Associative array/Creation

From Rosetta Code
Associative array/Creation
You are encouraged to solve this task according to the task description, using any language you may know.

The goal is to create an associative array (also known as a dictionary, map, or hash).

Related tasks:

See also



Because ActionScript does not have associative arrays in the normal sense, Object objects are used instead and keys are simply properties on those objects.

var map:Object = {key1: "value1", key2: "value2"};
trace(map['key1']); // outputs "value1"
// Dot notation can also be used
trace(map.key2); // outputs "value2"
// More keys and values can then be added
map['key3'] = "value3";
trace(map['key3']); // outputs "value3"

Note: The Object only supports String keys. To use an object as a key, try the flash.utils.Dictionary class.


Works with: GNAT version GPL 2007
with Ada.Containers.Ordered_Maps;
with Ada.Strings.Unbounded; use Ada.Strings.Unbounded;
with Ada.Text_IO;
procedure Associative_Array is
-- Instantiate the generic package Ada.Containers.Ordered_Maps
package Associative_Int is new Ada.Containers.Ordered_Maps(Unbounded_String, Integer);
use Associative_Int;
Color_Map : Map;
Color_Cursor : Cursor;
Success : Boolean;
Value : Integer;
-- Add values to the ordered map
Color_Map.Insert(To_Unbounded_String("Red"), 10, Color_Cursor, Success);
Color_Map.Insert(To_Unbounded_String("Blue"), 20, Color_Cursor, Success);
Color_Map.Insert(To_Unbounded_String("Yellow"), 5, Color_Cursor, Success);
-- retrieve values from the ordered map and print the value and key
-- to the screen
Value := Color_Map.Element(To_Unbounded_String("Red"));
Ada.Text_Io.Put_Line("Red:" & Integer'Image(Value));
Value := Color_Map.Element(To_Unbounded_String("Blue"));
Ada.Text_IO.Put_Line("Blue:" & Integer'Image(Value));
Value := Color_Map.Element(To_Unbounded_String("Yellow"));
Ada.Text_IO.Put_Line("Yellow:" & Integer'Image(Value));
end Associative_Array;


Aikido provides a native map for associative arrays. You can create them using a map literal and you can insert and remove items on the fly.

var names = {} // empty map
names["foo"] = "bar"
names[3] = 4
// initialized map
var names2 = {"foo": bar, 3:4}
// lookup map
var name = names["foo"]
if (typeof(name) == "none") {
println ("not found")
} else {
println (name)
// remove from map
delete names["foo"]


Aime records are heterogenous associative arrays. No creation procedure is required, declaration is fine.

record r;
r_put(r, "A", 33);		# an integer value
r_put(r, "C", 2.5); # a real value
r_put(r, "B", "associative"); # a string value

ALGOL 68[edit]

Translation of: C++
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
FORMAT color repr = $"16r"16r6d$;
# This is an associative array which maps strings to ints #
REF[]ITEM color map items := LOC[0]ITEM;
PROC color map find = (STRING color)REF COLOR:(
# linear search! #
FOR index FROM LWB key OF color map items TO UPB key OF color map items DO
IF color = key OF color map items[index] THEN
out := value OF color map items[index]; GO TO found
PROC color map = (STRING color)REF COLOR:(
REF COLOR out = color map find(color);
IF out :=: REF COLOR(NIL) THEN # extend color map array #
HEAP[UPB key OF color map items + 1]ITEM color map append;
color map append[:UPB key OF color map items] := color map items;
color map items := color map append;
value OF (color map items[UPB value OF color map items] := (color, 16r000000)) # black #
# First, populate it with some values #
color map("red") := 16rff0000;
color map("green") := 16r00ff00;
color map("blue") := 16r0000ff;
color map("my favourite color") := 16r00ffff;
# then, get some values out #
COLOR color := color map("green"); # color gets 16r00ff00 #
color := color map("black"); # accessing unassigned values assigns them to 16r0 #
# get some value out without accidentally inserting new ones #
REF COLOR value = color map find("green");
put(stand error, ("color not found!", new line))
printf(($"green: "f(color repr)l$, value))
# Now I changed my mind about my favourite color, so change it #
color map("my favourite color") := 16r337733;
# print out all defined colors #
FOR index FROM LWB color map items TO UPB color map items DO
ITEM item = color map items[index];
putf(stand error, ($"color map("""g""") = "f(color repr)l$, item))
FORMAT comma sep = $"("n(UPB color map items-1)(f(fmt)", ")f(fmt)")"$;
fmt := $""""g""""$;
printf(($g$,"keys: ", comma sep, key OF color map items, $l$));
fmt := color repr;
printf(($g$,"values: ", comma sep, value OF color map items, $l$))
green: 16r00ff00
color map("red") = 16rff0000
color map("green") = 16r00ff00
color map("blue") = 16r0000ff
color map("my favourite color") = 16r337733
color map("black") = 16r000000
keys: ("red", "green", "blue", "my favourite color", "black")
values: (16rff0000, 16r00ff00, 16r0000ff, 16r337733, 16r000000)


Apex provides a Map datatype that maps unique keys to a single value. Both keys and values can be any data type, including user-defined types. Like Java, equals and hashCode are used to determine key uniqueness for user-defined types. Uniqueness of sObject keys is determined by comparing field values.

Creating a new empty map of String to String:

// Cannot / Do not need to instantiate the algorithm implementation (e.g, HashMap).
Map<String, String> strMap = new Map<String, String>();
strMap.put('a', 'aval');
strMap.put('b', 'bval');
System.assert( strMap.containsKey('a') );
System.assertEquals( 'bval', strMap.get('b') );
// String keys are case-sensitive
System.assert( !strMap.containsKey('A') );

Creating a new map of String to String with values initialized:

Map<String, String> strMap = new Map<String, String>{
'a' => 'aval',
'b' => 'bval'
System.assert( strMap.containsKey('a') );
System.assertEquals( 'bval', strMap.get('b') );


Works with: Dyalog APL
⍝  Create a namespace ("hash")
X←⎕NS ⍬
⍝ Assign some names
⍝ Access the names
⍝ Or do it the array way
X.(foo this)
88 that
⍝ Namespaces are first class objects
sales ← ⎕NS ⍬
sales.(prices quantities) ← (100 98.4 103.4 110.16) (10 12 8 10)
sales.(revenue ← prices +.× quantities)

App Inventor[edit]

Associative arrays in App Inventor are lists of key:value 'pairs'.
When a list is organized as pairs, the lookup in pairs block can be used to retrieve an associated value from a key name.


True arrays[edit]

AutoHotkey_L has Objects which function as associative arrays.

associative_array := {key1: "value 1", "Key with spaces and non-alphanumeric characters !*+": 23}
MsgBox % associative_array.key1
. "`n" associative_array["Key with spaces and non-alphanumeric characters !*+"]

Legacy versions[edit]

AutoHotkey_Basic does not have typical arrays. However, variable names can be concatenated, simulating associative arrays.

arrayX1 = first
arrayX2 = second
arrayX3 = foo
arrayX4 = bar
Loop, 4
Msgbox % arrayX%A_Index%


Arrays in AWK are indeed associative arrays.

a["red"] = 0xff0000
a["green"] = 0x00ff00
a["blue"] = 0x0000ff
for (i in a) {
printf "%8s %06x\n", i, a[i]
# deleting a key/value
delete a["red"]
for (i in a) {
print i
# check if a key exists
print ( "red" in a ) # print 0
print ( "blue" in a ) # print 1


(("foo" 13)
("bar" 42)
("baz" 77)) ls2map !


global values$, keys$
dim values$[1]
dim keys$[1]
call updateKey("a","apple")
call updateKey("b","banana")
call updateKey("c","cucumber")
gosub show
print "I like to eat a " + getValue$("c") + " on my salad."
call deleteKey("b")
call updateKey("c","carrot")
call updateKey("e","endive")
gosub show
for t = 0 to countKeys()-1
print getKeyByIndex$(t) + " " + getValueByIndex$(t)
next t
subroutine updateKey(key$, value$)
# update or add an item
if i=-1 then
i = freeKey()
keys$[i] = key$
end if
values$[i] = value$
end subroutine
subroutine deleteKey(key$)
# delete by clearing the key
if i<>-1 then
keys$[i] = ""
end if
end subroutine
function freeKey()
# find index of a free element in the array
for n = 0 to keys$[?]-1
if keys$[n]="" then return n
next n
redim keys$[n+1]
redim values$[n+1]
return n
end function
function findKey(key$)
# return index or -1 if not found
for n = 0 to keys$[?]-1
if key$=keys$[n] then return n
next n
return -1
end function
function getValue$(key$)
# return a value by the key or "" if not existing
if i=-1 then
return ""
end if
return values$[i]
end function
function countKeys()
# return number of items
# remember to skip the empty keys (deleted ones)
k = 0
for n = 0 to keys$[?] -1
if keys$[n]<>"" then k++
next n
return k
end function
function getValueByIndex$(i)
# get a value by the index
# remember to skip the empty keys (deleted ones)
k = 0
for n = 0 to keys$[?] -1
if keys$[n]<>"" then
if k=i then return values$[k]
next n
return ""
end function
function getKeyByIndex$(i)
# get a key by the index
# remember to skip the empty keys (deleted ones)
k = 0
for n = 0 to keys$[?] -1
if keys$[n]<>"" then
if k=i then return keys$[k]
next n
return ""
end function
a apple
b banana
c cucumber

I like to eat a cucumber on my salad.
a apple
e endive
c carrot

Batch File[edit]

This is cheating, I'm sure of it.

@echo off
set array.wolf=3
set array.cow=4
for %%i in (dog cat wolf cow) do call :showit array.%%i !array.%%i!
set c=-27
call :mkarray sicko flu 5 measles 6 mumps 7 bromodrosis 8
for %%i in (flu measles mumps bromodrosis) do call :showit "sicko^&%%i" !sicko^&%%i!
goto :eof
set %1^&%2=%3
shift /2
shift /2
if "%2" neq "" goto :mkarray
goto :eof
echo %1 = %2
goto :eof
Output: = 1 = 2
array.wolf = 3
array.cow = 4
"sicko&flu" = 5
"sicko&measles" = 6
"sicko&mumps" = 7
"sicko&bromodrosis" = 8


      REM Store some values with their keys:
PROCputdict(mydict$, "FF0000", "red")
PROCputdict(mydict$, "00FF00", "green")
PROCputdict(mydict$, "0000FF", "blue")
REM Retrieve some values using their keys:
PRINT FNgetdict(mydict$, "green")
PRINT FNgetdict(mydict$, "red")
DEF PROCputdict(RETURN dict$, value$, key$)
IF dict$ = "" dict$ = CHR$(0)
dict$ += key$ + CHR$(1) + value$ + CHR$(0)
DEF FNgetdict(dict$, key$)
I% = INSTR(dict$, CHR$(0) + key$ + CHR$(1))
IF I% = 0 THEN = "" ELSE I% += LEN(key$) + 2
J% = INSTR(dict$, CHR$(0), I%)
= MID$(dict$, I%, J% - I%)


The hash is the only built-in Bracmat class. It is best used for e.g. a large dictionary, when manipulation of a very long list of key/value pairs with pattern matching would become too CPU-intensive. The same key can be stored with different values, as the example shows. If that is not desirable, the key (and its value) should be removed first.

& (myhash..insert)$(title."Some title")
& (myhash..insert)$(formula.a+b+x^7)
& (myhash..insert)$(fruit.apples oranges kiwis)
& (myhash..insert)$(meat.)
& (myhash..insert)$(fruit.melons bananas)
& out$(myhash..find)$fruit
& (myhash..remove)$formula
& (myhash..insert)$(formula.x^2+y^2)
& out$(myhash..find)$formula;
(fruit.melons bananas) (fruit.apples oranges kiwis)


h = [:]  #Empty hash
h[:a] = 1 #Assign value
h[:b] = [1 2 3] #Assign another value
h2 = [a: 1, b: [1 2 3], 10 : "ten"] #Initialized hash
h2[:b][2] #Returns 3


Solution is at Associative arrays/Creation/C.


The C++ standard defines std::map as a means of creating an association between a key of one arbitrary type and a value of another arbitrary type. This requires the inclusion of the standard header map.

#include <map>


To create a simple map whose key is of type A and whose value is of type B, one would define the variable like so:

std::map<A, B> exampleMap

If one wanted to us a key type of int and a value of double, you would define it like so:

std::map<int, double> exampleMap


Once we've created our map, we've got a couple different ways to insert the value. Let's use an example key of 7, and an exable value of 3.14.


The first method is using the [] operator.

exampleMap[7] = 3.14

Of course, you can use a variable (or any rvalue of the correct type) for the key or value parameters:

int myKey = 7;
double myValue = 3.14;
exampleMap[myKey] = myValue;


The second approach is a little more complicated. We have to use the pair<> template:

exampleMap.insert(std::pair<int, double>(7,3.14));

or by using make_pair to avoid repeating key/value types:



As with insertion, there are a couple ways we can retrieve the value.


We use it as an rvalue, supplying the correct key:

myValue = exampleMap[myKey]

If the value doesn't already exist, a default-constructed object of the value's type will be inserted using the key you specified, and that default value will be returned.


Alternatively, you can look up a value by using find(), storing its return value in an iterator, and comparing the iterator against the map's end() sentinal value:

double myValue = 0.0;
std::map<int, double>::iterator myIterator = exampleMap.find(myKey);
if(exampleMap.end() != myIterator)
// Return the value for that key.
myValue = myIterator->second;

The need for the ->second code is because our iterator points to a pair<>(), and our value is the second member of that pair.

This code assigns a 0 to myValue if the map contained a value.


This simple program creates a map, assigns a value to that map, retrieves a value from that map, and prints the value to STDOUT.

#include <map>
#include <iostreams>
int main()
// Create the map.
std::map<int, double> exampleMap;
// Choose our key
int myKey = 7;
// Choose our value
double myValue = 3.14;
// Assign a value to the map with the specified key.
exampleMap[myKey] = myValue;
// Retrieve the value
double myRetrievedValue = exampleMap[myKey];
// Display our retrieved value.
std::cout << myRetrievedValue << std::endl;
// main() must return 0 on success.
return 0;


Platform: .NET 1.x

System.Collections.HashTable map = new System.Collections.HashTable();
map["key1"] = "foo";

Platform: .NET 2.0

Dictionary<string, string> map = new Dictionary<string,string>();
map[ "key1" ] = "foo";
Works with: C# version 3.0+
var map = new Dictionary<string, string> {{"key1", "foo"}};


import ceylon.collection {
shared void run() {
// the easiest way is to use the map function to create
// an immutable map
value myMap = map {
"foo" -> 5,
"bar" -> 10,
"baz" -> 15,
"foo" -> 6 // by default the first "foo" will remain
// or you can use the HashMap constructor to create
// a mutable one
value myOtherMap = HashMap {
myOtherMap.put("baz", "baxx");
// there's also a sorted red/black tree map
value myTreeMap = naturalOrderTreeMap {
1 -> "won",
2 -> "too",
4 -> "fore"
for(num->homophone in myTreeMap) {
print("``num`` is ``homophone``");


In Chapel, associative arrays are regular arrays with a non-integer domain - values used as keys into the array. The creation of the domain is independent from the creation of the array, and in fact the same domain can be used for multiple arrays, creating associative arrays with identical sets of keys. When the domain is changed, all arrays that use it will be reallocated.

// arr is an array of string to int. any type can be used in both places.
var keys: domain(string);
var arr: [keys] int;
// keys can be added to a domain using +, new values will be initialized to the default value (0 for int)
keys += "foo";
keys += "bar";
keys += "baz";
// array access via [] or ()
arr["foo"] = 1;
arr["bar"] = 4;
arr("baz") = 6;
// write auto-formats domains and arrays
writeln("Keys: ", keys);
writeln("Values: ", arr);
// keys can be deleted using -
keys -= "bar";
writeln("Keys: ", keys);
writeln("Values: ", arr);
// chapel also supports array literals
var arr2 = [ "John" => 3, "Pete" => 14 ];
writeln("arr2 keys: ", arr2.domain);
writeln("arr2 values: ", arr2);
Keys: {foo, bar, baz}
Values: 1 4 6
Keys: {foo, baz}
Values: 1 6
arr2 keys: {John, Pete}
arr2 values: 3 14


{:key "value"
:key2 "value2"
:key3 "value3"}


<cfset myHash = structNew()>
<cfset myHash.key1 = "foo">
<cfset myHash["key2"] = "bar">
<cfset myHash.put("key3","java-style")>

In ColdFusion, a map is literally a java.util.HashMap, thus the above 3rd method is possible.

Common Lisp[edit]

;; default :test is #'eql, which is suitable for numbers only,
;; or for implementation identity for other types!
;; Use #'equalp if you want case-insensitive keying on strings.
(setf my-hash (make-hash-table :test #'equal))
(setf (gethash "H2O" my-hash) "Water")
(setf (gethash "HCl" my-hash) "Hydrochloric Acid")
(setf (gethash "CO" my-hash) "Carbon Monoxide")
;; That was actually a hash table, an associative array or
;; alist is written like this:
(defparameter *legs* '((cow . 4) (flamingo . 2) (centipede . 100)))
;; you can use assoc to do lookups and cons new elements onto it to make it longer.

Component Pascal[edit]

BlackBox Componente Builder
Using a handmade collections module with the following interface

DEFINITION Collections;
notFound = -1;
cap-, size-: INTEGER;
(h: Hash) ContainsKey (k: Boxes.Object): BOOLEAN, NEW;
(h: Hash) Get (k: Boxes.Object): Boxes.Object, NEW;
(h: Hash) IsEmpty (): BOOLEAN, NEW;
(h: Hash) Put (k, v: Boxes.Object): Boxes.Object, NEW;
(h: Hash) Remove (k: Boxes.Object): Boxes.Object, NEW;
(h: Hash) Reset, NEW
cap-, size-: INTEGER;
(hm: HashMap) ContainsKey (k: Boxes.Object): BOOLEAN, NEW;
(hm: HashMap) ContainsValue (v: Boxes.Object): BOOLEAN, NEW;
(hm: HashMap) Get (k: Boxes.Object): Boxes.Object, NEW;
(hm: HashMap) IsEmpty (): BOOLEAN, NEW;
(hm: HashMap) Keys (): POINTER TO ARRAY OF Boxes.Object, NEW;
(hm: HashMap) Put (k, v: Boxes.Object): Boxes.Object, NEW;
(hm: HashMap) Remove (k: Boxes.Object): Boxes.Object, NEW;
(hm: HashMap) Reset, NEW;
(hm: HashMap) Values (): POINTER TO ARRAY OF Boxes.Object, NEW
first-, last-: Node;
size-: INTEGER;
(ll: LinkedList) Add (item: Boxes.Object), NEW;
(ll: LinkedList) Append (item: Boxes.Object), NEW;
(ll: LinkedList) AsString (): POINTER TO ARRAY OF CHAR, NEW;
(ll: LinkedList) Contains (item: Boxes.Object): BOOLEAN, NEW;
(ll: LinkedList) Get (at: INTEGER): Boxes.Object, NEW;
(ll: LinkedList) IndexOf (item: Boxes.Object): INTEGER, NEW;
(ll: LinkedList) Insert (at: INTEGER; item: Boxes.Object), NEW;
(ll: LinkedList) IsEmpty (): BOOLEAN, NEW;
(ll: LinkedList) Remove (item: Boxes.Object), NEW;
(ll: LinkedList) RemoveAt (at: INTEGER), NEW;
(ll: LinkedList) Reset, NEW;
(ll: LinkedList) Set (at: INTEGER; item: Boxes.Object), NEW
size-, cap-: LONGINT;
(v: Vector) Add (item: Boxes.Object), NEW;
(v: Vector) AddAt (item: Boxes.Object; i: INTEGER), NEW;
(v: Vector) Contains (o: Boxes.Object): BOOLEAN, NEW;
(v: Vector) Get (i: LONGINT): Boxes.Object, NEW;
(v: Vector) IndexOf (o: Boxes.Object): LONGINT, NEW;
(v: Vector) Remove (o: Boxes.Object), NEW;
(v: Vector) RemoveIndex (i: LONGINT): Boxes.Object, NEW;
(v: Vector) Set (i: LONGINT; o: Boxes.Object): Boxes.Object, NEW;
(v: Vector) Trim, NEW
PROCEDURE NewHash (cap: INTEGER): Hash;
PROCEDURE NewHashMap (cap: INTEGER): HashMap;
PROCEDURE NewLinkedList (): LinkedList;
PROCEDURE NewVector (cap: INTEGER): Vector;
END Collections.

The program:

MODULE BbtAssociativeArrays;
IMPORT StdLog, Collections, Boxes;
hm : Collections.HashMap;
o : Boxes.Object;
keys, values: POINTER TO ARRAY OF Boxes.Object;
hm := Collections.NewHashMap(1009);
o := hm.Put(Boxes.NewString("first"),Boxes.NewInteger(1));
o := hm.Put(Boxes.NewString("second"),Boxes.NewInteger(2));
o := hm.Put(Boxes.NewString("third"),Boxes.NewInteger(3));
o := hm.Put(Boxes.NewString("one"),Boxes.NewInteger(1));
StdLog.String("size: ");StdLog.Int(hm.size);StdLog.Ln;
END BbtAssociativeArrays.

Execute:^Q BbtAssociativeArrays.Do

size:  4


void main() {
auto hash = ["foo":42, "bar":100];
assert("foo" in hash);


m = { => } # empty ordered map, future inserted keys will be ordered
h = { -> } # empty hash map, future inserted keys will not be ordered
m = { 'foo' => 42, 'bar' => 100 } # with ordered keys
h = { 'foo' -> 42, 'bar' -> 100 } # with unordered keys


program AssociativeArrayCreation;
uses Generics.Collections;
lDictionary: TDictionary<string, Integer>;
lDictionary := TDictionary<string, Integer>.Create;
lDictionary.Add('foo', 5);
lDictionary.Add('bar', 10);
lDictionary.Add('baz', 15);
lDictionary.AddOrSetValue('foo', 6); // replaces value if it exists


[].asMap()                             # immutable, empty
["one" => 1, "two" => 2] # immutable, 2 mappings
[].asMap().diverge() # mutable, empty
["one" => 2].diverge(String, float64) # mutable, initial contents,
# typed (coerces to float)


(lib 'hash) ;; needs hash.lib
(define H (make-hash)) ;; new hash table
;; keys may be symbols, numbers, strings ..
;; values may be any lisp object
(hash-set H 'simon 'antoniette)
→ antoniette
(hash-set H 'antoinette 'albert)
→ albert
(hash-set H "Elvis" 42)
(hash-ref H 'Elvis)
→ #f ;; not found. Elvis is not "Elvis"
(hash-ref H "Elvis")
(hash-ref H 'simon)
→ antoniette
(hash-count H)


#define system.
#define system'collections.
// --- Program ---
#symbol program =
// 1. Create
#var aMap := Dictionary new.
aMap@"key" := "foox".
aMap@"key" := "foo".
aMap@"key2":= "foo2".
aMap@"key3":= "foo3".
aMap@"key4":= "foo4".


Translation of: Erlang
defmodule RC do
def test_create do
IO.puts "< create >"
m = #=> creates an empty Map
m1 = Map.put(m,:foo,1)
m2 = Map.put(m1,:bar,2)
print_vals(%{m2 | foo: 3})
defp print_vals(m) do
IO.inspect m
Enum.each(m, fn {k,v} -> IO.puts "#{inspect k} => #{v}" end)
< create >
%{bar: 2, foo: 1}
:bar => 2
:foo => 1
%{bar: 2, foo: 3}
:bar => 2
:foo => 3

Emacs Lisp[edit]

(setq my-table (make-hash-table))
(puthash 'key 'value my-table)

make-hash-table compares keys with eql by default. This suits symbols and numbers (including floating point). For string keys an equal test can be used,

(setq my-table (make-hash-table :test 'equal))
(puthash "key" 123 my-table)

define-hash-table-test can create other key comparison types.


Erlang offers several associative array type data structures, this example uses the dictionary data structure.

test_create() ->
D = dict:new(),
D1 = dict:store(foo,1,D),
D2 = dict:store(bar,2,D1),
print_vals(D) ->
lists:foreach(fun (K) ->
io:format("~p: ~b~n",[K,dict:fetch(K,D)])
end, dict:fetch_keys(D)).
32> assoc:test_create().
bar: 2
foo: 1
bar: 2
foo: 3


.NET 3.5 Generic Dictionary (mutable)

let dic = System.Collections.Generic.Dictionary<string,string>() ;;
dic.Add("key","val") ;
dic.["key"] <- "new val" ;

Functional dictionary (immutable)

let d = [("key","val");("other key","other val")] |> Map.ofList
let newd = d.Add("new key","new val")
let takeVal (d:Map<string,string>) =
match d.TryFind("key") with
| Some(v) -> printfn "%s" v
| None -> printfn "not found"


Associative mappings follow the associative protocol. See the docs. Here's an example using a hashtable that can be run in the listener :

H{ { "one" 1 } { "two" 2 } }
{ [ "one" swap at . ]
[ 2 swap value-at . ]
[ "three" swap at . ]
[ [ 3 "three" ] dip set-at ]
[ "three" swap at . ] } cleave


Associative arrays are called 'maps' in Fantom:

class Main
public static Void main ()
// create a map which maps Ints to Strs, with given key-value pairs
Int:Str map := [1:"alpha", 2:"beta", 3:"gamma"]
// create an empty map
Map map2 := [:]
// now add some numbers mapped to their doubles
10.times |Int i|
map2[i] = 2*i


Works with: GNU Forth version 0.6.2

The Forth dictionary is normally only used for function and symbol definitions, but you can also define separate wordlists for holding functions or data. There is no special syntax in the language for this, but you can define your own. All of Forth's defining words are available for adding things to the wordlist, but CREATE is most generic.

: get ( key len table -- data )     \ 0 if not present
search-wordlist if
>body @
else 0 then ;
: put ( data key len table -- )
>r 2dup r@ search-wordlist if
r> drop nip nip
>body !
r> get-current >r set-current \ switch definition word lists
nextname create ,
r> set-current
then ;
wordlist constant bar
5 s" alpha" bar put
9 s" beta" bar put
2 s" gamma" bar put
s" alpha" bar get . \ 5
8 s" Alpha" bar put \ Forth dictionaries are normally case-insensitive
s" alpha" bar get . \ 8

This is not necessarily a good option in all Forths, as the dictionary may be implemented as a simple linked list (normally not a problem because the dictionary is only used for compiling and interactive interpretation). GNU Forth and many other hosted Forths use hash tables for the dictionary, so this is a fine choice. If you need case-sensitive keys, GNU Forth has table and table-find, replacing wordlist and search-wordlist, respectively.

(The use of nextname ( str len -- ) is a GNU Forth extension to create; there is no means in the ANS standard to use a string on the stack to create a dictionary entry.)

Hashtable for mapping strings to integer

include ffl/hct.fs
\ Create a hash table 'table' in the dictionary with a starting size of 10
10 hct-create htable
\ Insert entries
5 s" foo" htable hct-insert
10 s" bar" htable hct-insert
15 s" baz" htable hct-insert
\ Get entry from the table
s" bar" htable hct-get [IF]
.( Value:) . cr
.( Entry not present.) cr


Allowable key types are those with == and != operators. This includes is boolean, numeric, string, pointer, channel, and interface types. It also includes structs and arrays containing only these types. Disallowed as map keys are all slice, function, and map types.

// declare a nil map variable, for maps from string to int
var x map[string]int
// make an empty map
x = make(map[string]int)
// make an empty map with an initial capacity
x = make(map[string]int, 42)
// set a value
x["foo"] = 3
// getting values
y1 := x["bar"] // zero value returned if no map entry exists for the key
y2, ok := x["bar"] // ok is a boolean, true if key exists in the map
// removing keys
delete(x, "foo")
// make a map with a literal
x = map[string]int{
"foo": 2, "bar": 42, "baz": -1,


Create an empty map and add values

map = [:]
map[7] = 7
map['foo'] = 'foovalue'
map.put('bar', 'barvalue')
map.moo = 'moovalue'
assert 7 == map[7]
assert 'foovalue' ==
assert 'barvalue' == map['bar']
assert 'moovalue' == map.get('moo')

Create a pre-populated map and verify values

map = [7:7, foo:'foovalue', bar:'barvalue', moo:'moovalue']
assert 7 == map[7]
assert 'foovalue' ==
assert 'barvalue' == map['bar']
assert 'moovalue' == map.get('moo')


Create an empty array and add values:

arr := { => }
arr[ 10 ] := "Val_10"
arr[ "foo" ] := "foovalue"

Create and initialize array:

arr := hb_Hash( 10, "Val_10", "foo", "foovalue" )
// or
arr := { 10 => "Val_10", "foo" => "foovalue" }


Binary trees:

Works with: GHC
import Data.Map
dict = fromList [("key1","val1"), ("key2","val2")]
ans = Data.Map.lookup "key2" dict -- evaluates to Just "val2"

It is also possible to use association lists (lists of pairs). It is inefficient (O(n) lookup), but simple.

dict = [("key1","val1"), ("key2","val2")]
ans = lookup "key2" dict -- evaluates to Just "val2"

GHC also had an imperative hash table implementation in the Data.HashTable module, but was removed in GHC 7.8.

Other standard associatives arrays libraries are : Data.IntMap and Data.HasMap

Icon and Unicon[edit]

Icon and Unicon associative arrays are called tables. Any value may be used as a key including complex structures. Tables can have default values and they have no inherent size limitation growing from empty to whatever size is needed.

procedure main() 
local t
t := table()
t["foo"] := "bar"

Inform 7[edit]

The Inform 7 equivalent of an associative array is a relation between values.

Static relation[edit]

Hash Bar is a room.
Connection relates various texts to one number. The verb to be connected to implies the connection relation.
"foo" is connected to 12.
"bar" is connected to 34.
"baz" is connected to 56.
When play begins:
[change values]
now "bleck" is connected to 78;
[check values]
if "foo" is connected to 12, say "good.";
if "bar" is not connected to 56, say "good.";
[retrieve values]
let V be the number that "baz" relates to by the connection relation;
say "'baz' => [V].";
end the story.

Dynamic relation[edit]

Hash Bar is a room.
When play begins:
let R be a various-to-one relation of texts to numbers;
[initialize the relation]
now R relates "foo" to 12;
now R relates "bar" to 34;
now R relates "baz" to 56;
[check values]
if R relates "foo" to 12, say "good.";
if R does not relate "bar" to 56, say "good.";
[retrieve values]
let V be the number that "baz" relates to by R;
say "'baz' => [V].";
end the story.


{a: "a", b: "b"}


coclass 'assocArray'
encode=: 'z', (a.{~;48 65 97(+ i.)&.>10 26 26) {~ 62x #.inv 256x #. a.&i.
get=: ".@encode
has=: 0 <: nc@<@encode
set=:4 :'(encode x)=:y'

Example use:

   example=: conew 'assocArray'
'foo' set__example 1 2 3
1 2 3
'bar' set__example 4 5 6
4 5 6
get__example 'foo'
1 2 3
has__example 'foo'
bletch__example=: 7 8 9
get__example 'bletch'
7 8 9

Note that J's symbols ( might also be used for this purpose. However, symbols are not garbage collected within a J session (and, instead, a mechanism is provided to optionally preserve them across sessions).


Works with: Java version 1.5+

Defining the Map:

Map<String, Integer> map = new HashMap<String, Integer>();
map.put("foo", 5);
map.put("bar", 10);
map.put("baz", 15);
map.put("foo", 6);

"Putting" a value for a key that already exists ("map.put("foo", 6)" in this example) will replace and return the old value for the key.

Initializing a Map as a class member:

public static Map<String, Integer> map = new HashMap<String, Integer>(){{
put("foo", 5);
put("bar", 10);
put("baz", 15);
put("foo", 6);

Retrieving a value:

map.get("foo"); // => 6
map.get("invalid"); // => null

Note that it is possible to put null as a value, so null being returned by get is not sufficient for determining that the key is not in the Map. There is a containsKey method for that.

Iterate over keys:

for (String key: map.keySet()) 

Iterate over values:

for (int value: map.values())

Iterate over key, value pairs:

for (Map.Entry<String, Integer> entry: map.entrySet())
System.out.println(entry.getKey() + " => " + entry.getValue());


ECMAScript5.1 does not have associative arrays, however Objects (which are just an unordered bundle of name/value pairs) can be used like associative arrays. JavaScript Arrays may also be used, but Objects are the convention.

Javascript object property names (keys) are strings. Other types and expressions can be used with square bracket notation, they are evaluated and converted to strings and the result used as the property name. Using quotes on property names avoids potential collisions with reserved JavaScript key words.

var assoc = {};
assoc['foo'] = 'bar';
assoc['another-key'] = 3;
// dot notation can be used if the property name is a valid identifier
assoc.thirdKey = 'we can also do this!';
assoc[2] = "the index here is the string '2'";
//using JavaScript's object literal notation
var assoc = {
foo: 'bar',
'another-key': 3 //the key can either be enclosed by quotes or not
//iterating keys
for (var key in assoc) {
// hasOwnProperty() method ensures the property isn't inherited
if (assoc.hasOwnProperty(key)) {
alert('key:"' + key + '", value:"' + assoc[key] + '"');

ECMAScript 6 (ES6) offers both a map and a weak map implementation. While Objects must use strings, Maps may use objects, functions, and numbers as keys in addition to strings.

var map = new Map(),
fn = function () {},
obj = {};
map.set(fn, 123);
map.set(obj, 'abc');
map.set('key', 'val');
map.set(3, x => x + x);
map.get(fn); //=> 123
map.get(function () {}); //=> undefined because not the same function
map.get(obj); //=> 'abc'
map.get({}); //=> undefined because not the same object
map.get('key'); //=> 'val'
map.get(3); //=> (x => x + x)
map.size; //=> 4
//iterating using ES6 for..of syntax
for (var key of map.keys()) {
console.log(key + ' => ' + map.get(key));


Associative Arrays with String-Valued Keys[edit]

In jq, JSON objects can be used as associative arrays, it being understood that only strings can be used as keys. To avoid confusion, for the remainder of this section, we refer to JSON objects as such. Their type in jq is "object".

# An empty object:
# Its type:
{} | type
# "object"
# An object literal:
{"a": 97, "b" : 98}
# Programmatic object construction:
reduce ("a", "b", "c", "d") as $c ({}; . + { ($c) : ($c|explode[.0])} )
# {"a":97,"c":99,"b":98,"d":100}
# Same as above:
reduce range (97;101) as $i ({}; . + { ([$i]|implode) : $i })
# Addition of a key/value pair by assignment:
{}["A"] = 65 # in this case, the object being added to is {}
# Alteration of the value of an existing key:
{"A": 65}["A"] = "AA"

Associative Arrays with JSON-Valued Keys[edit]

In this subsection, we define addKey(key;value), getKey(key), and removeKey(key) to operate on a hash table for which the keys may be any JSON entities. This is done by defining a collisionless hash function.

def collisionless:
if type == "object" then with_entries(.value = (.value|collisionless))|tostring
elif type == "array" then map(collisionless)|tostring
else (type[0:1] + tostring)
# WARNING: addKey(key;value) will erase any previous value associated with key
def addKey(key;value):
if type == "object" then . + { (key|collisionless): value }
else {} | addKey(key;value)
def getKey(key): .[key|collisionless];
def removeKey(key): delpaths( [ [key|collisionless] ] );


{} | addKey(1;"one") | addKey(2; "two") | removeKey(1) | getKey(2)




We build dictionaries associating to some characters their code points, by listing the key/value pairs, through a dictionary comprehension, by creating an empty dictionary and filling it, by using the specific syntax associated to typed dictionaries and finally from two arrays using a constructor.

julia> hash = {'a' => 97, 'b' => 98}                  # list keys/values
julia> hash = {c => int(c) for c = 'a':'d'} # dict comprehension
julia> hash['é'] = 233 ; hash # add an element
julia> hash = Dict() # create an empty dict
julia> for c = 'a':'d' hash[c] = int(c) end ; hash # fill it
julia> hash = (Char=>Int64)['a' => 97, 'b' => 98] # create a typed dict
julia> hash["a"] = 1 # type mismatch
ERROR: no method convert(Type{Char}, ASCIIString)
in setindex! at dict.jl:533
julia> hash = Dict(['a','b','c'], [97,98,99]) # constructor
julia> typeof(hash) # type is infered correctly
Dict{Char,Int64} (constructor with 3 methods


Keys in a dictionary must be symbols (`symbol).

   / creating an dictionary
d1:.((`foo;1); (`bar;2); (`baz;3))
/ extracting a value

Another approach.

   d2: .()          / create empty dictionary

Extracting the keys and values.

   !d2              / the keys
`zero `one `two
d2[] / the values
0 1 2


Translation of: Java
fun main(args: Array<String>) {
// map definition:
val map = mapOf("foo" to 5,
"bar" to 10,
"baz" to 15,
"foo" to 6)
// retrieval:
println(map["foo"]) // => 6
println(map["invalid"]) // => null
// check keys:
println("foo" in map) // => true
println("invalid" in map) // => false
// iterate over keys:
for (k in map.keys) print("$k ")
// iterate over values:
for (v in map.values) print("$v ")
// iterate over key, value pairs:
for ((k, v) in map.entries) println("$k => $v")


: dip  swap '_ set execute _ ; : nip  swap drop ;
: first 0 extract nip ; : second 1 extract nip ;
: assoc-in swap keys eq ;
: assoc-index' over keys swap eq [1] index collapse ;
: at swap assoc-index' subscript collapse second ;
: delete-at swap assoc-index' first remove ;
: keys 1 transpose first ;
: set-at
over 'dup dip assoc-in '+ reduce if 'dup dip delete-at then
"swap 2 compress 1 compress" dip swap append ;
[['foo 5]]
10 'bar rot set-at
'bar over at .
'hello 'bar rot set-at
20 'baz rot set-at .


// In Lasso associative arrays are called maps
// Define an empty map
local(mymap = map)
// Define a map with content
local(mymap = map(
'one' = 'Monday',
'2' = 'Tuesday',
3 = 'Wednesday'
// add elements to an existing map
#mymap -> insert('fourth' = 'Thursday')
// retrieve a value from a map
#mymap -> find('2') // Tuesday
'<br />'
#mymap -> find(3) // Wednesday, found by the key not the position
'<br />'
// Get all keys from a map
#mymap -> keys // staticarray(2, fourth, one, 3)
'<br />'
// Iterate thru a map and get values
with v in #mymap do {^
'<br />'
// Tuesday<br />Thursday<br />Monday<br />Wednesday<br />
// Perform actions on each value of a map
#mymap -> foreach => {
#1 -> uppercase
#1 -> reverse
#mymap // map(2 = YADSEUT, fourth = YADSRUHT, one = YADNOM, 3 = YADSENDEW)


(let* ((my-dict (: dict new))
(my-dict (: dict store 'key-1 '"value 1" my-dict))
(my-dict (: dict store 'key-2 '"value 2" my-dict)))
(: io format '"size: ~p~n" (list (: dict size my-dict)))
(: io format '"some data: ~p~n" (list (: dict fetch 'key-1 my-dict))))

Liberty BASIC[edit]

Needs the sublist library from since LB does not have built-in associative arrays.

data "red", "255 50 50", "green", "50 255 50", "blue", "50 50 255"
data "my fave", "220 120 120", "black", "0 0 0"
myAssocList$ =""
for i =1 to 5
read k$
read dat$
call sl.Set myAssocList$, k$, dat$
next i
print " Key 'green' is associated with data item "; sl.Get$( myAssocList$, "green")
Key 'green' is associated with data item 50 255 50


props = [#key1: "value1", #key2: "value2"]
put props[#key2]
-- "value2"
put props["key2"]
-- "value2"
put props.key2
-- "value2"
put props.getProp(#key2)
-- "value2"


Livecode arrays are only associative, but can be accessed by ordinal if they are used as the key.

command assocArray
local tArray
put "value 1" into tArray["key 1"]
put 123 into tArray["key numbers"]
put "a,b,c" into tArray["abc"]
put "number of elements:" && the number of elements of tArray & return & \
"length of item 3:" && the length of tArray["abc"] & return & \
"keys:" && the keys of tArray
end assocArray


number of elements: 3
length of item 3: 5
keys: key numbers
key 1


UCB Logo has "property lists" which associate names with values. They have their own namespace.

pprop "animals "cat 5
pprop "animals "dog 4
pprop "animals "mouse 11
print gprop "animals "cat  ; 5
remprop "animals "dog
show plist "animals  ; [mouse 11 cat 5]


Lua tables are Hashes

hash = {}
hash[ "key-1" ] = "val1"
hash[ "key-2" ] = 1
hash[ "key-3" ] = {}

Returns nil on unknown key.


Maple tables are hashed arrays. A table can be constructed by using the table constructor.

> T := table( [ (2,3) = 4, "foo" = 1, sin(x) = cos(x) ] );
T := table(["foo" = 1, sin(x) = cos(x), (2, 3) = 4])
> T[2,3];
> T[sin(x)];
> T["foo"];

New entries are added by assignment.

> T[ "bar" ] := 2;
T["bar"] := 2
> T[ "bar" ];

Entries can be removed as follows.

> T[ "foo" ] := evaln( T[ "foo" ] );
T["foo"] := T["foo"]
> T[ "foo" ];

(The latter output indicates that T["foo"] is an unassigned name.)

Mathematica / Wolfram Language[edit]

a[2] = "string"; a["sometext"] = 23;

MATLAB / Octave[edit]

MATLAB/Octave: structs[edit]

Associative arrays are called structs. The following methods of creating hash are equivalent.

   hash.a = 1;
hash.b = 2;
hash.C = [3,4,5];


   hash = []; 
hash = setfield(hash,'a',1);
hash = setfield(hash,'b',2);
hash = setfield(hash,'C',[3,4,5]);


   hash.('a') = 1; 	
hash.('b') = 2;
hash.('C') = [3,4,5];
>>    disp(hash) 
  scalar structure containing the fields:
    a =  1
    b =  2
    C =

       3   4   5

Limitation: key must be a string containing only characters, digits and underscores, and the key string must start with a character.

MATLAB only: containers.Map[edit]

Use of containers.Map removes some restrictions on key types that structs have. Keys can all be numeric or all be strings. Values can be of any type. Key and value types cannot be changed after creation of the containers.Map object.

m = containers.Map({'a' 'b' 'C'}, [1 2 3]);

is equivalent to

m = containers.Map;
m('a') = 1;
m('b') = 2;
m('C') = 3;

since the KeyType defaults to 'char'. For numeric keys, the key and value types must be specified at creation.

m = containers.Map([51 72 37], {'fiftyone' 'seventytwo' 'thirtyseven'});

is equivalent to

m = containers.Map('KeyType', 'double', 'ValueType', 'any');
m(51) = 'fiftyone';
m(72) = 'seventytwo';
m(37) = 'thirtyseven';


>> m = containers.Map([51 72 37], {'fiftyone' 'seventytwo' 'thirtyseven'});
>> keys(m)

ans = 

    [37]    [51]    [72]

>> values(m)

ans = 

    'thirtyseven'    'fiftyone'    'seventytwo'


/* No need to declare anything, undeclared arrays are hashed */
h[1]: 6;
h[9]: 2;
[hashed, 1, [1], [9]]


This demonstrates two of several constructors, initializing the hashtable with a list of tuples or just specifying an initial capacity.

using System;
using System.Console;
using Nemerle.Collections;
module AssocArray
Main() : void
def hash1 = Hashtable([(1, "one"), (2, "two"), (3, "three")]);
def hash2 = Hashtable(3);
foreach (e in hash1)
hash2[e.Value] = e.Key;
WriteLine("Enter 1, 2, or 3:");
def entry = int.Parse(ReadLine());


/* NetRexx */
options replace format comments java crossref symbols
key0 = '0'
key1 = 'key0'
hash = '.' -- Initialize the associative array 'hash' to '.'
hash[key1] = 'value0' -- Set a specific key/value pair
say '<hash key="'key0'" value="'hash[key0]'" />' -- Display a value for a key that wasn't set
say '<hash key="'key1'" value="'hash[key1]'" />' -- Display a value for a key that was set
<hash key="0" value="." />
<hash key="key0" value="value0" />


import tables
hash = initTable[string, int]() # empty hash table
hash2 = {"key1": 1, "key2": 2}.toTable # hash table with two keys
hash3 = [("key1", 1), ("key2", 2)].toTable # hash table from tuple array
hash4 = @[("key1", 1), ("key2", 2)].toTable # hash table from tuple seq
value = hash2["key1"]
hash["spam"] = 1
hash["eggs"] = 2
hash.add("foo", 3)
echo "hash has ", hash.len, " elements"
echo "hash has key foo? ", hash.hasKey("foo")
echo "hash has key bar? ", hash.hasKey("bar")
echo "iterate pairs:" # iterating over (key, value) pairs
for key, value in hash:
echo key, ": ", value
echo "iterate keys:" # iterating over keys
for key in hash.keys:
echo key
echo "iterate values:" # iterating over values
for key in hash.values:
echo key
hash has 3 elements
hash has key foo? true
hash has key bar? false
iterate pairs:
eggs: 2
foo: 3
spam: 1
iterate keys:
iterate values:


Works with: oo2c Version 2
MODULE AssociativeArray;
Value = Boxed.LongInt;
assocArray: Dictionary.Dictionary(Key,Value);
iterK: Dictionary.IterKeys(Key,Value);
iterV: Dictionary.IterValues(Key,Value);
aux: Value;
k: Key;
assocArray := NEW(Dictionary.Dictionary(Key,Value));
aux := assocArray.Get("ten");
aux := assocArray.Get("eleven");
(* Iterate keys *)
iterK := assocArray.IterKeys();
WHILE (iterK.Next(k)) DO
(* Iterate values *)
iterV := assocArray.IterValues();
WHILE (iterV.Next(aux)) DO
END AssociativeArray.


Object parameters must be implicitly casted to the types expected by the method that's called.

Associative map[edit]

# create map
map := StringMap->New();
# insert
map->Insert("two", IntHolder->New(2)->As(Base));
map->Insert("thirteen", IntHolder->New(13)->As(Base));
map->Insert("five", IntHolder->New(5)->As(Base));
map->Insert("seven", IntHolder->New(7)->As(Base));
# find

Hash table[edit]

# create map
map := StringHash->New();
# insert
map->Insert("two", IntHolder->New(2)->As(Base));
map->Insert("thirteen", IntHolder->New(13)->As(Base));
map->Insert("five", IntHolder->New(5)->As(Base));
map->Insert("seven", IntHolder->New(7)->As(Base));
# find


Works with: Cocoa
Works with: GNUstep

You can use a NSDictionary to create an immutable hash. A dictionary can contain only objects; if you want store non objects like integer, you have to box it in NSNumber.

NSDictionary *dict = [NSDictionary dictionaryWithObjectsAndKeys:
@"Joe Doe", @"name",
[NSNumber numberWithUnsignedInt:42], @"age",
[NSNull null], @"extra",

The same as the above with the new literal syntax in clang 3.1+ / Apple LLVM Compiler 4.0+ (XCode 4.4+) :

NSDictionary *dict = @{
@"name": @"Joe Doe",
@"age": @42,
@"extra": [NSNull null],

To create a mutable dictionary, use NSMutableDictionary:

NSMutableDictionary *dict = [NSMutableDictionary dictionary];
[dict setObject:@"Joe Doe" forKey:@"name"];
[dict setObject:[NSNumber numberWithInt:42] forKey:@"age"];

You can access value with objectForKey:. If a key does not exists, nil is returned.

NSString *name = [dict objectForKey:@"name"];
unsigned age = [dict objectForKey:@"age"] unsignedIntValue];
id missing = [dict objectForKey:@"missing"];


Hash table[edit]

A simple idiom to create a hash table mapping strings to integers:

let hash = Hashtbl.create 0;;
List.iter (fun (key, value) -> Hashtbl.add hash key value)
["foo", 5; "bar", 10; "baz", 15];;

To retrieve a value:

let bar = Hashtbl.find hash "bar";; (* bar = 10 *)

To retrieve a value, returning a default if the key is not found:

let quux = try Hashtbl.find hash "quux" with Not_found -> some_value;;

Binary tree[edit]

A simple idiom to create a persistent binary tree mapping strings to integers:

module String = struct
type t = string
let compare =
module StringMap = Map.Make(String);;
let map =
(fun map (key, value) -> StringMap.add key value map)
["foo", 5; "bar", 10; "baz", 15]

To retrieve a value:

let bar = StringMap.find "bar" map;; (* bar = 10 *)

To retrieve a value, returning a default if the key is not found:

let quux = try StringMap.find "quux" map with Not_found -> some_value;;

Association list[edit]

Some list functions allow you to use a list as an associative map, although the access time is O(N) so a Hashtbl or binary tree should be used for larger data-sets.

let dict = ["foo", 5; "bar", 10; "baz", 15]
(* retrieve value *)
let bar_num = try List.assoc "bar" dict with Not_found -> 0;;
(* see if key exists *)
print_endline (if List.mem_assoc "foo" dict then "key found" else "key missing")


ooRexx has multiple classes that create index-to-item associative relationships.

  • Directory -- a mapping for a string index to an object instance
  • Table -- a mapping for an object index (of any class) to an object instance. Index equality is determined by the "==" method.
  • Relation -- a one-to-many mapping for an object index (of any class) to object instances. Index equality is determined by the "==" method.
  • IdentityTable -- a mapping for an object index (of any class) to an object instance. Index equality is determined by unique object identity rather than equality.
  • Stem -- The class backing ooRexx stem variables, which is also a first-class collection class.

All of the MapCollections are very similar in usage. We'll use Directory for the examples here.

Defining the map:

map = .directory~new
map["foo"] = 5
map["bar"] = 10
map["baz"] = 15
map["foo"] = 6

"Putting" a value for a key that already exists ("map["foo"] = 6" in this example) will replace and return the old value for the key.

Retrieving a value:

item = map["foo"] -- => 6
item = map["invalid"] -- => .nil

Note that it is possible to put .nil as a value, so .nil being returned as a value is not sufficient for determining that the key is not in the collection. There is a hasIndex method for that.

Iterate over keys:

loop key over map
say key

Iterate over values:

loop value over map~allItems
say value

Iterate over key, value pairs:

s = map~supplier
loop while s~available
say s~index "=>" s~item


Not very efficient but the 'find' method could be optimised very easily.

def n 200
Class AssociativeArray
indexbase 1
string s[n]
sys max
method find(string k) as sys
sys i,e
for i=1 to e step 2
if k=s[i] then return i
end method
method dat(string k) as string
sys i=find(k)
if i then return s[i+1]
end method
method dat(string k, d) as sys
sys i=find(k)
if i=0 then
if max>=n
print "Array overflow" : return 0
end if
end if
return i
end method
end class
AssociativeArray A
A.s<={"shoes","LC1", "ships","LC2", "sealingwax","LC3", "cabbages","LC4", "kings","LC5"}
print A.dat("ships") 'result LC2
A.dat("computers")="LC99" '
print A.dat("computers") 'result LC99


A mutable map is called a 'dictionary' in Oz:

Dict = {}
in := 5 := 10
Dict.baz := 15 := 20
{Inspect Dict}

'Records' can be consideres immutable maps:

Rec = name(foo:5 bar:10 baz:20)
{Inspect Rec}


Works with: PARI/GP version 2.8.1+

GP's associative arrays are called maps, and can be created like so:

M = Map();

They can be used as follows:

mapput(M, "key", "value");
mapput(M, 17, "different value");
mapput(M, "key2", Pi);
mapget(M, "key2") \\ returns Pi
mapisdefined(M, "key3") \\ returns 0
mapdelete(M, "key2");

In PARI the commands are gtomap, mapput, mapget, mapisdefined, and mapdelete. You can also use the solutions in Associative arrays/Creation/C.




# using => key does not need to be quoted unless it contains special chars
my %hash = (
key1 => 'val1',
'key-2' => 2,
three => -238.83,
4 => 'val3',
# using , both key and value need to be quoted if containing something non-numeric in nature
my %hash = (
'key1', 'val1',
'key-2', 2,
'three', -238.83,
4, 'val3',


print $hash{key1};
$hash{key1} = 'val1';
@hash{'key1', 'three'} = ('val1', -238.83);



my $hashref = {
key1 => 'val1',
'key-2' => 2,
three => -238.83,
4 => 'val3',


print $hash->{key1};
$hash->{key1} = 'val1';
@hash->{'key1', 'three'} = ('val1', -238.83);

Key Types[edit]

Keys are strings. Anything else is stringized in Perl's usual ways, which generally means integers work too, but for floating point care might be needed against round-off.

Various tie modules implement keys of other types, usually by constructing underlying string keys of suitable nature. For example Tie::RefHash allows objects (blessed or unblessed) as keys.

Perl 6[edit]

Works with: Rakudo version 2015.10-11

The fatarrow, =>, is no longer just a quoting comma; it now constructs a Pair object. But you can still define a hash with an ordinary list of even length.

my %h1 = key1 => 'val1', 'key-2' => 2, three => -238.83, 4 => 'val3';
my %h2 = 'key1', 'val1', 'key-2', 2, 'three', -238.83, 4, 'val3';

Creating a hash from two lists using a metaoperator.

my @a = 1..5;
my @b = 'a'..'e';
my %h = @a Z=> @b;

Hash elements and hash slices now use the same sigil as the whole hash. This is construed as a feature. Curly braces no longer auto-quote, but Perl 6's qw (shortcut < ... >) now auto-subscripts.

say %h1{'key1'};
say %h1<key1>;
%h1<key1> = 'val1';
%h1<key1 three> = 'val1', -238.83;

Special syntax is no longer necessary to access a hash stored in a scalar.

my $h = {key1 => 'val1', 'key-2' => 2, three => -238.83, 4 => 'val3'};
say $h<key1>;

Keys are of type Str or Int by default. The type of the key can be provided.

my %hash{Any}; # same as %hash{*}
class C {};
my %cash{C};
%cash{} = 1;


Associative arrays are supported via just eight simple routines, with no specialised syntax.
Any key can be mapped to any value, and both can be anything (integer|float|string|[nested]sequence, including 0|NULL).

The setd(key,val) procedure is self-explanatory, except for an optional third parameter which is explained below.

The getd(key) function returns the associated data or 0 for non-existent keys: if that might be a valid value see getd_index().

By default, all keys and values are entered into one central dictionary. You can create multiple dictionaries by calling integer tid=new_dict(), and pass that as an additional (final) parameter to the other routines (taking care not to miss any). When you have no further use for it, an entire dictionary can be removed by invoking destroy_dict(tid).

?getd("one") -- shows 1
?getd({3,4}) -- shows {5,"six"}
?getd(2) -- shows "duo"
?getd(2) -- shows 0


$array = array();
$array = []; // Simpler form of array initialization
$array['foo'] = 'bar';
$array['bar'] = 'foo';
echo($array['foo']); // bar
echo($array['moo']); // Undefined index
// Alternative (inline) way
$array2 = array('fruit' => 'apple',
'price' => 12.96,
'colour' => 'green');
// Another alternative (simpler) way
$array2 = ['fruit' => 'apple',
'price' => 12.96,
'colour' => 'green'];
// Check if key exists in the associative array
echo(isset($array['foo'])); // Faster, but returns false if the value of the element is set to null
echo(array_key_exists('foo', $array)); // Slower, but returns true if the value of the element is null

Iterate over key/value[edit]

foreach($array as $key => $value)
echo "Key: $key Value: $value";


Here we use symbol properties. Other possiblities could be index trees or association lists.

(put 'A 'foo 5)
(put 'A 'bar 10)
(put 'A 'baz 15)
(put 'A 'foo 20)
: (get 'A 'bar)
-> 10
: (get 'A 'foo)
-> 20
: (show 'A)
foo 20
bar 10
baz 15


*process source xref attributes or(!);
assocarr: Proc Options(main);
Dcl 1 aa,
2 an Bin Fixed(31) Init(0),
2 pairs(100),
3 key Char(10) Var,
3 val Char(10) Var;
Dcl hi Char(10) Value((high(10)));
Dcl i Bin Fixed(31);
Dcl k Char(10) Var;
Call aadd('1','spam');
Call aadd('2','eggs');
Call aadd('3','foo');
Call aadd('2','spam');
Call aadd('4','spam');
Put Skip(' ');
Put Edit('Iterate over keys')(Skip,a);
Do i=1 To an;
Put Edit('>'!!k!!'< => >'!!aacc(k)!!'<')(Skip,a);
aadd: Proc(k,v);
Dcl (k,v) Char(*) Var;
If aacc(k)^=hi Then
Put Edit('Key >',k,'< would be a duplicate, not added.')
Else Do;
Put Edit('added >'!!k!!'< -> '!!v!!'<')(Skip,a);
aacc: Proc(k) Returns(Char(10) Var);
Dcl k Char(*) Var;
Dcl v Char(10) Var;
Dcl i Bin Fixed(31);
Do i=1 To an;
If key(i)=k Then
added >1< -> spam<
added >2< -> eggs<
added >3< -> foo<
Key >2< would be a duplicate, not added.
added >4< -> spam<

Iterate over keys
>1< => >spam<
>2< => >eggs<
>3< => >foo<
>4< => >spam<  


PL/SQL allows associative arrays defined on two different keys types: Varchar2 or PLS/Integer

Associative Arrays are a PL/SQL only construct. Unlike Oracle Nested Tables or Varrays (the other two types of Oracle collections), associative arrays do not have a corresponding type which can be stored natively in the database. The following code will also show a workaround for this feature.

The following example code is a "record definition", which has nothing to do with associative arrays:-

type ThisIsNotAnAssocArrayType is record (
myShape VARCHAR2(20),
mySize number,
isActive BOOLEAN
assocArray ThisIsNotAnAssocArrayType ;
assocArray.myShape := 'circle';
dbms_output.put_line ('assocArray.myShape: ' || assocArray.myShape);
dbms_output.put_line ('assocArray.mySize: ' || assocArray.mySize);


;;; Create expandable hash table of initial size 50 and with default
;;; value 0 (default value is returned when the item is absent).
vars ht = newmapping([], 50, 0, true);
;;; Set value corresponding to string 'foo'
12 -> ht('foo');
;;; print it
ht('foo') =>
;;; Set value corresponding to vector {1 2 3}
17 -> ht({1 2 3});
;;; print it
ht({1 2 3}) =>
;;; Set value corresponding to number 42 to vector {0 1}
{0 1} -> ht(42);
;;; print it
ht(42) =>
;;; Iterate over keys printing keys and values.
procedure (key, value);
printf(value, '%p\t');
printf(key, '%p\n');


<</a 100 /b 200 /c 300>>
dup /a get =


mydictionary = (red=0xff0000, green=0x00ff00, blue=0x0000ff)
redblue = "purple"
mydictionary put(redblue, 0xff00ff)
255 == mydictionary("blue")
65280 == mydictionary("green")
16711935 == mydictionary("purple")


Am empty hash table can be created with

$hashtable = @{}

A hash table can be initialized with key/value pairs:

$hashtable = @{
"key1" = "value 1"
"key2" = 5

Individual values can be assigned or replaced by either using a property-style access method or indexing into the table with the given key:

$    = "bar"
$hashtable['bar'] = 42
$hashtable."a b" = 3.14 # keys can contain spaces, property-style access needs quotation marks, then
$hashtable[5] = 8 # keys don't need to be strings

Similarly, values can be retrieved using either syntax:

$hashtable.key1     # value 1
$hashtable['key2'] # 5

A shortcut to CREATE an OBJECT with empty fields:

$addressObj= "" | Select-Object nameStr,street1Str,street2Str,cityStr,stateStr,zipStr
$addressObj.nameStr = [string]"FirstName LastName"
$addressObj.street1Str= [string]"1 Main Street"
$addressObj.cityStr= [string]"Washington DC"
$addressObj.stateStr= [string]"DC"
$addressObj.zipStr= [string]"20009"


We use the facts table for this purpose.

?- mymap(key1,V).
V = value1


Hashes are a built-in type called Map in Purebasic.

NewMap dict.s()
dict("country") = "Germany"
Debug dict("country")


Hashes are a built-in type called dictionaries (or mappings) in Python.

hash = dict()  # 'dict' is the dictionary type.
hash = dict(red="FF0000", green="00FF00", blue="0000FF")
hash = { 'key1':1, 'key2':2, }
value = hash[key]

Numerous methods exist for the mapping type

# empty dictionary
d = {}
d['spam'] = 1
d['eggs'] = 2
# dictionaries with two keys
d1 = {'spam': 1, 'eggs': 2}
d2 = dict(spam=1, eggs=2)
# dictionaries from tuple list
d1 = dict([('spam', 1), ('eggs', 2)])
d2 = dict(zip(['spam', 'eggs'], [1, 2]))
# iterating over keys
for key in d:
print key, d[key]
# iterating over (key, value) pairs
for key, value in d.iteritems():
print key, value

Note: Python dictionary keys can be of any arbitrary "hashable" type. The following contains several distinct key value pairs:

myDict = { '1': 'a string', 1: 'an integer', 1.0: 'a floating point number', (1,): 'a tuple' }

(Some other languages such as awk and Perl evaluate all keys such that numerically or lexically equivalent expressions become identical entries in the hash or associative array).

User defined classes which implement the __hash__() special method can also be used as dictionary keys. It's the responsibility of the programmer to ensure the properties of the resultant hash value. The instance object's unique ID (accessible via the id() built-in function) is commonly used for this purpose.


R lacks a native representation of key-value pairs, but different structures allow named elements, which provide similar functionality.

environment example[edit]

> env <- new.env()
> env[["x"]] <- 123
> env[["x"]]
[1] 123
> index <- "1"
> env[[index]] <- "rainfed hay"
> env[[index]]
[1] "rainfed hay"
> env[["1"]]
[1] "rainfed hay"
> env
<environment: 0xb7cd560>
> print(env)
<environment: 0xb7cd560>

vector example[edit]

> x <- c(hello=1, world=2, "!"=3)
> print(x)
hello world     ! 
    1     2     3
> print(names(x))
[1] "hello" "world" "!"
[1] 1 2 3

list example[edit]

> a <- list(a=1, b=2, c=3.14, d="xyz")
> print(a)
[1] 1

[1] 2

[1] 3.14

[1] "xyz"
> print(names(a))
[1] "a" "b" "c" "d"
> print(unname(a))
[1] 1

[1] 2

[1] 3.14

[1] "xyz"


In Racket, hash tables are natively supported and encouraged over association lists in many cases. Data structures that behave like dictionaries support a unified interface.

#lang racket
;; a-lists
(define a-list '((a . 5) (b . 10)))
(assoc a-list 'a) ; => '(a . 5)
;; hash tables
(define table #hash((a . 5) (b . 10)))
(hash-ref table 'a) ; => 5
;; dictionary interface
(dict-ref a-list 'a) ; => 5
(dict-ref table 'a)  ; => 5


{ 'a' 1 'b' 2 'c' 3.14 'd' 'xyz' } as a_hash
a_hash print
hash (4 items)
a => 1
b => 2
c => 3.14
d => "xyz"
a_hash 'c' get # get key 'c'
6.28 a_hash 'c' set # set key 'c'
a_hash.'c' # get key 'c' shorthand
6.28 a_hash:'c' # set key 'c' shorthand

Null is returned for unknown keys.


with hashTable'
hashTable constant table
table %{ first = 100 }%
table %{ second = "hello, world!" keepString %}
table @" first" putn
table @" second" puts


version 1[edit]

Associative arrays are called stem variables in Rexx.

/* Rexx */
key0 = '0'
key1 = 'key0'
stem. = '.' /* Initialize the associative array 'stem' to '.' */
stem.key1 = 'value0' /* Set a specific key/value pair */
Say 'stem.key0= 'stem.key /* Display a value for a key that wasn't set */
Say 'stem.key1= 'stem.key1 /* Display a value for a key that was set */
stem.key0= .
stem.key1= value0

version 2[edit]

/*REXX program shows how to set/display values for an associative array.*/
│ The (below) two REXX statements aren't really necessary, but it │
│ shows how to define any and all entries in a associative array so │
│ that if a "key" is used that isn't defined, it can be displayed to │
│ indicate such, or its value can be checked to determine if a │
│ particular associative array element has been set (defined). │

stateC.=' [not defined yet] ' /*sets any/all state capitols. */
stateN.=' [not defined yet] ' /*sets any/all state names. */
│ In REXX, when a "key" is used, it's normally stored (internally) │
│ as uppercase characters (as in the examples below). Actually, any │
│ characters can be used, including blank(s) and non-displayable │
│ characters (including '00'x, 'ff'x, commas, periods, quotes, ...).│
stateC.nd='Bismarck'  ; stateN.nd='North Dakota''St. Paul'  ;'Minnesota'
stateC.dc='Washington'; stateN.dc='District of Columbia'
stateC.ri='Providence'; stateN.ri='Rhode Island and Providence Plantations'
say 'capital of California is'
say 'capital of Oklahoma is' stateC.ok
say 'capital of' stateN.yyy "is" stateC.yyy
/*stick a fork in it, we're done.*/
capital of California is Sacramento
capital of Oklahoma is  [not defined yet]
capital of Rhode Island and Providence Plantations is Providence


Associative arrays are called lists in RLaB.

x = <<>>; // create an empty list using strings as identifiers. = strtod("0xff0000"); // RLaB doesn't deal with hexadecimal numbers directly. Thus we = strtod("0x00ff00"); // convert it to real numbers using ''strtod'' function. = strtod("0x0000ff");
// print content of a list
for (i in members(x))
{ printf("%8s %06x\n", i, int(x.[i])); } // we have to use ''int'' function to convert reals to integers so "%x" format works
// deleting a key/value
clear (;
// we can also use numeric identifiers in the above example
xid = members(x); // this is a string array
for (i in 1:length(xid))
{ printf("%8s %06x\n", xid[i], int(x.[ xid[i] ])); }
// Finally, we can use numerical identifiers
// Note: ''members'' function orders the list identifiers lexicographically, in other words
// instead of, say, 1,2,3,4,5,6,7,8,9,10,11 ''members'' returns 1,10,11,2,3,4,5,6,7,8,9
x = <<>>; // create an empty list
for (i in 1:5)
{ x.[i] = i; } // assign to the element of list ''i'' the real value equal to i.


A hash object that returns nil for unknown keys

hash[777] # => nil
hash[666] # => 'devil'

A hash object that returns 'unknown key' for unknown keys'unknown key')
hash[777] # => 'unknown key'
hash[666] # => 'devil'

A hash object that returns "unknown key #{key}" for unknown keys{|h,k| "unknown key #{k}"}
hash[777] # => 'unknown key 777'
hash[666] # => 'devil'

A hash object that adds "key #{key} was added at #{}" to the hash the first time an unknown key is seen{|h,k|h[k]="key #{k} was added at #{}"}
hash[777] # => 'key 777 was added at Sun Apr 03 13:49:57 -0700 2011'
hash[555] # => 'key 555 was added at Sun Apr 03 13:50:01 -0700 2011'
hash[777] # => 'key 777 was added at Sun Apr 03 13:49:57 -0700 2011'


use std::collections::HashMap;
fn main() {
let mut olympic_medals = HashMap::new();
olympic_medals.insert("United States", (1072, 859, 749));
olympic_medals.insert("Soviet Union", (473, 376, 355));
olympic_medals.insert("Great Britain", (246, 276, 284));
olympic_medals.insert("Germany", (252, 260, 270));
println!("{:?}", olympic_medals);


class MAIN is
main is
-- creation of a map between strings and integers
map ::= #MAP{STR, INT};
-- add some values
map := map.insert("red", 0xff0000);
map := map.insert("green", 0xff00);
map := map.insert("blue", 0xff);
#OUT + map + "\n"; -- show the map...
-- test if "indexes" exist
#OUT + map.has_ind("red") + "\n";
#OUT + map.has_ind("carpet") + "\n";
-- retrieve a value by index
#OUT + map["green"] + "\n";


// immutable maps
var map = Map(1 -> 2, 3 -> 4, 5 -> 6)
map(3) // 4
map = map + (44 -> 99) // maps are immutable, so we have to assign the result of adding elements
map.isDefinedAt(33) // false
map.isDefinedAt(44) // true
// mutable maps (HashSets)
import scala.collection.mutable.HashMap
val hash = new HashMap[Int, Int]
hash(1) = 2
hash += (1 -> 2) // same as hash(1) = 2
hash += (3 -> 4, 5 -> 6, 44 -> 99)
hash(44) // 99
hash.contains(33) // false
hash.isDefinedAt(33) // same as contains
hash.contains(44) // true
// iterate over key/value
hash.foreach {e => println("key "+e._1+" value "+e._2)} // e is a 2 element Tuple
// same with for syntax
for((k,v) <- hash) println("key " + k + " value " + v)
// items in map where the key is greater than 3
map.filter {k => k._1 > 3} // Map(5 -> 6, 44 -> 99)
// same with for syntax
for((k, v) <- map; if k > 3) yield (k,v)


Scheme has association lists (alists), which are inefficient, ordered maps with arbitrary keys and values.

(define my-dict '((a b) (1 hello) ("c" (a b c)))
(assoc 'a my-dict) ; evaluates to '(a b)

Hash tables are provided by SRFI-69 [1]. Many Scheme implementation also provide native hash tables.

(define my-alist '((a b) (1 hello) ("c" (a b c)))
(define my-hash (alist->hash-table my-alist))

The R6RS standard specifies support for hashtables in the standard libraries document.

(import (rnrs base)
(rnrs hashtables (6)))
(define my-hash (make-hashtable equal-hash equal?))
(hashtable-set! my-hash 'a 'b)
(hashtable-set! my-hash 1 'hello)
(hashtable-set! my-hash "c" '(a b c))


Seed7 uses the type hash to support associative arrays.

$ include "seed7_05.s7i";
# Define hash type
const type: myHashType is hash [string] integer;
# Define hash table
var myHashType: aHash is myHashType.value;
const proc: main is func
var string: stri is "";
var integer: number is 0;
# Add elements
aHash @:= ["foo"] 42;
aHash @:= ["bar"] 100;
# Check presence of an element
if "foo" in aHash then
# Access an element
end if;
# Change an element
aHash @:= ["foo"] 7;
# Remove an element
excl(aHash, "foo");
# Loop over the hash values
for number range aHash do
end for;
# Loop over the hash keys
for key stri range aHash do
end for;
# Loop over hash keys and values
for number key stri range aHash do
writeln("key: " <& stri <& ", value: " <& number);
end for;
end func;


Associative arrays (referred to in SETL terminology as maps) are implemented as sets whose only members are tuples of length 2. Create such a set:

m := {['foo', 'a'], ['bar', 'b'], ['baz', 'c']};

We can then index the set, or map, with the first element of a constituent tuple to return that tuple's second element:

print( m('bar') );

If the map might contain more than one value associated with the same key, we can return the set of them (in this instance a unit set because the keys are in fact unique):

print( m{'bar'} );


var hash =
key1 => 'value1',
key2 => 'value2',
# Add a new key-value pair
hash{:key3} = 'value3';


Dictionary new*, 'MI' -> 'Michigan', 'MN' -> 'Minnesota'


states := Dictionary new.
states at: 'MI' put: 'Michigan'.
states at: 'MN' put: 'Minnesota'.


	t = table()
t<"red"> = "#ff0000"
t<"green"> = "#00ff00"
t<"blue"> = "#0000ff"
output = t<"red">
output = t<"blue">
output = t<"green">


REM Retrieve a KEY VALUE pair
SELECT aa.value_column FROM associative_array aa WHERE aa.key_column = 'KEY';


// make an empty map
var a = [String: Int]()
// or
var b: [String: Int] = [:]
// make an empty map with an initial capacity
var c = [String: Int](minimumCapacity: 42)
// set a value
c["foo"] = 3
// make a map with a literal
var d = ["foo": 2, "bar": 42, "baz": -1]


All arrays in Tcl are associative.

# Create one element at a time:
set hash(foo) 5
# Create in bulk:
array set hash {
foo 5
bar 10
baz 15
# Access one element:
set value $hash(foo)
# Output all values:
foreach key [array names hash] {
puts $hash($key)

Tcl also provides associative map values (called “dictionaries”) from 8.5 onwards.

Works with: Tcl version 8.5
# Create in bulk
set d [dict create foo 5 bar 10 baz 15]
# Create/update one element
dict set d foo 5
# Access one value
set value [dict get $d foo]
# Output all values
dict for {key value} $d {
puts $value
# Alternatively...
foreach value [dict values $d] {
puts $value
# Output the whole dictionary (since it is a Tcl value itself)
puts $d


Toka provides associative arrays via a library.

needs asarray
( create an associative array )
1024 cells is-asarray foo
( store 100 as the "first" element in the array )
100 " first" foo asarray.put
( store 200 as the "second" element in the array )
200 " second" foo asarray.put
( obtain and print the values )
" first" foo asarray.get .
" second" foo asarray.get .

UNIX Shell[edit]

Works with: ksh
typeset -A hash
hash=( [key1]=val1 [key2]=val2 )
echo "${hash[key3]}"
Works with: bash

assigning values is the same as ksh, but to declare the variable as an associative array:

declare -A hash


A key value file can be considered as an associative array

function init() {
cat <<EOF > $map
apple a
boy b
cat c
dog d
elephant e
function put() {
k=$1; v=$2;
del $k
echo $v $k >> $map
function get() {
for v in $(cat $map | grep "$k$"); do
echo $v
function del() {
mv $map $temp
cat $temp | grep -v "$k$" > $map
function dump() {
echo "-- Dump begin --"
cat $map
echo "-- Dump complete --"
get c
put c cow
get c


Library: Gee
using Gee;
void main(){
var map = new HashMap<string, int>(); // creates a HashMap with keys of type string, and values of type int
// two methods to set key,value pair
map["one"] = 1;
map["two"] = 2;
map.set("four", 4);
map.set("five", 5);
// two methods of getting key,value pair
stdout.printf("%d\n", map["one"]);
stdout.printf("%d\n", map.get("two"));
Compile with flag:
 --pkg gee-1.0 


See here in the MSDN the reference for the Dictionary object that can be used in VBA. The following example shows how to create a dictionary, add/remove keys, change a key or a value, and check the existence of a key.

Option Explicit
Sub Test()
Dim h As Object
Set h = CreateObject("Scripting.Dictionary")
h.Add "A", 1
h.Add "B", 2
h.Add "C", 3
Debug.Print h.Item("A")
h.Item("C") = 4
h.Key("C") = "D"
Debug.Print h.exists("C")
h.Remove "B"
Debug.Print h.Count
Debug.Print h.Count
End Sub

Vim Script[edit]

Dictionary keys are always strings.

" Creating a dictionary with some initial values
let dict = {"one": 1, "two": 2}
" Retrieving a value

let two_a = dict["two"]
let two_b = dict.two
let two_c = get(dict, "two", "default value for missing key")
" Modifying a value

let dict["one"] = 1.0
let dict.two = 2.0
" Adding a new value

let dict["three"] = 3
let dict.four = 4
" Removing a value

let one = remove(dict, "one")
unlet dict["two"]
unlet dict.three

Visual FoxPro[edit]

Visual FoxPro has a collection class which can be used for this.

LOCAL loCol As Collection, k, n, o
*!* Example using strings
loCol = NEWOBJECT("Collection")
loCol.Add("Apples", "A")
loCol.Add("Oranges", "O")
loCol.Add("Pears", "P")
n = loCol.Count
? "Items:", n
*!* Loop through the collection
k = 1
 ? o, loCol.GetKey(k)
k = k + 1
*!* Get an item by its key
? loCol("O")
*!* Example using objects
LOCAL loFruits As Collection
loFruits = NEWOBJECT("Collection")
loFruits.Add(CREATEOBJECT("fruit", "Apples"), "A")
loFruits.Add(CREATEOBJECT("fruit", "Oranges"), "O")
loFruits.Add(CREATEOBJECT("fruit", "Pears"), "P")
*!* Loop through the collection
k = 1
 ? o.Name, loFruits.GetKey(k)
k = k + 1
*!* Get an item name by its key
? loFruits("P").Name
DEFINE CLASS fruit As Custom
PROCEDURE Init(tcName As String)
THIS.Name = tcName


h <- (table 'a 1 'b 2)
h 'a
=> 1


XLISP refers to associative arrays as tables. The MAKE-TABLE function returns a new empty table, for instance:

(define starlings (make-table))

Values can then be inserted using TABLE-SET!:

(table-set! starlings "Common starling" "Sturnus vulgaris")
(table-set! starlings "Abbot's starling" "Poeoptera femoralis")
(table-set! starlings "Cape starling" "Lamprotornis nitens")

and retrieved using TABLE-REF with their keys:

(table-ref starlings "Cape starling")

Output in a REPL:

"Lamprotornis nitens"

Other functions provided for tables include MAP-OVER-TABLE-ENTRIES, which takes a table and a function of two arguments and applies the function to each entry (using the key and value as the two arguments), for instance:

(map-over-table-entries starlings (lambda (x y) (print (string-append x " (Linnaean name " y ")"))))

Output in a REPL:

"Abbott's starling (Linnaean name Poeoptera femoralis)" 
"Common starling (Linnaean name Sturnus vulgaris)" 
"Cape starling (Linnaean name Lamprotornis nitens)"


include c:\cxpl\stdlib;
char Dict(10,10);
int Entries;
proc AddEntry(Letter, Greek); \Insert entry into associative array
char Letter, Greek;
[Dict(Entries,0):= Letter;
StrCopy(Greek, @Dict(Entries,1));
Entries:= Entries+1; \(limit checks ignored for simplicity)
func Lookup(Greek); \Given Greek name return English letter
char Greek;
int I;
[for I:= 0, Entries-1 do
if StrCmp(Greek, @Dict(I,1)) = 0 then return Dict(I,0);
return ^?;
[Entries:= 0;
AddEntry(^A, "alpha");
AddEntry(^D, "delta");
AddEntry(^B, "beta");
AddEntry(^C, "gamma");
ChOut(0, Lookup("beta")); CrLf(0);
ChOut(0, Lookup("omega")); CrLf(0);

For greater speed a hashing algorithm should be used to look up items in a large dictionary, however hashing routines are not provided in the standard library.



zkl: Dictionary("one",1, "two",2, "three",3)
zkl: T("one",1, "two",2, "three",3).toDictionary()