

fun flatten(list: List<*>): List<*> {
    fun flattenElement(elem: Any?): Iterable<*> {
        return if (elem is List<*>)
            if (elem.isEmpty()) elem
            else flattenElement(elem.first()) + flattenElement(elem.drop(1))
        else listOf(elem)
    }
    return list.flatMap { elem -> flattenElement(elem) }
}

Output:

Nested    : [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
Flattened : [1, 2, 3, 4, 5, 6, 7, 8]

Lambdatalk

Lambdatalk doesn't have a builtin primitive flattening a multidimensionnal array.

1) Let's create this function

{def A.flatten
 {def A.flatten.r
  {lambda {:a}
   {if {A.empty? :a}
    then 
    else {let { {:b {A.first :a}}
              } {if {A.array? :b}
                 then {A.flatten.r :b}
                 else :b} }
         {A.flatten.r {A.rest :a}} }}}
 {lambda {:a}
  {A.new {A.flatten.r :a}}}}
-> A.flatten

and test it

{def list
 {A.new
  {A.new 1}
  2
  {A.new {A.new 3 4} 5}
  {A.new {A.new {A.new }}}
  {A.new {A.new {A.new 6}}}
  7
  8
  {A.new}
}}
->  [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]

{A.flatten {list}}
-> [1,2,3,4,5,6,7,8]

Lasso

Lasso Delve is a Lasso utility method explicitly for handling embedded arrays. With one array which contain other arrays, delve allows you to treat one array as a single series of elements, thus enabling easy access to an entire tree of values. www.lassosoft.com/lassoDocs/languageReference/obj/delve Lasso reference on Delve

local(original = json_deserialize('[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]'))

#original
'<br />'
(with item in delve(#original)
select #item) -> asstaticarray

array(array(1), 2, array(array(3, 4), 5), array(array(array())), array(array(array(6))), 7, 8, array())
staticarray(1, 2, 3, 4, 5, 6, 7, 8)

LFE

> (: lists flatten '((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ()))
(1 2 3 4 5 6 7 8)

Logo

to flatten :l
  if not list? :l [output :l]
  if empty? :l [output []]
  output sentence flatten first :l flatten butfirst :l
end

; using a template iterator (map combining results into a sentence)
to flatten :l
  output map.se [ifelse or not list? ? empty? ? [?] [flatten ?]] :l
end

make "a [[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []]
show flatten :a

Logtalk

flatten(List, Flatted) :-
    flatten(List, [], Flatted).

flatten(Var, Tail, [Var| Tail]) :-
    var(Var),
    !.
flatten([], Flatted, Flatted) :-
    !.
flatten([Head| Tail], List, Flatted) :-
    !,
    flatten(Tail, List, Aux),
    flatten(Head, Aux, Flatted).
flatten(Head, Tail, [Head| Tail]).

Lua

function flatten(list)
  if type(list) ~= "table" then return {list} end
  local flat_list = {}
  for _, elem in ipairs(list) do
    for _, val in ipairs(flatten(elem)) do
      flat_list[#flat_list + 1] = val
    end
  end
  return flat_list
end

test_list = {{1}, 2, {{3,4}, 5}, {{{}}}, {{{6}}}, 7, 8, {}}

print(table.concat(flatten(test_list), ","))

Maple

This can be accomplished using the Flatten command from the ListTools, or with a custom recursive procedure.

L := [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]:

with(ListTools):

Flatten(L);

Output:

                          [1, 2, 3, 4, 5, 6, 7, 8]

flatten := proc(x)
  `if`(type(x,'list'),seq(procname(i),i = x),x);
end proc:

L := [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]:

[flatten(L)];

Output:

                          [1, 2, 3, 4, 5, 6, 7, 8]

Mathematica / Wolfram Language

Flatten[{{1}, 2, {{3, 4}, 5}, {{{}}}, {{{6}}}, 7, 8, {}}]

Maxima

flatten([[[1, 2, 3], 4, [5, [6, 7]], 8], [[9, 10], 11], 12]);
/* [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12] */

Mercury

As with Haskell we need to use an algebraic data type.

:- module flatten_a_list.
:- interface.

:- import_module io.

:- pred main(io::di, io::uo) is det.

:- implementation.

:- import_module list.

:- type tree(T)
    --->    leaf(T)
    ;       node(list(tree(T))).

:- func flatten(tree(T)) = list(T).

flatten(leaf(X)) = [X].
flatten(node(Xs)) = condense(map(flatten, Xs)).

main(!IO) :-
    List = node([
        node([leaf(1)]),
        leaf(2),
        node([node([leaf(3), leaf(4)]), leaf(5)]),
        node([node([node([])])]),
        node([node([node([leaf(6)])])]),
        leaf(7),
        leaf(8),
        node([])
    ]),
    io.print_line(flatten(List), !IO).

:- end_module flatten_a_list.

Output:

    [1, 2, 3, 4, 5, 6, 7, 8]

min

Works with: min version 0.37.0

(
  (dup 'quotation? any?) 'flatten while
) ^deep-flatten

((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ()) deep-flatten puts!

Output:

(1 2 3 4 5 6 7 8)

Mirah

import java.util.ArrayList
import java.util.List
import java.util.Collection

def flatten(list: Collection) 
    flatten(list, ArrayList.new)
end
def flatten(source: Collection, result: List)

    source.each do |x|
        if x.kind_of?(Collection) 
            flatten(Collection(x), result)  
        else
            result.add(x)
            result  # if branches must return same type
        end 
    end
    result
end

# creating a list-of-list-of-list fails currently, so constructor calls are needed
source = [[1], 2, [[3, 4], 5], [[ArrayList.new]], [[[6]]], 7, 8, ArrayList.new]

puts flatten(source)

NewLISP

> (flat '((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ()))
(1 2 3 4 5 6 7 8)

NGS

Note that when kern method is called, the multi-dispatch tries to match kern parameters with given arguments last added F first: if x is an array, the second F kern is invoked, otherwise the first F kern is invoked.

NGS defines flatten as a shallow flatten, hence using flatten_r here.

F flatten_r(a:Arr)
	collector {
		local kern
		F kern(x) collect(x)
		F kern(x:Arr) x.each(kern)
		kern(a)
	}

echo(flatten_r([[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]))

Output:

[1,2,3,4,5,6,7,8]

Nim

Nim is statically-typed, so we need to use an object variant

type
  TreeList[T] = object
    case isLeaf: bool
    of true: data: T
    of false: list: seq[TreeList[T]]

proc L[T](list: varargs[TreeList[T]]): TreeList[T] =
  for x in list:
    result.list.add x

proc N[T](data: T): TreeList[T] =
  TreeList[T](isLeaf: true, data: data)

proc flatten[T](n: TreeList[T]): seq[T] =
  if n.isLeaf: result = @[n.data]
  else:
    for x in n.list:
      result.add flatten x

var x = L(L(N 1), N 2, L(L(N 3, N 4), N 5), L(L(L[int]())), L(L(L(N 6))), N 7, N 8, L[int]())
echo flatten(x)

Output:

@[1, 2, 3, 4, 5, 6, 7, 8]

Objective-C

Works with: Cocoa

#import <Foundation/Foundation.h>

@interface NSArray (FlattenExt)
@property (nonatomic, readonly) NSArray *flattened;
@end

@implementation NSArray (FlattenExt)
-(NSArray *) flattened {
    NSMutableArray *flattened = [[NSMutableArray alloc] initWithCapacity:self.count];
    
    for (id object in self) {
        if ([object isKindOfClass:[NSArray class]])
            [flattened addObjectsFromArray:((NSArray *)object).flattened];
        else
            [flattened addObject:object];
    }
    
    return [flattened autorelease];
}
@end

int main() {
    @autoreleasepool {
        NSArray *p = @[
		         @[ @1 ],
		         @2,
		         @[ @[@3, @4], @5],
		         @[ @[ @[ ] ] ],
		         @[ @[ @[ @6 ] ] ],
		         @7,
		         @8,
		         @[ ] ];
    
        for (id object in unflattened.flattened)
            NSLog(@"%@", object);
    
    }
    
    return 0;
}

OCaml

# let flatten = List.concat ;;
val flatten : 'a list list -> 'a list = <fun>

# let li = [[1]; 2; [[3;4]; 5]; [[[]]]; [[[6]]]; 7; 8; []] ;;
                ^^^
Error: This expression has type int but is here used with type int list

# (* use another data which can be accepted by the type system *)
  flatten [[1]; [2; 3; 4]; []; [5; 6]; [7]; [8]] ;;
- : int list = [1; 2; 3; 4; 5; 6; 7; 8]

Since OCaml is statically typed, it is not possible to have a value that could be both a list and a non-list. Instead, we can use an algebraic datatype:

# type 'a tree = Leaf of 'a | Node of 'a tree list ;;
type 'a tree = Leaf of 'a | Node of 'a tree list

# let rec flatten = function
     Leaf x -> [x]
   | Node xs -> List.concat (List.map flatten xs) ;;
val flatten : 'a tree -> 'a list = <fun>

# flatten (Node [Node [Leaf 1]; Leaf 2; Node [Node [Leaf 3; Leaf 4]; Leaf 5]; Node [Node [Node []]]; Node [Node [Node [Leaf 6]]]; Leaf 7; Leaf 8; Node []]) ;;
- : int list = [1; 2; 3; 4; 5; 6; 7; 8]

Oforth

[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []] expand println

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

Ol

(define (flatten x)
   (cond
      ((null? x)
         '())
      ((not (pair? x))
         (list x))
      (else
         (append (flatten (car x))
                 (flatten (cdr x))))))
 
(print
   (flatten '((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ())))

Output:

(1 2 3 4 5 6 7 8)

ooRexx

sub1 = .array~of(1)
sub2 = .array~of(3, 4)
sub3 = .array~of(sub2, 5)
sub4 = .array~of(.array~of(.array~new))
sub5 = .array~of(.array~of(.array~of(6)))
sub6 = .array~new

-- final list construction
list = .array~of(sub1, 2, sub3, sub4, sub5, 7, 8, sub6)

-- flatten
flatlist = flattenList(list)

say "["flatlist~toString("line", ", ")"]"

::routine flattenList
  use arg list
  -- we could use a list or queue, but let's just use an array
  accumulator = .array~new

  -- now go to the recursive processing version
  call flattenSublist list, accumulator

  return accumulator

::routine flattenSublist
  use arg list, accumulator

  -- ask for the items explicitly, since this will allow
  -- us to flatten indexed collections as well
  do item over list~allItems
      -- if the object is some sort of collection, flatten this out rather
      -- than add to the accumulator
      if item~isA(.collection) then call flattenSublist item, accumulator
      else accumulator~append(item)
  end

Oz

Oz has a standard library function "Flatten":

{Show {Flatten [[1] 2 [[3 4] 5] [[nil]] [[[6]]] 7 8 nil]}}

A simple, non-optimized implementation could look like this:

fun {Flatten2 Xs}
   case Xs of nil then nil
   [] X|Xr then
      {Append {Flatten2 X} {Flatten2 Xr}}
   else [Xs]
   end
end

PARI/GP

flatten(v)={
  my(u=[]);
  for(i=1,#v,
    u=concat(u,if(type(v[i])=="t_VEC",flatten(v[i]),v[i]))
  );
  u
};

Perl

sub flatten {
    map { ref eq 'ARRAY' ? flatten(@$_) : $_ } @_
}

my @lst = ([1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []);
print flatten(@lst), "\n";

Phix

standard builtin

?flatten({{1},2,{{3,4},5},{{{}}},{{{6}}},7,8,{}})

Output:

{1,2,3,4,5,6,7,8}

Phixmonti

1 2 3 10 20 30 3 tolist 4 5 6 3 tolist 2 tolist 1000 "Hello" 6 tolist
dup print nl flatten print

With syntactic sugar

include ..\Utilitys.pmt

( 1 2 3 ( ( 10 20 30 ) ( 4 5 6 ) ) 1000 "Hola" )
dup ? flatten ?

Output:

[1, 2, 3, [[10, 20, 30], [4, 5, 6]], 1000, "Hello"]
[1, 2, 3, 10, 20, 30, 4, 5, 6, 1000, "Hello"]

PHP

Works with: PHP version 4.x only, not 5.x

/* Note: This code is only for PHP 4.
   It won't work on PHP 5 due to the change in behavior of array_merge(). */
while (array_filter($lst, 'is_array'))
    $lst = call_user_func_array('array_merge', $lst);

Explanation: while $lst has any elements which are themselves arrays (i.e. $lst is not flat), we merge the elements all together (in PHP 4, array_merge() treated non-array arguments as if they were 1-element arrays; PHP 5 array_merge() no longer allows non-array arguments.), thus flattening the top level of any embedded arrays. Repeat this process until the array is flat.

Recursive

<?php
function flatten($ary) {
    $result = array();
    foreach ($ary as $x) {
        if (is_array($x))
            // append flatten($x) onto $result
            array_splice($result, count($result), 0, flatten($x));
        else
            $result[] = $x;
    }
    return $result;
}

$lst = array(array(1), 2, array(array(3, 4), 5), array(array(array())), array(array(array(6))), 7, 8, array());
var_dump(flatten($lst));
?>

Alternatively:

Works with: PHP version 5.3+

<?php
function flatten($ary) {
    $result = array();
    array_walk_recursive($ary, function($x, $k) use (&$result) { $result[] = $x; });
    return $result;
}

$lst = array(array(1), 2, array(array(3, 4), 5), array(array(array())), array(array(array(6))), 7, 8, array());
var_dump(flatten($lst));
?>

<?php
function flatten_helper($x, $k, $obj) {
    $obj->flattened[] = $x;
}

function flatten($ary) {
    $obj = (object)array('flattened' => array());
    array_walk_recursive($ary, 'flatten_helper', $obj);
    return $obj->flattened;
}

$lst = array(array(1), 2, array(array(3, 4), 5), array(array(array())), array(array(array(6))), 7, 8, array());
var_dump(flatten($lst));
?>

Using the standard library (warning: objects will also be flattened by this method):

<?php
$lst = array(array(1), 2, array(array(3, 4), 5), array(array(array())), array(array(array(6))), 7, 8, array());
$result = iterator_to_array(new RecursiveIteratorIterator(new RecursiveArrayIterator($lst)), false);
var_dump($result);
?>

Non-recursive

Function flat is iterative and flattens the array in-place.

<?php
function flat(&$ary) { // argument must be by reference or array will just be copied
    for ($i = 0; $i < count($ary); $i++) {
        while (is_array($ary[$i])) {
            array_splice($ary, $i, 1, $ary[$i]);
        }
    }
}

$lst = array(array(1), 2, array(array(3, 4), 5), array(array(array())), array(array(array(6))), 7, 8, array());
flat($lst);
var_dump($lst);
?>

PicoLisp

(de flatten (X)
   (make                               # Build a list
      (recur (X)                       # recursively over 'X'
         (if (atom X)
            (link X)                   # Put atoms into the result
            (mapc recurse X) ) ) ) )   # or recurse on sub-lists

or a more succint way using fish:

(de flatten (X)
   (fish atom X) )

Pike

There's a built-in function called Array.flatten() which does this, but here's a custom function:

array flatten(array a) {
	array r = ({ });
	
	foreach (a, mixed n) {
		if (arrayp(n)) r += flatten(n);
		else r += ({ n });
	}
	
	return r;
}

PL/I

The Translate(text,that,this) intrinsic function returns text with any character in text that is found in this (say the third) replaced by the corresponding third character in that. Suppose the availability of a function Replace(text,that,this) which returns text with all occurrences of this (a single text, possibly many characters) replaced by that, possibly zero characters. The Translate function does not change the length of its string, simply translate its characters in place.

list = translate (list, '  ', '[]' ); /*Produces "  1 , 2,   3,4 , 5 ,       ,    6   , 7, 8,     " */
list = Replace(list,'',' ');          /*Converts spaces to nothing. Same parameter order as Translate.*/
do while index(list,',,') > 0;        /*Is there a double comma anywhere?
  list = Replace(list,',',',,');      /*Yes. Convert double commas to single, nullifying empty lists*/
end;                                  /*And search afresh, in case of multiple commas in a row.*/
list = '[' || list || ']';            /*Repackage the list.*/

This is distinctly crude. A user-written Replace function is confronted by the requirement to specify a maximum size for its returned result, for instance Replace:Procedure(text,that,this) Returns(Character 200 Varying); which is troublesome for general use. The intrinsic function Translate has no such restriction.

An alternative would be to translate the commas into spaces also (thereby the null entry vanishes) then scan along the result.

PostScript

Library: initlib

/flatten {
    /.f {{type /arraytype eq} {{.f} map aload pop} ift}.
    [exch .f]
}.

[[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []] flatten

PowerShell

function flatten($a) {
    if($a.Count -gt 1) {
        $a | foreach{ $(flatten $_)}
    } else {$a}
}
$a = @(@(1), 2, @(@(3,4), 5), @(@(@())), @(@(@(6))), 7, 8, @())
"$(flatten $a)"

Output:

 
1 2 3 4 5 6 7 8

Prolog

flatten(List, FlatList) :-
	flatten(List, [], FlatList).

flatten(Var, T, [Var|T]) :-
	var(Var), !.
flatten([], T, T) :- !.
flatten([H|T], TailList, List) :- !,
	flatten(H, FlatTail, List),
	flatten(T, TailList, FlatTail).

flatten(NonList, T, [NonList|T]).

Python

Recursive

>>> def flatten(lst):
	return sum( ([x] if not isinstance(x, list) else flatten(x)
		     for x in lst), [] )

>>> lst = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
>>> flatten(lst)
[1, 2, 3, 4, 5, 6, 7, 8]

Recursive, generative and working with any type of iterable object

>>> def flatten(itr):
>>>    for x in itr:
>>>        try:
>>>            yield from flatten(x)
>>>        except:
>>>            yield x

>>> lst = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]

>>> list(flatten(lst))
[1, 2, 3, 4, 5, 6, 7, 8]

>>> tuple(flatten(lst))
(1, 2, 3, 4, 5, 6, 7, 8)

>>>for i in flatten(lst):
>>>    print(i)
1
2
3
4
5
6
7
8

Non-recursive

Function flat is iterative and flattens the list in-place. It follows the Python idiom of returning None when acting in-place:

>>> def flat(lst):
    i=0
    while i<len(lst):
        while True:
            try:
                lst[i:i+1] = lst[i]
            except (TypeError, IndexError):
                break
        i += 1
        
>>> lst = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
>>> flat(lst)
>>> lst
[1, 2, 3, 4, 5, 6, 7, 8]

Generative

This method shows a solution using Python generators.

flatten is a generator that yields the non-list values of its input in order. In this case, the generator is converted back to a list before printing.

>>> def flatten(lst):
     for x in lst:
         if isinstance(x, list):
             for x in flatten(x):
                 yield x
         else:
             yield x
 
 
>>> lst = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
>>> print list(flatten(lst)) 
[1, 2, 3, 4, 5, 6, 7, 8]

Functional Recursive

And, as the idea of Rosetta Code is to demonstrate how languages are similar as well as different, and to thus to 'aid a person with a grounding in one approach to a problem in learning another', here it is in terms of concatMap, which can be defined in any language, including mathematics, and which can be variously expressed in Python. (The fastest Python implementation of the concat component of the (concat . map) composition seems to be in terms of itertools.chain).

Works with: Python version 3.7

'''Flatten a nested list'''

from itertools import (chain)


# ----------------------- FLATTEN ------------------------

# flatten :: NestedList a -> [a]
def flatten(x):
    '''A list of atomic values resulting from fully
       flattening an arbitrarily nested list.
    '''
    return concatMap(flatten)(x) if (
        isinstance(x, list)
    ) else [x]


# ------------------------- TEST -------------------------
def main():
    '''Test: flatten an arbitrarily nested list.
    '''
    print(
        fTable(__doc__ + ':')(showList)(showList)(
            flatten
        )([
            [[[]]],
            [[1, 2, 3]],
            [[1], [[2]], [[[3, 4]]]],
            [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
        ])
    )


# ----------------------- GENERIC ------------------------

# compose (<<<) :: (b -> c) -> (a -> b) -> a -> c
def compose(g):
    '''Right to left function composition.'''
    return lambda f: lambda x: g(f(x))


# concatMap :: (a -> [b]) -> [a] -> [b]
def concatMap(f):
    '''A concatenated list over which a function has been mapped.
       The list monad can be derived by using a function f which
       wraps its output in a list,
       (using an empty list to represent computational failure).
    '''
    def go(xs):
        return chain.from_iterable(map(f, xs))
    return go


# fTable :: String -> (a -> String) ->
#                     (b -> String) ->
#        (a -> b) -> [a] -> String
def fTable(s):
    '''Heading -> x display function ->
                 fx display function ->
          f -> value list -> tabular string.'''
    def go(xShow, fxShow, f, xs):
        w = max(map(compose(len)(xShow), xs))
        return s + '\n' + '\n'.join([
            xShow(x).rjust(w, ' ') + (' -> ') + fxShow(f(x))
            for x in xs
        ])
    return lambda xShow: lambda fxShow: lambda f: lambda xs: go(
        xShow, fxShow, f, xs
    )


# showList :: [a] -> String
def showList(xs):
    '''Stringification of a list.'''
    return '[' + ','.join(str(x) for x in xs) + ']'


if __name__ == '__main__':
    main()

Output:

Flatten a nested list:
                                   [[[]]] -> []
                              [[1, 2, 3]] -> [1,2,3]
                   [[1],[[2]],[[[3, 4]]]] -> [1,2,3,4]
[[1],2,[[3, 4], 5],[[[]]],[[[6]]],7,8,[]] -> [1,2,3,4,5,6,7,8]

Functional Non-recursive

And, in contexts where it may be desirable to avoid not just recursion, but also:

1. mutation of the original list, and
2. dependence on error-events for evaluation control,

we can again use the universal concat . map composition (see the second recursive example above) by embedding it in a fold / reduction, and using it with a pure, but iteratively-implemented, until function.

( Note that the generic functions in the following example are curried, enabling not only more flexible composition, but also some simplifying reductions – here eliminating the need for two uses of Python's lambda keyword ):

Works with: Python version 3.7

'''Flatten a list'''

from functools import (reduce)
from itertools import (chain)


def flatten(xs):
    '''A flat list of atomic values derived
       from a nested list.
    '''
    return reduce(
        lambda a, x: a + list(until(every(notList))(
            concatMap(pureList)
        )([x])),
        xs, []
    )


# TEST ----------------------------------------------------
def main():
    '''From nested list to flattened list'''

    print(main.__doc__ + ':\n\n')
    xs = [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
    print(
        repr(xs) + ' -> ' + repr(flatten(xs))
    )


# GENERIC -------------------------------------------------

# concatMap :: (a -> [b]) -> [a] -> [b]
def concatMap(f):
    '''A concatenated list over which a function has been mapped.
       The list monad can be derived by using a function f which
       wraps its output in a list,
       (using an empty list to represent computational failure).
    '''
    return lambda xs: list(
        chain.from_iterable(map(f, xs))
    )


# every :: (a -> Bool) -> [a] -> Bool
def every(p):
    '''True if p(x) holds for every x in xs'''
    def go(p, xs):
        return all(map(p, xs))
    return lambda xs: go(p, xs)


# notList :: a -> Bool
def notList(x):
    '''True if the value x is not a list.'''
    return not isinstance(x, list)


# pureList :: a -> [b]
def pureList(x):
    '''x if x is a list, othewise [x]'''
    return x if isinstance(x, list) else [x]


# until :: (a -> Bool) -> (a -> a) -> a -> a
def until(p):
    '''The result of repeatedly applying f until p holds.
       The initial seed value is x.'''
    def go(f, x):
        v = x
        while not p(v):
            v = f(v)
        return v
    return lambda f: lambda x: go(f, x)


if __name__ == '__main__':
    main()

Output:

From nested list to flattened list:


[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []] -> [1, 2, 3, 4, 5, 6, 7, 8]

Q

Translation of: K

We repeatedly apply raze until the return value converges to a fixed value.

(raze/) ((1); 2; ((3;4); 5); ((())); (((6))); 7; 8; ())

Quackery

forward is flatten

[ [] swap
  witheach
    [ dup nest?
      if flatten
      join ] ]   resolves flatten ( [ --> [ )

Output:

/O> ' [ [ 1 ] 2 [ [ 3 4 ] 5 ] [ [ [ ] ] ] [ [ [ 6 ] ] ] 7 8 [ ] ] flatten
... 

Stack: [ 1 2 3 4 5 6 7 8 ]

R

x <- list(list(1), 2, list(list(3, 4), 5), list(list(list())), list(list(list(6))), 7, 8, list())

unlist(x)

Racket

Racket has a built-in flatten function:

#lang racket
(flatten '(1 (2 (3 4 5) (6 7)) 8 9))

Output:

'(1 2 3 4 5 6 7 8 9)

or, writing it explicitly with the same result:

#lang racket
(define (flatten l)
  (cond [(empty? l)      null]
        [(not (list? l)) (list l)]
        [else            (append (flatten (first l)) (flatten (rest l)))]))
(flatten '(1 (2 (3 4 5) (6 7)) 8 9))

Raku

(formerly Perl 6)

Works with: Rakudo Star version 2018.03

my @l = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []];

say .perl given gather @l.deepmap(*.take); # lazy recursive version

# Another way to do it is with a recursive function (here actually a Block calling itself with the &?BLOCK dynamic variable):

say { |(@$_ > 1 ?? map(&?BLOCK, @$_) !! $_) }(@l)

REBOL

flatten: func [
    "Flatten the block in place."
    block [any-block!]
][
    parse block [
        any [block: any-block! (change/part block first block 1) :block | skip]
    ]
    head block
]

Sample:

>> flatten [[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []]
== [1 2 3 4 5 6 7 8]

Red

flatten: function [
    "Flatten the block"
    block [any-block!]
][
    load form block
]

red>> flatten [[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []]
== [1 2 3 4 5 6 7 8]

;flatten a list to a string
>> blk: [1 2 ["test"] "a" [["bb"]] 3 4 [[[99]]]]
>> form blk
== "1 2 test a bb 3 4 99"

Refal

$ENTRY Go {
    , ((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ()): e.List
    = <Prout e.List ' -> ' <Flatten e.List>>
};

Flatten {
    = ;
    s.I e.X = s.I <Flatten e.X>;
    (e.X) e.Y = <Flatten e.X> <Flatten e.Y>;
};

Output:

((1 )2 ((3 4 )5 )((()))(((6 )))7 8 ()) -> 1 2 3 4 5 6 7 8

REXX

Translation of: PL/I

/*REXX program  (translated from PL/I)  flattens a list  (the data need not be numeric).*/
list= '[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]'  /*the list to be flattened.    */
say list                                                 /*display the original list.   */
 c= ','                                                  /*define a literal  (1 comma). */
cc= ',,'                                                 /*   "   "    "     (2 commas).*/
list= translate(list, , "[]")                            /*translate brackets to blanks.*/
list= space(list, 0)                                     /*Converts spaces to nulls.    */
                      do  while index(list, cc) > 0      /*any double commas ?          */
                      list= changestr(cc, list, c)       /*convert  ,,  to single comma.*/
                      end   /*while*/
list= strip(list, 'T', c)                                /*strip the last trailing comma*/
list = '['list"]"                                        /*repackage the list.          */
say list                                                 /*display the flattened list.  */

output:

[[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
[1,2,3,4,5,6,7,8]

Ring

aString = "[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]"
bString = ""
cString = ""
for n=1 to len(aString)
    if ascii(aString[n]) >= 48 and  ascii(aString[n]) <= 57
       bString = bString + ", " + aString[n]
    ok
next
cString = substr(bString,3,Len(bString)-2)
cString = '"' + cString + '"'
see cString + nl

"1, 2, 3, 4, 5, 6, 7, 8"

RPL

Soberly recursive.

Works with: Halcyon Calc version 4.2.7

≪ IF DUP SIZE THEN
   { } 1 LAST FOR j 
      OVER j GET 
      IF DUP TYPE 5 == THEN FLATL END
      + NEXT
   SWAP DROP END
≫ ‘FLATL’ STO

{{1} 2 {{3 4} 5} {{{}}} {{{6}}} 7 8 {}} FLATL

Output:

1: { 1 2 3 4 5 6 7 8 }

Ruby

flatten is a built-in method of Arrays

flat = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []].flatten
p flat  # => [1, 2, 3, 4, 5, 6, 7, 8]

The flatten method takes an optional argument, which dedicates the amount of levels to be flattened.

p flatten_once = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []].flatten(1)
# => [1, 2, [3, 4], 5, [[]], [[6]], 7, 8]

Rust

First we have to create a type that supports arbitrary nesting:

use std::{vec, mem, iter};

enum List<T> {
    Node(Vec<List<T>>),
    Leaf(T),
}

impl<T> IntoIterator for List<T> {
    type Item = List<T>;
    type IntoIter = ListIter<T>;
    fn into_iter(self) -> Self::IntoIter {
        match self {
            List::Node(vec) => ListIter::NodeIter(vec.into_iter()),
            leaf @ List::Leaf(_) => ListIter::LeafIter(iter::once(leaf)),
        }
    }
}

enum ListIter<T> {
    NodeIter(vec::IntoIter<List<T>>),
    LeafIter(iter::Once<List<T>>),
}

impl<T> ListIter<T> {
    fn flatten(self) -> Flatten<T> {
        Flatten {
            stack: Vec::new(),
            curr: self,
        }
    }
}

impl<T> Iterator for ListIter<T> {
    type Item = List<T>;
    fn next(&mut self) -> Option<Self::Item> {
        match *self {
            ListIter::NodeIter(ref mut v_iter) => v_iter.next(),
            ListIter::LeafIter(ref mut o_iter) => o_iter.next(),
        }
    }
}

struct Flatten<T> {
    stack: Vec<ListIter<T>>,
    curr: ListIter<T>,
}

// Flatten code is a little messy since we are shoehorning recursion into an Iterator
impl<T> Iterator for Flatten<T> {
    type Item = T;
    fn next(&mut self) -> Option<Self::Item> {
        loop {
            match self.curr.next() {
                Some(list) => {
                    match list {
                        node @ List::Node(_) => {
                            self.stack.push(node.into_iter());
                            let len = self.stack.len();
                            mem::swap(&mut self.stack[len - 1], &mut self.curr);
                        }
                        List::Leaf(item) => return Some(item),
                    }
                }
                None => {
                    if let Some(next) = self.stack.pop() {
                        self.curr = next;
                    } else {
                        return None;
                    }
                }
            }
        }
    }
}

use List::*;
fn main() {
    let list = Node(vec![Node(vec![Leaf(1)]),
                         Leaf(2),
                         Node(vec![Node(vec![Leaf(3), Leaf(4)]), Leaf(5)]),
                         Node(vec![Node(vec![Node(vec![])])]),
                         Node(vec![Node(vec![Node(vec![Leaf(6)])])]),
                         Leaf(7),
                         Leaf(8),
                         Node(vec![])]);

    for elem in list.into_iter().flatten() {
        print!("{} ", elem);
    }
    println!();

}

Output:

1 2 3 4 5 6 7 8

S-lang

define flatten ();

define flatten (list) {
    variable item,
        retval,
        val;
    if (typeof(list) != List_Type) {
        retval = list;
    } else {
        retval = {};
        foreach item (list) {
            foreach val (flatten(item)) {
                list_append(retval, val);
            }
        }
    }
    return retval;
}

Sample:

slsh> variable data = {{1}, 2, {{3,4}, 5}, {{{}}}, {{{6}}}, 7, 8, {}},
           result = flatten(data);                                    
slsh> print(result);              
{
1
2
3
4
5
6
7
8
}

Scala

def flatList(l: List[_]): List[Any] = l match {
  case Nil => Nil
  case (head: List[_]) :: tail => flatList(head) ::: flatList(tail)
  case head :: tail => head :: flatList(tail)
}

Sample:

scala> List(List(1), 2, List(List(3, 4), 5), List(List(List())), List(List(List(6))), 7, 8, List())
res10: List[Any] = List(List(1), 2, List(List(3, 4), 5), List(List(List())), List(List(List(6))), 7, 8, List())

scala> flatList(res10)
res12: List[Any] = List(1, 2, 3, 4, 5, 6, 7, 8)

Scheme

> (define (flatten x)
    (cond ((null? x) '())
          ((not (pair? x)) (list x))
          (else (append (flatten (car x))
                        (flatten (cdr x))))))

> (flatten '((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ()))
(1 2 3 4 5 6 7 8)

Shen

(define flatten
[] -> []
[X|Y] -> (append (flatten X) (flatten Y))
X -> [X])

(flatten [[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []])

Output:

[1 2 3 4 5 6 7 8]

Sidef

func flatten(a) {
    var flat = []
    a.each { |item|
        flat += (item.kind_of(Array) ? flatten(item) : [item])
    }
    return flat
}

var arr = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
say flatten(arr)      # used-defined function
say arr.flatten       # built-in Array method

Slate

s@(Sequence traits) flatten
[
  [| :out | s flattenOn: out] writingAs: s
].

s@(Sequence traits) flattenOn: w@(WriteStream traits)
[
  s do: [| :value |
    (value is: s)
      ifTrue: [value flattenOn: w]
      ifFalse: [w nextPut: value]].
].

Smalltalk

Works with: GNU Smalltalk

OrderedCollection extend [
  flatten [ |f|
    f := OrderedCollection new.
    self do: [ :i |
      i isNumber
        ifTrue: [ f add: i ]
        ifFalse: [ |t|
          t := (OrderedCollection withAll: i) flatten.
          f addAll: t
        ]
    ].
    ^ f
  ]
].


|list|
list := OrderedCollection 
          withAll: { {1} . 2 . { {3 . 4} . 5 } .
                     {{{}}} . {{{6}}} . 7 . 8 . {} }.

(list flatten) printNl.

Here is a non-OOP (but functional) version, which uses a block-closure as function (showing higher order features of Smalltalk):

flatDo := 
    [:element :action |
        element isCollection ifTrue:[
            element do:[:el | flatDo value:el value:action]
        ] ifFalse:[
            action value:element
        ].
    ].
    
collection := { 
                {1} . 2 . { {3 . 4} . 5 } .
                {{{}}} . {{{6}}} . 7 . 8 . {} 
              }.

newColl := OrderedCollection new.                         
flatDo 
    value:collection
    value:[:el | newColl add: el]

of course, many Smalltalk libraries already provide such functionality.

Works with: Smalltalk/X

Works with: Pharo

collection flatDo:[:el | newColl add:el]

Standard ML

In Standard ML, list must be homogeneous, but nested lists can be implemented as a tree-like data structure using a datatype statement:

datatype 'a nestedList =
	  L of 'a			(* leaf *)
	| N of 'a nestedList list	(* node *)

Flattening of this structure is similar to flatten trees:

fun flatten (L  x) = [x]
  | flatten (N xs) = List.concat (map flatten xs)

Output:

- flatten (N [ L 1, N [L 2, N []], L 3]);
val it = [1,2,3] : int list

Suneido

ob = [[1], 2, [[3,4], 5], [[[]]], [[[6]]], 7, 8, []]
ob.Flatten()

Output:

#(1, 2, 3, 4, 5, 6, 7, 8)

SuperCollider

SuperCollider has the method "flat", which completely flattens nested lists, and the method "flatten(n)" to flatten a certain number of levels.

a = [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []];
a.flatten(1); // answers [ 1, 2, [ 3, 4 ], 5, [ [  ] ], [ [ 6 ] ], 7, 8 ]
a.flat; // answers [ 1, 2, 3, 4, 5, 6, 7, 8 ]

Written as a function:

(
f = { |x|
	var res = res ?? List.new;
	if(x.isSequenceableCollection) {
		x.do { |each|
			res.addAll(f.(each))
		}
	} {
		res.add(x);
	};
	res
};
f.([[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]);
)

Swift

Recursive

func list(s: Any...) -> [Any] {
  return s
}

func flatten<T>(s: [Any]) -> [T] {
  var r = [T]()
  for e in s {
    switch e {
    case let a as [Any]:
      r += flatten(a)
    case let x as T:
      r.append(x)
    default:
      assert(false, "value of wrong type")
    }
  }
  return r
}

let s = list(list(1),
  2,
  list(list(3, 4), 5),
  list(list(list())),
  list(list(list(6))),
  7,
  8,
  list()
)
println(s)
let result : [Int] = flatten(s)
println(result)

Output:

[[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []]
[1 2 3 4 5 6 7 8]

More functionally:

Works with: Swift version 1.2+

func list(s: Any...) -> [Any] {
  return s
}

func flatten<T>(s: [Any]) -> [T] {
  return s.flatMap {
    switch $0 {
    case let a as [Any]:
      return flatten(a)
    case let x as T:
      return [x]
    default:
      assert(false, "value of wrong type")
    }
  }
}

let s = list(list(1),
  2,
  list(list(3, 4), 5),
  list(list(list())),
  list(list(list(6))),
  7,
  8,
  list()
)
println(s)
let result : [Int] = flatten(s)
println(result)

Output:

[[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []]
[1 2 3 4 5 6 7 8]

Non-recursive

Works with: Swift version 2.0+

func list(s: Any...) -> [Any]
{
    return s
}

func flatten<T>(array: [Any]) -> [T]
{
    var result: [T] = []
    var workstack: [(array: [Any], lastIndex: Int)] = [(array, 0)]
    
    workstackLoop: while !workstack.isEmpty
    {
        for element in workstack.last!.array.suffixFrom(workstack.last!.lastIndex)
        {
            workstack[workstack.endIndex - 1].lastIndex++
            
            if let element = element as? [Any]
            {
                workstack.append((element, 0))
                
                continue workstackLoop
            }
            
            result.append(element as! T)
        }
        
        workstack.removeLast()
    }
    
    return result
}

let input = list(list(1),
    2,
    list(list(3, 4), 5),
    list(list(list())),
    list(list(list(6))),
    7,
    8,
    list()
)

print(input)

let result: [Int] = flatten(input)

print(result)

Output:

[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
[1, 2, 3, 4, 5, 6, 7, 8]

Tailspin

templates flatten
  [ $ -> # ] !
  when <[]> do
    $... -> #
  otherwise
    $ !
end flatten

[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []] -> flatten -> !OUT::write

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

Tcl

proc flatten list {
    for {set old {}} {$old ne $list} {} {
        set old $list
        set list [join $list]
    }
    return $list
}

puts [flatten {{1} 2 {{3 4} 5} {{{}}} {{{6}}} 7 8 {}}]
# ===> 1 2 3 4 5 6 7 8

Note that because lists are not syntactically distinct from strings, it is probably a mistake to use this procedure with real (especially non-numeric) data. Also note that there are no parentheses around the outside of the list when printed; this is just a feature of how Tcl regards lists, and the value is a proper list (it can be indexed into with lindex, iterated over with foreach, etc.)

Another implementation that's slightly more terse:

proc flatten {data} {
    while { $data != [set data [join $data]] } { }
    return $data
}
puts [flatten {{1} 2 {{3 4} 5} {{{}}} {{{6}}} 7 8 {}}]
# ===> 1 2 3 4 5 6 7 8

Trith

[[1] 2 [[3 4] 5] [[[]]] [[[6]]] 7 8 []] flatten

TXR

An important builtin.

@(bind foo ((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ()))
@(bind bar foo)
@(flatten bar)

Run:

$ txr -a 5 flatten.txr  # show variable bindings in array notation to depth 5
foo[0][0]="1"
foo[1]="2"
foo[2][0][0]="3"
foo[2][0][1]="4"
foo[2][1]="5"
foo[4][0][0][0]="6"
foo[5]="7"
foo[6]="8"
bar[0]="1"
bar[1]="2"
bar[2]="3"
bar[3]="4"
bar[4]="5"
bar[5]="6"
bar[6]="7"
bar[7]="8"

VBScript

Working on embedded arrays as that's about the closest we get to lists.

Implementation

class flattener
	dim separator 
	
	sub class_initialize
		separator = ","
	end sub
	
	private function makeflat( a )
		dim i
		dim res
		for i = lbound( a ) to ubound( a ) 
			if isarray( a( i ) ) then
				res = res & makeflat( a( i ) )
			else
				res = res & a( i ) & separator
			end if
		next
		makeflat = res
	end function

	public function flatten( a )
		dim res
		res = makeflat( a )
		res = left( res, len( res ) - len(separator))
		res = split( res, separator )
		flatten = res
	end function
	
	public property let itemSeparator( c )
		separator = c
	end property
end class

Invocation

dim flat
set flat = new flattener
flat.itemSeparator = "~"
wscript.echo join( flat.flatten( array( array( 1 ),2,array(array(3,4),5),array(array(array())),array(array(array(6))),7,8,array())), "!")

Output:

1!2!3!4!5!6!7!8

Alternative (classless) Version

Works with: Windows Script Host version *

' Flatten the example array...
a = FlattenArray(Array(Array(1), 2, Array(Array(3,4), 5), Array(Array(Array())), Array(Array(Array(6))), 7, 8, Array()))

' Print the list, comma-separated...
WScript.Echo Join(a, ",")

Function FlattenArray(a)
	If IsArray(a) Then DoFlatten a, FlattenArray: FlattenArray = Split(Trim(FlattenArray))
End Function

Sub DoFlatten(a, s)
	For i = 0 To UBound(a)
		If IsArray(a(i)) Then DoFlatten a(i), s Else s = s & a(i) & " "
	Next
End Sub

V (Vlang)

Translation of: PL/I

fn main() {
	arr := "[[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]"
	println(convert(arr))
}

fn convert(arr string) []int {
	mut new_arr := []int{}
	for value in arr.replace_each(["[","","]",""]).split_any(", ") {if value !="" {new_arr << value.int()}}
	return new_arr
}

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

Wart

Here's how Wart implements flatten:

def (flatten seq acc)
  if no.seq
       acc
     ~list?.seq
       (cons seq acc)
     :else
       (flatten car.seq (flatten cdr.seq acc))

Output:

(flatten '((1) 2 ((3 4) 5) ((())) (((6))) 7 8 ()))
=> (1 2 3 4 5 6 7 8)

WDTE

let a => import 'arrays';
let s => import 'stream';

let flatten array =>
  a.stream array
  -> s.flatMap (@ f v => v {
      reflect 'Array' => a.stream v -> s.flatMap f;
    })
  -> s.collect
  ;

Usage:

flatten [[1]; 2; [[3; 4]; 5]; [[[]]]; [[[6]]]; 7; 8; []] -- io.writeln io.stdout;

Output:

[1; 2; 3; 4; 5; 6; 7; 8]

Wren

Library: Wren-seq

A method already exists for this operation in the above module.

import "./seq" for Lst

var a = [[1], 2, [[3, 4], 5], [[[]]], [[[6]]], 7, 8, []]
System.print(Lst.flatten(a))

Output:

[1, 2, 3, 4, 5, 6, 7, 8]

zkl

fcn flatten(list){ list.pump(List,
    fcn(i){ if(List.isType(i)) return(Void.Recurse,i,self.fcn); i}) }

flatten(L(L(1), L(2), L(L(3,4), 5), L(L(L())), L(L(L(6))), 7, 8, L()))
//-->L(1,2,3,4,5,6,7,8)

This works by recursively writing the contents of lists to a new list. If a list is recursive or cyclic, it will blow the stack and throw an exception.