Singly-linked list/Element definition
You are encouraged to solve this task according to the task description, using any language you may know.
[edit] ACL2
The built in pair type, cons, is sufficient for defining a linked list. ACL2 does not have mutable variables, so functions must instead return a copy of the original list.
(let ((elem 8)
(next (list 6 7 5 3 0 9)))
(cons elem next))
Output:
(8 6 7 5 3 0 9)
[edit] ActionScript
package
{
public class Node
{
public var data:Object = null;
public var link:Node = null;
public function Node(obj:Object)
{
data = obj;
}
}
}
[edit] Ada
type Link;
type Link_Access is access Link;
type Link is record
Next : Link_Access := null;
Data : Integer;
end record;
[edit] ALGOL 68
MODE DATA = STRUCT ( ... );
MODE LINK = STRUCT (
REF LINK next,
DATA value
);
[edit] AutoHotkey
element = 5 ; data
element_next = element2 ; link to next element
[edit] AWK
Awk only has global associative arrays, which will be used for the list. Numerical indexes into the array will serve as node pointers. A list element will have the next node pointer separated from the value by the pre-defined SUBSEP value. A function will be used to access a node's next node pointer or value given a node pointer (array index). The first array element will serve as the list head.
BEGIN {
NIL = 0
HEAD = 1
LINK = 1
VALUE = 2
delete list
initList()
}
function initList() {
delete list
list[HEAD] = makeNode(NIL, NIL)
}
function makeNode(link, value) {
return link SUBSEP value
}
function getNode(part, nodePtr, linkAndValue) {
split(list[nodePtr], linkAndValue, SUBSEP)
return linkAndValue[part]
}
[edit] BBC BASIC
DIM node{pNext%, iData%}
[edit] Bracmat
Data mutation is not Bracmatish, but it can be done. Here is a datastructure for a mutable data value and for a mutable reference.
link =
(next=)
(data=)
Example of use:
new$link:?link1
& new$link:?link2
& first thing:?(link1..data)
& secundus:?(link2..data)
& '$link2:(=?(link1..next))
& !(link1..next..data)
The last line returns
secundus
[edit] C
struct link {
struct link *next;
int data;
};
[edit] C++
The simplest C++ version looks basically like the C version:
struct link
{
link* next;
int data;
};
Initialization of links on the heap can be simplified by adding a constructor:
struct link
{
link* next;
int data;
link(int a_data, link* a_next = 0): next(a_next), data(a_data) {}
};
With this constructor, new nodes can be initialized directly at allocation; e.g. the following code creates a complete list with just one statement:
link* small_primes = new link(2, new link(3, new link(5, new link(7))));
However, C++ also allows to make it generic on the data type (e.g. if you need large numbers, you might want to use a larger type than int, e.g. long on 64-bit platforms, long long on compilers that support it, or even a bigint class).
template<typename T> struct link
{
link* next;
T data;
link(T a_data, link* a_next = 0): next(a_next), data(a_data) {}
};
Note that the generic version works for any type, not only integral types.
[edit] C#
class Link
{
public int item;
public Link next;
}
[edit] Common Lisp
The built-in cons type is used to construct linked lists. Using another type would be unidiomatic and inefficient.
(cons 1 (cons 2 (cons 3 nil)) => (1 2 3)
[edit] Clean
import StdMaybe
:: Link t = { next :: Maybe (Link t), data :: t }
[edit] D
Generic template-based node element.
struct SLinkedNode(T) {
T data;
typeof(this)* next;
}
void main() {
alias SLinkedNode!int N;
N* n = new N(10);
}
Also the Phobos library contains a singly-linked list, std.container.SList. Tango contains tango.util.collection.LinkSeq.
[edit] Delphi
A simple one way list. I use a generic pointer for the data that way it can point to any structure, individual variable or whatever. Note that in Standard Pascal, there are no generic pointers, therefore one has to settle for a specific data type there.
Type
pOneWayList = ^OneWayList;
OneWayList = record
pData : pointer ;
Next : pOneWayList ;
end;
[edit] E
interface LinkedList guards LinkedListStamp {}
def empty implements LinkedListStamp {
to null() { return true }
}
def makeLink(value :int, var next :LinkedList) {
def link implements LinkedListStamp {
to null() { return false }
to value() { return value }
to next() { return next }
to setNext(new) { next := new }
}
return link
}
[edit] Factor
TUPLE: linked-list data next ;
: <linked-list> ( data -- linked-list )
linked-list new swap >>data ;
[edit] Fantom
class Node
{
const Int value // keep value fixed
Node? successor // allow successor to change, also, can be 'null', for end of list
new make (Int value, Node? successor := null)
{
this.value = value
this.successor = successor
}
}
[edit] Forth
Idiomatically,
0 value numbers
: push ( n -- )
here swap numbers , , to numbers ;
NUMBERS is the head of the list, initially nil (= 0); PUSH adds an element to the list; list cells have the structure {Link,Number}. Speaking generally, Number can be anything and list cells can be as long as desired (e.g., {Link,N1,N2} or {Link,Count,"a very long string"}), but the link is always first - or rather, a link always points to the next link, so that NEXT-LIST-CELL is simply fetch (@). Some operations:
: length ( list -- u )
0 swap begin dup while 1 under+ @ repeat drop ;
: head ( list -- x )
cell+ @ ;
: .numbers ( list -- )
begin dup while dup head . @ repeat drop ;
Higher-order programming, simple continuations, and immediate words can pull out the parallel code of LENGTH and .NUMBERS . Anonymous and dynamically allocated lists are as straightforward.
[edit] Fortran
In ISO Fortran 95 or later:
type node
real :: data
type( node ), pointer :: next => null()
end type node
!
!. . . .
!
type( node ) :: head
[edit] Go
type Ele struct {
Data interface{}
Next *Ele
}
func (e *Ele) Append(data interface{}) *Ele {
if e.Next == nil {
e.Next = &Ele{data, nil}
} else {
tmp := &Ele{data, e.Next}
e.Next = tmp
}
return e.Next
}
func (e *Ele) String() string {
return fmt.Sprintf("Ele: %v", e.Data)
}
[edit] Groovy
Solution:
class ListNode {
Object payload
ListNode next
String toString() { "${payload} -> ${next}" }
}
Test:
def n1 = new ListNode(payload:25)
n1.next = new ListNode(payload:88)
println n1
Output:
25 -> 88 -> null
[edit] Haskell
This task is not idiomatic for Haskell. Usually, all data in pure functional programming is immutable, and deconstructed through Pattern Matching. The Prelude already contains a parametrically polymorphic list type that can take any data member type, including numeric values. These lists are then used very frequently. Because of this, lists have additional special syntactic sugar.
An equivalent declaration for such a list type without the special syntax would look like this:
data List a = Nil | Cons a (List a)
A declaration like the one required in the task, with an integer as element type and a mutable link, would be
data IntList s = Nil | Cons Integer (STRef s (IntList s))
but that would be really awkward to use.
[edit] Icon and Unicon
The Icon version works in both Icon and Unicon. Unicon also permits a class-based definition.
[edit] Icon
record Node (value, successor)
[edit] Unicon
class Node (value, successor)
initially (value, successor)
self.value := value
self.successor := successor
end
With either the record or the class definition, new linked lists are easily created and manipulated:
procedure main ()
n := Node(1, Node (2))
write (n.value)
write (n.successor.value)
end
[edit] J
This task is not idomatic in J -- J has lists natively and while using lists to emulate lists is quite possible, it creates additional overhead at every step of the way. (J's native lists are probably best thought of as arrays with values all adjacent to each other, though they also support constant time append.)
However, for illustrative purposes:
list=: 0 2$0
list
This creates and then displays an empty list, with zero elements. The first number in an item is (supposed to be) the index of the next element of the list (_ for the final element of the list). The second number in an item is the numeric value stored in that list item. The list is named and names are mutable in J which means links are mutable.
To create such a list with one element which contains number 42, we can do the following:
list=: ,: _ 42
list
_ 42
Now list contains one item, with index of the next item and value.
Note: this solution exploits the fact that, in this numeric case, data types for index and for node content are the same. If we need to store, for example, strings in the nodes, we should do something different, for example:
list=: 0 2$a: NB. creates list with 0 items
list
list=: ,: (<_) , <'some text' NB. creates list with 1 item
list
+-+---------+
|_|some text|
+-+---------+
[edit] Java
The simplest Java version looks basically like the C++ version:
class Link
{
Link next;
int data;
}
Initialization of links on the heap can be simplified by adding a constructor:
class Link
{
Link next;
int data;
Link(int a_data, Link a_next) { next = a_next; data = a_data; }
}
With this constructor, new nodes can be initialized directly at allocation; e.g. the following code creates a complete list with just one statement:
Link small_primes = new Link(2, new Link(3, new Link(5, new Link(7, null))));
However, Java also allows to make it generic on the data type. This will only work on reference types, not primitive types like int or float (wrapper classes like Integer and Float are available).
class Link<T>
{
Link<T> next;
T data;
Link(T a_data, Link<T> a_next) { next = a_next; data = a_data; }
}
[edit] JavaScript
function LinkedList(value, next) {
this._value = value;
this._next = next;
}
LinkedList.prototype.value = function() {
if (arguments.length == 1)
this._value = arguments[0];
else
return this._value;
}
LinkedList.prototype.next = function() {
if (arguments.length == 1)
this._next = arguments[0];
else
return this._next;
}
// convenience function to assist the creation of linked lists.
function createLinkedListFromArray(ary) {
var head = new LinkedList(ary[0], null);
var prev = head;
for (var i = 1; i < ary.length; i++) {
var node = new LinkedList(ary[i], null);
prev.next(node);
prev = node;
}
return head;
}
var head = createLinkedListFromArray([10,20,30,40]);
[edit] Logo
As with other list-based languages, simple lists are represented easily in Logo.
fput item list ; add item to the head of a list
first list ; get the data
butfirst list ; get the remainder
bf list ; contraction for "butfirst"
These return modified lists, but you can also destructively modify lists. These are normally not used because you might accidentally create cycles in the list.
.setfirst list value
.setbf list remainder
[edit] Mathematica
Append[{}, x]
-> {x}
[edit] Modula-2
TYPE
Link = POINTER TO LinkRcd;
LinkRcd = RECORD
Next: Link;
Data: INTEGER
END;
[edit] Objective-C
This implements a class which has the primitive basic Objective-C class Object as parent.
#import <objc/Object.h>
@interface RCListElement : Object
{
RCListElement *next;
id datum;
}
+ (RCListElement *)new;
- (RCListElement *)next;
- (id)datum;
- (RCListElement *)setNext: (RCListElement *)nx;
- (void)setDatum: (id)d;
@end
@implementation RCListElement
+ (RCListElement *)new
{
RCListElement *m = [super new];
[m setNext: nil];
[m setDatum: nil];
return m;
}
- (RCListElement *)next
{
return next;
}
- (id)datum
{
return datum;
}
- (RCListElement *)setNext: (RCListElement *)nx
{
RCListElement *p;
p = next;
next = nx;
return p;
}
- (void)setDatum: (id)d
{
datum = d;
}
@end
[edit] OCaml
This task is not idiomatic for OCaml. OCaml already contains a built-in parametrically polymorphic list type that can take any data member type, including numeric values. These lists are then used very frequently. Because of this, lists have additional special syntactic sugar. OCaml's built-in lists, like most functional data structures, are immutable, and are deconstructed through Pattern Matching.
An equivalent declaration for such a list type without the special syntax would look like this:
type 'a list = Nil | Cons of 'a * 'a list
A declaration like the one required in the task, with an integer as element type and a mutable link, would be
type int_list = Nil | Cons of int * int_list ref
but that would be really awkward to use.
[edit] ooRexx
The simplest ooRexx version is similar in form to the Java or C++ versions:
list = .linkedlist~new
index = list~insert("abc") -- insert a first item, keeping the index
list~insert("def") -- adds to the end
list~insert("123", .nil) -- adds to the begining
list~insert("456", index) -- inserts between "abc" and "def"
list~remove(index) -- removes "abc"
say "Manual list traversal"
index = list~first -- demonstrate traversal
loop while index \== .nil
say index~value
index = index~next
end
say
say "Do ... Over traversal"
do value over list
say value
end
-- the main list item, holding the anchor to the links.
::class linkedlist
::method init
expose anchor
-- create this as an empty list
anchor = .nil
-- return first link element
::method first
expose anchor
return anchor
-- return last link element
::method last
expose anchor
current = anchor
loop while current \= .nil
-- found the last one
if current~next == .nil then return current
current = current~next
end
-- empty
return .nil
-- insert a value into the list, using the convention
-- followed by the built-in list class. If the index item
-- is omitted, add to the end. If the index item is .nil,
-- add to the end. Otherwise, just chain to the provided link.
::method insert
expose anchor
use arg value
newLink = .link~new(value)
-- adding to the end
if arg() == 1 then do
if anchor == .nil then anchor = newLink
else self~last~insert(newLink)
end
else do
use arg ,index
if index == .nil then do
if anchor \== .nil then newLink~next = anchor
anchor = newLink
end
else index~insert(newLink)
end
-- the link item serves as an "index"
return newLink
-- remove a link from the chain
::method remove
expose anchor
use strict arg index
-- handle the edge case
if index == anchor then anchor = anchor~next
else do
-- no back link, so we need to scan
previous = self~findPrevious(index)
-- invalid index, don't return any item
if previous == .nil then return .nil
previous~next = index~next
end
-- belt-and-braces, remove the link and return the value
index~next = .nil
return index~value
-- private method to find a link predecessor
::method findPrevious private
expose anchor
use strict arg index
-- we're our own precessor if this first
if index == anchor then return self
current = anchor
loop while current \== .nil
if current~next == index then return current
current = current~next
end
-- not found
return .nil
-- helper method to allow DO ... OVER traversal
::method makearray
expose anchor
array = .array~new
current = anchor
loop while current \= .nil
array~append(current~value)
current = current~next
end
return array
::class link
::method init
expose value next
-- by default, initialize both data and next to empty.
use strict arg value = .nil, next = .nil
-- allow external access to value and next link
::attribute value
::attribute next
::method insert
expose next
use strict arg newNode
newNode~next = next
next = newNode
A link element can hold a reference to any ooRexx object.
[edit] Pascal
type
PLink = ^TLink;
TLink = record
FNext: PLink;
FData: integer;
end;
[edit] Perl
Just use an array. You can traverse and splice it any way. Linked lists are way too low level.
However, if all you got is an algorithm in a foreign language, you can use references to accomplish the translation.
my %node = (
data => 'say what',
next => \%foo_node,
);
$node{next} = \%bar_node; # mutable
[edit] Perl 6
The Pair constructor is exactly equivalent to a cons cell.
my $elem = 42 => $nextelem;
[edit] PicoLisp
In PicoLisp, the singly-linked list is the most important data structure. Many built-in functions deal with linked lists. A list consists of interconnected "cells". Cells are also called "cons pairs", because they are constructed with the function 'cons'.
Each cell consists of two parts: A CAR and a CDR. Both may contain (i.e. point to) arbitrary data (numbers, symbols, other cells, or even to itself). In the case of a linked list, the CDR points to the rest of the list.
The CAR of a cell can be manipulated with 'set' and the CDR with 'con'.
[edit] Pop11
List are built in into Pop11, so normally on would just use them:
;;; Use shorthand syntax to create list.
lvars l1 = [1 2 three 'four'];
;;; Allocate a single list node, with value field 1 and the link field
;;; pointing to empty list
lvars l2 = cons(1, []);
;;; print first element of l1
front(l1) =>
;;; print the rest of l1
back(l1) =>
;;; Use index notation to access third element
l1(3) =>
;;; modify link field of l2 to point to l1
l1 -> back(l2);
;;; Print l2
l2 =>
If however one wants to definite equivalent user-defined type, one can do this:
uses objectclass;
define :class ListNode;
slot value = [];
slot next = [];
enddefine;
;;; Allocate new node and assign to l1
newListNode() -> l1;
;;; Print it
l1 =>
;;; modify value
1 -> value(l1);
l1 =>
;;; Allocate new node with initialized values and assign to link field
;;; of l1
consListNode(2, []) -> next(l1);
l1 =>
[edit] PureBasic
Structure MyData
*next.MyData
Value.i
EndStructure
[edit] Python
The Node class implements also iteration for more Pythonic iteration over linked lists.
class LinkedList(object):
"""USELESS academic/classroom example of a linked list implemented in Python.
Don't ever consider using something this crude! Use the built-in list() type!
"""
class Node(object):
def __init__(self, item):
self.value = item
self.next = None
def __init__(self, item=None):
if item is not None:
self.head = Node(item); self.tail = self.head
else:
self.head = None; self.tail = None
def append(self, item):
if not self.head:
self.head = Node(item)
self.tail = self.head
elif self.tail:
self.tail.next = Node(item)
self.tail = self.tail.next
else:
self.tail = Node(item)
def __iter__(self):
cursor = self.head
while cursor:
yield cursor.value
cursor = cursor.next
Note: As explained in this class' docstring implementing linked lists and nodes in Python is an utterly pointless academic exercise. It may give on the flavor of the elements that would be necessary in some other programming languages (e.g. using Python as "executable psuedo-code"). Adding methods for finding, counting, removing and inserting elements is left as an academic exercise to the reader. For any practical application use the built-in list() or dict() types as appropriate.
[edit] Racket
Unlike other Lisp dialects, Racket's cons cells are immutable, so they cannot be used to satisfy this task. However, Racket also includes mutable pairs which are still the same old mutable singly-linked lists.
#lang racket
(mcons 1 (mcons 2 (mcons 3 '()))) ; a mutable list
[edit] REXX
The REXX language doesn't have any native linked lists, but they can be created easily.
The values of a REXX linked list can be anything (including any type/kind of number, of course).
/*REXX program to show how to create and show a single-linked list. */
@.=0 /*define a null linked list (so far). */
call set@ ,3 /*build linked list of 12 proth primes.*/
call set@ ,5
call set@ ,13
call set@ ,17
call set@ ,41
call set@ ,97
call set@ ,113
call set@ ,193
call set@ ,241
call set@ ,257
call set@ ,353
call set@ ,449
w=length(@._last) /*use width of the last item number. */
do j=1 for @._last /*show all entries of the linked list.*/
say "item" right(j,w) '=' right(@.j._value,@.max_width)
end /*j*/
exit /*stick a fork in it, we're done. */
/*───────────────────────────────SET@ subroutine────────────────────────*/
set@: procedure expose @.; parse arg #,y
if arg(1,'o') then do /*if 1st arg omitted, then add to list*/
_=@._last+1 /*bump the last ptr in the linked list*/
@._last=_ /*define the next item in linked list.*/
#=_ /*point to this item in linked list.*/
end
@.#._value=y /*set the item to the value specified.*/
@.max_width=max(@.max_width,length(y)) /*set maximum width of any value.*/
if #\==1 then do /*if not the first item, link it. */
prev=#-1 /*figure out what the previous item is*/
@.prev._next=# /*now, link the previous item to here.*/
end
return /*return to the invoker of this sub. *//
output
item 1 = 3 item 2 = 5 item 3 = 13 item 4 = 17 item 5 = 41 item 6 = 97 item 7 = 113 item 8 = 193 item 9 = 241 item 10 = 257 item 11 = 353 item 12 = 449
[edit] Ruby
class ListNode
attr_accessor :value, :succ
def initialize(value, succ=nil)
self.value = value
self.succ = succ
end
def each(&b)
yield self
succ.each(&b) if succ
end
include Enumerable
def self.from_array(ary)
head = self.new(ary[0], nil)
prev = head
ary[1..-1].each do |val|
node = self.new(val, nil)
prev.succ = node
prev = node
end
head
end
end
list = ListNode.from_array([1,2,3,4])
[edit] Rust
Rust's enum types make the definition of a singly-linked list trivial.
enum SingleLinkedList<T> {
Node(T, @mut SingleLinkedList<T>),
None
}
The @mut indicates that a Node holds a mutable, managed, link to the next node/list.
[edit] Run BASIC
data = 10
link = 10
dim node{data,link}
[edit] Scala
class Node(n: Int, link: Node) {
var data = n
var next = link
}
The one below is more inline with the built-in definition
class Node {
var data: Int
var next = this
def this(n: Int, link: Node) {
this()
if (next != null){
data = n
next = link
}
}
[edit] Scheme
Scheme, like other Lisp dialects, has extensive support for singly-linked lists. The element of such a list is known as a cons-pair, because you use the cons function to construct it:
(cons value next)
The value and next-link parts of the pair can be deconstructed using the car and cdr functions, respectively:
(car my-list) ; returns the first element of the list
(cdr my-list) ; returns the remainder of the list
Each of these parts are mutable and can be set using the set-car! and set-cdr! functions, respectively:
(set-car! my-list new-elem)
(set-cdr! my-list new-next)
[edit] Tcl
While it is highly unusual to implement linked lists in Tcl, since the language has a built-in list type (that internally uses arrays of references), it is possible to simulate it with objects.
oroo::class create List {
variable content next
constructor {value {list ""}} {
set content $value
set next $list
}
method value args {
set content {*}$args
}
method attach {list} {
set next $list
}
method detach {} {
set next ""
}
method next {} {
return $next
}
method print {} {
for {set n [self]} {$n ne ""} {set n [$n next]} {
lappend values [$n value]
}
return $values
}
}
[edit] X86 Assembly
struct link
.next: resd 1
.data: resd 1
endstruc
Of course, ASM not natively having structures we can simply do..
link resb 16
Which would reserve 16 bytes(2 dwords). We could just simply think of it in the form of a structure.
link struct
next dd ?
data dd ?
link ends
struc link next,data
{
.next dd next
.data dd data
}
[edit] XPL0
include c:\cxpl\codes; \intrinsic 'code' declarations
def IntSize=4; \number of bytes in an integer
def Size=10; \number of nodes in this linked list
int Link, List, Node;
[Link:= 0; \build linked list, starting at the end
for Node:= 0 to Size-1 do
[List:= Reserve(IntSize*2); \get some memory to hold link and data
List(0):= Link;
List(1):= Node*Node; \insert example data
Link:= List; \Link now points to newly created node
];
Node:= List; \traverse the linked list
repeat IntOut(0, Node(1)); CrLf(0); \display the example data
Node:= Node(0); \move to next node
until Node=0; \end of the list
]
- Programming Tasks
- Data Structures
- ACL2
- ActionScript
- Ada
- ALGOL 68
- AutoHotkey
- AWK
- BBC BASIC
- Bracmat
- C
- C++
- C sharp
- Common Lisp
- Clean
- D
- Delphi
- E
- Factor
- Fantom
- Forth
- Fortran
- Go
- Groovy
- Haskell
- Icon
- Unicon
- J
- Java
- JavaScript
- Logo
- Mathematica
- Modula-2
- Objective-C
- OCaml
- OoRexx
- Pascal
- Perl
- Perl 6
- PicoLisp
- Pop11
- PureBasic
- Python
- Racket
- REXX
- Ruby
- Rust
- Run BASIC
- Scala
- Scheme
- Tcl
- TclOO
- X86 Assembly
- XPL0
- GUISS/Omit
