Associative array/Creation

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

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


Related tasks:


See also



11l

V dict = [‘key1’ = 1, ‘key2’ = 2]
V value2 = dict[‘key2’]

8th

8th has 'maps' as built-in data types, and can use JSON to describe them:

{ "one" : 1, "two" : "bad" }

Alternatively, they can be created in code:

m:new "one" 1 m:! "two" "bad" m:!

AArch64 Assembly

Works with: as version Raspberry Pi 3B version Buster 64 bits
or android 64 bits with application Termux
/* ARM assembly AARCH64 Raspberry PI 3B or android 64 bits */
/*  program hashmap64.s   */ 

/*******************************************/
/* Constantes file                         */
/*******************************************/
/* for this file see task include a file in language AArch64 assembly*/
.include "../includeConstantesARM64.inc"
.equ MAXI, 10
.equ HEAPSIZE,20000
.equ LIMIT, 10                  // key characters number for compute hash
.equ COEFF, 80                  // filling rate 80 = 80%


/*******************************************/
/* Structures                               */
/********************************************/
/* structure hashMap   */
    .struct  0
hash_count:                       //  stored values counter
    .struct  hash_count + 8
hash_key:                         //  key
    .struct  hash_key + (8 * MAXI)
hash_data:                        // data
    .struct  hash_data + (8 * MAXI)
hash_fin:
/*********************************/
/* Initialized data              */
/*********************************/
.data
szMessFin:          .asciz "End program.\n"
szCarriageReturn:   .asciz "\n"
szMessNoP:          .asciz "Key not found !!!\n"
szKey1:             .asciz "one"
szData1:            .asciz "Ceret"
szKey2:             .asciz "two"
szData2:            .asciz "Maureillas"
szKey3:             .asciz "three"
szData3:            .asciz "Le Perthus"
szKey4:             .asciz "four"
szData4:            .asciz "Le Boulou"

.align 4
iptZoneHeap:    .quad sZoneHeap            // start heap address
iptZoneHeapEnd: .quad sZoneHeap + HEAPSIZE // end heap address
/*********************************/
/* UnInitialized data            */
/*********************************/
.bss
tbHashMap1:       .skip hash_fin         // hashmap
sZoneHeap:        .skip HEAPSIZE         // heap
/*********************************/
/*  code section                 */
/*********************************/
.text
.global main 
main:                          // entry of program 
    
    ldr x0,qAdrtbHashMap1
    bl hashInit                // init hashmap
    ldr x0,qAdrtbHashMap1
    ldr x1,qAdrszKey1          // store key one
    ldr x2,qAdrszData1
    bl hashInsert
    cmp x0,#0                  // error ?
    bne 100f
    ldr x0,qAdrtbHashMap1
    ldr x1,qAdrszKey2         // store key two
    ldr x2,qAdrszData2
    bl hashInsert
    cmp x0,#0
    bne 100f
    
    ldr x0,qAdrtbHashMap1
    ldr x1,qAdrszKey3          // store key three
    ldr x2,qAdrszData3
    bl hashInsert
    cmp x0,#0
    bne 100f
    ldr x0,qAdrtbHashMap1
    ldr x1,qAdrszKey4          // store key four
    ldr x2,qAdrszData4
    bl hashInsert
    cmp x0,#0
    bne 100f
    
    ldr x0,qAdrtbHashMap1
    ldr x1,qAdrszKey2          // remove key two
    bl hashRemoveKey
    cmp x0,#0
    bne 100f
    
    ldr x0,qAdrtbHashMap1
    ldr x1,qAdrszKey1          // search key one
    bl searchKey
    cmp x0,#-1
    beq 1f
    bl affichageMess
    ldr x0,qAdrszCarriageReturn
    bl affichageMess
    b 2f
1:
    ldr x0,qAdrszMessNoP
    bl affichageMess
2:
    ldr x0,qAdrtbHashMap1
    ldr x1,qAdrszKey2          // search key two
    bl searchKey
    cmp x0,#-1
    beq 3f
    bl affichageMess
    ldr x0,qAdrszCarriageReturn
    bl affichageMess
    b 4f
3:
    ldr x0,qAdrszMessNoP
    bl affichageMess
4:
    ldr x0,qAdrtbHashMap1
    ldr x1,qAdrszKey4          // search key four
    bl searchKey
    cmp x0,#-1
    beq 5f
    bl affichageMess
    ldr x0,qAdrszCarriageReturn
    bl affichageMess
    b 6f
5:
    ldr x0,qAdrszMessNoP
    bl affichageMess
6:
    ldr x0,qAdrszMessFin
    bl affichageMess
 
100:                                  // standard end of the program 
    mov x0, #0                        // return code
    mov x8, #EXIT                     // request to exit program
    svc #0                            // perform the system call
 
qAdrszCarriageReturn:     .quad szCarriageReturn
qAdrszMessFin:            .quad szMessFin
qAdrtbHashMap1:           .quad tbHashMap1
qAdrszKey1:               .quad szKey1
qAdrszData1:              .quad szData1
qAdrszKey2:               .quad szKey2
qAdrszData2:              .quad szData2
qAdrszKey3:               .quad szKey3
qAdrszData3:              .quad szData3
qAdrszKey4:               .quad szKey4
qAdrszData4:              .quad szData4
qAdrszMessNoP:            .quad szMessNoP
/***************************************************/
/*     init hashMap               */
/***************************************************/
// x0 contains address to hashMap
hashInit:
    stp x1,lr,[sp,-16]!         // save  registres
    stp x2,x3,[sp,-16]!         // save  registres
    mov x1,#0
    mov x2,#0
    str x2,[x0,#hash_count]      // init counter
    add x0,x0,#hash_key         // start zone key/value
1:
    lsl x3,x1,#3
    add x3,x3,x0
    str x2,[x3,#hash_key]
    str x2,[x3,#hash_data]
    add x1,x1,#1
    cmp x1,#MAXI
    blt 1b
100:
    ldp x2,x3,[sp],16         // restaur des  2 registres
    ldp x1,lr,[sp],16         // restaur des  2 registres
    ret
/***************************************************/
/*     insert key/datas               */
/***************************************************/
// x0 contains address to hashMap
// x1 contains address to key
// x2 contains address to datas
hashInsert:
    stp x1,lr,[sp,-16]!         // save  registres
    stp x2,x3,[sp,-16]!         // save  registres
    stp x4,x5,[sp,-16]!         // save  registres
    stp x6,x7,[sp,-16]!         // save  registres
    mov x6,x0                   // save address 
    bl hashIndex                // search void key or identical key
    cmp x0,#0                   // error ?
    blt 100f
    
    ldr x3,qAdriptZoneHeap
    ldr x3,[x3]
    ldr x7,qAdriptZoneHeapEnd
    ldr x7,[x7]
    sub x7,x7,#50
    lsl x0,x0,#3               // 8 bytes
    add x5,x6,#hash_key        // start zone key/value
    ldr x4,[x5,x0]
    cmp x4,#0                  // key already stored ?
    bne 1f
    ldr x4,[x6,#hash_count]    // no -> increment counter
    add x4,x4,#1
    cmp x4,#(MAXI * COEFF / 100)
    bge 98f
    str x4,[x6,#hash_count]
1:
    str x3,[x5,x0]            // store heap key address in hashmap
    mov x4,#0
2:                            // copy key loop in heap
    ldrb w5,[x1,x4]
    strb w5,[x3,x4]
    cmp w5,#0
    add x4,x4,#1
    bne 2b
    add x3,x3,x4
    cmp x3,x7
    bge 99f
    add x1,x6,#hash_data
    str x3,[x1,x0]           // store heap data address in hashmap
    mov x4,#0
3:                            // copy data loop in heap
    ldrb w5,[x2,x4]
    strb w5,[x3,x4]
    cmp w5,#0
    add x4,x4,#1
    bne 3b
    add x3,x3,x4
    cmp x3,x7
    bge 99f
    ldr x0,qAdriptZoneHeap
    str x3,[x0]               // new heap address
    
    mov x0,#0                 // insertion OK
    b 100f
98:                           // error hashmap
    adr x0,szMessErrInd
    bl affichageMess
    mov x0,#-1
    b 100f
99:                           // error heap
    adr x0,szMessErrHeap
    bl affichageMess
    mov x0,#-1
100:
    ldp x6,x7,[sp],16         // restaur des  2 registres
    ldp x4,x5,[sp],16         // restaur des  2 registres
    ldp x2,x3,[sp],16         // restaur des  2 registres
    ldp x1,lr,[sp],16         // restaur des  2 registres
    ret
szMessErrInd:          .asciz "Error : HashMap size Filling rate Maxi !!\n"
szMessErrHeap:         .asciz "Error : Heap size  Maxi !!\n"
.align 4
qAdriptZoneHeap:       .quad iptZoneHeap
qAdriptZoneHeapEnd:    .quad iptZoneHeapEnd
/***************************************************/
/*     search void index in hashmap              */
/***************************************************/
// x0 contains hashMap address 
// x1 contains key address
hashIndex:
    stp x1,lr,[sp,-16]!         // save  registres
    stp x2,x3,[sp,-16]!         // save  registres
    stp x4,x5,[sp,-16]!         // save  registres
    add x4,x0,#hash_key
    mov x2,#0             // index
    mov x3,#0             // characters sum 
1:                        // loop to compute characters sum 
    ldrb w0,[x1,x2]
    cmp w0,#0             // string end ?
    beq 2f
    add x3,x3,x0          // add to sum
    add x2,x2,#1
    cmp x2,#LIMIT
    blt 1b
2:
    mov x5,x1             // save key address
    mov x0,x3
    mov x1,#MAXI
    udiv x2,x0,x1
    msub x3,x2,x1,x0           // compute remainder -> x3
    mov x1,x5             // key address
    
3:
    ldr x0,[x4,x3,lsl #3] // loak key for computed index 
    cmp x0,#0             // void key ?
    beq 4f 
    bl comparStrings      // identical key ?
    cmp x0,#0
    beq 4f                // yes
    add x3,x3,#1          // no search next void key
    cmp x3,#MAXI          // maxi ?
    csel x3,xzr,x3,ge     // restart to index 0
    b 3b
4:
    mov x0,x3             // return index void array or key equal
100:
    ldp x4,x5,[sp],16         // restaur des  2 registres
    ldp x2,x3,[sp],16         // restaur des  2 registres
    ldp x1,lr,[sp],16         // restaur des  2 registres
    ret

/***************************************************/
/*     search key in hashmap              */
/***************************************************/
// x0 contains hash map address
// x1 contains key address
searchKey:
    stp x1,lr,[sp,-16]!         // save  registres
    stp x2,x3,[sp,-16]!         // save  registres
    mov x2,x0
    bl hashIndex
    lsl x0,x0,#3
    add x1,x0,#hash_key
    ldr x1,[x2,x1]
    cmp x1,#0
    beq 2f
    add x1,x0,#hash_data
    ldr x0,[x2,x1]
    b 100f
2:
    mov x0,#-1
100:
    ldp x2,x3,[sp],16         // restaur des  2 registres
    ldp x1,lr,[sp],16         // restaur des  2 registres
    ret
/***************************************************/
/*     remove key in hashmap              */
/***************************************************/
// x0 contains hash map address
// x1 contains key address
hashRemoveKey:
    stp x1,lr,[sp,-16]!         // save  registres
    stp x2,x3,[sp,-16]!         // save  registres
    mov x2,x0
    bl hashIndex
    lsl x0,x0,#3
    add x1,x0,#hash_key
    ldr x3,[x2,x1]
    cmp x3,#0
    beq 2f
    str xzr,[x2,x1]            // raz key address
    add x1,x0,#hash_data
    str xzr,[x2,x1]            // raz datas address
    mov x0,0
    b 100f
2:
    adr x0,szMessErrRemove
    bl affichageMess
    mov x0,#-1
100:
    ldp x2,x3,[sp],16         // restaur des  2 registres
    ldp x1,lr,[sp],16         // restaur des  2 registres
    ret 
szMessErrRemove:         .asciz "\033[31mError remove key !!\033[0m\n"
.align 4
/************************************/	   
/* Strings case sensitive comparisons  */
/************************************/	  
/* x0 et x1 contains the address of strings */
/* return 0 in x0 if equals */
/* return -1 if string x0 < string x1 */
/* return 1  if string x0 > string x1 */
comparStrings:
    stp x1,lr,[sp,-16]!  // save  registres
    stp x2,x3,[sp,-16]!  // save  registres
    stp x4,x5,[sp,-16]!  // save  registres
    mov x2,#0            // characters counter
1:
    ldrb w3,[x0,x2]      // byte string 1
    ldrb w4,[x1,x2]      // byte string 2
    cmp w3,w4
    blt 2f
    bgt 3f
    cmp w3,#0            // 0 end string ?
    beq 4f
    add x2,x2,#1         // else add 1 in counter
    b 1b                 // and loop
2:
    mov x0,#-1           // smaller
    b 100f
3:
    mov x0,#1            // greather
    b 100f
4:
    mov x0,#0           // equals
100:  
    ldp x4,x5,[sp],16   // restaur des  2 registres
    ldp x2,x3,[sp],16   // restaur des  2 registres
    ldp x1,lr,[sp],16   // restaur des  2 registres
    ret
/********************************************************/
/*        File Include fonctions                        */
/********************************************************/
/* for this file see task include a file in language AArch64 assembly */
.include "../includeARM64.inc"

ActionScript

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.

Ada

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;
begin

   -- 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

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

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

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
main:(

  MODE COLOR = BITS;
  FORMAT color repr = $"16r"16r6d$;

  # This is an associative array which maps strings to ints #
  MODE ITEM = STRUCT(STRING key, COLOR value);
  REF[]ITEM color map items := LOC[0]ITEM;

  PROC color map find = (STRING color)REF COLOR:(
    REF COLOR out;
# 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
      FI
    OD;
      NIL EXIT
    found:
      out
  );

  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 #
    ELSE
      out
    FI
  );
  
  # 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");
  IF value :=: REF COLOR(NIL) THEN
    put(stand error, ("color not found!", new line))
  ELSE
    printf(($"green: "f(color repr)l$, value))
  FI;
  
  # 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))
  OD;

  FORMAT fmt;
  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, $l$))

)
Output:
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

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') );

APL

Works with: Dyalog APL
⍝  Create a namespace ("hash")
      X⎕NS 
      
      ⍝ Assign some names
      X.this'that'
      X.foo88
      
      ⍝  Access the names
      X.this
that
      
      ⍝  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)
      sales.revenue
4109.6
Works with: GNU APL
      ⍝ Assign some names
      X.this'that'
      X.foo88
      
      ⍝  Access the names
      X.this
that

      ⍝  ..or access via 'array index' syntax
      X['this']
that
      
      ⍝  Or do it the array way
      X.(foo)
88

      ⍝ GNU APL does not support multiple assoc. array indices however
      X.(foo this)
VALUE ERROR
      X.(foo this)
             ^

      (sales.prices sales.quantities)  (100 98.4 103.4 110.16) (10 12 8 10)
      sales.revenue  sales.prices +.× sales.quantities
      sales.revenue
4109.6

App Inventor

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.
<VIEW BLOCKS AND ANDROID APP>

ARM Assembly

Works with: as version Raspberry Pi
or android 32 bits with application Termux
/* ARM assembly Raspberry PI  or android 32 bits */
/*  program hashmap.s   */ 

/*      */

/* REMARK 1 : this program use routines in a include file 
   see task Include a file language arm assembly 
   for the routine affichageMess conversion10 
   see at end of this program the instruction include */
/* for constantes see task include a file in arm assembly */
/************************************/
/* Constantes                       */
/************************************/
.include "../constantes.inc"
.equ MAXI, 10                   @ size hashmap
.equ HEAPSIZE,20000
.equ LIMIT, 10                  @ key characters number for compute index
.equ COEFF, 80                  @ filling rate 80 = 80%


/*******************************************/
/* Structures                               */
/********************************************/
/* structure hashMap   */
    .struct  0
hash_count:                       //  stored values counter
    .struct  hash_count + 4
hash_key:                         //  key
    .struct  hash_key + (4 * MAXI)
hash_data:                        // data
    .struct  hash_data + (4 * MAXI)
hash_fin:
/*********************************/
/* Initialized data              */
/*********************************/
.data
szMessFin:          .asciz "End program.\n"
szCarriageReturn:   .asciz "\n"
szMessNoP:          .asciz "Key not found !!!\n"
szKey1:             .asciz "one"
szData1:            .asciz "Ceret"
szKey2:             .asciz "two"
szData2:            .asciz "Maureillas"
szKey3:             .asciz "three"
szData3:            .asciz "Le Perthus"
szKey4:             .asciz "four"
szData4:            .asciz "Le Boulou"

.align 4
iptZoneHeap:    .int sZoneHeap            // start heap address
iptZoneHeapEnd: .int sZoneHeap + HEAPSIZE // end heap address
/*********************************/
/* UnInitialized data            */
/*********************************/
.bss
//sZoneConv:        .skip 24
tbHashMap1:       .skip hash_fin         @ hashmap
sZoneHeap:        .skip HEAPSIZE         @ heap
/*********************************/
/*  code section                 */
/*********************************/
.text
.global main 
main:                          @ entry of program 
    
    ldr r0,iAdrtbHashMap1
    bl hashInit                @ init hashmap
    ldr r0,iAdrtbHashMap1
    ldr r1,iAdrszKey1          @ store key one
    ldr r2,iAdrszData1
    bl hashInsert
    cmp r0,#0                  @ error ?
    bne 100f
    ldr r0,iAdrtbHashMap1
    ldr r1,iAdrszKey2          @ store key two
    ldr r2,iAdrszData2
    bl hashInsert
    cmp r0,#0
    bne 100f
    ldr r0,iAdrtbHashMap1
    ldr r1,iAdrszKey3          @ store key three
    ldr r2,iAdrszData3
    bl hashInsert
    cmp r0,#0
    bne 100f
    ldr r0,iAdrtbHashMap1
    ldr r1,iAdrszKey4          @ store key four
    ldr r2,iAdrszData4
    bl hashInsert
    cmp r0,#0
    bne 100f
    
    ldr r0,iAdrtbHashMap1
    ldr r1,iAdrszKey2          @ remove key two
    bl hashRemoveKey
    cmp r0,#0
    bne 100f
    
    ldr r0,iAdrtbHashMap1
    ldr r1,iAdrszKey1         @ search key
    bl searchKey
    cmp r0,#-1
    beq 1f
    bl affichageMess
    ldr r0,iAdrszCarriageReturn
    bl affichageMess
    b 2f
1:
    ldr r0,iAdrszMessNoP
    bl affichageMess
2:
    ldr r0,iAdrtbHashMap1
    ldr r1,iAdrszKey2
    bl searchKey
    cmp r0,#-1
    beq 3f
    bl affichageMess
    ldr r0,iAdrszCarriageReturn
    bl affichageMess
    b 4f
3:
    ldr r0,iAdrszMessNoP
    bl affichageMess
4:
    ldr r0,iAdrtbHashMap1
    ldr r1,iAdrszKey4
    bl searchKey
    cmp r0,#-1
    beq 5f
    bl affichageMess
    ldr r0,iAdrszCarriageReturn
    bl affichageMess
    b 6f
5:
    ldr r0,iAdrszMessNoP
    bl affichageMess
6:
    ldr r0,iAdrszMessFin
    bl affichageMess
 
100:                                  @ standard end of the program 
    mov r0, #0                        @ return code
    mov r7, #EXIT                     @ request to exit program
    svc #0                            @ perform the system call
 
iAdrszCarriageReturn:     .int szCarriageReturn
iAdrszMessFin:            .int szMessFin
iAdrtbHashMap1:           .int tbHashMap1
iAdrszKey1:               .int szKey1
iAdrszData1:              .int szData1
iAdrszKey2:               .int szKey2
iAdrszData2:              .int szData2
iAdrszKey3:               .int szKey3
iAdrszData3:              .int szData3
iAdrszKey4:               .int szKey4
iAdrszData4:              .int szData4
iAdrszMessNoP:            .int szMessNoP
/***************************************************/
/*     init hashMap               */
/***************************************************/
// r0 contains address to hashMap
hashInit:
    push {r1-r3,lr}             @ save  registers 
    mov r1,#0
    mov r2,#0
    str r2,[r0,#hash_count]      @ init counter
    add r0,r0,#hash_key         @ start zone key/value
1:
    lsl r3,r1,#3
    add r3,r3,r0
    str r2,[r3,#hash_key]
    str r2,[r3,#hash_data]
    add r1,r1,#1
    cmp r1,#MAXI
    blt 1b
100:
    pop {r1-r3,pc}             @ restaur registers
/***************************************************/
/*     insert key/datas               */
/***************************************************/
// r0 contains address to hashMap
// r1 contains address to key
// r2 contains address to datas
hashInsert:
    push {r1-r8,lr}             @ save  registers 
    mov r6,r0                   @ save address 
    bl hashIndex                @ search void key or identical key
    cmp r0,#0                   @ error ?
    blt 100f
    
    ldr r3,iAdriptZoneHeap
    ldr r3,[r3]
    ldr r8,iAdriptZoneHeapEnd
    ldr r8,[r8]
    sub r8,r8,#50
    lsl r0,r0,#2               @ 4 bytes
    add r7,r6,#hash_key        @ start zone key/value
    ldr r4,[r7,r0]
    cmp r4,#0                  @ key already stored ?
    bne 1f
    ldr r4,[r6,#hash_count]    @ no -> increment counter
    add r4,r4,#1
    cmp r4,#(MAXI * COEFF / 100)
    bge 98f
    str r4,[r6,#hash_count]
1:
    str r3,[r7,r0]
    mov r4,#0
2:                            @ copy key loop in heap
    ldrb r5,[r1,r4]
    strb r5,[r3,r4]
    cmp r5,#0
    add r4,r4,#1
    bne 2b
    add r3,r3,r4
    cmp r3,r8
    bge 99f
    add r7,r6,#hash_data
    str r3,[r7,r0]
    mov r4,#0
3:                            @ copy data loop in heap
    ldrb r5,[r2,r4]
    strb r5,[r3,r4]
    cmp r5,#0
    add r4,r4,#1
    bne 3b
    add r3,r3,r4
    cmp r3,r8
    bge 99f
    ldr r0,iAdriptZoneHeap
    str r3,[r0]               @ new heap address
    
    mov r0,#0                 @ insertion OK
    b 100f
98:                           @ error hashmap
    adr r0,szMessErrInd
    bl affichageMess
    mov r0,#-1
    b 100f
99:                           @ error heap
    adr r0,szMessErrHeap
    bl affichageMess
    mov r0,#-1
100:
    pop {r1-r8,lr}             @ restaur registers
    bx lr                      @ return
szMessErrInd:          .asciz "Error : HashMap size Filling rate Maxi !!\n"
szMessErrHeap:         .asciz "Error : Heap size  Maxi !!\n"
.align 4
iAdriptZoneHeap:       .int iptZoneHeap
iAdriptZoneHeapEnd:    .int iptZoneHeapEnd
/***************************************************/
/*     search void index in hashmap              */
/***************************************************/
// r0 contains hashMap address 
// r1 contains key address
hashIndex:
    push {r1-r4,lr}       @ save  registers 
    add r4,r0,#hash_key
    mov r2,#0             @ index
    mov r3,#0             @ characters sum 
1:                        @ loop to compute characters sum 
    ldrb r0,[r1,r2]
    cmp r0,#0             @ string end ?
    beq 2f
    add r3,r3,r0          @ add to sum
    add r2,r2,#1
    cmp r2,#LIMIT
    blt 1b
2:
    mov r5,r1             @ save key address
    mov r0,r3
    mov r1,#MAXI
    bl division           @ compute remainder -> r3
    mov r1,r5             @ key address
    
3:
    ldr r0,[r4,r3,lsl #2] @ loak key for computed index 
    cmp r0,#0             @ void key ?
    beq 4f 
    bl comparStrings      @ identical key ?
    cmp r0,#0
    beq 4f                @ yes
    add r3,r3,#1          @ no search next void key
    cmp r3,#MAXI          @ maxi ?
    movge r3,#0           @ restart to index 0
    b 3b
4:
    mov r0,r3             @ return index void array or key equal
100:
    pop {r1-r4,pc}        @ restaur registers

/***************************************************/
/*     search key in hashmap              */
/***************************************************/
// r0 contains hash map address
// r1 contains key address
searchKey:
    push {r1-r2,lr}           @ save  registers 
    mov r2,r0
    bl hashIndex
    lsl r0,r0,#2
    add r1,r0,#hash_key
    ldr r1,[r2,r1]
    cmp r1,#0
    moveq r0,#-1
    beq 100f
    add r1,r0,#hash_data
    ldr r0,[r2,r1]
100:
    pop {r1-r2,pc}            @ restaur registers
/***************************************************/
/*     remove key in hashmap              */
/***************************************************/
// r0 contains hash map address
// r1 contains key address
hashRemoveKey:                  @ INFO: hashRemoveKey
    push {r1-r3,lr}             @ save  registers 
    mov r2,r0
    bl hashIndex
    lsl r0,r0,#2
    add r1,r0,#hash_key
    ldr r3,[r2,r1]
    cmp r3,#0
    beq  2f
    add r3,r2,r1
    mov r1,#0                   @ raz key address
    str r1,[r3]   
    add r1,r0,#hash_data
    add r3,r2,r1
    mov r1,#0
    str r1,[r3]                 @ raz datas address
    mov r0,#0
    b 100f
2:
    adr r0,szMessErrRemove
    bl affichageMess
    mov r0,#-1
100:
    pop {r1-r3,pc}            @ restaur registers
szMessErrRemove:         .asciz "\033[31mError remove key !!\033[0m\n"
.align 4
/************************************/   
/* Strings case sensitive comparisons  */
/************************************/	  
/* r0 et r1 contains the address of strings */
/* return 0 in r0 if equals */
/* return -1 if string r0 < string r1 */
/* return 1  if string r0 > string r1 */
comparStrings:
    push {r1-r4}         @ save des registres
    mov r2,#0            @ characters counter
1:
    ldrb r3,[r0,r2]      @ byte string 1
    ldrb r4,[r1,r2]      @ byte string 2
    cmp r3,r4
    movlt r0,#-1         @ smaller
    movgt r0,#1	         @ greather
    bne 100f             @ not equals
    cmp r3,#0            @ 0 end string ?
    moveq r0,#0          @ equals
    beq 100f             @ end string
    add r2,r2,#1         @ else add 1 in counter
    b 1b                 @ and loop
100:
    pop {r1-r4}
    bx lr   
/***************************************************/
/*      ROUTINES INCLUDE                           */
/***************************************************/
.include "../affichage.inc"

Arturo

; create a dictionary
d: #[
	name: 		"john"
	surname: 	"doe"
	age: 		34
]

print d
Output:
[name:john surname:doe age:34]

ATS

ATS, like many languages, does not have high level stuff such as maps built into the language itself, so library support or one's own code is what one needs.

In the following, persistence means that past values of the map are retained, if they continue to have references to them. This property sometimes is called immutability, but I feel that term implies too much.

Persistent association lists

The following implementation includes set, get, and delete, and also "generators", which are like iterators except they are closures rather than regular objects.

(*------------------------------------------------------------------*)

#define ATS_DYNLOADFLAG 0

#include "share/atspre_staload.hats"

(*------------------------------------------------------------------*)
(* Interface                                                        *)

(* You can put the interface in a .sats file. You will have to remove
   the word "extern". *)

typedef alist_t (key_t  : t@ype+,
                 data_t : t@ype+,
                 size   : int) =
  list (@(key_t, data_t), size)
typedef alist_t (key_t  : t@ype+,
                 data_t : t@ype+) =
  [size : int]
  alist_t (key_t, data_t, size)

extern prfun
lemma_alist_t_param :
  {size  : int} {key_t : t@ype} {data_t : t@ype}
  alist_t (key_t, data_t, size) -<prf> [0 <= size] void

extern fun {key_t : t@ype}     (* Implement key equality with this. *)
alist_t$key_eq : (key_t, key_t) -<> bool

(* alist_t_nil: create an empty association list. *)
extern fun
alist_t_nil :
  {key_t : t@ype} {data_t : t@ype}
  () -<> alist_t (key_t, data_t, 0)

(* alist_t_set: add an association, deleting old associations with an
   equal key. *)
extern fun {key_t  : t@ype}
           {data_t : t@ype}
alist_t_set {size : int}
            (alst : alist_t (key_t, data_t, size),
             key  : key_t,
             data : data_t) :<>
    [sz : int | 1 <= sz]
    alist_t (key_t, data_t, sz)

(* alist_t_get: find an association and return its data, if
   present. *)
extern fun {key_t  : t@ype}
           {data_t : t@ype}
alist_t_get {size : int}
            (alst : alist_t (key_t, data_t, size),
             key  : key_t) :<>
    Option data_t

(* alist_t_delete: delete all associations with key. *)
extern fun {key_t  : t@ype}
           {data_t : t@ype}
alist_t_delete {size : int}
               (alst : alist_t (key_t, data_t, size),
                key  : key_t ) :<>
    [sz : int | 0 <= sz]
    alist_t (key_t, data_t, sz)

(* alist_t_make_pairs_generator: make a closure that returns
   the association pairs, one by one. This is a form of iterator.
   Analogous generators can be made for the keys or data values
   alone. *)
extern fun {key_t  : t@ype}
           {data_t : t@ype}
alist_t_make_pairs_generator
          {size : int}
          (alst : alist_t (key_t, data_t, size)) :<!wrt>
    () -<cloref,!refwrt> Option @(key_t, data_t)

(*------------------------------------------------------------------*)
(* Implementation                                                   *)

#define NIL list_nil ()
#define :: list_cons

primplement
lemma_alist_t_param alst =
  lemma_list_param alst

implement
alist_t_nil () =
  NIL

implement {key_t} {data_t}
alist_t_set (alst, key, data) =
  @(key, data) :: alist_t_delete (alst, key)

implement {key_t} {data_t}
alist_t_get (alst, key) =
  let
    fun
    loop {n : nat}
         .<n>.                  (* <-- proof of termination *)
         (lst : alist_t (key_t, data_t, n)) :<>
        Option data_t =
      case+ lst of
      | NIL => None ()
      | head :: tail =>
        if alist_t$key_eq (key, head.0) then
          Some (head.1)
        else
          loop tail

    prval _ = lemma_alist_t_param alst
  in
    loop alst
  end

implement {key_t} {data_t}
alist_t_delete (alst, key) =
  let
    fun
    delete {n : nat}
           .<n>.                (* <-- proof of termination *)
           (lst : alist_t (key_t, data_t, n)) :<>
        [m : nat] alist_t (key_t, data_t, m) =
      (* This implementation is *not* tail recursive, but has the
         minor advantage of preserving the order of entries without
         doing a lot of work. *)
      case+ lst of
      | NIL => lst
      | head :: tail =>
        if alist_t$key_eq (key, head.0) then
          delete tail
        else
          head :: delete tail

    prval _ = lemma_alist_t_param alst
  in
    delete alst
  end

implement {key_t} {data_t}
alist_t_make_pairs_generator alst =
  let
    typedef alist_t = [sz : int] alist_t (key_t, data_t, sz)

    val alst_ref = ref alst

    (* Cast the ref to a pointer so it can be enclosed in the
       closure. *)
    val alst_ptr = $UNSAFE.castvwtp0{ptr} alst_ref
  in
    lam () =>
      let
        val alst_ref = $UNSAFE.castvwtp0{ref alist_t} alst_ptr
      in
        case+ !alst_ref of
        | NIL => None ()
        | head :: tail =>
          begin
            !alst_ref := tail;
            (* For a keys generator, change the following line to
               "Some (head.0)"; for a data values generator, change
               it to "Some (head.1)". *)
            Some head
          end
      end
  end

(*------------------------------------------------------------------*)
(* Demonstration program                                            *)

implement
alist_t$key_eq<string> (s, t) =
  s = t

typedef s2i_alist_t = alist_t (string, int)

fn
s2i_alist_t_set (map  : s2i_alist_t,
                 key  : string,
                 data : int) :<> s2i_alist_t =
  alist_t_set<string><int> (map, key, data)

fn
s2i_alist_t_set_ref (map  : &s2i_alist_t >> _,
                     key  : string,
                     data : int) :<!wrt> void =
  (* Update a reference to a persistent alist. *)
  map := s2i_alist_t_set (map, key, data)

fn
s2i_alist_t_get (map  : s2i_alist_t,
                 key  : string) :<> Option int =
  alist_t_get<string><int> (map, key)

extern fun {}        (* {} = a template without template parameters *)
s2i_alist_t_get_dflt$dflt :<> () -> int
fn {}                (* {} = a template without template parameters *)
s2i_alist_t_get_dflt (map  : s2i_alist_t,
                      key  : string) : int =
  case+ s2i_alist_t_get (map, key) of
  | Some x => x
  | None () => s2i_alist_t_get_dflt$dflt<> ()

overload [] with s2i_alist_t_set_ref
overload [] with s2i_alist_t_get_dflt

implement
main0 () =
  let
    implement s2i_alist_t_get_dflt$dflt<> () = 0
    var map = alist_t_nil ()
    var gen : () -<cloref1> @(string, int)
    var pair : Option @(string, int)
  in
    map["one"] := 1;
    map["two"] := 2;
    map["three"] := 3;
    println! ("map[\"one\"] = ", map["one"]);
    println! ("map[\"two\"] = ", map["two"]);
    println! ("map[\"three\"] = ", map["three"]);
    println! ("map[\"four\"] = ", map["four"]);
    gen := alist_t_make_pairs_generator<string><int> map;
    for (pair := gen (); option_is_some pair; pair := gen ())
      println! (pair)
  end

(*------------------------------------------------------------------*)
Output:
$ patscc -O2 -DATS_MEMALLOC_GCBDW association_lists-postiats.dats -lgc && ./a.out
map["one"] = 1
map["two"] = 2
map["three"] = 3
map["four"] = 0
Some((three, 3))
Some((two, 2))
Some((one, 1))

Persistent hashmaps based on AVL trees

Translation of: Scheme
Library: xxHash

The following implementation includes set and get, and also "generators", which are like iterators except they are closures rather than regular objects. A delete function could easily be added.

To run this demonstration you need the xxHash C library. I have written an implementation of SpookyHash in ATS, but using a C library is fine, and requires less ATS code.

Because the hash table is an AVL tree, performance will tend to be logarithmic. This assumes a good hash function, of course. With a bad hash function, performance may be linear.

(*------------------------------------------------------------------*)

#define ATS_DYNLOADFLAG 0

#include "share/atspre_staload.hats"

(*------------------------------------------------------------------*)
(* String hashing using XXH3_64bits from the xxHash suite.          *)

#define ATS_EXTERN_PREFIX "hashmaps_postiats_"

%{^ /* Embedded C code. */

#include <xxhash.h>

ATSinline() atstype_uint64
hashmaps_postiats_mem_hash (atstype_ptr data, atstype_size len)
{
  return (atstype_uint64) XXH3_64bits (data, len);
}

%}

extern fn mem_hash : (ptr, size_t) -<> uint64 = "mac#%"

fn
string_hash (s : string) :<> uint64 =
  let
    val len = string_length s
  in
    mem_hash ($UNSAFE.cast{ptr} s, len)
  end

(*------------------------------------------------------------------*)
(* A trimmed down version of the AVL trees from the AVL Tree task.  *)

datatype bal_t =
| bal_minus1
| bal_zero
| bal_plus1

datatype avl_t (key_t  : t@ype+,
                data_t : t@ype+,
                size   : int) =
| avl_t_nil (key_t, data_t, 0)
| {size_L, size_R : nat}
  avl_t_cons (key_t, data_t, size_L + size_R + 1) of
    (key_t, data_t, bal_t,
     avl_t (key_t, data_t, size_L),
     avl_t (key_t, data_t, size_R))
typedef avl_t (key_t  : t@ype+,
               data_t : t@ype+) =
  [size : int] avl_t (key_t, data_t, size)

extern fun {key_t : t@ype}
avl_t$compare (u : key_t, v : key_t) :<> int

#define NIL avl_t_nil ()
#define CONS avl_t_cons
#define LNIL list_nil ()
#define :: list_cons
#define F false
#define T true

typedef fixbal_t = bool

prfn
lemma_avl_t_param {key_t : t@ype} {data_t : t@ype} {size : int}
                  (avl : avl_t (key_t, data_t, size)) :<prf>
    [0 <= size] void =
  case+ avl of NIL => () | CONS _ => ()

fn {}
minus_neg_bal (bal : bal_t) :<> bal_t =
  case+ bal of
  | bal_minus1 () => bal_plus1
  | _ => bal_zero ()

fn {}
minus_pos_bal (bal : bal_t) :<> bal_t =
  case+ bal of
  | bal_plus1 () => bal_minus1
  | _ => bal_zero ()

fn
avl_t_is_empty {key_t : t@ype} {data_t : t@ype} {size   : int}
               (avl : avl_t (key_t, data_t, size)) :<>
    [b : bool | b == (size == 0)] bool b =
  case+ avl of
  | NIL => T
  | CONS _ => F

fn
avl_t_isnot_empty {key_t : t@ype} {data_t : t@ype} {size   : int}
                  (avl : avl_t (key_t, data_t, size)) :<>
    [b : bool | b == (size <> 0)] bool b =
  ~avl_t_is_empty avl

fn {key_t : t@ype} {data_t : t@ype}
avl_t_search_ref {size  : int}
                 (avl   : avl_t (key_t, data_t, size),
                  key   : key_t,
                  data  : &data_t? >> opt (data_t, found),
                  found : &bool? >> bool found) :<!wrt>
    #[found : bool] void =
  let
    fun
    search (p     : avl_t (key_t, data_t),
            data  : &data_t? >> opt (data_t, found),
            found : &bool? >> bool found) :<!wrt,!ntm>
        #[found : bool] void =
      case+ p of
      | NIL =>
        {
          prval _ = opt_none {data_t} data
          val _ = found := F
        }
      | CONS (k, d, _, left, right) =>
        begin
          case+ avl_t$compare<key_t> (key, k) of
          | cmp when cmp < 0 => search (left, data, found)
          | cmp when cmp > 0 => search (right, data, found)
          | _ =>
            {
              val _ = data := d
              prval _ = opt_some {data_t} data
              val _ = found := T
            }
        end
  in
    $effmask_ntm search (avl, data, found)
  end

fn {key_t : t@ype} {data_t : t@ype}
avl_t_search_opt {size : int}
                 (avl  : avl_t (key_t, data_t, size),
                  key  : key_t) :<>
    Option (data_t) =
  let
    var data : data_t?
    var found : bool?
    val _ = $effmask_wrt avl_t_search_ref (avl, key, data, found)
  in
    if found then
      let
        prval _ = opt_unsome data
      in
        Some {data_t} data
      end
    else
      let
        prval _ = opt_unnone data
      in
        None {data_t} ()
      end
  end

fn {key_t : t@ype} {data_t : t@ype}
avl_t_insert_or_replace {size : int}
                        (avl  : avl_t (key_t, data_t, size),
                         key  : key_t,
                         data : data_t) :<>
    [sz : pos] (avl_t (key_t, data_t, sz), bool) =
  let
    fun
    search {size   : nat}
           (p      : avl_t (key_t, data_t, size),
            fixbal : fixbal_t,
            found  : bool) :<!ntm>
        [sz : pos]
        (avl_t (key_t, data_t, sz), fixbal_t, bool) =
      case+ p of
      | NIL => (CONS (key, data, bal_zero, NIL, NIL), T, F)
      | CONS (k, d, bal, left, right) =>
        case+ avl_t$compare<key_t> (key, k) of
        | cmp when cmp < 0 =>
          let
            val (p1, fixbal, found) = search (left, fixbal, found)
          in
            case+ (fixbal, bal) of
            | (F, _) => (CONS (k, d, bal, p1, right), F, found)
            | (T, bal_plus1 ()) =>
              (CONS (k, d, bal_zero (), p1, right), F, found)
            | (T, bal_zero ()) =>
              (CONS (k, d, bal_minus1 (), p1, right), fixbal, found)
            | (T, bal_minus1 ()) =>
              let
                val+ CONS (k1, d1, bal1, left1, right1) = p1
              in
                case+ bal1 of
                | bal_minus1 () =>
                  let
                    val q = CONS (k, d, bal_zero (), right1, right)
                    val q1 = CONS (k1, d1, bal_zero (), left1, q)
                  in
                    (q1, F, found)
                  end
                | _ =>
                  let
                    val p2 = right1
                    val- CONS (k2, d2, bal2, left2, right2) = p2
                    val q = CONS (k, d, minus_neg_bal bal2,
                                  right2, right)
                    val q1 = CONS (k1, d1, minus_pos_bal bal2,
                                   left1, left2)
                    val q2 = CONS (k2, d2, bal_zero (), q1, q)
                  in
                    (q2, F, found)
                  end
              end
          end
        | cmp when cmp > 0 =>
          let
            val (p1, fixbal, found) = search (right, fixbal, found)
          in
            case+ (fixbal, bal) of
            | (F, _) => (CONS (k, d, bal, left, p1), F, found)
            | (T, bal_minus1 ()) =>
              (CONS (k, d, bal_zero (), left, p1), F, found)
            | (T, bal_zero ()) =>
              (CONS (k, d, bal_plus1 (), left, p1), fixbal, found)
            | (T, bal_plus1 ()) =>
              let
                val+ CONS (k1, d1, bal1, left1, right1) = p1
              in
                case+ bal1 of
                | bal_plus1 () =>
                  let
                    val q = CONS (k, d, bal_zero (), left, left1)
                    val q1 = CONS (k1, d1, bal_zero (), q, right1)
                  in
                    (q1, F, found)
                  end
                | _ =>
                  let
                    val p2 = left1
                    val- CONS (k2, d2, bal2, left2, right2) = p2
                    val q = CONS (k, d, minus_pos_bal bal2,
                                  left, left2)
                    val q1 = CONS (k1, d1, minus_neg_bal bal2,
                                   right2, right1)
                    val q2 = CONS (k2, d2, bal_zero (), q, q1)
                  in
                    (q2, F, found)
                  end
              end
          end
        | _ => (CONS (key, data, bal, left, right), F, T)
  in
    if avl_t_is_empty avl then
      (CONS (key, data, bal_zero, NIL, NIL), F)
    else
      let
        prval _ = lemma_avl_t_param avl
        val (avl, _, found) = $effmask_ntm search (avl, F, F)
      in
        (avl, found)
      end
  end

fn {key_t : t@ype} {data_t : t@ype}
avl_t_insert {size : int}
             (avl  : avl_t (key_t, data_t, size),
              key  : key_t,
              data : data_t) :<>
    [sz : pos] avl_t (key_t, data_t, sz) =
  (avl_t_insert_or_replace<key_t><data_t> (avl, key, data)).0

fun {key_t : t@ype} {data_t : t@ype}
push_all_the_way_left (stack : List (avl_t (key_t, data_t)),
                       p     : avl_t (key_t, data_t)) :
    List0 (avl_t (key_t, data_t)) =
  let
    prval _ = lemma_list_param stack
  in
    case+ p of
    | NIL => stack
    | CONS (_, _, _, left, _) =>
      push_all_the_way_left (p :: stack, left)
  end

fun {key_t : t@ype} {data_t : t@ype}
update_generator_stack (stack     : List (avl_t (key_t, data_t)),
                        right     : avl_t (key_t, data_t)) :
    List0 (avl_t (key_t, data_t)) =
  let
    prval _ = lemma_list_param stack
  in
    if avl_t_is_empty right then
      stack
    else
      push_all_the_way_left<key_t><data_t> (stack, right)
  end

fn {key_t : t@ype} {data_t : t@ype}
avl_t_make_data_generator {size : int}
                          (avl  : avl_t (key_t, data_t, size)) :
    () -<cloref1> Option data_t =
  let
    typedef avl_t = avl_t (key_t, data_t)

    val stack = push_all_the_way_left<key_t><data_t> (LNIL, avl)
    val stack_ref = ref stack

    (* Cast stack_ref to its (otherwise untyped) pointer, so it can be
       enclosed within ‘generate’. *)
    val p_stack_ref = $UNSAFE.castvwtp0{ptr} stack_ref

    fun
    generate () :<cloref1> Option data_t =
      let
        (* Restore the type information for stack_ref. *)
        val stack_ref =
          $UNSAFE.castvwtp0{ref (List avl_t)} p_stack_ref

        var stack : List0 avl_t = !stack_ref
        var retval : Option data_t
      in
        begin
          case+ stack of
          | LNIL => retval := None ()
          | p :: tail =>
            let
              val- CONS (_, d, _, left, right) = p
            in
              retval := Some d;
              stack :=
                update_generator_stack<key_t><data_t> (tail, right)
            end
        end;
        !stack_ref := stack;
        retval
      end
  in
    generate
  end

(*------------------------------------------------------------------*)
(* Hashmaps implemented with AVL trees and association lists.       *)

(* The interface  - - - - - - - - - - - - - - - - - - - - - - - - - *)

typedef hashmap_t (key_t  : t@ype+,
                   data_t : t@ype+) =
  avl_t (uint64, List1 @(key_t, data_t))

(* For simplicity, let us support only 64-bit hashes. *)
extern fun {key_t : t@ype}  (* Implement a hash function with this. *)
hashmap_t$hashfunc : key_t -<> uint64

extern fun {key_t : t@ype}     (* Implement key equality with this. *)
hashmap_t$key_eq : (key_t, key_t) -<> bool

extern fun
hashmap_t_nil :
  {key_t : t@ype} {data_t : t@ype}
  () -<> hashmap_t (key_t, data_t)

extern fun {key_t  : t@ype}
           {data_t : t@ype}
hashmap_t_set (map  : hashmap_t (key_t, data_t),
               key  : key_t,
               data : data_t) :<>
    hashmap_t (key_t, data_t)

extern fun {key_t  : t@ype}
           {data_t : t@ype}
hashmap_t_get (map : hashmap_t (key_t, data_t),
               key : key_t) :<>
    Option data_t

(*
  Notes:
    * Generators for hashmap_t produce their output in unspecified
      order.
    * Generators for keys and data values can be made by analogy to
      the following generator for pairs, or can be written in terms
      of the generator for pairs. (The former approach seems better;
      it might copy less data.)
*)
extern fun {key_t  : t@ype}
           {data_t : t@ype}
hashmap_t_make_pairs_generator (map : hashmap_t (key_t, data_t)) :
    () -<cloref1> Option @(key_t, data_t)

(* The implementation - - - - - - - - - - - - - - - - - - - - - - - *)

implement
avl_t$compare<uint64> (u, v) =
  if u < v then
    ~1
  else if v < u then
    1
  else
    0

implement
hashmap_t_nil () =
  avl_t_nil ()

fun {key_t  : t@ype}
    {data_t : t@ype}
remove_association {n : nat} .<n>.
                   (lst : list (@(key_t, data_t), n),
                    key : key_t) :<>
    List0 @(key_t, data_t) =
  (* This implementation uses linear stack space, and so presumes the
     list is not extremely long. It preserves the order of the list,
     although doing so is not necessary for persistence. (You might
     wish to think about that, taking into account that the
     order of traversal through a hashmap usually is considered
     "unspecified".) *)
  case+ lst of
  | list_nil () => lst
  | list_cons (head, tail) =>
    if hashmap_t$key_eq<key_t> (key, head.0) then
      tail                      (* Assume there is only one match. *)
    else
      list_cons (head, remove_association (tail, key))

fun {key_t  : t@ype}
    {data_t : t@ype}
find_association {n : nat} .<n>.
                 (lst : list (@(key_t, data_t), n),
                  key : key_t) :<>
    List0 @(key_t, data_t) =
  (* This implementation is tail recursive. It will not build up the
     stack. *)
  case+ lst of
  | list_nil () => lst
  | list_cons (head, tail) =>
    if hashmap_t$key_eq<key_t> (key, head.0) then
      lst
    else
      find_association (tail, key)

implement {key_t} {data_t}
hashmap_t_set (map, key, data) =
  let
    typedef lst_t = List1 @(key_t, data_t) (* Association list. *)
    val hash = hashmap_t$hashfunc<key_t> key
    val lst_opt = avl_t_search_opt<uint64><lst_t> (map, hash)
    val lst =
      begin
        case+ lst_opt of
        | Some lst =>
          (* There is already an association list for this hash value.
             Remove any association already in it. *)
          remove_association<key_t><data_t> (lst, key)
        | None () =>
          (* Start a new association list. *)
          list_nil ()
      end : List0 @(key_t, data_t)
    val lst = list_cons (@(key, data), lst)
  in
    avl_t_insert<uint64><lst_t> (map, hash, lst)
  end

implement {key_t} {data_t}
hashmap_t_get (map, key) =
  let
    typedef lst_t = List1 @(key_t, data_t) (* Association list. *)
    val hash = hashmap_t$hashfunc<key_t> key
    val lst_opt = avl_t_search_opt<uint64><lst_t> (map, hash)
  in
    case+ lst_opt of
    | None () => None{data_t} ()
    | Some lst =>
      begin
        case+ find_association<key_t><data_t> (lst, key) of
        | list_nil () => None{data_t} ()
        | list_cons (@(_, data), _) => Some{data_t} data
      end
  end

implement {key_t} {data_t}
hashmap_t_make_pairs_generator (map) =
  let
    typedef pair_t = @(key_t, data_t)
    typedef lst_t = List1 pair_t
    typedef lst_t_0 = List0 pair_t

    val avl_gen = avl_t_make_data_generator<uint64><lst_t> (map)

    val current_alist_ref : ref lst_t_0 = ref (list_nil ())
    val current_alist_ptr =
      $UNSAFE.castvwtp0{ptr} current_alist_ref
  in
    lam () =>
      let
        val current_alist_ref =
          $UNSAFE.castvwtp0{ref lst_t_0} current_alist_ptr
      in
        case+ !current_alist_ref of
        | list_nil () =>
          begin
            case+ avl_gen () of
            | None () => None ()
            | Some lst =>
              begin
                case+ lst of
                | list_cons (head, tail) =>
                  begin
                    !current_alist_ref := tail;
                    Some head
                  end
              end
          end
        | list_cons (head, tail) =>
          begin
            !current_alist_ref := tail;
            Some head
          end
      end
  end

(*------------------------------------------------------------------*)

implement
hashmap_t$hashfunc<string> (s) =
  string_hash s

implement
hashmap_t$key_eq<string> (s, t) =
  s = t

typedef s2i_hashmap_t = hashmap_t (string, int)

fn
s2i_hashmap_t_set (map  : s2i_hashmap_t,
                   key  : string,
                   data : int) :<> s2i_hashmap_t =
  hashmap_t_set<string><int> (map, key, data)

fn
s2i_hashmap_t_set_ref (map  : &s2i_hashmap_t >> _,
                       key  : string,
                       data : int) :<!wrt> void =
  (* Update a reference to a persistent hashmap. *)
  map := s2i_hashmap_t_set (map, key, data)

fn
s2i_hashmap_t_get (map  : s2i_hashmap_t,
                   key  : string) :<> Option int =
  hashmap_t_get<string><int> (map, key)

extern fun {}        (* {} = a template without template parameters *)
s2i_hashmap_t_get_dflt$dflt :<> () -> int
fn {}                (* {} = a template without template parameters *)
s2i_hashmap_t_get_dflt (map  : s2i_hashmap_t,
                        key  : string) : int =
  case+ s2i_hashmap_t_get (map, key) of
  | Some x => x
  | None () => s2i_hashmap_t_get_dflt$dflt<> ()

overload [] with s2i_hashmap_t_set_ref
overload [] with s2i_hashmap_t_get_dflt

implement
main0 () =
  let
    implement s2i_hashmap_t_get_dflt$dflt<> () = 0
    var map = hashmap_t_nil ()
    var gen : () -<cloref1> @(string, int)
    var pair : Option @(string, int)
  in
    map["one"] := 1;
    map["two"] := 2;
    map["three"] := 3;
    println! ("map[\"one\"] = ", map["one"]);
    println! ("map[\"two\"] = ", map["two"]);
    println! ("map[\"three\"] = ", map["three"]);
    println! ("map[\"four\"] = ", map["four"]);
    gen := hashmap_t_make_pairs_generator<string><int> map;
    for (pair := gen (); option_is_some pair; pair := gen ())
      println! (pair)
  end

(*------------------------------------------------------------------*)
Output:
$ patscc -O2 -DATS_MEMALLOC_GCBDW hashmaps-postiats.dats -lxxhash -lgc && ./a.out
map["one"] = 1
map["two"] = 2
map["three"] = 3
map["four"] = 0
Some((two, 2))
Some((one, 1))
Some((three, 3))

AutoHotkey

True arrays

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

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%

AutoIt

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.

; Associative arrays in AutoIt.
; All the required functions are below the examples.

; Initialize an error handler to deal with any COM errors..
global $oMyError = ObjEvent("AutoIt.Error", "AAError")

; first example, simple.
global $simple

; Initialize your array ...
AAInit($simple)

AAAdd($simple, "Appple", "fruit")
AAAdd($simple, "Dog", "animal")
AAAdd($simple, "Silicon", "tetravalent metalloid semiconductor")

ConsoleWrite("It is well-known that Silicon is a " & AAGetItem($simple, "Silicon") & "." & @CRLF)
ConsoleWrite(@CRLF)


; A more interesting example..

$ini_path = "AA_Test.ini"
; Put this prefs section in your ini file..
; [test]
; foo=foo value
; foo2=foo2 value
; bar=bar value
; bar2=bar2 value


global $associative_array
AAInit($associative_array)

; We are going to convert this 2D array into a cute associative array where we
; can access the values by simply using their respective key names..
$test_array = IniReadSection($ini_path, "test")

for $z = 1 to 2 ; do it twice, to show that the items are *really* there!
	for $i = 1 to $test_array[0][0]
		$key_name = $test_array[$i][0]
		ConsoleWrite("Adding '" & $key_name & "'.." & @CRLF)
		; key already exists in "$associative_array", use the pre-determined value..
		if AAExists($associative_array, $key_name) then
			$this_value = AAGetItem($associative_array, $key_name)
			ConsoleWrite("key_name ALREADY EXISTS! : =>" & $key_name & "<=" & @CRLF)
		else
			$this_value = $test_array[$i][1]
			; store left=right value pair in AA
			if $this_value then
				AAAdd($associative_array, $key_name, $this_value)
			endif
		endif
	next
next

ConsoleWrite(@CRLF & "Array Count: =>" & AACount($associative_array) & "<=" & @CRLF)
AAList($associative_array)

ConsoleWrite(@CRLF & "Removing 'foo'..")
AARemove($associative_array, "foo")

ConsoleWrite(@CRLF & "Array Count: =>" & AACount($associative_array) & "<=" & @CRLF)
AAList($associative_array)


AAWipe($associative_array)


; end



func AAInit(ByRef $dict_obj)
$dict_obj = ObjCreate("Scripting.Dictionary")
endfunc

; Adds a key and item pair to a Dictionary object..
func AAAdd(ByRef $dict_obj, $key, $val)
    $dict_obj.Add($key, $val)
    If @error Then return SetError(1, 1, -1)
endfunc

; Removes a key and item pair from a Dictionary object..
func AARemove(ByRef $dict_obj, $key)
	$dict_obj.Remove($key)
	If @error Then return SetError(1, 1, -1)
endfunc

; Returns true if a specified key exists in the associative array, false if not..
func AAExists(ByRef $dict_obj, $key)
	return $dict_obj.Exists($key)
endfunc

; Returns a value for a specified key name in the associative array..
func AAGetItem(ByRef $dict_obj, $key)
	return $dict_obj.Item($key)
endfunc

; Returns the total number of keys in the array..
func AACount(ByRef $dict_obj)
	return $dict_obj.Count 
endfunc

; List all the "Key" > "Item" pairs in the array..
func AAList(ByRef $dict_obj)
ConsoleWrite("AAList: =>" & @CRLF)
	local $k = $dict_obj.Keys ; Get the keys
	; local $a = $dict_obj.Items ; Get the items (for reference)
	for $i = 0 to AACount($dict_obj) -1 ; Iterate the array
		ConsoleWrite($k[$i] & " ==> " & AAGetItem($dict_obj, $k[$i]) & @CRLF)
	next
endfunc

; Wipe the array, obviously.
func AAWipe(ByRef $dict_obj)
	$dict_obj.RemoveAll()
endfunc

; Oh oh!
func AAError()
	Local $err = $oMyError.number
	If $err = 0 Then $err = -1
	SetError($err)  ; to check for after this function returns
endfunc

;; End AA Functions.

AWK

Arrays in AWK are indeed associative arrays.

BEGIN {
  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
}

Babel

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

BASIC

BaCon

DECLARE associative ASSOC STRING

associative("abc") = "first three"
associative("xyz") = "last three"

PRINT associative("xyz")
Output:
prompt$ ./assoc
last three

String keys, with ASSOC to a given data type. Sizing is dynamic.

BASIC256

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

end

show:
for t = 0 to countKeys()-1
   print getKeyByIndex$(t) + " " + getValueByIndex$(t)
next t
print
return

subroutine updateKey(key$, value$)
   # update or add an item
   i=findKey(key$)
   if i=-1 then
      i = freeKey()
      keys$[i] = key$
   end if
   values$[i] = value$
end subroutine

subroutine deleteKey(key$)
   # delete by clearing the key
   i=findKey(key$)
   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
   i=findKey(key$)
   if i=-1 then
      return ""
   end if
   return values$[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]
         k++
      endif
   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]
         k++
      endif
   next n
   return ""
end function
Output:
a apple
b banana
c cucumber

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

BBC BASIC

      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")
      END
      
      DEF PROCputdict(RETURN dict$, value$, key$)
      IF dict$ = "" dict$ = CHR$(0)
      dict$ += key$ + CHR$(1) + value$ + CHR$(0)
      ENDPROC
      
      DEF FNgetdict(dict$, key$)
      LOCAL I%, J%
      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%)

Batch File

This is cheating, I'm sure of it.

::assocarrays.cmd
@echo off
setlocal ENABLEDELAYEDEXPANSION
set array.dog=1
set array.cat=2
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!
endlocal
goto :eof

:mkarray
set %1^&%2=%3
shift /2
shift /2
if "%2" neq "" goto :mkarray
goto :eof

:showit
echo %1 = %2
goto :eof
Output:
array.dog = 1
array.cat = 2
array.wolf = 3
array.cow = 4
"sicko&flu" = 5
"sicko&measles" = 6
"sicko&mumps" = 7
"sicko&bromodrosis" = 8

Bracmat

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.

  new$hash:?myhash
& (myhash..insert)$(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;
Output:
(fruit.melons bananas) (fruit.apples oranges kiwis)
formula.x^2+y^2

Brat

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

C

Solution is at Associative arrays/Creation/C.

C#

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"}};

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>

Creation

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

Insertion

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.

Operator[]

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;

insert()

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:

exampleMap.insert(std::make_pair(7,3.14));

Lookup

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

operator[]

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.

find()

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.

Example

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;
}

Ceylon

import ceylon.collection {

	ArrayList,
	HashMap,
	naturalOrderTreeMap
}

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 {
		"foo"->"bar"
	};
	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``");
	}
	
}

Chapel

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);
Output:
Keys: {foo, bar, baz}
Values: 1 4 6
Keys: {foo, baz}
Values: 1 6
arr2 keys: {John, Pete}
arr2 values: 3 14

Clojure

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

ColdFusion

<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

;; 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

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

DEFINITION Collections;

	IMPORT Boxes;

	CONST
		notFound = -1;

	TYPE
		Hash = POINTER TO RECORD 
			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
		END;

		HashMap = POINTER TO RECORD 
			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
		END;

		LinkedList = POINTER TO RECORD 
			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
		END;

		Vector = POINTER TO RECORD 
			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
		END;

	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;

PROCEDURE Do*;
VAR
	hm : Collections.HashMap;
	o : Boxes.Object;
	keys, values: POINTER TO ARRAY OF Boxes.Object;
	i: INTEGER;
	
BEGIN
	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 Do;

END BbtAssociativeArrays.

Execute:^Q BbtAssociativeArrays.Do

Output:
size:  4

Crystal

hash1 = {"foo" => "bar"}

# hash literals that don't perfectly match the intended hash type must be given an explicit type specification
# the following would fail without `of String => String|Int32`
hash2 : Hash(String, String|Int32) = {"foo" => "bar"} of String => String|Int32

D

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

Dao

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

Dart

main() {
	var rosettaCode = { // Type is inferred to be Map<String, String>
		'task': 'Associative Array Creation'
	};

	rosettaCode['language'] = 'Dart';

	// The update function can be used to update a key using a callback
	rosettaCode.update( 'is fun',  // Key to update
		(value) => "i don't know", // New value to use if key is present
		ifAbsent: () => 'yes!' // Value to use if key is absent
	);
	
	assert( rosettaCode.toString() == '{task: Associative Array Creation, language: Dart, is fun: yes!}');
	
	// If we type the Map with dynamic keys and values, it is like a JavaScript object
	Map<dynamic, dynamic> jsObject = {
		'key': 'value',
		1: 2,
		1.5: [ 'more', 'stuff' ],
		#doStuff: () => print('doing stuff!') // #doStuff is a symbol, only one instance of this exists in the program. Would be :doStuff in Ruby
	};

	print( jsObject['key'] );
	print( jsObject[1] );
	
	for ( var value in jsObject[1.5] )
		print('item: $value');

	jsObject[ #doStuff ](); // Calling the function
	
	print('\nKey types:');
	jsObject.keys.forEach( (key) => print( key.runtimeType ) );
}
Output:
value
2
item: more
item: stuff
doing stuff!

Key types:
String
int
double
Symbol

Delphi

program AssociativeArrayCreation;

{$APPTYPE CONSOLE}

uses Generics.Collections;

var
  lDictionary: TDictionary<string, Integer>;
begin
  lDictionary := TDictionary<string, Integer>.Create;
  try
    lDictionary.Add('foo', 5);
    lDictionary.Add('bar', 10);
    lDictionary.Add('baz', 15);
    lDictionary.AddOrSetValue('foo', 6); // replaces value if it exists
  finally
    lDictionary.Free;
  end;
end.

Diego

Diego has in-built hash and dict (short for 'dictionary') objects which function as associative arrays. The only exception is that hash can only accept uuid datatypes for keys. Diego also has hash_ verb and _hash posit, used to hash an object/command.

To create a dictionary associative array:

use_namespace(rosettacode)_me();

    add_dict(tanzanianBanknoteObverseDipiction)_keys(500,1000,2000,5000,10000)_values(Karume,Nyerere,lion,black rhinoceros,elephant);

reset_ns[];

To create a hash associative array:

use_namespace(rosettacode)_me();

    add_hash(tanzanianBanknoteReverseDipiction)_values(
        University of Dar es Salaam\, Central Hall Building,
        Ikulu\, Dar es Salaam,
        Ngome Kongwe\, Zanzibar,
        Geita Cyanid Leaching Plant,
        Bank of Tanzania Headquarters\, Dar es Salaam
    );

reset_ns[];

DuckDB

DuckDB has several ways to support associative arrays, but there are important differences.

Associative Arrays as Tables

The following is an adaptation of the snippet at #SQL on this page.

# Create a table to associate a key with 0 or more values
CREATE TABLE associative_array (KEY_COLUMN VARCHAR(10), VALUE_COLUMN VARCHAR(100));
# Insert a Key Value pair
INSERT INTO associative_array (KEY_COLUMN, VALUE_COLUMN) VALUES ('KEY', 'VALUE');
# Retrieve a key value pair
SELECT aa.value_column FROM associative_array aa where aa.key_column = 'KEY';

In practice, one is more likely to write the following near-equivalent:

CREATE TABLE associative_array (KEY_COLUMN VARCHAR, VALUE_COLUMN VARCHAR);
# Insert a Key Value pair
INSERT INTO associative_array VALUES ('KEY', 'VALUE');
# Retrieve a key value pair
SELECT aa.value_column FROM associative_array aa where aa.key_column = 'KEY';

However this does not enforce any of the constraints that one might expect for an associative array. The following CREATE statement would be appropriate if keys must be unique and neither keys nor values can be NULL

CREATE TABLE aa (KEY_COLUMN VARCHAR UNIQUE NOT NULL, VALUE_COLUMN VARCHAR NOT NULL);

For example:

INSERT into aa VALUES ('k', 1), ('l', 2);
from aa;
INSERT into aa VALUES ('k', 3);
Output:
┌────────────┬──────────────┐
│ KEY_COLUMN │ VALUE_COLUMN │
│  varchar   │   varchar    │
├────────────┼──────────────┤
│ k          │ 1            │
│ l          │ 2            │
└────────────┴──────────────┘

Constraint Error: Duplicate key "KEY_COLUMN: k" violates unique constraint. If this is an unexpected constraint violation please double check with the known index limitations section in our documentation (https://duckdb.org/docs/sql/indexes).

STRUCTs and MAPs

In the following, we will ignore so-called "unnamed structs".

The main differences between STRUCTs and MAPs are as follows:

  • in a STRUCT, the keys must be strings, and their case is ignored;
  • the keys within a single MAP entity must have the same type, and the same is true of all the values, though the type of the keys may differ

from the type of the values.

Rows and STRUCTs

Each row of a table can be viewed as an associative array. In fact, a row is essentially a STRUCT, and DuckDB makes it easy to convert both ways. For example:

SELECT aa FROM aa LIMIT 1;
Output:
┌──────────────────────────────────────────────────┐
│                        aa                        │
│ struct(key_column varchar, value_column varchar) │
├──────────────────────────────────────────────────┤
│ {'KEY_COLUMN': k, 'VALUE_COLUMN': 1}             │
└──────────────────────────────────────────────────┘

Conversely, to convert a STRUCT to a table, one can use `unnest()`, e.g.:

select unnest({'KEY_COLUMN': 'k', 'VALUE_COLUMN': 1});
Output:
┌────────────┬──────────────┐
│ KEY_COLUMN │ VALUE_COLUMN │
│  varchar   │    int32     │
├────────────┼──────────────┤
│ k          │            1 │
└────────────┴──────────────┘

For STRUCTs that are cells in a table, one can use the pattern exemplified by:

SELECT info.* FROM (SELECT {'a': 1, 'b': 2 } as info);

JSON objects

DuckDB also supports JSON objects, even without the JSON extension.

The keys of JSON objects must be strings, but otherwise JSON objects provide a very flexible way to represent associative arrays in DuckDB. Here's an example illustrating that DuckDB upholds the principle that keys for JSON objects are case-sensitive:

SELECT  '{"a": 1, "A": 2}'::JSON as j;
Output:
┌───────────────┐
│       j       │
│     json      │
├───────────────┤
│ {"a":1,"A":2} │
└───────────────┘

Dyalect

Dyalect has a Tuple data type which allows to add labels to values:

var t = (x: 1, y: 2, z: 3)
print(t.Keys().ToArray())
Output:
[x, y, z]

E

[].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)

EasyLang

# use array of array for this
proc hashGet ind$ . ar$[][] item$ .
   for i to len ar$[][]
      if ar$[i][1] = ind$
         item$ = ar$[i][2]
         return
      .
   .
   item$ = ""
.
proc hashSet ind$ val$ . ar$[][] .
   for i to len ar$[][]
      if ar$[i][1] = ind$
         ar$[i][2] = val$
         return
      .
   .
   ar$[][] &= [ ind$ val$ ]
.
clothing$[][] = [ [ "type" "t-shirt" ] [ "color" "red" ] ]
clothing$[][] &= [ "size" "xl" ]
# 
hashSet "color" "green" clothing$[][]
hashGet "color" clothing$[][] col$
print col$

EchoLisp

(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)
     42
(hash-ref H 'Elvis)
     #f ;; not found. Elvis is not "Elvis"
(hash-ref H "Elvis")
     42
(hash-ref H 'simon)
     antoniette
(hash-count H)
     3

Ecstasy

Defining the Map:

Map<String, Int> map = new HashMap();
map["foo"] = 5;      // or: map.put("foo", 5)
map["bar"] = 10;
map["baz"] = 15;
map["foo"] = 6;      // replaces previous value of 5

Literal map notation:

Map<String, Int> map = ["foo"=6, "bar"=10, "baz"=15];

Retrieving a value:

Int? mightBeNull = map["foo"];
Int neverNull = map.getOrDefault("foo", 0);
if (Int n := map.get("foo")) {
    // if "foo" is in the map, then the variable "n" is set to its value
} else {
    // if "foo" is not in the map, then the variable "n" is not defined
}

Iterate over keys:

for (String key : map) {
    // the variable "key" is defined here
}

Iterate over values:

for (Int value : map.values) {
    // the variable "value" is defined here
}

Iterate over key, value pairs:

for ((String key, Int value) : map) {
    // the variables "key" and "value" are defined here
}

Elena

ELENA 5.0:

import system'collections;
 
public program()
{
    // 1. Create
    var map := Dictionary.new();
    map["key"] := "foox";
    map["key"] := "foo";
    map["key2"]:= "foo2";
    map["key3"]:= "foo3";
    map["key4"]:= "foo4";
}

Strong typed dictionary

import system'collections;

public program()
{
    // 1. Create
    auto map := new Map<string,string>();
    map["key"] := "foox";
    map["key"] := "foo";
    map["key2"]:= "foo2";
    map["key3"]:= "foo3";
    map["key4"]:= "foo4";
}

Elixir

Map literals

An empty map:

%{}

A map with key-value pairs of different types:

%{"one" => :two, 3 => "four"}

Maps with atoms as keys:

%{a: 1, b: 2}
# equivalent to
%{:a => 1, :b => 2}

Programmatic map creation

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

RC.test_create
Output:
< create Map.new >
%{bar: 2, foo: 1}
:bar => 2
:foo => 1
%{bar: 2, foo: 3}
:bar => 2
:foo => 3

Emacs Lisp

(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.

EMal

Map empty = Map(int, text) # creates an empty map
writeLine(empty)
var longFruit = Map(int, text).of(1, "banana") # creates a map with the pair 1 => "banana"
longFruit[2] = "melon" # associates a key of 2 with "melon"
longFruit.insert(3, "avocado")
writeLine(longFruit) # prints the map
var shortFruit = int%text[4 => "kiwi", 5 => "apple"] # map creation using arrow notation
writeLine(shortFruit[5]) # retrieves the value with a key of 5 and prints it out
writeLine(shortFruit.length) # prints the number of entries
writeLine(shortFruit) # prints the map
writeLine(text%text["Italy" => "Rome", "France" => "Paris", "Germany" => "Berlin", "Spain" => "Madrid"])
Output:
[]
[1:banana,2:melon,3:avocado]
apple
2
[4:kiwi,5:apple]
[Italy:Rome,France:Paris,Germany:Berlin,Spain:Madrid]

Erlang

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

-module(assoc).
-compile([export_all]).

test_create() ->   
    D = dict:new(),
    D1 = dict:store(foo,1,D),
    D2 = dict:store(bar,2,D1),
    print_vals(D2),
    print_vals(dict:store(foo,3,D2)).

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

F#

.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"

Factor

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

Fantom

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 
    }

  }
}

Forth

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 !
  else
    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
[ELSE]
  .( Entry not present.) cr
[THEN]
Works with: 4tH version 3.64

Similar functionality is available in 4tH. A binary search is applied to search values, hence hashtables can be quite compact.

include lib/hash.4th
include lib/hashkey.4th
 
\ Create a hash table 'table' with a starting size of 10

['] fnv1a is hash                      \ set hash method
10 constant /htable                    \ determine the size
/htable array htable                   \ allocate the table
latest /htable hashtable               \ turn it into a hashtable
 
\ Insert entries
 
 5 s" foo" htable put
10 s" bar" htable put
15 s" baz" htable put
 
\ Get entry from the table
 
s" bar" htable get error? if
  .( Entry not present.) cr drop
else
  .( Value: ) . cr
then

FreeBASIC

Uses unions to store the keys and associated values, and FreeBASIC's ability to resize arrays makes adding new entries easy.

#define max(a, b) Iif(a>b,a,b)

enum datatype
    'for this demonstration we'll allow these five data types
    BOOL
    STRNG
    BYYTE
    INTEG
    FLOAT
end enum

union value
    bool as boolean
    strng as string*32
    byyte as byte
    integ as integer
    float as double
end union

type dicitem
    'one part of the dictionary entry, either the key or the value
    datatype as datatype   'need to keep track of what kind of data it is
    value as value
end type

type dicentry
    'a dic entry has two things, a key and a value
    key as dicitem
    value as dicitem
end type

sub add_dicentry( Dic() as dicentry, entry as dicentry )
    redim preserve Dic(0 to max(ubound(Dic)+1,0))
    Dic(ubound(Dic)) = entry
    return
end sub

redim as dicentry Dictionary(-1)  'initialise a dictionary with no entries as yet

dim as dicentry thing1, thing2

'generate some test dictionary entries
with thing1
    with .key
        .datatype = STRNG
        .value.strng = "Cat"
    end with
    with .value
        .datatype = STRNG
        .value.strng = "Mittens"
    end with
end with

with thing2
    with .key
        .datatype = integ
        .value.integ = 32767
    end with
    with .value
        .datatype = float
        .value.float = 2.718281828
    end with
end with

add_dicentry( Dictionary(), thing1 )
add_dicentry( Dictionary(), thing2 )

print Dictionary(0).value.value.strng
print Dictionary(1).key.value.integ
Output:
Mittens                         
32767

Free Pascal

FPC 3.2.0.+. Similar to Delphi.

program AssociativeArrayCreation;
{$IFDEF FPC}{$MODE DELPHI}{$ENDIF} 
{$IFDEF WINDOWS}{$APPTYPE CONSOLE}{$ENDIF}
uses Generics.Collections; 
var
  lDictionary: TDictionary<string, Integer>;
begin
  lDictionary := TDictionary<string, Integer>.Create;
  try
    lDictionary.Add('foo', 5);
    lDictionary.Add('bar', 10);
    lDictionary.Add('baz', 15);
    lDictionary.AddOrSetValue('foo', 6); // replaces value if it exists
  finally
    lDictionary.Free;
  end;
end.

FPC 2.4+. Using FGL instead of rtl-generics:

program AssociativeArrayCreation;
{$IFDEF WINDOWS}{$APPTYPE CONSOLE}{$ENDIF}
{$MODE DELPHI} 
uses fgl;
 
var
  lDictionary: TFPGMap<string, Integer>;
begin
  lDictionary := TFPGMap<string, Integer>.Create;
  try
    lDictionary.Add('foo', 5);
    lDictionary.Add('bar', 10);
    lDictionary.Add('baz', 15);
  finally
    lDictionary.Free;
  end;
end.

Futhark

let associative_array = {key1=1,key2=2}


FutureBasic

void local fn DoIt
  CFDictionaryRef dict1 = fn DictionaryWithObjects( @"Alpha", @"A", @"Bravo", @"B", @"Charlie", @"C", @"Delta", @"D", NULL ) // must be NULL-terminated
  
  CFDictionaryRef dict2 = @{@"A":@"Alpha", @"B":@"Bravo", @"C":@"Charlie", @"D":@"Delta"} // shorthand syntax
  
  CFMutableDictionaryRef dict3 = fn MutableDictionaryNew
  MutableDictionarySetObjectForKey( dict3, @"Alpha", @"A" )
  MutableDictionarySetObjectForKey( dict3, @"Bravo", @"B" )
  MutableDictionarySetObjectForKey( dict3, @"Charlie", @"C" )
  MutableDictionarySetObjectForKey( dict3, @"Delta", @"D" )
  
  CFMutableDictionaryRef dict4 = fn MutableDictionaryWithDictionary( @{@"A":@"Alpha", @"B":@"Bravo", @"C":@"Charlie", @"D":@"Delta"} )
  
  print fn DictionaryObjectForKey( dict1, @"A" )
  print dict1[@"B"]
  print dict2[@"C"]
  print dict3[@"D"]
  print dict4
end fn

fn DoIt

HandleEvents
Output:
Alpha
Bravo
Charlie
Delta
{
    A = Alpha;
    B = Bravo;
    C = Charlie;
    D = Delta;
}

Gambas

See [Associative array/Iteration]

Go

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,
}

Gosu

As an OOP language with generics Gosu can use any variety of Map classes. In addition Gosu provides associative array syntax for all objects.

// empty map
var emptyMap = new HashMap<String, Integer>()

// map initialization
var map = {"Scott"->50, "Carson"->40, "Luca"->30, "Kyle"->38}

// map key/value assignment
map["Scott"] = 51

// get a value
var x = map["Scott"]
 
// remove an entry
map.remove("Scott")
 
// loop and maps
for(entry in map.entrySet()) {
  print("Key: ${entry.Key}, Value: ${entry.Value}") 
}

// functional iteration
map.eachKey(\ k ->print(map[k]))
map.eachValue(\ v ->print(v))
map.eachKeyAndValue(\ k, v -> print("Key: ${v}, Value: ${v}"))
var filtered = map.filterByValues(\ v ->v < 50)

// any object can be treated as an associative array
class Person {
  var name: String
  var age: int 
}
// access properties on Person dynamically via associative array syntax
var scott = new Person()
scott["name"] = "Scott"
scott["age"] = 29

Groovy

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' == map.foo
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' == map.foo
assert 'barvalue' == map['bar']
assert 'moovalue' == map.get('moo')

Harbour

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" }

Haskell

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

hexiscript

let d         dict 2  # Initial estimated size
let d["test"] "test"  # Strings can be used as index
let d[123]    123     # Integers can also be used as index

println d["test"]
println d[123]

Icon and Unicon

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"
   write(t["foo"])
end

Inform 7

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

Static relation

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

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.

Insitux

It is possible to use any value type for both keys and values for a dictionary in Insitux.

{
  :a "value"    ;keyword key, string value
  :b 123        ;keyword key, number value
  456 [1 2 3]   ;number key, vector value
  [5 6 7] :b    ;vector key, keyword value
  {:a 1} {:b 2} ;dictionary key, dictionary value
}
;use dictionary as function for lookup; commas for readability, treated as white-space
> ({:a 1, :b 2, :c 3} :b)
2
;extend existing dictionary by using it as a function with two arguments
> ({:a 1, :b 2, :c 3} :b 3)
{:a 1, :b 3, :c 3}

Ioke

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

J

Usually, in J, you would use a named pair of (same length) lists for this purpose - one of keys, one of values. There are a number of details here that vary with the the intended use patterns. (First you get it working and then if you run into bottlenecks you rebuild things to relieve the problems).

However, it's also possible to use the symbol table itself to hold the names. The symbol table has limitations (can only accept syntactically valid names), but we can turn arbitrary strings into valid symbols using base 62 encode and prefixing with a letter (hypothetically speaking, base 64 encode would let us build longer names than base 62, because of computational complexity issues - but the J symbol table also comes with a name length limit - 255 characters - and does not support 64 different characters in names):

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'
1
   bletch__example=: 7 8 9
   get__example 'bletch'
7 8 9
   codestroy__example''

Note that J's symbols (http://www.jsoftware.com/help/dictionary/dsco.htm) 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).

Jakt

fn main() {
    let dictionary = ["foo": 1, "bar": 2]
    println("{}", dictionary)
}

Java

Java offers an associative array, referred to as a Map, as part of the Collections Framework.
There are numerous implementing classes, each which provide unique functionality for various tasks.

It's worth noting that Java also offers the Dictionary class, which appears to be less preferred, according to Java.
"[The Map] interface takes the place of the Dictionary class ...".

The most generalized Map would be the HashMap, which is a basic, unordered, set of keys and values.
There is also the LinkedHashMap, which will preserve the order of input.
There is the TreeMap, which is used to store the keys in a specific order, using the key's compareTo method. Optionally, you could provide your own comparator using the Comparator interface, which I'll demonstrate below.
There are numerous other implementing classes, which can be found under the Map Javadoc, Map (Java SE 20 & JDK 20).

Here is a basic implementation of a HashMap.

Map<String, Integer> map = new HashMap<>();

To add a key and value pair, you use the Map.put method.
A useful feature of Map.put is that if the key already exists, thus is being overridden, it will return the overridden value, otherwise null.

map.put("rosetta", 100);
map.put("code", 200);

To get a value, you use the Map.get method, specifying the key as the parameter.
If the specified key does not exist, null is returned.

int valueA = map.get("rosetta");
int valueB = map.get("code");

To mutate a key's value, you use the Map.replace method.

map.replace("rosetta", 300);

Alternately, you can replace the value, only if it is of a current value.
So, in this example it will return true, only if the key "rosetta" had the current value of 100.

boolean replaced = map.replace("rosetta", 100, 300);

To check for the existence of a key, you use the containsKey method.

boolean contains = map.containsKey("rosetta");

And to check for the existence of a value, you use the containsValue method.

boolean contains = map.containsValue(100);

A LinkedHashMap is exactly the same as a HashMap, except it will preserve the order in which the keys were added.

Map<String, Integer> map = new LinkedHashMap<>();
map.put("rosetta", 100);
map.put("code", 200);

A TreeMap is useful for when you want the keys in a specific order.
By default, it will use the key's compareTo method, to determine the order.
So, if you're key is a String, the order will be ascending, similar to an actual dictionary.

Map<String, Integer> map = new TreeMap<>();
map.put("rosetta", 100);
map.put("code", 200);

You could, optionally, specify a comparator by implementing a Comparator interface.
A TreeMap, conveniently, only requires you to implement the compare method of the Comparator, so the implementation can be done as an anonymous class.

Comparator<String> comparator = new Comparator<String>() {
    public int compare(String stringA, String stringB) {
        if (stringA.compareTo(stringB) > 0) {
            return -1;
        } else if (stringA.compareTo(stringB) < 0) {
            return 1;
        }
        return 0;
    }
};

Which you then supply as an argument to the constructor.

Map<String, Integer> map = new TreeMap<>(comparator);

To make things even simpler, you could use a lambda for the anonymous class.

Comparator<String> comparator = (stringA, stringB) -> {
    if (stringA.compareTo(stringB) > 0) {
        return -1;
    } else if (stringA.compareTo(stringB) < 0) {
        return 1;
    }
    return 0;
};

JavaScript

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));
}

jq

Associative Arrays with String-Valued Keys

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

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)
   end;

# 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)
  end;

def getKey(key): .[key|collisionless];

def removeKey(key): delpaths( [ [key|collisionless] ] );

Example:

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

produces:

"two"

Jsish

From Javascript. jsish warns of duplicate var, in this case the assoc variable is reused.

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'";
;assoc;
 
//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)) {
    puts('key:"' + key + '", value:"' + assoc[key] + '"');
  }
}
;assoc;

/*
=!EXPECTSTART!=
associativeArray.jsi:12: warn: duplicate var: assoc
assoc ==> { 2:"the index here is the string \'2\'", "another-key":3, foo:"bar", thirdKey:"we can also do this!" }
key:"another-key", value:"3"
key:"foo", value:"bar"
assoc ==> { "another-key":3, foo:"bar" }
=!EXPECTEND!=
*/
Output:
prompt$ jsish -u associativeArray.jsi
[PASS] associativeArray.jsi

Julia

Works with: Julia version 0.6

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.

dict = Dict('a' => 97, 'b' => 98) # list keys/values
# Dict{Char,Int64} with 2 entries:
#   'b' => 98
#   'a' => 97

dict = Dict(c => Int(c) for c = 'a':'d') # dict comprehension
# Dict{Char,Int64} with 4 entries:
#   'b' => 98
#   'a' => 97
#   'd' => 100
#   'c' => 99

dict['é'] = 233; dict # add an element
# Dict{Char,Int64} with 3 entries:
#   'b' => 98
#   'a' => 97
#   'é' => 233

emptydict = Dict() # create an empty dict
# Dict{Any,Any} with 0 entries

dict["a"] = 1 # type mismatch
# ERROR: MethodError: Cannot `convert` an object of type String to an object of type Char

typeof(dict) # type is infered correctly
# Dict{Char,Int64}

K

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

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

   / extracting a value
   d1[`bar]
2

Another approach.

   d2: .()          / create empty dictionary
   d2[`"zero"]:0
   d2[`"one"]:1
   d2[`"two"]:2

   d2
.((`zero;0;)
  (`one;1;)
  (`two;2;))

Extracting the keys and values.

   !d2              / the keys
`zero `one `two

   d2[]             /  the values
0 1 2

Kotlin

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 ")
    println()

    // iterate over values:
    for (v in map.values) print("$v ")
    println()

    // iterate over key, value pairs:
    for ((k, v) in map) println("$k => $v")
}

Lambdatalk

Associative arrays are not currently built in the JS.js kernel of lambdatalk but are added via the lib_hash library page.

1) a (currently) reduced set of functions:

HASH: [5] [H.lib, H.new, H.disp, H.get, H.set!]

2) building an associative array: {def H key value | ...}

{def capitals
 {H.new nyk New York, United States | 
        lon London, United Kingdom |
        par Paris, France |
        mos Moscou, Russia }}
-> capitals

3) displaying: {H.disp hash}
{H.disp {capitals}}    
-> 
[
nyk: New York, United States
lon: London, United Kingdom
par: Paris, France
mos: Moscou, Russia
]

4) getting a value from a key: {H.get hash key}

{H.get {capitals} nyk} -> New York, United States
{H.get {capitals} lon} -> London, United Kingdom
{H.get {capitals} par} -> Paris, France
{H.get {capitals} mos} -> Moscou, Russia

5) adding a new (key,value): {H.set! hash key value}

{H.disp {H.set! {capitals} bar Barcelona, Catalunya}}
-> 
[
nyk: New York, United States
lon: London, United Kingdom
par: Paris, France
mos: Moscou, Russia
bar: Barcelona, Catalunya
]

6) editing a key

{H.disp 
  {H.set! {capitals}
          nyk 
          {H.get {capitals} nyk} of America}} // adding "of America" to nyk
-> 
[
nyk: New York, United States of America
lon: London, United Kingdom
par: Paris, France
mos: Moscou, Russia
bar: Barcelona, Catalunya
]

Lang5

: 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 .

langur

Hash keys in langur may be numbers or strings. Number keys are simplified, so that 1.0 is the same key as 1.

var h = {1: "abc", "1": 789}

# may assign with existing or non-existing hash key (if hash is mutable)
h[7] = 49

writeln h[1]
writeln h[7]
writeln h["1"]

# using an alternate value in case of invalid index; prevents exception
writeln h[1; 42]
writeln h[2; 42]
Output:
abc
49
789
abc
42

Lasso

// 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 {^
	#v
	'<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)

LFE

(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

Needs the sublist library from http://basic.wikispaces.com/SubList+Library 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

Lingo

props = [#key1: "value1", #key2: "value2"]

put props[#key2]
-- "value2"
put props["key2"]
-- "value2"
put props.key2
-- "value2"
put props.getProp(#key2)
-- "value2"

LiveCode

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

Output

number of elements: 3
length of item 3: 5
keys: key numbers
abc
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]

LOLCODE

BUKKITs are associative arrays

HAI 1.2
I HAS A Hash ITZ A BUKKIT
Hash HAS A key1 ITZ "val1" BTW This works for identifier-like keys, like obj.key in JavaScript
Hash HAS A SRS "key-2" ITZ 1 BTW Non-identifier keys need the SRS
VISIBLE Hash'Z SRS "key-2"
KTHXBYE
Output:
1

Lua

Lua tables are Hashes

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

Returns nil on unknown key.

M2000 Interpreter

Μ2000 has Inventory object to use it as a Map. All keys converted to strings. If a key has no value then key is the value until we place one. A special type of Inventory is the Inventory Queue, where we can use same keys, and we can't delete except from the last append.

Inventory A="100":=1, "200":=5, 10:=500,  20:="Hello There"
Print len(A)
Print A(100)=1, A(200)=5, A$(20)="Hello There"
Return A, 100:=3, 200:=7
\\ print all elements
Print A
For i=0 to Len(A)-1 {
      \\ Key, Value by current order (using !)
      Print Eval$(A, i), A$(i!)
}
\\ Iterator
Append A, "End":=5000
N=Each(A)
While N {
      Print Eval$(A, N^), A$(N^!)
}
Print Len(A)=5
Delete A, "100", 10, 20
Print Len(A)=2
If Exist(A, "End") Then Print Eval(A)=5000

Maple

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];
                                   4

> T[sin(x)];
                                 cos(x)

> T["foo"];
                                   1

New entries are added by assignment.

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

> T[ "bar" ];     
                                   2

Entries can be removed as follows.

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

> T[ "foo" ];
                                T["foo"]

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

Mathematica / Wolfram Language

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

MATLAB / Octave

MATLAB/Octave: structs

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

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

alternatively

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

or

   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

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';

Usage:

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

ans = 

    [37]    [51]    [72]

>> values(m)

ans = 

    'thirtyseven'    'fiftyone'    'seventytwo'

Maxima

/* No need to declare anything, undeclared arrays are hashed */

h[1]: 6;
h[9]: 2;

arrayinfo(h);
[hashed, 1, [1], [9]]

min

Works with: min version 0.19.6
{1 :one 2 :two 3 :three}

MiniScript

A map literal in MiniScript is enclosed in curly braces, with key:value pairs separated by commas. Keys and values may be any type. Retrieval or assignment is by putting the key in square brackets. As syntactic sugar, when a string key follows the rules of a MiniScript identifier (starts with a letter and contains only letters, numbers, and underscores), you may also access it with dot syntax.

map = { 3: "test", "foo": 42 }

print map[3]
map[3] = "more tests"
print map[3]
print map["foo"]
print map.foo  // same as map["foo"] (only for string keys that are valid identifiers)

Nemerle

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());
        WriteLine(hash1[entry]);
    }
}

NetRexx

/* 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
Output:
<hash key="0" value="." />
<hash key="key0" value="value0" />

Nim

import tables

var
  hash = initTable[string, int]() # empty hash table
  hash1: Table[string, int]       # empty hash table (implicit initialization).
  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["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
Output:
hash has 3 elements
hash has key foo? true
hash has key bar? false
iterate pairs:
eggs: 2
foo: 3
spam: 1
iterate keys:
eggs
foo
spam
iterate values:
2
3
1

Oberon-2

Works with: oo2c Version 2
MODULE AssociativeArray;
IMPORT
  ADT:Dictionary,
  Object:Boxed,
  Out;
TYPE
  Key = STRING;
  Value = Boxed.LongInt;
  
VAR
  assocArray: Dictionary.Dictionary(Key,Value);
  iterK: Dictionary.IterKeys(Key,Value);
  iterV: Dictionary.IterValues(Key,Value);
  aux: Value;
  k: Key;
  
BEGIN
  assocArray := NEW(Dictionary.Dictionary(Key,Value));
  assocArray.Set("ten",NEW(Value,10));
  assocArray.Set("eleven",NEW(Value,11));
  
  aux := assocArray.Get("ten");
  Out.LongInt(aux.value,0);Out.Ln;
  aux := assocArray.Get("eleven");
  Out.LongInt(aux.value,0);Out.Ln;Out.Ln;
  
  (* Iterate keys *)
  iterK := assocArray.IterKeys();
  WHILE (iterK.Next(k)) DO
    Out.Object(k);Out.Ln
  END;
  
  Out.Ln;
  
  (* Iterate values *)
  iterV := assocArray.IterValues();
  WHILE (iterV.Next(aux)) DO
    Out.LongInt(aux.value,0);Out.Ln
  END
  
END AssociativeArray.

Objeck

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

Associative map

# 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
map->Find("thirteen")->As(IntHolder)->GetValue()->PrintLine();
map->Find("seven")->As(IntHolder)->GetValue()->PrintLine();

Hash table

# 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
map->Find("thirteen")->As(IntHolder)->GetValue()->PrintLine();
map->Find("seven")->As(IntHolder)->GetValue()->PrintLine();

Objective-C

Works with: Cocoa

and

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",
    nil];

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"];

OCaml

Hash table

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

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

module String = struct
   type t = string
   let compare = Pervasives.compare
end
module StringMap = Map.Make(String);;

let map =
  List.fold_left
    (fun map (key, value) -> StringMap.add key value map)
    StringMap.empty
    ["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

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")

Ol

Associative arrays in Otus Lisp has name "fixed function" (aka "ff") and fully conforms the functional paradigm. It means that there are no ability to change the existing associative array, only get new changed one.

You can use only values as keys (atomic numbers, constants) and, as exception, symbols (symbols are references, but unique). No strings, lists, vectors and other objects can be used directly. In such cases use hashes or similar mechanisms.

;;; empty associative array
#empty
; or short form
#e

;;; creating the new empty associative array
(define empty-map #empty)

;;; creating associative array with values
(define my-map (pairs->ff '(
   (1 . 100)
   (2 . 200)
   (7 . 777))))
;;; or in short form (available from Ol version 2.1)
(define my-map {
   1 100
   2 200
   7 777})

;;; add new key-value pair to the existing associative array
(define my-new-map (put my-map 'the-key 'the-value))

;;; print our arrays
(print empty-map)
; ==> #()

(print my-map)
; ==> #((1 . 100) (2 . 200) (7 . 777))

(print my-new-map)
; ==> #((1 . 100) (2 . 200) (7 . 777) (the-key . the-value))

ooRexx

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
end

Iterate over values:

loop value over map~allItems
    say value
end

Iterate over key, value pairs:

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

OxygenBasic

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
  e=max*2
  for i=1 to e step 2
    if k=s[i] then return i
  next
  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 
    max+=1
    i=max*2-1
    s[i]=k
  end if
  s[i+1]=d
  return i
  end method

end class


'====
'TEST
'====

AssociativeArray A

'fill
A.s<={"shoes","LC1",  "ships","LC2",  "sealingwax","LC3",  "cabbages","LC4",  "kings","LC5"}
A.max=5
'access
print A.dat("ships")       'result LC2
A.dat("computers")="LC99"  '
print A.dat("computers")   'result LC99

Oz

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

declare
  Dict = {Dictionary.new}
in
  Dict.foo := 5
  Dict.bar := 10
  Dict.baz := 15
  Dict.foo := 20

  {Inspect Dict}

'Records' can be consideres immutable maps:

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

PARI/GP

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.

PascalABC.NET

begin
  var zoo := new Dictionary<string,integer>;
  zoo['crocodile'] := 2;
  zoo['jiraffe'] := 3;
  zoo['behemoth'] := 1;
  zoo.Println;
end.
Output:
(crocodile,2) (jiraffe,3) (behemoth,1)

Perl

Hash

Definition:

# 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',
);

Use:

print $hash{key1};

$hash{key1} = 'val1';

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

HashRef

Definition:

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

Use:

print $hashref->{key1};

$hashref->{key1} = 'val1';

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

Key Types

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.

Phix

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).

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

Phixmonti

include ..\Utilitys.pmt

def getd    /# dict key -- dict data #/
    swap 1 get rot find nip
    dup if
        swap 2 get rot get nip
    else
        drop "Unfound"
    endif
enddef

def setd    /# dict ( key data ) -- dict #/
    1 get var ikey
    2 get var idata
    drop
    1 get ikey find var p drop
    p if
        2 get idata p set 2 set
    else
        2 get idata 0 put 2 set
        1 get ikey 0 put 1 set
    endif
enddef

( ( ) ( ) )

( 1 "one" ) setd
( "two" 2 ) setd
( PI PI ) setd

1 getd print nl
"two" getd print nl
PI getd tostr print nl
3 getd print

PHP

$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

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

Picat

Associative arrays are called "map" in Picat.

go =>

   % Create an empty map
   Map = new_map(),
   println(Map),

   % add some data
   Map.put(a,1),
   Map.put("picat",2),
   Map.put("picat",3), % overwrite values
   % Add a new value (a long list of different stuff)
   Map.put([a,list,of,different,"things",[including, lists],3.14159],2),
   println(Map),

   println(a=Map.get(a)), % get a value
   println(b=Map.get(b,'default value')), % the second argument to get/2 is the default value

   % create a map from a list of values
   Map2 = new_map([K=V : {K,V} in zip([a,b,c,d,e,f,g,h],1..8)]),
   println(Map2),
   println(h=Map2.get(h)),

   % Check if there is a value in the map
   if not Map2.has_key(z) then
     println("no key 'z'")
   end,

   % keys() and value() returns unsorted list of elements
   % so we sort them.
   println(keys=Map2.keys().sort()),
   println(values=Map2.values().sort()),

   % Print the values for the keys that are even
   println([K : K=V in Map2, V mod 2=0].sort), 

   nl.
Output:
(map)[]
(map)[a = 1,picat = 3,[a,list,of,different,things,[including,lists],3.14159] = 2]
a = 1
b = default value
(map)[b = 2,a = 1,e = 5,h = 8,f = 6,g = 7,d = 4,c = 3]
h = 8
no key 'z'
keys = abcdefgh
values = [1,2,3,4,5,6,7,8]
bdfh

PicoLisp

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)
A NIL
   foo 20
   bar 10
   baz 15

Pike

Pike provides a built in type called mapping for associative arrays. Any data type including user created classes can be used as indices (aka keys in some other languages) and values.

There are two main ways of indexing a mapping; a[b] or a->b, with variants for "safe" indexing that will not throw error when indexing nonexistent indices. See the documentation for indexing.

indices() and values() can be used to enumerate the contents of an existing mapping.

mapping m = ([ "apple": "fruit", 17: "seventeen" ]);
write("indices: %O\nvalues: %O\n17: %O\n",
      indices(m),
      values(m),
      m[17]);
Output:
indices: ({
    17,
    "apple"
})
values: ({
    "seventeen",
    "fruit"
})
17: "seventeen"

Since any data type can be used nested structures of arbitrary size can be constructed.

mapping m2 = ([ "car": ([ "ford":17, "volvo":42 ]) ]);
write("#ford: %O, #volvo: %O\n",
      m2->car->ford,
      m2["car"]["volvo"]);
Output:
#ford: 17, #volvo: 42

PL/I

*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;
   k=key(i);
   Put Edit('>'!!k!!'< => >'!!aacc(k)!!'<')(Skip,a);
   End;

 aadd: Proc(k,v);
 Dcl (k,v) Char(*) Var;
 If aacc(k)^=hi Then
   Put Edit('Key >',k,'< would be a duplicate, not added.')
           (Skip,a,a,a);
 Else Do;
   an+=1;
   key(an)=k;
   val(an)=v;
   Put Edit('added >'!!k!!'< -> '!!v!!'<')(Skip,a);
   End;
 End;

 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
     Return(val(i));
   End;
 Return(hi);
 End;

 End;
Output:
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

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:-

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

    dbms_output.put_line ('assocArray.myShape: ' || assocArray.myShape);
    dbms_output.put_line ('assocArray.mySize: ' || assocArray.mySize);
END;
/

Pop11

;;; 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.  
 appproperty(ht,
    procedure (key, value);
      printf(value, '%p\t');
      printf(key, '%p\n');
     endprocedure);

PostScript

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

Potion

mydictionary = (red=0xff0000, green=0x00ff00, blue=0x0000ff)

redblue = "purple"
mydictionary put(redblue, 0xff00ff)

255 == mydictionary("blue")
65280 == mydictionary("green")
16711935 == mydictionary("purple")

PowerShell

An empty hash table can be created with:

$hashtable = @{}

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

$hashtable = @{
    "key1" = "value 1"
    key2 = 5            # if the key name has no spaces, no quotes are needed.
}

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

$hashtable.foo    = "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

NB. PowerShell compares strings as case-insensitive, that means the hashtable keys 'a' and 'A' are considered the same key. This happens when @{} is turned into a hashtable, but can be overridden by an explicit long-form:

# Case insensitive keys, both end up as the same key:
$h=@{}
$h['a'] = 1
$h['A'] = 2
$h

Name                           Value                                                                                              
----                           -----                                                                                              
a                              2     

# Case sensitive keys:
$h = New-Object -TypeName System.Collections.Hashtable
$h['a'] = 1
$h['A'] = 2
$h

Name                           Value                                                                                              
----                           -----                                                                                              
A                              2                                                                                                  
a                              1

Similarly, values can be retrieved using either syntax:

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

It is common to see a hashtable literal used to create an object, by casting it to a new type:

$obj = [PSCustomObject]@{
    "key1" = "value 1"
    key2 = 5            
}

This is a convenience syntax, has less code and runs faster than other ways to create objects.

Prolog

We use the facts table for this purpose.

mymap(key1,value1).
mymap(key2,value2).

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

PureBasic

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

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

Python

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 https://docs.python.org/3/library/stdtypes.html#mapping-types-dict

# 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

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

environment example

> 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

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

list example

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

$b
[1] 2

$c
[1] 3.14

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

[[2]]
[1] 2

[[3]]
[1] 3.14

[[4]]
[1] "xyz"

Racket

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

Raku

(formerly Perl 6)

Works with: Rakudo version 2018.03

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 Raku'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{C.new} = 1;

Raven

{ '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.


Relation

relation key, value
insert "foo", "bar"
insert "bar", "foo"
insert "fruit", "apple"
assert unique key
print
select key == "fruit"
print

assert will show an error, if a key is used twice. However, it not stop the insertion.

key value
foo bar
bar foo
fruit apple
key value
fruit apple

Retro

with hashTable'
hashTable constant table

table %{ first = 100 }%
table %{ second = "hello, world!" keepString %}

table @" first" putn
table @" second" puts

REXX

version 1

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    */
Output:
stem.key0= .
stem.key1= value0

version 2

/*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.ca='Sacramento'; stateN.ca='California'
stateC.nd='Bismarck'  ; stateN.nd='North Dakota'
stateC.mn='St. Paul'  ; stateN.mn='Minnesota'
stateC.dc='Washington'; stateN.dc='District of Columbia'
stateC.ri='Providence'; stateN.ri='Rhode Island and Providence Plantations'

say 'capital of California is' stateC.ca
say 'capital of Oklahoma is' stateC.ok
yyy='RI'
say 'capital of' stateN.yyy "is" stateC.yyy
                                       /*stick a fork in it, we're done.*/
Output:
capital of California is Sacramento
capital of Oklahoma is  [not defined yet]
capital of Rhode Island and Providence Plantations is Providence

Ring

# Project  Associative array/Creation

myarray = [["one",1],
                 ["two",2],
                 ["three",3]]
see find(myarray,"two",1) + nl           
see find(myarray,2,2) + nl

Output:

2
2

RLaB

Associative arrays are called lists in RLaB.

x = <<>>;  // create an empty list using strings as identifiers.
x.red   = strtod("0xff0000");  // RLaB doesn't deal with hexadecimal numbers directly. Thus we
x.green = strtod("0x00ff00");  // convert it to real numbers using ''strtod'' function.
x.blue  = 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 (x.red);

// 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.

Ruby

Hash literals

Ruby has literal syntax for Hash objects.

A Hash with symbols as keys:

{:name => 'Zeus', :planet => 'Jupiter'}

Shorthand for symbol keys:

{name: 'Zeus', planet: 'Jupiter'}

A Hash with keys and values of arbitrary types:

{1 => 'two', three: 4}

An empty Hash:

{}

Customizing the default value

A Hash object that returns nil for unknown keys:

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

A Hash object that returns 'unknown key' for unknown keys:

hash = Hash.new('unknown key')
hash[666] = 'devil'
hash[777]  # => 'unknown key'
hash[666]  # => 'devil'

A Hash object that returns "unknown key #{key}" for unknown keys:

hash = Hash.new { |h, k| "unknown key #{k}" }
hash[666] = 'devil'
hash[777]  # => 'unknown key 777'
hash[666]  # => 'devil'

A Hash object that adds "key #{key} was added at #{Time.now}" to the hash the first time an unknown key is seen:

hash = Hash.new { |h, k| h[k] = "key #{k} was added at #{Time.now}" }
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'

Rust

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);
}

Sather

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";
  end;
end;

Scala

// 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

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.

#!r6rs

(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))

A persistent associative array from scratch

Works with: CHICKEN version 5.3.0
Library: r7rs
Library: srfi-1
Library: srfi-151

Most implementations of associative arrays—including those for Scheme—are for ‘mutable’ arrays, whose previous values are effectively lost whenever an insertion is done. Here instead is a persistent (‘immutable’) implementation, using code from the AVL Tree task. (So performance will be on average logarithmic. Just be aware of that.)

That there are so many implementations of associative arrays for Scheme is partly because making an implementation from scratch is fairly easy. But many approaches are difficult to use if the goal is persistent associative arrays. For instance, if you use a classical hash table, inserting an association would require copying an entire array.

Associate arrays are not part of the Scheme language itself, but are compiler/interpreter or library add-ons. So I feel justified in presenting this sketch of yet another library add-on.

(cond-expand
  (r7rs)
  (chicken (import r7rs)))

(define-library (avl-trees)

  ;;
  ;; This library implements ‘persistent’ (that is, ‘immutable’) AVL
  ;; trees for R7RS Scheme.
  ;;
  ;; References:
  ;;
  ;;   * Niklaus Wirth, 1976. Algorithms + Data Structures =
  ;;     Programs. Prentice-Hall, Englewood Cliffs, New Jersey.
  ;;
  ;;   * Niklaus Wirth, 2004. Algorithms and Data Structures. Updated
  ;;     by Fyodor Tkachov, 2014.
  ;;
  ;; THIS IS A TRIMMED-DOWN VERSION OF MY SOLUTION TO THE AVL TREES
  ;; TASK: https://rosettacode.org/wiki/AVL_tree#Scheme
  ;;

  (export avl)
  (export avl?)
  (export avl-empty?)
  (export avl-insert)
  (export avl-search-values)
  (export avl-check-usage)

  (import (scheme base))
  (import (scheme case-lambda))
  (import (scheme process-context))
  (import (scheme write))

  (begin

    (define-syntax avl-check-usage
      (syntax-rules ()
        ((_ pred msg)
         (or pred (usage-error msg)))))

    (define-record-type <avl>
      (%avl key data bal left right)
      avl?
      (key %key)
      (data %data)
      (bal %bal)
      (left %left)
      (right %right))

    (define (avl)
      (%avl #f #f #f #f #f))

    (define (avl-empty? tree)
      (avl-check-usage
       (avl? tree)
       "avl-empty? expects an AVL tree as argument")
      (not (%bal tree)))

    (define (avl-search-values pred<? tree key)
      ;; Return two values: the data matching the key, or #f is the
      ;; key is not found; and a second value that is either #f or #t,
      ;; depending on whether the key is found.
      (define (search p)
        (if (not p)
            (values #f #f)
            (let ((k (%key p)))
              (cond ((pred<? key k) (search (%left p)))
                    ((pred<? k key) (search (%right p)))
                    (else (values (%data p) #t))))))
      (avl-check-usage
       (procedure? pred<?)
       "avl-search-values expects a procedure as first argument")
      (if (avl-empty? tree)
          (values #f #f)
          (search tree)))

    (define (avl-insert pred<? tree key data)

      (define (search p fix-balance?)
        (cond
         ((not p)
          ;; The key was not found. Make a new node and set
          ;; fix-balance?
          (values (%avl key data 0 #f #f) #t))

         ((pred<? key (%key p))
          ;; Continue searching.
          (let-values (((p1 fix-balance?)
                        (search (%left p) fix-balance?)))
            (cond
             ((not fix-balance?)
              (let ((p^ (%avl (%key p) (%data p) (%bal p)
                              p1 (%right p))))
                (values p^ #f)))
             (else
              ;; A new node has been inserted on the left side.
              (case (%bal p)
                ((1)
                 (let ((p^ (%avl (%key p) (%data p) 0
                                 p1 (%right p))))
                   (values p^ #f)))
                ((0)
                 (let ((p^ (%avl (%key p) (%data p) -1
                                 p1 (%right p))))
                   (values p^ fix-balance?)))
                ((-1)
                 ;; Rebalance.
                 (case (%bal p1)
                   ((-1)
                    ;; A single LL rotation.
                    (let* ((p^ (%avl (%key p) (%data p) 0
                                     (%right p1) (%right p)))
                           (p1^ (%avl (%key p1) (%data p1) 0
                                      (%left p1) p^)))
                      (values p1^ #f)))
                   ((0 1)
                    ;; A double LR rotation.
                    (let* ((p2 (%right p1))
                           (bal2 (%bal p2))
                           (p^ (%avl (%key p) (%data p)
                                     (- (min bal2 0))
                                     (%right p2) (%right p)))
                           (p1^ (%avl (%key p1) (%data p1)
                                      (- (max bal2 0))
                                      (%left p1) (%left p2)))
                           (p2^ (%avl (%key p2) (%data p2) 0
                                      p1^ p^)))
                      (values p2^ #f)))
                   (else (internal-error))))
                (else (internal-error)))))))

         ((pred<? (%key p) key)
          ;; Continue searching.
          (let-values (((p1 fix-balance?)
                        (search (%right p) fix-balance?)))
            (cond
             ((not fix-balance?)
              (let ((p^ (%avl (%key p) (%data p) (%bal p)
                              (%left p) p1)))
                (values p^ #f)))
             (else
              ;; A new node has been inserted on the right side.
              (case (%bal p)
                ((-1)
                 (let ((p^ (%avl (%key p) (%data p) 0
                                 (%left p) p1)))
                   (values p^ #f)))
                ((0)
                 (let ((p^ (%avl (%key p) (%data p) 1
                                 (%left p) p1)))
                   (values p^ fix-balance?)))
                ((1)
                 ;; Rebalance.
                 (case (%bal p1)
                   ((1)
                    ;; A single RR rotation.
                    (let* ((p^ (%avl (%key p) (%data p) 0
                                     (%left p) (%left p1)))
                           (p1^ (%avl (%key p1) (%data p1) 0
                                      p^ (%right p1))))
                      (values p1^ #f)))
                   ((-1 0)
                    ;; A double RL rotation.
                    (let* ((p2 (%left p1))
                           (bal2 (%bal p2))
                           (p^ (%avl (%key p) (%data p)
                                     (- (max bal2 0))
                                     (%left p) (%left p2)))
                           (p1^ (%avl (%key p1) (%data p1)
                                      (- (min bal2 0))
                                      (%right p2) (%right p1)))
                           (p2^ (%avl (%key p2) (%data p2) 0
                                      p^ p1^)))
                      (values p2^ #f)))
                   (else (internal-error))))
                (else (internal-error)))))))

         (else
          ;; The key was found; p is an existing node.
          (values (%avl key data (%bal p) (%left p) (%right p))
                  #f))))

      (avl-check-usage
       (procedure? pred<?)
       "avl-insert expects a procedure as first argument")
      (if (avl-empty? tree)
          (%avl key data 0 #f #f)
          (let-values (((p fix-balance?) (search tree #f)))
            p)))

    (define (internal-error)
      (display "internal error\n" (current-error-port))
      (emergency-exit 123))

    (define (usage-error msg)
      (display "Procedure usage error:\n" (current-error-port))
      (display "  " (current-error-port))
      (display msg (current-error-port))
      (newline (current-error-port))
      (exit 1))

    )) ;; end library (avl-trees)


(define-library (associative-arrays)

  ;;
  ;; Persistent associative arrays for R7RS Scheme.
  ;;
  ;; The story:
  ;; 
  ;; An implementation of associative arrays, where keys are compared
  ;; with an ‘equal to’ predicate, typically has three parts:
  ;;
  ;;    * a hash function, which converts a key to a hash value; and
  ;;      the hash value either has a ‘less than’ predicate or can be
  ;;      put in a radix tree;
  ;;
  ;;    * a table keyed by the hash values;
  ;;
  ;;    * a way to resolve hash value collisions.
  ;;
  ;; At one extreme is the association list, which can be viewed as
  ;; having a hash function that *always* collides. At a nearly
  ;; opposite extreme are ideal hash trees, which never have
  ;; collisions, but which, towards that end, require hash values to
  ;; ‘grow’ on the fly.
  ;;
  ;; Perhaps the simplest form of associative array having all three
  ;; parts is ‘separate chaining’: the hash function generates an
  ;; integer modulo some table size; the table itself is an array of
  ;; that size; and collisions are resolved by falling back to an
  ;; association list.
  ;;
  ;; Below I use my solution to the AVL Tree task
  ;; (https://rosettacode.org/wiki/AVL_tree#Scheme) to implement
  ;; *persistent* (that is, ‘immutable’) associative arrays. The hash
  ;; function is whatever you want, as long as it produces (what
  ;; Scheme regards as) a real number. Hash value collisions are
  ;; resolved by falling back to association lists.
  ;;

  (export assoc-array)
  (export assoc-array?)
  (export assoc-array-set)
  (export assoc-array-ref)

  (import (scheme base))
  (import (scheme case-lambda))
  (import (scheme write))
  (import (avl-trees))

  (cond-expand
    (chicken (import (only (srfi 1) alist-delete)))
    ;; Insert whatever you need here for your Scheme.
    (else))

  (begin

    (define-record-type <assoc-array>
      (%assoc-array hashfunc pred=? default table)
      assoc-array?
      (hashfunc %hashfunc)
      (pred=? %pred=?)
      (default %default)
      (table %table))

    (define assoc-array
      ;; Create an associative array.
      (case-lambda
        ((hashfunc)
         (let ((pred=? equal?)
               (default #f))
           (assoc-array hashfunc pred=? default)))
        ((hashfunc pred=?)
         (let ((default #f))
           (assoc-array hashfunc pred=? default)))
        ((hashfunc pred=? default)
         (%assoc-array hashfunc pred=? default (avl)))))

    (define (assoc-array-set array key data)
      ;; Produce a new associative array that is the same as the input
      ;; array except for the given key-data association. The input
      ;; array is left unchanged (which is why the procedure is called
      ;; ‘assoc-array-set’ rather than ‘assoc-array-set!’).
      (let ((hashfunc (%hashfunc array))
            (pred=? (%pred=? array))
            (default (%default array))
            (table (%table array)))
        (let ((hash-value (hashfunc key)))
          ;; The following could be made more efficient by combining
          ;; the ‘search’ and ‘insert’ operations for the AVL tree.
          (let*-values
              (((alst found?) (avl-search-values < table hash-value)))
            (cond
             (found?
              ;; Add a new entry to the association list. Removal of
              ;; any old associations with the key is not strictly
              ;; necessary, but without it the associative array will
              ;; grow every time you replace an
              ;; association. (Alternatively, you could occasionally
              ;; clean the associative array of shadowed key
              ;; associations.)
              (let* ((alst (alist-delete key alst pred=?))
                     (alst `((,key . ,data) . ,alst))
                     (table (avl-insert < table hash-value alst)))
                (%assoc-array hashfunc pred=? default table)))
             (else
              ;; Start a new association list.
              (let* ((alst `((,key . ,data)))
                     (table (avl-insert < table hash-value alst)))
                (%assoc-array hashfunc pred=? default table))))))))

    (define (assoc-array-ref array key)
      ;; Return the data associated with the key. If the key is not in
      ;; the table, return the associative array’s default data.
      (let* ((hashfunc (%hashfunc array))
             (hash-value (hashfunc key)))
        (let*-values
            (((alst found?)
              (avl-search-values < (%table array) hash-value)))
          (if found?
              (let ((pair (assoc key alst (%pred=? array))))
                (if pair
                    (cdr pair)
                    (%default array)))
              (%default array)))))

    )) ;; end library (associative-arrays)


(cond-expand
  (DEMONSTRATION
   (begin
     (import (scheme base))
     (import (scheme write))
     (import (srfi 151))
     (import (associative-arrays))

     ;; I like SpookyHash, but for this demonstration I shall use the
     ;; simpler ‘ElfHash’ and define it only for strings. See
     ;; https://en.wikipedia.org/w/index.php?title=PJW_hash_function&oldid=997863283
     (define (hashfunc s)
       (let ((n (string-length s))
             (h 0))
         (do ((i 0 (+ i 1)))
             ((= i n))
           (let* ((ch
                   ;; If the character is outside the 8-bit range,
                   ;; probably I should break it into four bytes, each
                   ;; incorporated separately into the hash. For this
                   ;; demonstration, I shall simply discard the higher
                   ;; bits.
                   (bitwise-and (char->integer (string-ref s i))
                                #xFF))
                  (h^ (+ (arithmetic-shift h 4) ch))
                  (high^ (bitwise-and h^ #xF0000000)))
             (unless (zero? high^)
               (set! h^
                 (bitwise-xor h^ (arithmetic-shift high^ -24))))
             (set! h (bitwise-and h^ (bitwise-not high^)))))
         h))

     (let* ((a1 (assoc-array hashfunc))
            (a2 (assoc-array-set a1 "A" #\A))
            (a3 (assoc-array-set a2 "B" #x42)) ; ASCII ‘B’.
            (a4 (assoc-array-set a3 "C" "C")))
       (write (assoc-array-ref a1 "A")) (newline)
       (write (assoc-array-ref a1 "B")) (newline)
       (write (assoc-array-ref a1 "C")) (newline)
       (write (assoc-array-ref a2 "A")) (newline)
       (write (assoc-array-ref a2 "B")) (newline)
       (write (assoc-array-ref a2 "C")) (newline)
       (write (assoc-array-ref a3 "A")) (newline)
       (write (assoc-array-ref a3 "B")) (newline)
       (write (assoc-array-ref a3 "C")) (newline)
       (write (assoc-array-ref a4 "A")) (newline)
       (write (assoc-array-ref a4 "B")) (newline)
       (write (assoc-array-ref a4 "C")) (newline))

     ))
  (else))
Output:
$ csc -DDEMONSTRATION -R r7rs -X r7rs associative_array-scheme.scm && ./associative_array-scheme
#f
#f
#f
#\A
#f
#f
#\A
66
#f
#\A
66
"C"

It is easy to add generators to this implementation, by which I mean ‘iterators’ that are themselves executable procedures.

(Afterword. To be honest, I would not use the term ‘associative array’, because ‘array’ implies uniformly constant time access, which even a traditional hash table generally cannot provide. I would call these things ‘maps’—or, if the word ‘map’ is heavily used for something else [such as Scheme’s map procedures], then I would call them ‘dictionaries’.)

Seed7

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
  local
    var string: stri is "";
    var integer: number is 0;
  begin
    # Add elements
    aHash @:= ["foo"] 42;
    aHash @:= ["bar"] 100;

    # Check presence of an element
    if "foo" in aHash then

      # Access an element
      writeln(aHash["foo"]);
    end if;

    # Change an element
    aHash @:= ["foo"] 7;

    # Remove an element
    excl(aHash, "foo");

    # Loop over the hash values
    for number range aHash do
      writeln(number);
    end for;

    # Loop over the hash keys
    for key stri range aHash do
      writeln(stri);
    end for;

    # Loop over hash keys and values
    for number key stri range aHash do
      writeln("key: " <& stri <& ", value: " <& number);
    end for;
  end func;

SenseTalk

Associative arrays in SenseTalk are called property lists, or objects.

put {} into emptyPlist
put an empty property list into emptyPlist2

put {first:"Albert", last:"Einstein"} into Einstein
set Einstein's occupation to "Physicist"
put 1879 into Einstein.yearBorn
put "!!!" after the occupation of Einstein
put Einstein
Output:
{first:"Albert", last:"Einstein", occupation:"Physicist!!!", yearBorn:1879}

SETL

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') );
Output:
b

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'} );
Output:
{b}

SETL4

* Iterate over key-value pairs of a map

    map = new('map 1:one 2:two 3:three')
    
visit(domain(map),'expr to evaluate for each member')
    
visit(range(map),'expr to evaluate for each member')
    
    next 
        this = next(map)             :f(done)
        out(show(this) ':' show(get(map,this)) :next)
    done
    
    loop(d = domain(map)
    
    next
        out('next domain entry',next(d))    :s(next)
    done
    
    loop(d = range(map)
    
    next
        out('next domain entry',next(d))    :s(next)
    done

Sidef

var hash = Hash(
    key1 => 'value1',
    key2 => 'value2',
)

# Add a new key-value pair
hash{:key3} = 'value3'

Slate

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

Smalltalk

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

alternative:

Dictionary withAssociations:{ 'MI' -> 'Michigan' . 'MN' -> 'Minnesota'}

SNOBOL4

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

	output = t<"red">
	output = t<"blue"> 
	output = t<"green">
end

SQL

REM Create a table to associate keys with values
CREATE TABLE  associative_array ( KEY_COLUMN VARCHAR2(10), VALUE_COLUMN VARCHAR2(100)); .
REM Insert a Key Value Pair
INSERT (KEY_COLUMN, VALUE_COLUMN) VALUES ( 'VALUE','KEY');.
REM Retrieve a key value pair
SELECT aa.value_column FROM associative_array aa where aa.key_column = 'KEY';

SQL PL

Works with: Db2 LUW

version 9.7 or higher.

With SQL PL:

--#SET TERMINATOR @

SET SERVEROUTPUT ON @

BEGIN
 DECLARE TYPE ASSOC_ARRAY AS VARCHAR(20) ARRAY [VARCHAR(20)];
 DECLARE HASH ASSOC_ARRAY;
 SET HASH['key1'] = 'val1';
 SET HASH['key-2'] = 2;
 SET HASH['three'] = -238.83;
 SET HASH[4] = 'val3';

 CALL DBMS_OUTPUT.PUT_LINE(HASH['key1']);
 CALL DBMS_OUTPUT.PUT_LINE(HASH['key-2']);
 CALL DBMS_OUTPUT.PUT_LINE(HASH['three']);
 CALL DBMS_OUTPUT.PUT_LINE(HASH[4]);
 CALL DBMS_OUTPUT.PUT_LINE(HASH['5']);
END@

Output:

db2 -td@
db2 => BEGIN
...
db2 (cont.) => END @
END
DB21034E  The command was processed as an SQL statement because it was not a 
valid Command Line Processor command.  During SQL processing it returned:
SQL20439N  Array index with value "5" is out of range or does not exist.  
SQLSTATE=2202E

val1
2
-238.83
val3

Stata

mata
a=asarray_create()

// Add entries
asarray(a,"one",1)
asarray(a,"two",2)

// Check existence of key
asarray_contains(a,"two")

// Get a vector of all keys
asarray_keys(a)

// Number of entries
asarray_elements(a)

// End Mata session
end

Swift

// 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]

Symsyn

Symsyn implements Hash as a list of strings. Each name/value or key/value pair is stored in another string. The name/value pair is in the format 'name=value', the '=' is reserved.

 #+ 'name=bob' $hash   | add to hash
 #? 'name' $hash $S    | find 'name' and return 'bob' in $S
 #- 'name' $hash       | delete 'name=bob' from hash

Tcl

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

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

Works with: ksh
typeset -A hash
hash=( [key1]=val1 [key2]=val2 )
hash[key3]=val3
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

UnixPipes

A key value file can be considered as an associative array

map='p.map'

function init() {
cat <<EOF > $map
apple a
boy b
cat c
dog d
elephant e
EOF
}

function put() {
    k=$1; v=$2;
    del $k
    echo $v $k >> $map
 }

function get() {
    k=$1
    for v in $(cat $map | grep "$k$"); do
        echo $v
        break
    done
 }

function del() {
    k=$1
    temp=$(mktemp)
    mv $map $temp
    cat $temp | grep -v "$k$" > $map
}

function dump() {
    echo "-- Dump begin --"
    cat $map
    echo "-- Dump complete --"
}

init
get c
put c cow
get c
dump

Vala

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 

VBA

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
    h.RemoveAll
    Debug.Print h.Count
End Sub

Vim Script

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

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

LOCAL loCol As Collection, k, n, o
CLEAR 
*!* 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
FOR EACH o IN loCol FOXOBJECT
    ? o, loCol.GetKey(k)
    k = k + 1 
ENDFOR	
*!* 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
FOR EACH o IN loFruits FOXOBJECT
    ? o.Name, loFruits.GetKey(k)
    k = k + 1 
ENDFOR
*!* Get an item name by its key
? loFruits("P").Name


DEFINE CLASS fruit As Custom
PROCEDURE Init(tcName As String)
THIS.Name = tcName
ENDPROC
ENDDEFINE

V (Vlang)

fn main() {
    // make empty map
    mut my_map := map[string]int{}

    //s et value
    my_map['foo'] = 3

    // getting values
    y1 := my_map['foo']

    // remove keys
    my_map.delete('foo')

    // make map with values
    my_map = {
        'foo': 2
        'bar': 42
        'baz': -1
    }
}

Wart

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

Wren

Wren has a Map class built in.

var fruit = {}                   // creates an empty map
fruit[1] = "orange"              // associates a key of 1 with "orange"
fruit[2] = "apple"               // associates a key of 2 with "apple"
System.print(fruit[1])           // retrieves the value with a key of 1 and prints it out
fruit.remove(1)                  // removes the entry with a key of 1 from the map
System.print(fruit)              // prints the rest of the map
System.print()
var capitals = {                 // creates a new map with three entries
    "France": "Paris",
    "Germany": "Berlin",
    "Spain": "Madrid"
}
capitals["Russia"] = "Moscow"    // adds another entry
System.print(capitals["France"]) // retrieves the "France" entry and prints out its capital
capitals.remove("France")        // removes the "France" entry
System.print(capitals)           // prints all remaining entries
System.print(capitals.count)     // prints the number of remaining entries
System.print(capitals["Sweden"]) // prints the entry for Sweden (null as there isn't one)
Output:
orange
{2: apple}

Paris
{Russia: Moscow, Germany: Berlin, Spain: Madrid}
3
null

XLISP

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)"

XPL0

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.

Output:
B
?

zkl

zkl: Dictionary("one",1, "two",2, "three",3)
D(two:2,three:3,one:1)

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