Singly-linked list/Element insertion: Difference between revisions

Content added Content deleted

Inline

Revision as of 20:25, 22 September 2008

Using the link element defined in Singly-Linked List (element), define a method to insert an element into a singly-linked list following a given element.

Using this method, insert an element C into a list comprised of elements A->B, following element A.

C

Define the method:

void insert_append (link *anchor, link *newlink) {
  newbie->next = anchor->next;
  anchor->next = newlink;
}

Note that in a production implementation, one should check anchor and newlink to ensure they're valid values. (I.e., not NULL.)

And now on to the code.

Create our links.

link *a, *b, *c;
a = malloc(sizeof(link));
b = malloc(sizeof(link));
c = malloc(sizeof(link));
a->data = 1;
b->data = 2;
c->data = 3;

Prepare our initial list

insert_append (a, b);

Insert element c after element a

insert_append (a, c);

Remember to free the memory once we're done.

free (a);
free (b);
free (c);

C++

This uses the generic version of the link node. Of course, normally this would be just some implementation detail inside some list class, not to be used directly by client code.

template<typename T> void insert_after(link<T>* list_node, link<T>* new_node)
{
  new_node->next = list_node->next;
  list_node->next = new_node;
};

Here's the example code using that method:

The following code creates the links. As numeric values I've just taken the corresponding character values.

link<int>* a = new link<int>('A', new link<int>('B'));
link<int>* c = new link<int>('C');

Now insert c after a:

insert_after(a, c);

Finally destroy the list:

while (a)
{
  link<int>* tmp = a;
  a = a->next;
  delete tmp;
}

A simple insertion into a one way list. I use a generic pointer for the data that way it can point to any structure, individual variable or whatever. NOTE: For original versions of Turbo Pascal, substitute the MemAvail Function for the Try Except block as this does not exist in this version of the pascal language. Also, Turbo Pascal doesn't have C++-style comments, therefore those have to be replaced with Pascal style comments, i.e. { ... } or (* ... *).


// Using the same type defs from the one way list example.

Type

  // The pointer to the list structure
  pOneWayList = ^OneWayList;

  // The list structure
  OneWayList = record
                 pData : pointer ;
                 Next  : pOneWayList ;
               end;

// I will illustrate a simple function that will return a pointer to the 
// new node or it will return NIL.  In this example I will always insert
// right, to keep the code clear.  Since I am using a function all operations
// for the new node will be conducted on the functions result.  This seems
// somewhat counter intuitive, but it is the simplest way to accomplish this.

Function InsertNode(VAR CurrentNode:pOneWayList): pOneWayList
begin

    // I try not to introduce different parts of the language, and keep each
    // example to just the code required.  in this case it is important to use
    // a try/except block.  In any OS that is multi-threaded and has many apps
    // running at the same time, you cannot rely on a call to check memory available
    // and then attempting to allocate.  In the time between the two, another 
    // program may have grabbed the memory you were trying to get.

    Try
      // Try to allocate enough memory for a variable the size of OneWayList
      GetMem(Result,SizeOf(OneWayList));
    Except
      On EOutOfMemoryError do
         begin
           Result := NIL
           exit;
         end;
    end;

    // Initialize the variable.
    Result.Next  := NIL ;
    Reuslt.pdata := NIL ;

    // Ok now we will insert to the right.

    // Is the Next pointer of CurrentNode Nil?  If it is we are just tacking
    // on to the end of the list.

    if CurrentNode.Next = NIL then
       CurrentNode.Next := Result
    else
      // We are inserting into the middle of this list
      Begin
         Result.Next      := CurrentNode.Next ;
         CurrentNode.Next := result ;
      end;
end;

E

def insertAfter(head :LinkedList ? (!head.null()),
                new  :LinkedList ? (new.next().null())) {
    new.setNext(head.next())
    head.setNext(new)
}

def a := makeLink(1, empty)
def b := makeLink(2, empty)
def c := makeLink(3, empty)

insertAfter(a, b)
insertAfter(a, c)

var x := a
while (!x.null()) {
    println(x.value())
    x := x.next()
}

Forth

Using the linked list concept described in the Singly-Linked_List_(element) topic:

\ Create the list and some list elements
create A   0 , char A ,
create B   0 , char B ,
create C   0 , char C ,

Now insert b after a and c after b, giving a->b->c

B A chain
C B chain

Here is an abbreviated version of the definition of 'chain' from the other article:

: chain ( a b -- )   2dup  @ swap !  ! ;

And here is a function to walk a list, calling an XT on each data cell:

: walk ( a xt -- )
   >r begin ?dup while
     dup cell+ @ r@ execute
   @ repeat r> drop ;

Testing code:

A ' emit walk ABC ok

Fortran

In ISO Fortran 95 or later:

   subroutine addAfter(nodeBefore,value)
      type (node), intent(inout) :: nodeBefore
      type (node), pointer :: newNode
      
      allocate(newNode)
      newNode%data = value
      newNode%next => nodeBefore%next
      nodeBefore%next => newNode

   end subroutine addAfter

Haskell

This kind of list manipulation is unidiomatic Haskell. But you can try the following:

insertAfter a b (c:cs) | a==c = a : b : cs
                       | otherwise = c : insertAfter a b cs
insertAfter _ _ [] = error "Can't insert"

OCaml

This kind of list manipulation is unidiomatic OCaml. But you can try the following: <ocaml>let rec insert_after a b = function

  c :: cs when a = c -> a :: b :: cs
| c :: cs            -> c :: insert_after a b cs
| []                 -> raise Not_found</ocaml>

Pascal

Note: This code uses only Standard Pascal features. For code using features only available in modern Pascal versions, see above under "[Delphi / Object Pascal / >>Turbo Pascal<<]"

Since Standard Pascal doesn't know a generic pointer type, and also no generic types, one has to settle for a specific data type fore the linked list. Since the task mentions node names "A", "B", "C", here a char is chosen. Of course any data type (including pointers to a specific data type) could have been used here.

type
  pCharNode = ^CharNode;
  CharNode = record
               data: char;
               next: pCharNode;
             end;

(* This procedure inserts a node (newnode) directly after another node which is assumed to already be in a list.
  It does not allocate a new node, but takes an already allocated node, thus allowing to use it (together with
  a procedure to remove a node from a list) for splicing a node from one list to another. *)
procedure InsertAfter(listnode, newnode: pCharNode);
begin
  newnode^.next := listnode^.next;
  listnode^.next := newnode;
end;

Usage example:

var
  A, B: pCharNode;
begin
  (* build the two-component list A->C manually *)
  new(A);
  A^.data := 'A';
  new(A^.next);
  A^.next^.data := 'C';
  A^.next^.next := nil;

  (* create the node to be inserted. The initialization of B^.next isn't strictly necessary
    (it gets overwritten anyway), but it's good style not to leave any values undefined. *)
  new(B);
  node^.data := 'B';
  node^.next := nil;

  (* call the above procedure to insert node B after node A *)
  InsertAfter(A, B);

  (* delete the list *)
  while A <> nil do
   begin
    B := A;
    A := A^.next;
    dispose(B);
   end
end.

Perl

If you don't really need the constant-time insertion property of singly linked lists, just use an array. You can traverse and splice it any way.

my @l  = ($A, $B);
push @l, $C, splice @l, 1;

However, if you really need a linked list, or all you got is an algorithm in a foreign language, you can use references to accomplish the translation.

sub insert_after {
  # first argument: node to insert after
  # second argument: node to insert
  $_[1]{next} = $_[0]{next};
  $_[0]{next} = $_[1];
}

my %B = (
    data => 3,
    next => undef, # not a circular list
);
my %A = (
    data => 1,
    next => \%B,
);
my %C = (
    data => 2,
);
insert_after \%A, \%C;

Note that you don't have to name your new nodes. The following works just as well:

insert_after \%A, { data => 2 };

Note the curly braces instead of round parentheses.

It is straightforward to extend the function to take an arbitrary number of list nodes to insert:

sub insert_after {
  my $node = $_[0];
  my $next = $node->{next};
  shift;
  while (defined $_[0]) {
    $node->{next} = $_[0];
    $node = $node->{next};
    shift;
  }
  $node->{next} = $next;
}

With this, it's rather easy to build a list:

my %list = ( data => 'A' );
insert_after \%list, { data => 'B' }, { data => 'C' };

List handling is simplified if the variables themselves contain references. For example:

my $list2;

# create a new list ('A'. 'B', 'C') and store it in $list2
insert_after $list2 = { data => 'A' }, { data => 'B' }, { data => 'C' };

# append two new nodes ('D', 'E') after the first element
insert_after $list2, { data => 'A2' }, { data => 'A3' };

# append new nodes ('A2a', 'A2b') after the second element (which now is 'A2')
insert_after $list2->{next}, { data => 'A2a' }, { data => 'A2b' };

Pop11

In Pop11 one normally uses built-in lists:

define insert_into_list(anchor, x);
   cons(x, back(anchor)) -> back(anchor);
enddefine;
;;; Build inital list
lvars l1 = cons("a", []);
insert_into_list(l1, "b");
;;; insert c
insert_into_list(l1,  "c");

If one wants one can use user-defined list node (for convenience we repeat definition of list node):

uses objectclass;
define :class ListNode;
    slot value = [];
    slot next = [];
enddefine;

define insert_into_List(anchor, x);
   consListNode(x, next(anchor)) -> next(anchor);
enddefine;
;;; Build inital list
lvars l2 = consListNode("a", []);
insert_into_List(l2, "b");
;;; insert c
insert_into_List(l2,  "c");

Note that user-defined case differs from built-in case only because of names.

Python

def insert(lst, at, item):
    for node in lst:
        if node.value == at:
            node.next = Node(item, node.next)
            return
    raise ValueError(str(at))

Example usage:

>>> lst = Node('A', Node('B'))
>>> insert(lst, 'A', 'C')
>>> lst.next.value
'C'

Ruby

 class ListNode
   def insertAfter(a,b)
     if a==value
       self.succ = ListNode.new(b,succ)
     else
       succ.insertAfter(a,b)
     end
   end
 end

 list = ListNode.new(:a,ListNode.new(:b))
 list.insertAfter(:a,:c)

Scheme

Non-mutating: <scheme>(define (insert-after a b lst)

 (if (null? lst)
     lst       ; This should be an error, but we will just return the list untouched
     (let ((c (car lst))
           (cs (cdr lst)))
       (if (equal? a c)
           (cons a (cons b cs))
           (cons c (insert-after a b cs))))))</scheme>

Mutating: <scheme>(define (insert-after! a b lst)

 (let ((pos (member a lst)))
   (if pos
       (set-cdr! pos (cons b (cdr pos))))))</scheme>