Roman numerals/Encode

From Rosetta Code
Task
Roman numerals/Encode
You are encouraged to solve this task according to the task description, using any language you may know.
Task

Create a function taking a positive integer as its parameter and returning a string containing the Roman numeral representation of that integer. Modern Roman numerals are written by expressing each digit separately, starting with the left most digit and skipping any digit with a value of zero.


In Roman numerals:

  • 1990 is rendered: 1000=M, 900=CM, 90=XC; resulting in MCMXC
  • 2008 is written as 2000=MM, 8=VIII; or MMVIII
  • 1666 uses each Roman symbol in descending order: MDCLXVI



11l

Translation of: Python
V anums = [1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1]
V rnums = ‘M CM D CD C XC L XL X IX V IV I’.split(‘ ’)

F to_roman(=x)
   V ret = ‘’
   L(a, r) zip(:anums, :rnums)
      (V n, x) = divmod(x, a)
      ret ‘’= r * n
   R ret

V test = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 25, 30, 40,
          50, 60, 69, 70, 80, 90, 99, 100, 200, 300, 400, 500, 600, 666, 700, 800, 900, 1000,
          1009, 1444, 1666, 1945, 1997, 1999, 2000, 2008, 2010, 2011, 2500, 3000, 3999]
L(val) test
   print(val‘ - ’to_roman(val))

360 Assembly

*        Roman numerals Encode     - 11/05/2020
ROMAENC  CSECT
         USING  ROMAENC,R13        base register
         B      72(R15)            skip savearea
         DC     17F'0'             savearea
         SAVE   (14,12)            save previous context
         ST     R13,4(R15)         link backward
         ST     R15,8(R13)         link forward
         LR     R13,R15            set addressability
         LA     R6,1               i=1
   DO  WHILE=(C,R6,LE,=A(8))       do i=1 to hbound(nums)
         LR     R1,R6                i
         SLA    R1,1                 ~
         LH     R8,NUMS-2(R1)        n=nums(i)
         MVC    PG,=CL80'.... :'     clear buffer
         LA     R9,PG                @pg
         XDECO  R8,XDEC              edit n
         MVC    0(4,R9),XDEC+8       output n
         LA     R9,7(R9)             @pg+=7
         LA     R7,1                 j=1
   DO  WHILE=(C,R7,LE,=A(13))        do j=1 to 13
         LR     R1,R7                  j
         SLA    R1,1                   ~
         LH     R3,ARABIC-2(R1)        aj=arabic(j)
   DO WHILE=(CR,R8,GE,R3)              while n>=aj 
         LR     R1,R7                    j
         SLA    R1,1                     ~
         LA     R4,ROMAN-2(R1)           roman(j)
         MVC    0(2,R9),0(R4)            output roman(j)
       IF   CLI,1(R9),NE,C' ' THEN       if roman(j)[2]=' ' then
         LA     R9,2(R9)                   @pg+=2
       ELSE     ,                        else
         LA     R9,1(R9)                   @pg+=1
       ENDIF    ,                        endif
         SR     R8,R3                    n-=aj
   ENDDO        ,                      endwile
         LA     R7,1(R7)               j++
   ENDDO        ,                    enddo j
         XPRNT  PG,L'PG              print buffer
         LA     R6,1(R6)             i++
   ENDDO        ,                  enddo i
         L      R13,4(0,R13)       restore previous savearea pointer
         RETURN (14,12),RC=0       restore registers from calling save
ARABIC   DC     H'1000',H'900',H'500',H'400',H'100',H'90'
         DC     H'50',H'40',H'10',H'9',H'5',H'4',H'1'
ROMAN    DC     CL2'M',CL2'CM',CL2'D',CL2'CD',CL2'C',CL2'XC'
         DC     CL2'L',CL2'XL',CL2'X',CL2'IX',CL2'V',CL2'IV',CL2'I'
NUMS     DC   H'14',H'16',H'21',H'888',H'1492',H'1999',H'2020',H'3999'
PG       DS     CL80               buffer
XDEC     DS     CL12               temp for xdeco
         REGEQU
         END    ROMAENC
Output:
  14 : XIV
  16 : XVI
  21 : XXI
 888 : DCCCLXXXVIII
1492 : MCDXCII
1999 : MCMXCIX
2020 : MMXX
3999 : MMMCMXCIX

8080 Assembly

		org	100h
		jmp	test
		;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
		;; Takes a 16-bit integer in HL, and stores it
		;; as a 0-terminated string starting at BC.
		;; On exit, all registers destroyed; BC pointing at
		;; end of string. 
mkroman:	push	h	; put input on stack
		lxi	h,mkromantab
mkromandgt:	mov	a,m	; scan ahead to next entry
		ana	a
		inx	h
		jnz	mkromandgt
		xthl		; load number
		mov	a,h	; if zero, we're done
		ora	l
		jz	mkromandone
		xthl		; load next entry from table
		mov	e,m	; de = number
		inx	h
		mov	d,m
		inx	h
		xthl		; load number
		xra	a	; find how many we need
subtract:	inr	a	; with trial subtraction
		dad	d
		jc	subtract
		push	psw	; keep counter
		mov	a,d	; we subtracted one too many
		cma		; so we need to add one back
		mov	d,a
		mov	a,e
		cma
		mov	e,a
		inx	d
		dad	d
		pop	d	; restore counter (into D)
		xthl		; load table pointer
stringouter:	dcr	d	; do we need to include one?
		jz	mkromandgt
		push	h	; keep string location
stringinner:	mov	a,m	; copy string into target
		stax	b
		ana	a	; done yet?
		jz	stringdone
		inx	h
		inx	b	; copy next character
		jmp	stringinner
stringdone:	pop	h	; restore string location
		jmp	stringouter
mkromandone:	pop	d	; remove temporary variable from stack
		ret
mkromantab:	db	0
		db	18h,0fch,'M',0		; The value for each entry
		db	7ch,0fch,'CM',0		; is stored already negated
		db	0ch,0feh,'D',0		; so that it can be immediately
		db	70h,0feh,'CD',0		; added using `dad'.
		db	9ch,0ffh,'C',0		; This also has the convenient
		db	0a6h,0ffh,'XC',0	; property of not having any
		db	0ceh,0ffh,'L',0		; zero bytes except the string
		db	0d8h,0ffh,'XL',0	; and row terminators.
		db	0f6h,0ffh,'X',0
		db	0f7h,0ffh,'IX',0
		db	0fbh,0ffh,'V',0
		db	0fch,0ffh,'IV',0
		db	0ffh,0ffh,'I',0
		;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
		;; Test code
test:		mvi	c,10	; read string from console
		lxi	d,dgtbufdef
		call	5
		lxi	h,0	; convert to integer
		lxi	b,dgtbuf
readdgt:	ldax	b
		ana	a
		jz	convert
		dad	h	; hl *= 10
		mov	d,h
		mov	e,l
		dad	h
		dad	h
		dad	d
		sui	'0'
		mov	e,a
		mvi	d,0
		dad	d
		inx	b
		jmp	readdgt
convert:	lxi	b,romanbuf	; convert to roman
		call	mkroman
		mvi	a,'$'	; switch string terminator
		stax	b
		mvi	c,9	; output result
		lxi	d,romanbuf
		jmp	5
nl:		db	13,10,'$'
dgtbufdef:	db	5,0
dgtbuf:		ds	6
romanbuf:

8086 Assembly

Main and Supporting Functions

The main program and test values: 70,1776,2021,3999,4000

	mov ax,0070h
	call EncodeRoman
	mov si,offset StringRam
	call PrintString
	call NewLine	
	
	mov ax,1776h
	call EncodeRoman
	mov si,offset StringRam
	call PrintString
	call NewLine

	mov ax,2021h
	call EncodeRoman
	mov si,offset StringRam
	call PrintString
	call NewLine

	mov ax,3999h
	call EncodeRoman
	mov si,offset StringRam
	call PrintString
	call NewLine
	
	mov ax,4000h
	call EncodeRoman
	mov si,offset StringRam

	
	ReturnToDos     ;macro that calls the int that exits dos

The EncodeRoman routine:

;ROMAN NUMERALS MODULE

EncodeRoman:
;takes a BCD value in AX and stores its Roman numeral equivalent in ram.

	call UnpackBCD
	
	cmp dh,03h
	jng continue_EncodeRoman
	;roman numerals only go up to 3999.
	jmp errorhandler_encodeRoman_inputTooBig
continue_EncodeRoman:
	mov si,offset StringRam
	;using SI as destination of roman numerals.
	push ax	
	push cx
		mov ch,0
		mov cl,dh					;loop counter
		cmp dh,0
		jz skipThousands
encodeRoman_handleThousands:
		mov al,"M"
		mov [ds:si],al				;store in string ram
		inc si
		; call PrintChar
		loop encodeRoman_handleThousands
skipThousands:
	pop cx
	pop ax
	
encodeRoman_HandleHundreds:
	pushall
		mov bh,0
		mov bl,dl  ;use bx as an offset into Roman_Lookup_Master
		SHL bl,1
		SHL bl,1	;multiply by 2, we are indexing into a table with 4 bytes per row.
		mov di,offset Roman_Lookup_Master
		mov cx,4
getChar_Hundreds:		
		mov al,[bx+es:di] ;get first char index
		push bx
		push di
			mov di,offset Roman_Hund
			mov bl,al
			mov al,[bx+es:di]
			cmp al,0
			jz skipNullChar_RomanHund
			mov [ds:si],al	;store in ram
			inc si
			; call PrintChar			
skipNullChar_RomanHund:
		pop di
		pop bx
		inc di
		loop getChar_Hundreds
	popall
	
	
encodeRoman_HandleTens:
	pushall
		mov bh,0
		mov bl,ah  ;use bx as an offset into Roman_Lookup_Master
		SHL bl,1
		SHL bl,1	;multiply by 2, we are indexing into a table with 4 bytes per row.
		mov di,offset Roman_Lookup_Master
		mov cx,4
getChar_Tens:		
		mov al,[bx+es:di] ;get first char index
		push bx
		push di
			mov di,offset Roman_Tens
			mov bl,al
			mov al,[bx+es:di]
			cmp al,0
			jz skipNullChar_RomanTens
			mov [ds:si],al	;store in ram
			inc si
			; call PrintChar			
skipNullChar_RomanTens:
		pop di
		pop bx
		inc di
		loop getChar_Tens
	popall
	
encodeRoman_HandleOnes:
	pushall
		mov bh,0
		mov bl,al  ;use bx as an offset into Roman_Lookup_Master
		SHL bl,1
		SHL bl,1	;multiply by 2, we are indexing into a table with 4 bytes per row.
		mov di,offset Roman_Lookup_Master
		mov cx,4
getChar_Ones:		
		mov al,[bx+es:di] ;get first char index
		push bx
		push di
			mov di,offset Roman_Ones
			mov bl,al
			mov al,[bx+es:di]
			cmp al,0
			jz skipNullChar_RomanOnes
			mov [ds:si],al	;store in ram
			inc si
			; call PrintChar			
skipNullChar_RomanOnes:
		pop di
		pop bx
		inc di
		loop getChar_Ones
	popall
	
	mov al,0
	mov [ds:si],al ;place a null terminator at the end of the string.
	ret

			
errorhandler_encodeRoman_inputTooBig:
	push ds
	push ax
		LoadSegment ds,ax,@data
		mov al,01h
		mov byte ptr [ds:error_code],al
		mov ax, offset EncodeRoman
		mov word ptr [ds:error_routine],ax
		
		LoadSegment ds,ax,@code
		mov si,offset Roman_Error
		call PrintString
	pop ax
	pop ds
	stc		;set carry, allowing program to branch if error occurred.
	ret

	

Roman_Lookup_Master db 0,0,0,0	;0
		    db 0,0,0,1	;1
		    db 0,0,1,1	;2
		    db 0,1,1,1	;3
		    db 0,0,1,2	;4
		    db 0,0,0,2	;5
		    db 0,0,2,1	;6
		    db 0,2,1,1	;7
		    db 2,1,1,1	;8
		    db 0,0,1,3	;9

Roman_Ones	    db 0,"IVX"  ;the same pattern is used regardless of what power of 10 we're working with
Roman_Tens	    db 0,"XLC"
Roman_Hund	    db 0,"CDM"

Roman_Error	    db "ERROR: BAD INPUT",0


UnpackBCD:
;converts a "packed" BCD value in AX to an "unpacked" value in DX.AX
;DX is the high byte, AX is the low byte.
;CLOBBERS DX AND AX.
	mov dx,0
	mov dl,ah
	mov ah,0
	push cx
		mov cl,4
		rol dx,cl
		;BEFORE: DX = 00XYh
		;AFTER:  DX = 0XY0h
		ror dl,cl	;DX = 0X0Yh
		
		rol ax,cl	
		;BEFORE: AX = 00XYh
		;AFTER:  AX = 0XY0h
		ror al,cl	;AX = 0X0Yh
	pop cx
	ret

Macros used:

pushall macro
push ax
push bx
push cx
push dx
push ds
push es
push di
;I forgot SI in this macro, but once you add it in the code stops working! So I left it out.
endm

popall macro
pop di
pop es
pop ds
pop dx
pop cx
pop bx
pop ax
endm

Output

Output:
LXX
MDCCLXXVI
MMXXI
MMMCMXCIX
ERROR: BAD INPUT

Action!

DEFINE PTR="CARD"
CARD ARRAY arabic=[1000 900 500 400 100 90 50 40 10 9 5 4 1]
PTR ARRAY roman(13)

PROC InitRoman()
  roman(0)="M" roman(1)="CM" roman(2)="D" roman(3)="CD"
  roman(4)="C" roman(5)="XC" roman(6)="L" roman(7)="XL"
  roman(8)="X" roman(9)="IX" roman(10)="V" roman(11)="IV" roman(12)="I"
RETURN

PROC EncodeRomanNumber(CARD n CHAR ARRAY res)
  BYTE i,len
  CHAR ARRAY tmp

  res(0)=0 len=0
  FOR i=0 TO 12
  DO
    WHILE arabic(i)<=n
    DO
      tmp=roman(i)
      SAssign(res,tmp,len+1,len+1+tmp(0))
      len==+tmp(0)
      n==-arabic(i)
    OD
  OD
  res(0)=len
RETURN

PROC Main()
  CARD ARRAY data=[1990 2008 5555 1666 3888 3999]
  BYTE i
  CHAR ARRAY r(20)

  InitRoman()
  FOR i=0 TO 5
  DO
    EncodeRomanNumber(data(i),r)
    PrintF("%U=%S%E",data(i),r)
  OD
RETURN
Output:

Screenshot from Atari 8-bit computer

1990=MCMXC
2008=MMVIII
5555=MMMMMDLV
1666=MDCLXVI
3888=MMMDCCCLXXXVIII
3999=MMMCMXCIX

ActionScript

function arabic2roman(num:Number):String {
	var lookup:Object = {M:1000, CM:900, D:500, CD:400, C:100, XC:90, L:50, XL:40, X:10, IX:9, V:5, IV:4, I:1};
	var roman:String = "", i:String;
	for (i in lookup) {
		while (num >= lookup[i]) {
			roman += i;
			num -= lookup[i];
		}
	}
	return roman;
}
trace("1990 in roman is " + arabic2roman(1990));
trace("2008 in roman is " + arabic2roman(2008));
trace("1666 in roman is " + arabic2roman(1666));
Output:
1990 in roman is MCMXC
2008 in roman is MMVIII
1666 in roman is MDCLXVI

And the reverse:

function roman2arabic(roman:String):Number {
	var romanArr:Array = roman.toUpperCase().split('');
	var lookup:Object = {I:1, V:5, X:10, L:50, C:100, D:500, M:1000};
	var num:Number = 0, val:Number = 0;
	while (romanArr.length) {
		val = lookup[romanArr.shift()];
		num += val * (val < lookup[romanArr[0]] ? -1 : 1);
	}
	return num;
}
trace("MCMXC in arabic is " + roman2arabic("MCMXC"));
trace("MMVIII in arabic is " + roman2arabic("MMVIII"));
trace("MDCLXVI in arabic is " + roman2arabic("MDCLXVI"));
Output:
MCMXC in arabic is 1990
MMVIII in arabic is 2008
MDCLXVI in arabic is 1666

Ada

with Ada.Text_IO;  use Ada.Text_IO;

procedure Roman_Numeral_Test is
   function To_Roman (Number : Positive) return String is
      subtype Digit is Integer range 0..9;
      function Roman (Figure : Digit; I, V, X : Character) return String is
      begin
         case Figure is
            when 0 => return "";
            when 1 => return "" & I;
            when 2 => return I & I;
            when 3 => return I & I & I;
            when 4 => return I & V;
            when 5 => return "" & V;
            when 6 => return V & I;
            when 7 => return V & I & I;
            when 8 => return V & I & I & I;
            when 9 => return I & X;
         end case;
      end Roman;
   begin
      pragma Assert (Number >= 1 and Number < 4000);
      return
         Roman (Number / 1000,       'M', ' ', ' ') &
         Roman (Number / 100 mod 10, 'C', 'D', 'M') &
         Roman (Number / 10 mod 10,  'X', 'L', 'C') &
         Roman (Number mod 10,       'I', 'V', 'X');
   end To_Roman;
begin
   Put_Line (To_Roman (1999));
   Put_Line (To_Roman (25));
   Put_Line (To_Roman (944));
end Roman_Numeral_Test;
Output:
 MCMXCIX
 XXV
 CMXLIV

ALGOL 68

Works with: ALGOL 68 version Revision 1 - no extensions to language used
Works with: ALGOL 68G version Any - tested with release 1.18.0-9h.tiny
Works with: ELLA ALGOL 68 version Any (with appropriate job cards) - tested with release 1.8-8d
[]CHAR roman =        "MDCLXVmdclxvi"; # UPPERCASE for thousands #
[]CHAR adjust roman = "CCXXmmccxxii";
[]INT arabic =       (1000000, 500000, 100000, 50000, 10000, 5000, 1000, 500, 100, 50, 10, 5, 1);
[]INT adjust arabic = (100000, 100000,  10000, 10000,  1000, 1000,  100, 100,  10, 10,  1, 1, 0);

PROC arabic to roman = (INT dclxvi)STRING: (
  INT in := dclxvi; # 666 #
  STRING out := "";
  FOR scale TO UPB roman WHILE in /= 0 DO
    INT multiples = in OVER arabic[scale];
    in -:= arabic[scale] * multiples;
    out +:= roman[scale] * multiples;
    IF in >= -adjust arabic[scale] + arabic[scale] THEN
      in -:= -adjust arabic[scale] + arabic[scale];
      out +:=  adjust roman[scale] +  roman[scale]
    FI
  OD;
  out
);

main:(
  []INT test = (1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,25,30,40,50,60,69,70,
     80,90,99,100,200,300,400,500,600,666,700,800,900,1000,1009,1444,1666,1945,1997,1999,
     2000,2008,2500,3000,4000,4999,5000,6666,10000,50000,100000,500000,1000000,max int);
  FOR key TO UPB test DO
    INT val = test[key];
    print((val, " - ", arabic to roman(val), new line))
  OD
)
Output:
(last example is manually wrapped)
         +1 - i
         +2 - ii
         +3 - iii
         +4 - iv
         +5 - v
         +6 - vi
         +7 - vii
         +8 - viii
         +9 - ix
        +10 - x
        +11 - xi
        +12 - xii
        +13 - xiii
        +14 - xiv
        +15 - xv
        +16 - xvi
        +17 - xvii
        +18 - xviii
        +19 - xix
        +20 - xx
        +25 - xxv
        +30 - xxx
        +40 - xl
        +50 - l
        +60 - lx
        +69 - lxix
        +70 - lxx
        +80 - lxxx
        +90 - xc
        +99 - xcix
       +100 - c
       +200 - cc
       +300 - ccc
       +400 - cd
       +500 - d
       +600 - dc
       +666 - dclxvi
       +700 - dcc
       +800 - dccc
       +900 - cm
      +1000 - m
      +1009 - mix
      +1444 - mcdxliv
      +1666 - mdclxvi
      +1945 - mcmxlv
      +1997 - mcmxcvii
      +1999 - mcmxcix
      +2000 - mm
      +2008 - mmviii
      +2500 - mmd
      +3000 - mmm
      +4000 - mV
      +4999 - mVcmxcix
      +5000 - V
      +6666 - Vmdclxvi
     +10000 - X
     +50000 - L
    +100000 - C
    +500000 - D
   +1000000 - M
+2147483647 - MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
              MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMCDLXXXmmmdcxlvii

ALGOL W

Works with: awtoc version any - tested with release Mon Apr 27 14:25:27 NZST 2009
BEGIN

PROCEDURE ROMAN (INTEGER VALUE NUMBER; STRING(15) RESULT CHARACTERS; INTEGER RESULT LENGTH);
    COMMENT
         Returns the Roman number of an integer between 1 and 3999.
         "MMMDCCCLXXXVIII" (15 characters long) is the longest Roman number under 4000;
    BEGIN
        INTEGER PLACE, POWER;

        PROCEDURE APPEND (STRING(1) VALUE C);
            BEGIN CHARACTERS(LENGTH|1) := C; LENGTH := LENGTH + 1 END;

        PROCEDURE I; APPEND(CASE PLACE OF ("I","X","C","M"));
        PROCEDURE V; APPEND(CASE PLACE OF ("V","L","D"));
        PROCEDURE X; APPEND(CASE PLACE OF ("X","C","M"));

        ASSERT (NUMBER >= 1) AND (NUMBER < 4000);

        CHARACTERS := "               ";  
        LENGTH := 0;
        POWER := 1000;  
        PLACE := 4;
        WHILE PLACE > 0 DO
            BEGIN
                CASE NUMBER DIV POWER + 1 OF BEGIN
                    BEGIN            END;
                    BEGIN I          END;
                    BEGIN I; I       END;
                    BEGIN I; I; I    END;
                    BEGIN I; V       END;
                    BEGIN V          END;
                    BEGIN V; I       END;
                    BEGIN V; I; I    END;
                    BEGIN V; I; I; I END;
                    BEGIN I; X       END
                END;
                NUMBER := NUMBER REM POWER;
                POWER := POWER DIV 10;
                PLACE := PLACE - 1
            END
    END ROMAN;

INTEGER I;
STRING(15) S;

ROMAN(1, S, I);    WRITE(S, I);
ROMAN(3999, S, I); WRITE(S, I);
ROMAN(3888, S, I); WRITE(S, I);
ROMAN(2009, S, I); WRITE(S, I);
ROMAN(405, S, I);  WRITE(S, I);
END.
Output:
I                           1
MMMCMXCIX                   9
MMMDCCCLXXXVIII            15
MMIX                        4
CDV                         3

APL

Works with: Dyalog APL
toRoman{
    ⍝ Digits and corresponding values
    ds((⊢≠⊃)⊆⊢)' M CM D CD C XC L XL X IX V IV I'
    vs1000, ,100 10 1∘.×9 5 4 1
    ⍝ Input ≤ 0 is invalid
    0:⎕SIGNAL 11
    {   0=d⊃⍸vs⍵:    ⍝ Find highest digit in number
        (dds),∇⍵-dvs  ⍝ While one exists, add it and subtract from number
    }
}
Output:
      toRoman¨ 1990 2008 1666 2021
 MCMXC  MMVIII  MDCLXVI  MMXXI 

AppleScript

Translation of: JavaScript

(ES6 version)

Translation of: Haskell

(mapAccumL version)

------------------ ROMAN INTEGER STRINGS -----------------

-- roman :: Int -> String
on roman(n)
    set kvs to {["M", 1000], ["CM", 900], ["D", 500], ¬
        ["CD", 400], ["C", 100], ["XC", 90], ["L", 50], ¬
        ["XL", 40], ["X", 10], ["IX", 9], ["V", 5], ¬
        ["IV", 4], ["I", 1]}
    
    script stringAddedValueDeducted
        on |λ|(balance, kv)
            set {k, v} to kv
            set {q, r} to quotRem(balance, v)
            if q > 0 then
                {r, concat(replicate(q, k))}
            else
                {r, ""}
            end if
        end |λ|
    end script
    
    concat(snd(mapAccumL(stringAddedValueDeducted, n, kvs)))
end roman


--------------------------- TEST -------------------------
on run
    
    map(roman, [2016, 1990, 2008, 2000, 1666])
    
    --> {"MMXVI", "MCMXC", "MMVIII", "MM", "MDCLXVI"}
end run


---------------- GENERIC LIBRARY FUNCTIONS ---------------

-- concat :: [[a]] -> [a] | [String] -> String
on concat(xs)
    script append
        on |λ|(a, b)
            a & b
        end |λ|
    end script
    
    if length of xs > 0 and ¬
        class of (item 1 of xs) is string then
        set unit to ""
    else
        set unit to {}
    end if
    foldl(append, unit, xs)
end concat

-- foldl :: (a -> b -> a) -> a -> [b] -> a
on foldl(f, startValue, xs)
    tell mReturn(f)
        set v to startValue
        set lng to length of xs
        repeat with i from 1 to lng
            set v to |λ|(v, item i of xs, i, xs)
        end repeat
        return v
    end tell
end foldl

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

-- 'The mapAccumL function behaves like a combination of map and foldl; 
-- it applies a function to each element of a list, passing an 
-- accumulating parameter from left to right, and returning a final 
-- value of this accumulator together with the new list.' (see Hoogle)

-- mapAccumL :: (acc -> x -> (acc, y)) -> acc -> [x] -> (acc, [y])
on mapAccumL(f, acc, xs)
    script
        on |λ|(a, x)
            tell mReturn(f) to set pair to |λ|(item 1 of a, x)
            [item 1 of pair, (item 2 of a) & {item 2 of pair}]
        end |λ|
    end script
    
    foldl(result, [acc, {}], xs)
end mapAccumL

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

--  quotRem :: Integral a => a -> a -> (a, a)
on quotRem(m, n)
    {m div n, m mod n}
end quotRem

-- Egyptian multiplication - progressively doubling a list, appending
-- stages of doubling to an accumulator where needed for binary 
-- assembly of a target length

-- replicate :: Int -> a -> [a]
on replicate(n, a)
    set out to {}
    if n < 1 then return out
    set dbl to {a}
    
    repeat while (n > 1)
        if (n mod 2) > 0 then set out to out & dbl
        set n to (n div 2)
        set dbl to (dbl & dbl)
    end repeat
    return out & dbl
end replicate

-- snd :: (a, b) -> b
on snd(xs)
    if class of xs is list and length of xs = 2 then
        item 2 of xs
    else
        missing value
    end if
end snd
Output:
{"MMXVI", "MCMXC", "MMVIII", "MM", "MDCLXVI"}

Arturo

Translation of: Nim
nums: [[1000 "M"] [900 "CM"] [500 "D"] [400 "CD"] [100 "C"] [90 "XC"] 
        [50 "L"] [40 "XL"] [10 "X"] [9 "IX"] [5 "V"] [4 "IV"] [1 "I"])
 
toRoman: function [x][
   ret: ""
   idx: 0
   initial: x
   loop nums 'num [
      d: num\0
      l: num\1

      i: 0 
      while [i<initial/d] [
         ret: ret ++ l
         i: i+1
      ]

      initial: mod initial d
   ]
   return ret
]
 
loop [1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 25 30 40
      50 60 69 70 80 90 99 100 200 300 400 500 600 666 700 800 900
      1000 1009 1444 1666 1945 1997 1999 2000 2008 2010 2011 2500
      3000 3999] 'n 
   -> print [n "->" toRoman n]
Output:
1 -> I 
2 -> II 
3 -> III 
4 -> IV 
5 -> V 
6 -> VI 
7 -> VII 
8 -> VIII 
9 -> IX 
10 -> X 
11 -> XI 
12 -> XII 
13 -> XIII 
14 -> XIV 
15 -> XV 
16 -> XVI 
17 -> XVII 
18 -> XVIII 
19 -> XIX 
20 -> XX 
25 -> XXV 
30 -> XXX 
40 -> XL 
50 -> L 
60 -> LX 
69 -> LXIX 
70 -> LXX 
80 -> LXXX 
90 -> XC 
99 -> XCIX 
100 -> C 
200 -> CC 
300 -> CCC 
400 -> CD 
500 -> D 
600 -> DC 
666 -> DCLXVI 
700 -> DCC 
800 -> DCCC 
900 -> CM 
1000 -> M 
1009 -> MIX 
1444 -> MCDXLIV 
1666 -> MDCLXVI 
1945 -> MCMXLV 
1997 -> MCMXCVII 
1999 -> MCMXCIX 
2000 -> MM 
2008 -> MMVIII 
2010 -> MMX 
2011 -> MMXI 
2500 -> MMD 
3000 -> MMM 
3999 -> MMMCMXCIX

AutoHotkey

Translation of: C++
MsgBox % stor(444)

stor(value)
{
  romans = M,CM,D,CD,C,XC,L,XL,X,IX,V,IV,I
  M := 1000
  CM := 900
  D := 500
  CD := 400
  C := 100
  XC := 90
  L := 50
  XL := 40
  X := 10
  IX := 9
  V := 5
  IV := 4
  I := 1
  Loop, Parse, romans, `,
  {
    While, value >= %A_LoopField%
    {
      result .= A_LoopField
      value := value - (%A_LoopField%)
    }
  }
  Return result . "O" 
}

Autolisp

(defun c:roman() (romanNumber (getint "\n Enter number > "))
(defun romanNumber (n / uni dec hun tho nstr strlist nlist rom)
  (if (and (> n 0) (<= n 3999))
    (progn
       (setq 
         UNI (list "" "I" "II" "III" "IV" "V" "VI" "VII" "VIII" "IX")
         DEC (list "" "X" "XX" "XXX" "XL" "L" "LX" "LXX" "LXXX" "XC")
         HUN (list "" "C" "CC" "CCC" "CD" "D" "DC" "DCC" "DCCC" "CM")
         THO (list "" "M" "MM" "MMM")
         nstr (itoa n)
       )
       (while (> (strlen nstr) 0) (setq strlist (append strlist (list (substr nstr 1 1))) nstr (substr nstr 2 (strlen nstr))))
       (setq nlist (mapcar 'atoi strlist))
       (cond
          ((> n 999)(setq rom(strcat(nth (car nlist) THO)(nth (cadr nlist) HUN)(nth (caddr nlist) DEC) (nth (last nlist)UNI ))))
          ((and (> n 99)(<= n 999))(setq rom(strcat (nth (car nlist) HUN)(nth (cadr nlist) DEC) (nth (last nlist)UNI ))))
          ((and (> n 9)(<= n 99))(setq rom(strcat (nth (car nlist) DEC) (nth (last nlist)UNI ))))
          ((<= n 9)(setq rom(nth (last nlist)UNI)))
          )
        )
        (princ "\nNumber out of range!")
    )
rom
)
Output:
1577 "MDLXXVII" 
3999 "MMMCMXCIX"
888 "DCCCLXXXVIII" 
159 "CLIX" 

AWK

# syntax: GAWK -f ROMAN_NUMERALS_ENCODE.AWK
BEGIN {
    leng = split("1990 2008 1666",arr," ")
    for (i=1; i<=leng; i++) {
      n = arr[i]
      printf("%s = %s\n",n,dec2roman(n))
    }
    exit(0)
}
function dec2roman(number,  v,w,x,y,roman1,roman10,roman100,roman1000) {
    number = int(number) # force to integer
    if (number < 1 || number > 3999) { # number is too small | big
      return
    }
    split("I II III IV V VI VII VIII IX",roman1," ")   # 1 2 ... 9
    split("X XX XXX XL L LX LXX LXXX XC",roman10," ")  # 10 20 ... 90
    split("C CC CCC CD D DC DCC DCCC CM",roman100," ") # 100 200 ... 900
    split("M MM MMM",roman1000," ")                    # 1000 2000 3000
    v = (number - (number % 1000)) / 1000
    number = number % 1000
    w = (number - (number % 100)) / 100
    number = number % 100
    x = (number - (number % 10)) / 10
    y = number % 10
    return(roman1000[v] roman100[w] roman10[x] roman1[y])
}
Output:
1990 = MCMXC
2008 = MMVIII
1666 = MDCLXVI

BASIC

Applesoft BASIC

 1 N = 1990: GOSUB 5: PRINT N" = "V$
 2 N = 2008: GOSUB 5: PRINT N" = "V$
 3 N = 1666: GOSUB 5: PRINT N" = "V$;
 4  END 
 5 V = N:V$ = "": FOR I = 0 TO 12: FOR L = 1 TO 0 STEP 0:A =  VAL ( MID$ ("1E3900500400100+90+50+40+10+09+05+04+01",I * 3 + 1,3))
 6 L = (V - A) >  = 0:V$ = V$ +  MID$ ("M.CMD.CDC.XCL.XLX.IXV.IVI",I * 2 + 1,(I -  INT (I / 2) * 2 + 1) * L):V = V - A * L: NEXT L,I
 7  RETURN

ASIC

Translation of: DWScript
REM Roman numerals/Encode
DIM Weights(12)
DIM Symbols$(12)
DATA 1000, "M", 900, "CM", 500, "D", 400, "CD", 100, "C", 90, "XC", 50, "L"
DATA 40, "XL", 10, "X", 9, "IX", 5, "V", 4, "IV", 1, "I"
REM 3888 or MMMDCCCLXXXVIII (15 chars) is the longest string properly encoded 
REM with these symbols.
FOR J = 0 TO 12
  READ Weights(J)
  READ Symbols$(J)
NEXT J

AValue = 1990
GOSUB ToRoman: 
PRINT Roman$ 
REM MCMXC
AValue = 2022
GOSUB ToRoman:
PRINT Roman$ 
REM MMXXII
AValue = 3888
GOSUB ToRoman:
PRINT Roman$ 
REM MMMDCCCLXXXVIII
END

ToRoman:
REM Result: Roman$
Roman$ = ""
I = 0
Loop:
  IF (I > 12 THEN ExitToRoman:
  IF AValue <= 0 THEN ExitToRoman:
  WHILE AValue >= Weights(I)
    Roman$ = Roman$ + Symbols$(I)
    AValue = AValue - Weights(I)
  WEND
  I = I + 1
  GOTO Loop:
ExitToRoman:
RETURN

BaCon

OPTION BASE 1

GLOBAL roman$[] = { "M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I" }
GLOBAL number[] = { 1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1 }

FUNCTION toroman$(value)

    LOCAL result$

    DOTIMES UBOUND(number)
        WHILE value >= number[_]
            result$ = result$ & roman$[_]
            DECR value, number[_]
        WEND
    DONE

    RETURN result$

ENDFUNC

PRINT toroman$(1990)
PRINT toroman$(2008)
PRINT toroman$(1666)
Output:
MCMXC
MMVIII
MDCLXVI

BASIC256

Works with: BASIC256
    
print 1666+" = "+convert$(1666)
print 2008+" = "+convert$(2008)
print 1001+" = "+convert$(1001)
print 1999+" = "+convert$(1999)

function convert$(value)
convert$=""
arabic = {1000, 900, 500, 400, 100, 90, 50,  40,  10,  9,  5,   4,  1 }
roman$ = {"M", "CM", "D","CD", "C","XC","L","XL","X","IX","V","IV","I"}
   for i = 0 to 12
           while value >= arabic[i]
	    convert$ += roman$[i]
	    value  = value - arabic[i]
	 end while
    next i
end function
Output:
1666 = MDCLXVI
2008 = MMVIII
1001 = MI
1999 = MCMXCIX

BBC BASIC

      PRINT ;1999, FNroman(1999)
      PRINT ;2012, FNroman(2012)
      PRINT ;1666, FNroman(1666)
      PRINT ;3888, FNroman(3888)
      END
      
      DEF FNroman(n%)
      LOCAL i%, r$, arabic%(), roman$()
      DIM arabic%(12), roman$(12)
      arabic%() = 1,   4,   5,   9,  10,  40,  50,  90, 100, 400, 500, 900,1000
      roman$() = "I","IV", "V","IX", "X","XL", "L","XC", "C","CD", "D","CM", "M"
      FOR i% = 12 TO 0 STEP -1
        WHILE n% >= arabic%(i%)
          r$ += roman$(i%)
          n% -= arabic%(i%)
        ENDWHILE
      NEXT
      = r$
Output:
1999      MCMXCIX
2012      MMXII
1666      MDCLXVI
3888      MMMDCCCLXXXVIII

Chipmunk Basic

Works with: Chipmunk Basic version 3.6.4
Translation of: GW-BASIC
100 cls
110 dim arabic(12), roman$(12)
120 for j = 0 to 12 : read arabic(j),roman$(j) : next j
130 data 1000,"M", 900,"CM", 500,"D", 400,"CD", 100,"C", 90,"XC"
140 data 50,"L",40,"XL",10,"X",9,"IX",5,"V",4,"IV",1,"I"
187 avalor = 1990 : print avalor "= "; : gosub 220 : print roman$ ' MCMXC
188 avalor = 2008 : print avalor "= "; : gosub 220 : print roman$ ' MMXXII
189 avalor = 1666 : print avalor "= "; : gosub 220 : print roman$ ' MDCLXVI
200 end
210 rem Encode to Roman
220 roman$ = "" : i = 0
230 while (i <= 12) and (avalor > 0)
240  while avalor >= arabic(i)
250   roman$ = roman$+roman$(i)
260   avalor = avalor-arabic(i)
270  wend
280  i = i+1
290 wend
300 return
Output:
1990 = MCMXC
2008 = MMVIII
1666 = MDCLXVI

Commodore BASIC

Works with: Commodore BASIC version 7.0

C-128 version:

100 DIM RN$(12),NV(12)
110 FOR I=0 TO 12
120 : READ RN$(I), NV(I)
130 NEXT I
140 DATA M,1000, CM,900, D,500, CD,400
150 DATA C, 100, XC, 90, L, 50, XL, 40
160 DATA X,  10, IX,  9, V,  5, IV,  4
170 DATA I,   1
180 PRINT CHR$(19);CHR$(19);CHR$(147);CHR$(18);
190 PRINT "*****    ROMAN NUMERAL ENCODER     *****";CHR$(27);"T"
200 DO
210 : PRINT "ENTER NUMBER (0 TO QUIT):";
220 : OPEN 1,0:INPUT#1,AN$:CLOSE 1:PRINT
230 : AN=VAL(AN$):IF AN=0 THEN EXIT
240 : RN$=""
250 : DO WHILE AN > 0
260 :   FOR I=0 TO 12
270 :     IF AN >= NV(I) THEN BEGIN
280 :       RN$ = RN$+ RN$(I)
290 :       AN = AN - NV(I)
300 :       GOTO 330
310 :     BEND
320 :   NEXT I
330 : LOOP
340 : PRINT RN$;CHR$(13)
350 LOOP
Works with: Commodore BASIC version 3.5

C-16/116/Plus-4 version (BASIC 3.5 has DO/LOOP but not BEGIN/BEND)

100 DIM RN$(12),NV(12)
110 FOR I=0 TO 12
120 : READ RN$(I), NV(I)
130 NEXT I
140 DATA M,1000, CM,900, D,500, CD,400
150 DATA C, 100, XC, 90, L, 50, XL, 40
160 DATA X,  10, IX,  9, V,  5, IV,  4
170 DATA I,   1
180 PRINT CHR$(19);CHR$(19);CHR$(147);CHR$(18);
190 PRINT "*****    ROMAN NUMERAL ENCODER     *****";CHR$(27);"T"
200 DO
210 : PRINT "ENTER NUMBER (0 TO QUIT):";
220 : OPEN 1,0:INPUT#1,AN$:CLOSE 1:PRINT
230 : AN=VAL(AN$):IF AN=0 THEN EXIT
240 : RN$=""
250 : DO WHILE AN > 0
260 :   FOR I=0 TO 12
270 :     IF AN < NV(I) THEN 320
280 :     RN$ = RN$+ RN$(I)
290 :     AN = AN - NV(I)
300 :     I = 12
320 :   NEXT I
330 : LOOP
340 : PRINT RN$;CHR$(13)
350 LOOP
Works with: Commodore BASIC version 2.0

This version works on any Commodore, though the title banner should be adjusted to match the color and screen width of the particular machine.

100 DIM RN$(12),NV(12)
110 FOR I=0 TO 12
120 : READ RN$(I), NV(I)
130 NEXT I
140 DATA M,1000, CM,900, D,500, CD,400
150 DATA C, 100, XC, 90, L, 50, XL, 40
160 DATA X,  10, IX,  9, V,  5, IV,  4
170 DATA I,   1
180 PRINT CHR$(19);CHR$(19);CHR$(147);CHR$(18);
190 PRINT "*****    ROMAN NUMERAL ENCODER     *****";
200 REM BEGIN MAIN LOOP
210 : PRINT "NUMBER (0 TO QUIT):";
220 : OPEN 1,0:INPUT#1,AN$:CLOSE 1:PRINT
230 : AN=VAL(AN$):IF AN=0 THEN END
240 : RN$=""
250 : IF AN <= 0 THEN 340
260 :   FOR I=0 TO 12
270 :     IF AN < NV(I) THEN 320
280 :     RN$ = RN$+ RN$(I)
290 :     AN = AN - NV(I)
300 :     I = 12
320 :   NEXT I
330 : GOTO 250
340 : PRINT RN$;CHR$(13)
350 GOTO 210

The output is the same for all the above versions:

Output:
*****    ROMAN NUMERAL ENCODER     *****

ENTER NUMBER (0 TO QUIT):2009
MMIX

ENTER NUMBER (0 TO QUIT):1666
MDCLXVI

ENTER NUMBER (0 TO QUIT):3888
MMMDCCCLXXXVIII

ENTER NUMBER (0 TO QUIT):0

READY.

FreeBASIC

Works with: FreeBASIC
DIM SHARED arabic(0 TO 12) AS Integer  => {1000, 900, 500, 400, 100, 90, 50,  40,  10,  9,  5,   4,  1 }
DIM SHARED  roman(0 TO 12) AS String*2 => {"M", "CM", "D","CD", "C","XC","L","XL","X","IX","V","IV","I"}

FUNCTION toRoman(value AS Integer) AS String
    DIM i 	AS Integer
    DIM result  AS String
    
    FOR i = 0 TO 12
        DO WHILE value >= arabic(i)
	    result = result + roman(i)
	    value  = value - arabic(i)
	LOOP
    NEXT i
    toRoman = result
END FUNCTION

'Testing
PRINT "2009 = "; toRoman(2009)
PRINT "1666 = "; toRoman(1666)
PRINT "3888 = "; toRoman(3888)
Output:
 2009 = MMIX
 1666 = MDCLXVI
 3888 = MMMDCCCLXXXVIII

Another solution:

' FB 1.05.0 Win64

Function romanEncode(n As Integer) As String
  If n < 1 OrElse n > 3999 Then Return "" '' can only encode numbers in range 1 to 3999
  Dim roman1(0 To 2) As String = {"MMM", "MM", "M"}
  Dim roman2(0 To 8) As String = {"CM", "DCCC", "DCC", "DC", "D", "CD", "CCC", "CC", "C"}
  Dim roman3(0 To 8) As String = {"XC", "LXXX", "LXX", "LX", "L", "XL", "XXX", "XX", "X"}
  Dim roman4(0 To 8) As String = {"IX", "VIII", "VII", "VI", "V", "IV", "III", "II", "I"}
  Dim As Integer thousands, hundreds, tens, units
  thousands = n \ 1000
  n Mod= 1000
  hundreds = n \ 100
  n Mod= 100
  tens = n \ 10
  units = n Mod 10
  Dim roman As String = ""
  If thousands > 0  Then roman += roman1(3 - thousands)
  If hundreds > 0   Then roman += roman2(9 - hundreds)
  If tens > 0       Then roman += roman3(9 - tens)
  If units > 0      Then roman += roman4(9 - units)
  Return roman
End Function

Dim a(2) As Integer = {1990, 2008, 1666}
For i As Integer = 0 To 2
  Print a(i); " => "; romanEncode(a(i))
Next

Print
Print "Press any key to quit"
Sleep
Output:
 1990 => MCMXC
 2008 => MMVIII
 1666 => MDCLXVI

FutureBasic

window 1

local fn DecimaltoRoman( decimal as short ) as Str15
  short arabic(12)
  Str15 roman(12)
  long  i
  Str15 result : result = ""
  
  arabic(0) = 1000 : arabic(1) = 900 : arabic(2) = 500 : arabic(3) = 400
  arabic(4) = 100  : arabic(5) = 90  : arabic(6) = 50  : arabic(7)  = 40
  arabic(8) = 10   : arabic(9) = 9   : arabic(10) = 5  : arabic(11) = 4: arabic(12) = 1
  
  roman(0) = "M" : roman(1) = "CM" : roman(2) = "D"  : roman(3)  = "CD"
  roman(4) = "C" : roman(5) = "XC" : roman(6) = "L"  : roman(7)  = "XL"
  roman(8) = "X" : roman(9) = "IX" : roman(10) = "V" : roman(11) = "IV" : roman(12) = "I"
  
  for i = 0 to 12
    while ( decimal >= arabic(i) )
      result = result + roman(i)
      decimal = decimal - arabic(i)
    wend
  next i
  if result == "" then result = "Zepherium"
end fn = result

print "1990 = "; fn DecimaltoRoman( 1990 )
print "2008 = "; fn DecimaltoRoman( 2008 )
print "2016 = "; fn DecimaltoRoman( 2016 )
print "1666 = "; fn DecimaltoRoman( 1666 )
print "3888 = "; fn DecimaltoRoman( 3888 )
print "1914 = "; fn DecimaltoRoman( 1914 )
print "1000 = "; fn DecimaltoRoman( 1000 )
print " 513 = "; fn DecimaltoRoman(  513 )
print "  33 = "; fn DecimaltoRoman(   33 )

HandleEvents

Output:

1990 = MCMXC
2008 = MMVIII
2016 = MMXVI
1666 = MDCLXVI
3888 = MMMDCCCLXXXVIII
1914 = MCMXIV
1000 = M
 513 = DXIII
  33 = XXXIII

Gambas

Translation of: FreeBASIC
Public Sub Main() 
  
  'Testing
  Print "2009 = "; toRoman(2009) 
  Print "1666 = "; toRoman(1666) 
  Print "3888 = "; toRoman(3888)  

End 

Function toRoman(value As Integer) As String 

  Dim result As String 
  Dim arabic As Integer[] = [1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1]
  Dim roman As String[] = ["M", "CM", "D", "CD", "C", "XC", "L" , "XL", "X", "IX", "V", "IV", "I"]

  
  For i As Integer = 0 To arabic.Max
    Do While value >= arabic[i] 
      result &= roman[i] 
      value -= arabic[i] 
    Loop 
  Next
  Return result 

End Function
Output:
Same as FreeBASIC entry.

GW-BASIC

Translation of: DWScript
Works with: BASICA
 
10 REM Roman numerals/Encode
20 DIM WEIGHTS%(12), SYMBOLS$(12)
30 FOR J% = 0 TO 12: READ WEIGHTS%(J%), SYMBOLS$(J%): NEXT J%
40 DATA 1000, "M", 900, "CM", 500, "D", 400, "CD", 100, "C", 90, "XC"
50 DATA 50, "L", 40, "XL", 10, "X", 9, "IX", 5, "V", 4, "IV", 1, "I"
60 REM 3888 or MMMDCCCLXXXVIII (15 chars) is
70 REM the longest string properly encoded
80 REM with these symbols.
90 AVALUE% = 1990: GOSUB 1000: PRINT ROMAN$ ' MCMXC
100 AVALUE% = 2022: GOSUB 1000: PRINT ROMAN$ ' MMXXII
110 AVALUE% = 3888: GOSUB 1000: PRINT ROMAN$ ' MMMDCCCLXXXVIII
120 END
990 REM Encode to roman
1000 ROMAN$ = "": I% = 0
1010 WHILE (I% <= 12) AND (AVALUE% > 0)
1020  WHILE AVALUE% >= WEIGHTS%(I%)
1030   ROMAN$ = ROMAN$ + SYMBOLS$(I%)
1040   AVALUE% = AVALUE% - WEIGHTS%(I%)
1050  WEND
1060  I% = I% + 1
1070 WEND
1080 RETURN
Output:
MCMXC
MMXXII
MMMDCCCLXXXVIII

IS-BASIC

100 PROGRAM "Roman.bas"
110 DO 
120   PRINT :INPUT PROMPT "Enter an arabic number: ":N
130   IF N<1 THEN EXIT DO
140   PRINT TOROMAN$(N)
150 LOOP 
160 DEF TOROMAN$(X)
170   IF X>3999 THEN
180     LET TOROMAN$="Too big."
190     EXIT DEF
200   END IF 
210   RESTORE 
220   LET SUM$=""
230   FOR I=1 TO 13
240     READ ARABIC,ROMAN$
250     DO WHILE X>=ARABIC
260       LET SUM$=SUM$&ROMAN$
270       LET X=X-ARABIC
280     LOOP 
290   NEXT 
300   LET TOROMAN$=SUM$
310 END DEF 
320 DATA 1000,"M",900,"CM",500,"D",400,"CD",100,"C",90,"XC"
330 DATA 50,"L",40,"XL",10,"X",9,"IX",5,"V",4,"IV",1,"I"

Liberty BASIC

Works with: Just BASIC
    dim arabic( 12)
    for i =0 to 12
        read k
        arabic( i) =k
    next i
    data 1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1

    dim roman$( 12)
    for i =0 to 12
        read k$
        roman$( i) =k$
    next i
    data "M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I"

    print 2009, toRoman$( 2009)
    print 1666, toRoman$( 1666)
    print 3888, toRoman$( 3888)

    end

function toRoman$( value)
    i       =0
    result$ =""
    for i = 0 to 12
        while value >=arabic( i)
            result$ = result$ + roman$( i)
            value   = value   - arabic( i)
        wend
    next i
    toRoman$ =result$ 
end function
2009          MMIX
1666          MDCLXVI
3888          MMMDCCCLXXXVIII

Microsoft Small Basic

Translation of: DWScript
arabicNumeral = 1990
ConvertToRoman() 
TextWindow.WriteLine(romanNumeral) 'MCMXC
arabicNumeral = 2018
ConvertToRoman() 
TextWindow.WriteLine(romanNumeral) 'MMXVIII
arabicNumeral = 3888
ConvertToRoman() 
TextWindow.WriteLine(romanNumeral) 'MMMDCCCLXXXVIII
 
Sub ConvertToRoman
  weights[0] = 1000
  weights[1] =  900
  weights[2] =  500
  weights[3] =  400
  weights[4] =  100
  weights[5] =   90
  weights[6] =   50
  weights[7] =   40
  weights[8] =   10
  weights[9] =    9
  weights[10] =   5
  weights[11] =   4
  weights[12] =   1
  symbols[0] = "M"
  symbols[1] = "CM"
  symbols[2] = "D"
  symbols[3] = "CD"
  symbols[4] = "C"
  symbols[5] = "XC"
  symbols[6] = "L"
  symbols[7] = "XL"
  symbols[8] = "X"
  symbols[9] = "IX"
  symbols[10] = "V"
  symbols[11] = "IV"
  symbols[12] = "I"
  romanNumeral = ""
  i = 0
  While (i <= 12) And (arabicNumeral > 0)
    While arabicNumeral >= weights[i]
      romanNumeral = Text.Append(romanNumeral, symbols[i])
      arabicNumeral = arabicNumeral - weights[i]
    EndWhile
    i = i + 1
  EndWhile
EndSub
Output:
MCMXC
MMXVIII
MMMDCCCLXXXVIII

Nascom BASIC

Translation of: DWScript
Works with: Nascom ROM BASIC version 4.7
10 REM Roman numerals/Encode
20 DIM WEIGHTS(12),SYMBOLS$(12)
30 FOR I=0 TO 12
40 READ WEIGHTS(I),SYMBOLS$(I)
50 NEXT I
60 DATA 1000,M,900,CM,500,D,400,CD,100,C,90,XC
70 DATA 50,L,40,XL,10,X,9,IX,5,V,4,IV,1,I
80 REM ** 3888 or MMMDCCCLXXXVIII (15 chars) is
90 REM    the longest string properly encoded
100 REM   with these symbols.
110 V=1990:GOSUB 500
120 PRINT ROMAN$:REM MCMXC
130 V=2022:GOSUB 500
140 PRINT ROMAN$:REM MMXXII
150 V=3888:GOSUB 500
160 PRINT ROMAN$:REM MMMDCCCLXXXVIII
170 END
490 REM ** Encode to roman
500 ROMAN$=""
510 I=0
520 IF I>12 OR V<=0 THEN RETURN
530 IF V<WEIGHTS(I) THEN 570
540 ROMAN$=ROMAN$+SYMBOLS$(I)
550 V=V-WEIGHTS(I)
560 GOTO 530
570 I=I+1
580 GOTO 520
590 RETURN
Output:
MCMXC
MMXXII
MMMDCCCLXXXVIII

PowerBASIC

Translation of: BASIC
Works with: PB/Win version 8+
Works with: PB/CC version 5
FUNCTION toRoman(value AS INTEGER) AS STRING
    DIM arabic(0 TO 12) AS INTEGER
    DIM roman(0 TO 12) AS STRING
    ARRAY ASSIGN arabic() = 1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1
    ARRAY ASSIGN roman() = "M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I"

    DIM i AS INTEGER
    DIM result AS STRING

    FOR i = 0 TO 12
        DO WHILE value >= arabic(i)
            result = result & roman(i)
            value = value - arabic(i)
        LOOP
    NEXT i
    toRoman = result
END FUNCTION

FUNCTION PBMAIN
    'Testing
    ? "2009 = " & toRoman(2009)
    ? "1666 = " & toRoman(1666)
    ? "3888 = " & toRoman(3888)
END FUNCTION

PureBasic

#SymbolCount = 12 ;0 based count
DataSection
  denominations:
  Data.s "M","CM","D","CD","C","XC","L","XL","X","IX","V","IV","I" ;0-12
  
  denomValues:
  Data.i  1000,900,500,400,100,90,50,40,10,9,5,4,1 ;values in decending sequential order
EndDataSection

;-setup
Structure romanNumeral
  symbol.s 
  value.i
EndStructure
  
Global Dim refRomanNum.romanNumeral(#SymbolCount)

Restore denominations
For i = 0 To #SymbolCount
  Read.s refRomanNum(i)\symbol
Next

Restore denomValues
For i = 0 To #SymbolCount
  Read refRomanNum(i)\value
Next  

Procedure.s decRoman(n)
  ;converts a decimal number to a roman numeral
  Protected roman$, i
  
  For i = 0 To #SymbolCount
    Repeat
      If n >= refRomanNum(i)\value
        roman$ + refRomanNum(i)\symbol
        n - refRomanNum(i)\value
      Else
        Break
      EndIf
    ForEver
  Next

  ProcedureReturn roman$
EndProcedure

If OpenConsole()

  PrintN(decRoman(1999)) ;MCMXCIX
  PrintN(decRoman(1666)) ;MDCLXVI
  PrintN(decRoman(25))   ;XXV
  PrintN(decRoman(954))  ;CMLIV

  Print(#CRLF$ + #CRLF$ + "Press ENTER to exit")
  Input()
  CloseConsole()
EndIf

QBasic

DIM SHARED arabic(0 TO 12)
DIM SHARED roman$(0 TO 12)

FUNCTION toRoman$ (value)
    LET result$ = ""
    FOR i = 0 TO 12
        DO WHILE value >= arabic(i)
           LET result$ = result$ + roman$(i)
           LET value = value - arabic(i)
        LOOP
    NEXT i
    toRoman$ = result$
END FUNCTION

FOR i = 0 TO 12
    READ arabic(i), roman$(i)
NEXT i

DATA 1000, "M", 900, "CM", 500, "D", 400, "CD", 100, "C", 90, "XC"
DATA 50, "L", 40, "XL", 10, "X", 9, "IX", 5, "V",  4, "IV", 1, "I"

'Testing
PRINT "2009 = "; toRoman$(2009)
PRINT "1666 = "; toRoman$(1666)
PRINT "3888 = "; toRoman$(3888)

Run BASIC

[loop]
input "Input value:";val$
print roman$(val$)
goto [loop]

' ------------------------------
' Roman numerals
' ------------------------------
FUNCTION roman$(val$)
a2r$ = "M:1000,CM:900,D:500,CD:400,C:100,XC:90,L:50,XL:40,X:10,IX:9,V:5,IV:4,I:1"
v = val(val$)
for i = 1 to 13
  r$  = word$(a2r$,i,",")
  a   = val(word$(r$,2,":"))
  while v >= a 
    roman$ = roman$ + word$(r$,1,":")
    v      = v - a
  wend
next i
END FUNCTION

TI-83 BASIC

PROGRAM:DEC2ROM
:"="→Str1
:Lbl ST
:ClrHome
:Disp "NUMBER TO"
:Disp "CONVERT:"
:Input A
:If fPart(A) or A≠abs(A)
:Then
:Goto PI
:End
:A→B
:While B≥1000
:Str1+"M"→Str1
:B-1000→B
:End
:If B≥900
:Then
:Str1+"CM"→Str1
:B-900→B
:End
:If B≥500
:Then
:Str1+"D"→Str1
:B-500→B
:End
:If B≥400
:Then
:Str1+"CD"?Str1
:B-400→B
:End
:While B≥100
:Str1+"C"→Str1
:B-100→B
:End
:If B≥90
:Then
:Str1+"XC"→Str1
:B-90→B
:End
:If B≥50
:Then
:Str1+"L"→Str1
:B-50→B
:End
:If B≥40
:Then
:Str1+"XL"→Str1
:B-40→B
:End
:While B≥10
:Str1+"X"→Str1
:B-10→B
:End
:If B≥9
:Then
:Str1+"IX"→Str1
:B-9→B
:End
:If B≥5
:Then
:Str1+"V"→Str1
:B-5→B
:End
:If B≥4
:Then
:Str1+"IV"→Str1
:B-4→B
:End
:While B>0
:Str1+"I"→Str1
:B-1→B
:End
:ClrHome
:Disp A
:Disp Str1
:Stop
:Lbl PI
:ClrHome
:Disp "THE NUMBER MUST"
:Disp "BE A POSITIVE"
:Disp "INTEGER."
:Pause 
:Goto ST

True BASIC

OPTION BASE 0
DIM arabic(12), roman$(12)

FOR i = 0 to 12
    READ arabic(i), roman$(i)
NEXT i

DATA 1000, "M", 900, "CM", 500, "D", 400, "CD", 100, "C", 90, "XC"
DATA 50, "L", 40, "XL", 10, "X", 9, "IX", 5, "V",  4, "IV", 1, "I"

FUNCTION toRoman$(value)
    LET result$ = ""
    FOR i = 0 TO 12
        DO WHILE value >= arabic(i)
           LET result$ = result$ & roman$(i)
           LET value = value - arabic(i)
        LOOP
    NEXT i
    LET toRoman$ = result$
END FUNCTION

!Testing
PRINT "2009 = "; toRoman$(2009)
PRINT "1666 = "; toRoman$(1666)
PRINT "3888 = "; toRoman$(3888)
END

uBasic/4tH

Translation of: BBC Basic
Push 1, 4, 5, 9, 10, 40, 50, 90, 100, 400, 500, 900, 1000
                                       ' Initialize array
For i = 12 To 0 Step -1
  @(i) = Pop()
Next
                                       ' Calculate and print numbers
Print 1999, : Proc _FNroman (1999)
Print 2014, : Proc _FNroman (2014)
Print 1666, : Proc _FNroman (1666)
Print 3888, : Proc _FNroman (3888)

End

_FNroman Param (1)                     ' ( n --)
  Local (1)                            ' Define b@
                                       ' Try all numbers in array
  For b@ = 12 To 0 Step -1
    Do While a@ > @(b@) - 1            ' Several occurences of same number?
      GoSub ((b@ + 1) * 10)            ' Print roman digit
      a@ = a@ - @(b@)                  ' Decrement number
    Loop
  Next

  Print                                ' Terminate line
Return
                                       ' Print roman digits
 10 Print "I";  : Return
 20 Print "IV"; : Return
 30 Print "V";  : Return
 40 Print "IX"; : Return
 50 Print "X";  : Return
 60 Print "XL"; : Return
 70 Print "L";  : Return
 80 Print "XC"; : Return
 90 Print "C";  : Return
100 Print "CD"; : Return
110 Print "D";  : Return
120 Print "CM"; : Return
130 Print "M";  : Return

Visual Basic

Translation of: BASIC
Function toRoman(value) As String
    Dim arabic As Variant
    Dim roman As Variant

    arabic = Array(1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1)
    roman = Array("M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I")

    Dim i As Integer, result As String

    For i = 0 To 12
        Do While value >= arabic(i)
            result = result + roman(i)
            value = value - arabic(i)
        Loop
    Next i

    toRoman = result
End Function

Sub Main()
    MsgBox toRoman(Val(InputBox("Number, please")))
End Sub

XBasic

Translation of: DWScript
Works with: Windows XBasic
PROGRAM "romanenc"
VERSION "0.0000"

DECLARE FUNCTION Entry()
INTERNAL FUNCTION ToRoman$(aValue%%)

' 3888 or MMMDCCCLXXXVIII (15 chars) is the longest string properly encoded with these symbols.

FUNCTION Entry()
  PRINT ToRoman$(1990) ' MCMXC
  PRINT ToRoman$(2018) ' MMXVIII
  PRINT ToRoman$(3888) ' MMMDCCCLXXXVIII
END FUNCTION

FUNCTION ToRoman$(aValue%%)
  DIM weights%%[12]
  DIM symbols$[12]

  weights%%[0] = 1000
  weights%%[1] = 900
  weights%%[2] = 500
  weights%%[3] = 400
  weights%%[4] = 100
  weights%%[5] = 90
  weights%%[6] = 50
  weights%%[7] = 40
  weights%%[8] = 10
  weights%%[9] = 9
  weights%%[10] = 5
  weights%%[11] = 4
  weights%%[12] = 1

  symbols$[0] = "M"
  symbols$[1] = "CM"
  symbols$[2] = "D"
  symbols$[3] = "CD"
  symbols$[4] = "C"
  symbols$[5] = "XC"
  symbols$[6] = "L"
  symbols$[7] = "XL"
  symbols$[8] = "X"
  symbols$[9] = "IX"
  symbols$[10] = "V"
  symbols$[11] = "IV"
  symbols$[12] = "I"

  destination$ = ""
  i@@ = 0
  DO WHILE (i@@ <= 12) AND (aValue%% > 0)
    DO WHILE aValue%% >= weights%%[i@@]
      destination$ = destination$ + symbols$[i@@]
      aValue%% = aValue%% - weights%%[i@@]
    LOOP
    i@@ = i@@ + 1
  LOOP
  RETURN destination$
END FUNCTION
END PROGRAM
Output:
MCMXC
MMXVIII
MMMDCCCLXXXVIII

Yabasic

roman$ = "M, CM, D, CD, C, XC, L, XL, X, IX, V, IV, I"
decml$ = "1000, 900, 500, 400, 100, 90, 50,  40,  10,  9,  5,   4,  1"
 
sub toRoman$(value)
    local res$, i, roman$(1), decml$(1), long
	
    long = token(roman$, roman$(), ", ")
    long = token(decml$, decml$(), ", ")
	
    for i=1 to long
        while(value >= val(decml$(i)))
            res$ = res$ + roman$(i)
            value = value - val(decml$(i))
        wend
    next i
    return res$
end sub
 
print 400, " ", toRoman$(400)
print 1990, " ", toRoman$(1990)
print 2008, " ", toRoman$(2008)
print 2009, " ", toRoman$(2009)
print 1666, " ", toRoman$(1666)
print 3888, " ", toRoman$(3888)
//Output:
// 400 = CD
// 1990 = MCMXC
// 2008 = MMVIII
// 2009 = MMIX
// 1666 = MDCLXVI
// 3888 = MMMDCCCLXXXVIII

ZX Spectrum Basic

 10 DATA 1000,"M",900,"CM"
 20 DATA 500,"D",400,"CD"
 30 DATA 100,"C",90,"XC"
 40 DATA 50,"L",40,"XL"
 50 DATA 10,"X",9,"IX"
 60 DATA 5,"V",4,"IV",1,"I"
 70 INPUT "Enter an arabic number: ";V
 80 LET VALUE=V
 90 LET V$=""
100 FOR I=0 TO 12
110 READ A,R$
120 IF V<A THEN GO TO 160
130 LET V$=V$+R$
140 LET V=V-A
150 GO TO 120
160 NEXT I
170 PRINT VALUE;"=";V$

Batch File

Translation of: BASIC
@echo off
setlocal enabledelayedexpansion

set cnt=0&for %%A in (1000,900,500,400,100,90,50,40,10,9,5,4,1) do (set arab!cnt!=%%A&set /a cnt+=1)
set cnt=0&for %%R in (M,CM,D,CD,C,XC,L,XL,X,IX,V,IV,I) do (set rom!cnt!=%%R&set /a cnt+=1)

::Testing
call :toRoman 2009
echo 2009 = !result!
call :toRoman 1666
echo 1666 = !result!
call :toRoman 3888
echo 3888 = !result!
pause>nul
exit/b 0

::The "function"...
:toRoman
set value=%1
set result=

for /l %%i in (0,1,12) do (
	set a=%%i
	call :add_val
)
goto :EOF

:add_val
if !value! lss !arab%a%! goto :EOF
set result=!result!!rom%a%!
set /a value-=!arab%a%!
goto add_val
Output:
2009 = MMIX
1666 = MDCLXVI
3888 = MMMDCCCLXXXVIII

BCPL

get "libhdr"

let toroman(n, v) = valof
$(  let extract(n, val, rmn, v) = valof
    $(  while n >= val
        $(  n := n - val;
            for i=1 to rmn%0 do v%(v%0+i) := rmn%i
            v%0 := v%0 + rmn%0
        $)
        resultis n
    $)
    
    v%0 := 0
    n := extract(n, 1000, "M",  v)
    n := extract(n,  900, "CM", v)
    n := extract(n,  500, "D",  v)
    n := extract(n,  400, "CD", v)
    n := extract(n,  100, "C",  v)
    n := extract(n,   90, "XC", v)
    n := extract(n,   50, "L",  v)
    n := extract(n,   40, "XL", v)
    n := extract(n,   10, "X",  v)
    n := extract(n,    9, "IX", v)
    n := extract(n,    5, "V",  v)
    n := extract(n,    4, "IV", v)
    n := extract(n,    1, "I",  v)
    resultis v
$)

let show(n) be
$(  let v = vec 50
    writef("%I4 = %S*N", n, toroman(n, v))
$)

let start() be
$(  show(1666)
    show(2008)
    show(1001)
    show(1999)
    show(3888)
    show(2021)
$)
Output:
1666 = MDCLXVI
2008 = MMVIII
1001 = MI
1999 = MCMXCIX
3888 = MMMDCCCLXXXVIII
2021 = MMXXI

Befunge

Reads the number to convert from standard input. No range validation is performed.

&>0\0>00p:#v_$ >:#,_ $ @
4-v >5+#:/#<\55+%:5/\5%:
vv_$9+00g+5g\00g8+>5g\00
g>\20p>:10p00g \#v _20gv
> 2+ v^-1g01\g5+8<^ +9 _
        IVXLCDM
Output:
1666
MDCLXVI

BQN

Translation of: APL
ToRomanR  {
  ds  1¨(¯1+`=)" I IV V IX X XL L XC C CD D CM M"
  vs  1e3˜ 1459×⌜˜103
  R  {
    𝕨𝕊0: "";
    (ds·𝕊𝕩-⊑vs) 1-˜vs𝕩
  }
}
Example use:
   ToRoman¨ 1990‿2008‿1666‿2021
⟨ "MCMXC" "MMVIII" "MDCLXVI" "MMXXI" ⟩

Bracmat

( ( encode
  =   indian roman cifr tenfoldroman letter tenfold
    .   !arg:#?indian
      & :?roman
      &   whl
        ' ( @(!indian:#%?cifr ?indian)
          & :?tenfoldroman
          &   whl
            ' ( !roman:%?letter ?roman
              &     !tenfoldroman
                    (       (I.X)
                            (V.L)
                            (X.C)
                            (L.D)
                            (C.M)
                        : ? (!letter.?tenfold) ?
                      & !tenfold
                    | "*"
                    )
                : ?tenfoldroman
              )
          & !tenfoldroman:?roman
          & ( !cifr:9&!roman I X:?roman
            |   !cifr:~<4
              &     !roman
                    (!cifr:4&I|)
                    V
                : ?roman
              & !cifr+-5:?cifr
              & ~
            |   whl
              ' ( !cifr+-1:~<0:?cifr
                & !roman I:?roman
                )
            )
          )
      & ( !roman:? "*" ?&~`
        | str$!roman
        )
  )
& 1990 2008 1666 3888 3999 4000:?NS
&   whl
  ' ( !NS:%?N ?NS
    &   out
      $ ( encode$!N:?K&!N !K
        | str$("Can't convert " !N " to Roman numeral")
        )
    )
);
Output:
1990 MCMXC
2008 MMVIII
1666 MDCLXVI
3888 MMMDCCCLXXXVIII
3999 MMMCMXCIX
Can't convert 4000 to Roman numeral

C

Naive solution

This solution is a smart but does not return the number written as a string.

#include <stdio.h>


int main() {
    int arabic[] = {1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1};

    // There is a bug: "XL\0" is translated into sequence 58 4C 00 00, i.e. it is 4-bytes long...
    // Should be "XL" without \0 etc.
    //
    char roman[13][3] = {"M\0", "CM\0", "D\0", "CD\0", "C\0", "XC\0", "L\0", "XL\0", "X\0", "IX\0", "V\0", "IV\0", "I\0"};
    int N;

    printf("Enter arabic number:\n");
    scanf("%d", &N);
    printf("\nRoman number:\n");

    for (int i = 0; i < 13; i++) {
        while (N >= arabic[i]) {
            printf("%s", roman[i]);
            N -= arabic[i];
        }
    }
    return 0;
}
Output:
Enter arabic number:
215

Roman number:
CCXV

Not thread-safe

#define _CRT_SECURE_NO_WARNINGS

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

int RomanNumerals_parseInt(const char* string)
{
    int value;
    return scanf("%u", &value) == 1 && value > 0 ? value : 0;
}

const char* RomanNumerals_toString(int value)
{
#define ROMAN_NUMERALS_MAX_OUTPUT_STRING_SIZE 64
    static buffer[ROMAN_NUMERALS_MAX_OUTPUT_STRING_SIZE];

    const static int maxValue = 5000;
    const static int minValue = 1;

    const static struct Digit {
        char string[4]; // It's better to use 4 than 3 (aligment).
        int  value;
    } digits[] = {
        {"M", 1000}, {"CM", 900}, {"D", 500 }, {"CD", 400 },
        {"C", 100 }, {"XC", 90 }, {"L",  50 }, {"XL", 40}, 
        {"X", 10}, {"IX", 9}, {"V", 5}, {"IV", 4}, {"I", 1 }, 
        {"?", 0}
    };

    *buffer = '\0'; // faster than memset(buffer, 0, sizeof(buffer));
    if (minValue <= value && value <= maxValue)
    {
        struct Digit* digit = &digits[0];

        while (digit->value)
        {
            while (value >= digit->value)
            {
                value -= digit->value;
                // It is not necessary - total length would not be exceeded...
                // if (strlen(buffer) + strlen(digit->string) < sizeof(buffer))
                strcat(buffer, digit->string);
            }
            digit++;
        }
    }
    return buffer;
}


int main(int argc, char* argv[])
{
    if (argc < 2)
    {
        // Blanks are needed for a consistient blackground on some systems.
        // BTW, puts append an extra newline at the end.
        //
        puts("Write given numbers as Roman numerals. \n"
             "                                       \n"
             "Usage:                                 \n"
             "    roman n1 n2 n3 ...                 \n"
             "                                       \n"
             "where n1 n2 n3 etc. are Arabic numerals\n");

        int numbers[] = { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1498, 2022 };
        for (int i = 0; i < sizeof(numbers) / sizeof(int); i++)
        {
            printf("%4d = %s\n", 
                numbers[i], RomanNumerals_toString(numbers[i]));
        }
    }
    else
    {
        for (int i = 1; i < argc; i++)
        {
            int number = RomanNumerals_parseInt(argv[i]);
            if (number)
            {
                puts(RomanNumerals_toString(number));
            }
            else
            {
                puts("???");
            }
        }
    }

    return 0;
}
Output:
Write given numbers as Roman numerals.

Usage:
    roman n1 n2 n3 ...

where n1 n2 n3 etc. are Arabic numerals

   1 = I
   2 = II
   3 = III
   4 = IV
   5 = V
   6 = VI
   7 = VII
   8 = VIII
   9 = IX
  10 = X
1498 = MCDXCVIII
2022 = MMXXII

C#

using System;
class Program
{
    static uint[] nums = { 1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1 };
    static string[] rum = { "M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I" };

    static string ToRoman(uint number)
    {
        string value = "";
        for (int i = 0; i < nums.Length && number != 0; i++)
        {
            while (number >= nums[i])
            {
                number -= nums[i];
                value += rum[i];
            }
        }
        return value;
    }

    static void Main()
    {
        for (uint number = 1; number <= 1 << 10; number *= 2)
        {
            Console.WriteLine("{0} = {1}", number, ToRoman(number));
        }
    }
}

One-liner Mono REPL

Func<int, string> toRoman = (number) =>
  new Dictionary<int, string>
  {
    {1000, "M"},
    {900, "CM"},
    {500, "D"},
    {400, "CD"},
    {100, "C"},
    {90, "XC"},
    {50, "L"},
    {40, "XL"},
    {10, "X"},
    {9, "IX"},
    {5, "V"},
    {4, "IV"},
    {1, "I"}
  }.Aggregate(new string('I', number), (m, _) => m.Replace(new string('I', _.Key), _.Value));
Output:
1 = I
2 = II
4 = IV
8 = VIII
16 = XVI
32 = XXXII
64 = LXIV
128 = CXXVIII
256 = CCLVI
512 = DXII
1024 = MXXIV

C++

C++ 98

#include <iostream>
#include <string>

std::string to_roman(int value)
{
  struct romandata_t { int value; char const* numeral; };
  static romandata_t const romandata[] =
     { 1000, "M",
        900, "CM",
        500, "D",
        400, "CD",
        100, "C",
         90, "XC",
         50, "L",
         40, "XL",
         10, "X",
          9, "IX",
          5, "V",
          4, "IV",
          1, "I",
          0, NULL }; // end marker

  std::string result;
  for (romandata_t const* current = romandata; current->value > 0; ++current)
  {
    while (value >= current->value)
    {
      result += current->numeral;
      value  -= current->value;
    }
  }
  return result;
}

int main()
{
  for (int i = 1; i <= 4000; ++i)
  {
    std::cout << to_roman(i) << std::endl;
  }
}

C++ 11

#include <iostream>
#include <string>

std::string to_roman(int x) {
     if (x <= 0)
         return "Negative or zero!";
     auto roman_digit = [](char one, char five, char ten, int x) {
        if (x <= 3)
            return std::string().assign(x, one);
        if (x <= 5)
            return std::string().assign(5 - x, one) + five;
        if (x <= 8)
            return five + std::string().assign(x - 5, one);
        return std::string().assign(10 - x, one) + ten;
    };
    if (x >= 1000)
        return x - 1000 > 0 ? "M" + to_roman(x - 1000) : "M";
    if (x >= 100) {
        auto s = roman_digit('C', 'D', 'M', x / 100);
        return x % 100 > 0 ? s + to_roman(x % 100) : s;
    }
    if (x >= 10) {
        auto s = roman_digit('X', 'L', 'C', x / 10);
        return x % 10 > 0 ? s + to_roman(x % 10) : s;
    }
    return roman_digit('I', 'V', 'X', x);
}

int main() {
    for (int i = 0; i < 2018; i++)
        std::cout << i << " --> " << to_roman(i) << std::endl;
}

Ceylon

shared void run() {
	
	class Numeral(shared Character char, shared Integer int) {}
	
	value tiers = [
		[Numeral('I', 1),   Numeral('V', 5),   Numeral('X', 10)],
		[Numeral('X', 10),  Numeral('L', 50),  Numeral('C', 100)],
		[Numeral('C', 100), Numeral('D', 500), Numeral('M', 1k)]
	];
	
	String toRoman(Integer hindu, Integer tierIndex = 2) {
		
		assert (exists tier = tiers[tierIndex]);
		
		" Finds if it's a two character numeral like iv, ix, xl, xc, cd and cm."
		function findTwoCharacterNumeral() =>
			if (exists bigNum = tier.rest.find((numeral) => numeral.int - tier.first.int <= hindu < numeral.int))
			then [tier.first, bigNum]
			else null;
		
		if (hindu <= 0) {
			// if it's zero then we are done!
			return "";
		} 
		else if (exists [smallNum, bigNum] = findTwoCharacterNumeral()) {
			value twoCharSymbol = "``smallNum.char````bigNum.char``";
			value twoCharValue = bigNum.int - smallNum.int;
			return "``twoCharSymbol````toRoman(hindu - twoCharValue, tierIndex)``";
		} 
		else if (exists num = tier.reversed.find((Numeral elem) => hindu >= elem.int)) {
			return "``num.char````toRoman(hindu - num.int, tierIndex)``";
		} 
		else {
			// nothing was found so move to the next smaller tier!
			return toRoman(hindu, tierIndex - 1);
		}
	}
	
	assert (toRoman(1) == "I");
	assert (toRoman(2) == "II");
	assert (toRoman(4) == "IV");
	assert (toRoman(1666) == "MDCLXVI");
	assert (toRoman(1990) == "MCMXC");
	assert (toRoman(2008) == "MMVIII");
}

Clojure

The easiest way is to use the built-in cl-format function

(def arabic->roman 
  (partial clojure.pprint/cl-format nil "~@R"))

(arabic->roman 147)
;"CXXIII"
(arabic->roman 99)
;"XCIX"
Alternatively:
(def roman-map
  (sorted-map
    1    "I", 4    "IV", 5   "V", 9   "IX",
    10   "X", 40   "XL", 50  "L", 90  "XC",
    100  "C", 400  "CD", 500 "D", 900 "CM"
    1000 "M"))

(defn int->roman [n]
  {:pre (integer? n)}
  (loop [res (StringBuilder.), n n]
    (if-let [v (roman-map n)]
      (str (.append res v))
      (let [[k v] (->> roman-map keys (filter #(> n %)) last (find roman-map))]
        (recur (.append res v) (- n k))))))

(int->roman 1999)
; "MCMXCIX"


An alternate implementation:

(defn a2r [a]
  (let [rv '(1000 500 100 50 10 5 1)
        rm (zipmap rv "MDCLXVI")
        dv (->> rv (take-nth 2) next #(interleave % %))]
    (loop [a a rv rv dv dv r nil]
      (if (<= a 0)
        r
        (let [v (first rv)
              d (or (first dv) 0)
              l (- v d)]
          (cond
            (= a v) (str r (rm v))
            (= a l) (str r (rm d) (rm v))
            (and (> a v) (> a l)) (recur (- a v) rv dv (str r (rm v)))
            (and (< a v) (< a l)) (recur a (rest rv) (rest dv) r)
            :else (recur (- a l) (rest rv) (rest dv) (str r (rm d) (rm v)))))))))

Usage:

(a2r 1666)
"MDCLXVI"

(map a2r [1000 1 389 45])
("M" "I" "CCCLXXXIX" "XLV")

An alternate implementation:

(def roman-map
  (sorted-map-by >
                 1    "I", 4    "IV", 5   "V", 9   "IX",
                 10   "X", 40   "XL", 50  "L", 90  "XC",
                 100  "C", 400  "CD", 500 "D", 900 "CM"
                 1000 "M"))

(defn a2r
  ([r]
   (reduce str (a2r r (keys roman-map))))
  ([r n]
   (when-not (empty? n)
     (let [e (first n)
           v (- r e)
           roman (roman-map e)]
       (cond
         (< v 0) (a2r r (rest n))
         (= v 0) (cons roman [])
         (>= v e) (cons roman (a2r v n))
         (< v e) (cons roman (a2r v (rest n))))))))

Usage:

(a2r 1666)
"MDCLXVI"

(map a2r [1000 1 389 45])
("M" "I" "CCCLXXXIX" "XLV")

CLU

roman = cluster is encode
    rep = null
    
    dmap = struct[v: int, s: string]
    darr = array[dmap]
    own chunks: darr := darr$
       [dmap${v: 1000, s: "M"},
        dmap${v:  900, s: "CM"},
        dmap${v:  500, s: "D"},
        dmap${v:  400, s: "CD"},
        dmap${v:  100, s: "C"},
        dmap${v:   90, s: "XC"},
        dmap${v:   50, s: "L"},
        dmap${v:   40, s: "XL"},
        dmap${v:   10, s: "X"},
        dmap${v:    9, s: "IX"},
        dmap${v:    5, s: "V"},
        dmap${v:    4, s: "IV"},
        dmap${v:    1, s: "I"}]
      
    largest_chunk = proc (i: int) returns (int, string)
        for chunk: dmap in darr$elements(chunks) do
            if chunk.v <= i then return (chunk.v, chunk.s) end
        end            
        return (0, "")
    end largest_chunk
    
    encode = proc (i: int) returns (string)
        result: string := ""
        while i > 0 do
            val: int chunk: string
            val, chunk := largest_chunk(i)
            result := result || chunk
            i := i - val
        end
        return (result)
    end encode 
end roman

start_up = proc ()
    po: stream := stream$primary_output()
    tests: array[int] := array[int]$[1666, 2008, 1001, 1999, 3888, 2021]
    
    for test: int in array[int]$elements(tests) do
        stream$putl(po, int$unparse(test) || " = " || roman$encode(test))
    end
end start_up
Output:
1666 = MDCLXVI
2008 = MMVIII
1001 = MI
1999 = MCMXCIX
3888 = MMMDCCCLXXXVIII
2021 = MMXXI

COBOL

IDENTIFICATION DIVISION.
PROGRAM-ID. TOROMAN.
DATA DIVISION.
working-storage section.
  01 ws-number pic 9(4) value 0.
  01 ws-save-number pic 9(4).
  01 ws-tbl-def.
    03 filler pic x(7) value '1000M  '.
    03 filler pic x(7) value '0900CM '.
    03 filler pic x(7) value '0500D  '.
    03 filler pic x(7) value '0400CD '.
    03 filler pic x(7) value '0100C  '.
    03 filler pic x(7) value '0090XC '.
    03 filler pic x(7) value '0050L  '.
    03 filler pic x(7) value '0040XL '.
    03 filler pic x(7) value '0010X  '.
    03 filler pic x(7) value '0009IX '.
    03 filler pic x(7) value '0005V  '.
    03 filler pic x(7) value '0004IV '.
    03 filler pic x(7) value '0001I  '.
  01  filler redefines ws-tbl-def.
    03 filler occurs 13 times indexed by rx.
      05 ws-tbl-divisor    pic 9(4).
      05 ws-tbl-roman-ch   pic x(1) occurs 3 times indexed by cx.
  01 ocx pic 99.
  01 ws-roman.
    03 ws-roman-ch         pic x(1) occurs 16 times.
PROCEDURE DIVISION.
  accept ws-number
  perform
  until ws-number = 0
    move ws-number to ws-save-number
    if ws-number > 0 and ws-number < 4000
      initialize ws-roman
      move 0 to ocx
      perform varying rx from 1 by +1
      until ws-number = 0
        perform until ws-number < ws-tbl-divisor (rx)
          perform varying cx from 1 by +1 
  		  until ws-tbl-roman-ch (rx, cx) = spaces
            compute ocx = ocx + 1
            move ws-tbl-roman-ch (rx, cx) to ws-roman-ch (ocx)
          end-perform
          compute ws-number = ws-number - ws-tbl-divisor (rx)
        end-perform
      end-perform
      display 'inp=' ws-save-number ' roman=' ws-roman
    else
      display 'inp=' ws-save-number ' invalid'
    end-if
    accept ws-number
  end-perform
  .
Output:
(input was supplied via STDIN)
inp=0111 roman=CXI             
inp=2234 roman=MMCCXXXIV       
inp=0501 roman=DI              
inp=0010 roman=X               
inp=0040 roman=XL              
inp=0050 roman=L               
inp=0066 roman=LXVI            
inp=0666 roman=DCLXVI          
inp=5666 invalid
inp=3333 roman=MMMCCCXXXIII    
inp=3888 roman=MMMDCCCLXXXVIII 
inp=3999 roman=MMMCMXCIX       
inp=3345 roman=MMMCCCXLV      

CoffeeScript

decimal_to_roman = (n) ->
  # This should work for any positive integer, although it
  # gets a bit preposterous for large numbers.
  if n >= 4000
    thousands = decimal_to_roman n / 1000
    ones = decimal_to_roman n % 1000
    return "M(#{thousands})#{ones}"
    
  s = ''
  translate_each = (min, roman) ->
    while n >= min
      n -= min
      s += roman
  translate_each 1000, "M"
  translate_each  900, "CM"
  translate_each  500, "D"
  translate_each  400, "CD"
  translate_each  100, "C"
  translate_each   90, "XC"
  translate_each   50, "L"
  translate_each   40, "XL"
  translate_each   10, "X"
  translate_each    9, "IX"
  translate_each    5, "V"
  translate_each    4, "IV"
  translate_each    1, "I"
  s
  
###################
tests =
  IV: 4
  XLII: 42
  MCMXC: 1990
  MMVIII: 2008
  MDCLXVI: 1666
  'M(IV)': 4000
  'M(VI)IX': 6009
  'M(M(CXXIII)CDLVI)DCCLXXXIX': 123456789
  'M(MMMV)I': 3005001

for expected, decimal of tests
  roman = decimal_to_roman(decimal)
  if roman == expected
    console.log "#{decimal} = #{roman}"
  else
    console.log "error for #{decimal}: #{roman} is wrong"

Common Lisp

(defun roman-numeral (n)
  (format nil "~@R" n))

Cowgol

include "cowgol.coh";
include "argv.coh";

# Encode the given number as a Roman numeral
sub decimalToRoman(num: uint16, buf: [uint8]): (rslt: [uint8]) is
    # return the start of the buffer for easy printing
    rslt := buf;
    
    # Add string to buffer
    sub Add(str: [uint8]) is
        while [str] != 0 loop
            [buf] := [str];
            buf := @next buf;
            str := @next str;
        end loop;
    end sub;
    
    # Table of Roman numerals
    record Roman is
        value: uint16;
        string: [uint8];
    end record;
    
    var numerals: Roman[] := {
        {1000, "M"}, {900, "CM"}, {500, "D"}, {400, "CD"},
        {100, "C"}, {90, "XC"}, {50, "L"}, {40, "XL"},
        {10, "X"}, {9, "IX"}, {5, "V"}, {4, "IV"},
        {1, "I"}
    };
    
    var curNum := &numerals as [Roman];
    while num != 0 loop
        while num >= curNum.value loop
            Add(curNum.string);
            num := num - curNum.value;
        end loop;
        curNum := @next curNum;
    end loop;
    
    [buf] := 0; # terminate the string
end sub;

# Read numbers from the command line and print the corresponding Roman numerals
ArgvInit();
var buffer: uint8[100];
loop
    var argmt := ArgvNext();
    if argmt == (0 as [uint8]) then
        break;
    end if;
    
    var dummy: [uint8];
    var number: int32;
    (number, dummy) := AToI(argmt);
    
    print(decimalToRoman(number as uint16, &buffer as [uint8]));
    print_nl();
end loop;
Output:
$ ./romanenc.386 1990 2008 1666
MCMXC
MMVIII
MDCLXVI

D

string toRoman(int n) pure nothrow
in {
    assert(n < 5000);
} body {
    static immutable weights = [1000, 900, 500, 400, 100, 90,
                                50, 40, 10, 9, 5, 4, 1];
    static immutable symbols = ["M","CM","D","CD","C","XC","L",
                                "XL","X","IX","V","IV","I"];

    string roman;
    foreach (i, w; weights) {
        while (n >= w) {
            roman ~= symbols[i];
            n -= w;
        }
        if (n == 0)
            break;
    }
    return roman;
} unittest {
    assert(toRoman(455)  == "CDLV");
    assert(toRoman(3456) == "MMMCDLVI");
    assert(toRoman(2488) == "MMCDLXXXVIII");
}

void main() {}

Delphi

Translation of: DWScript
program RomanNumeralsEncode;

{$APPTYPE CONSOLE}

function IntegerToRoman(aValue: Integer): string;
var
  i: Integer;
const
  WEIGHTS: array[0..12] of Integer = (1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1);
  SYMBOLS: array[0..12] of string = ('M', 'CM', 'D', 'CD', 'C', 'XC', 'L', 'XL', 'X', 'IX', 'V', 'IV', 'I');
begin
  for i := Low(WEIGHTS) to High(WEIGHTS) do
  begin
    while aValue >= WEIGHTS[i] do
    begin
      Result := Result + SYMBOLS[i];
      aValue := aValue - WEIGHTS[i];
    end;
    if aValue = 0 then
      Break;
  end;
end;

begin
  Writeln(IntegerToRoman(1990)); // MCMXC
  Writeln(IntegerToRoman(2008)); // MMVIII
  Writeln(IntegerToRoman(1666)); // MDCLXVI
end.

DWScript

Translation of: D
const weights = [1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1];
const symbols = ["M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I"];

function toRoman(n : Integer) : String;
var
   i, w : Integer;
begin
   for i := 0 to weights.High do begin
      w := weights[i];
      while n >= w do begin
         Result += symbols[i];
         n -= w;
      end;
      if n = 0 then Break;
   end;
end;

PrintLn(toRoman(455));
PrintLn(toRoman(3456));
PrintLn(toRoman(2488));

EasyLang

func$ dec2rom dec .
   values[] = [ 1000 900 500 400 100 90 50 40 10 9 5 4 1 ]
   symbol$[] = [ "M" "CM" "D" "CD" "C" "XC" "L" "XL" "X" "IX" "V" "IV" "I" ]
   for i = 1 to len values[]
      while dec >= values[i]
         rom$ &= symbol$[i]
         dec -= values[i]
      .
   .
   return rom$
.
print dec2rom 1990
print dec2rom 2008
print dec2rom 1666

ECL

RomanEncode(UNSIGNED Int) := FUNCTION
  SetWeights := [1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1];
  SetSymbols := ['M', 'CM', 'D', 'CD', 'C', 'XC', 'L', 'XL', 'X', 'IX', 'V', 'IV', 'I'];
  ProcessRec := RECORD
    UNSIGNED val;
    STRING Roman;
  END;
  dsWeights  := DATASET(13,TRANSFORM(ProcessRec,SELF.val := Int, SELF := []));

  SymbolStr(i,n,STRING s) := CHOOSE(n+1,'',SetSymbols[i],SetSymbols[i]+SetSymbols[i],SetSymbols[i]+SetSymbols[i]+SetSymbols[i],s);
	
  RECORDOF(dsWeights) XF(dsWeights L, dsWeights R, INTEGER C) := TRANSFORM
    ThisVal := IF(C=1,R.Val,L.Val);
    IsDone := ThisVal = 0;
    SELF.Roman := IF(IsDone,L.Roman,L.Roman + SymbolStr(C,ThisVal DIV SetWeights[C],L.Roman));
    SELF.val := IF(IsDone,0,ThisVal - ((ThisVal DIV SetWeights[C])*SetWeights[C]));
  END;
  i := ITERATE(dsWeights,XF(LEFT,RIGHT,COUNTER));
  RETURN i[13].Roman;
END;

RomanEncode(1954);  //MCMLIV 
RomanEncode(1990 ); //MCMXC  
RomanEncode(2008 ); //MMVIII 
RomanEncode(1666);  //MDCLXVI

Eiffel

class
	APPLICATION

create
	make

feature {NONE} -- Initialization

	make
		local
			numbers: ARRAY [INTEGER]
		do
			numbers := <<1990, 2008, 1666, 3159, 1977, 2010>>
				   -- "MCMXC", "MMVIII", "MDCLXVI", "MMMCLIX", "MCMLXXVII", "MMX"
			across numbers as n loop
				print (n.item.out + " in decimal Arabic numerals is " +
				       decimal_to_roman (n.item) + " in Roman numerals.%N")
			end
		end

feature -- Roman numerals

	decimal_to_roman (a_int: INTEGER): STRING
		-- Representation of integer `a_int' as Roman numeral
		require
			a_int > 0
		local
			dnums: ARRAY[INTEGER]
			rnums: ARRAY[STRING]

			dnum: INTEGER
			rnum: STRING

			i: INTEGER
		do
			dnums := <<1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1>>
			rnums := <<"M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I">>

			dnum := a_int
			rnum := ""

			from
				i := 1
			until
				i > dnums.count or dnum <= 0
			loop
				from
				until
					dnum < dnums[i]
				loop
					dnum := dnum - dnums[i]
					rnum := rnum + rnums[i]
				end
				i := i + 1
			end

			Result := rnum
		end
end

Ela

Translation of: Haskell
open number string math

digit x y z k = 
  [[x],[x,x],[x,x,x],[x,y],[y],[y,x],[y,x,x],[y,x,x,x],[x,z]] :
  (toInt k - 1)
 
toRoman 0 = ""
toRoman x | x < 0     = fail "Negative roman numeral"
          | x >= 1000 = 'M' :: toRoman (x - 1000)
          | x >= 100  = let (q,r) = x `divrem` 100 in 
                        digit 'C' 'D' 'M' q ++ toRoman r
          | x >= 10   = let (q,r) = x `divrem` 10 in
                        digit 'X' 'L' 'C' q ++ toRoman r
          | else = digit 'I' 'V' 'X' x

map (join "" << toRoman) [1999,25,944]
Output:
["MCMXCIX","XXV","CMXLIV"]

Elena

Translation of: C#

ELENA 6.x :

import system'collections;
import system'routines;
import extensions;
import extensions'text;
 
static RomanDictionary = Dictionary.new()
                            .setAt(1000, "M")
                            .setAt(900, "CM")
                            .setAt(500, "D")
                            .setAt(400, "CD")
                            .setAt(100, "C")
                            .setAt(90, "XC")
                            .setAt(50, "L")
                            .setAt(40, "XL")
                            .setAt(10, "X")
                            .setAt(9, "IX")
                            .setAt(5, "V")
                            .setAt(4, "IV")
                            .setAt(1, "I");
 
extension op
{
   toRoman()
      = RomanDictionary.accumulate(new StringWriter("I", self), (m,kv => m.replace(new StringWriter("I",kv.Key).Value, kv.Value)));
}
 
public program()
{
    console.printLine("1990 : ", 1990.toRoman());
    console.printLine("2008 : ", 2008.toRoman());
    console.printLine("1666 : ", 1666.toRoman())
}
Output:
1990 : MCMXC
2008 : MMVIII
1666 : MDCLXVI

Elixir

Translation of: Erlang
defmodule Roman_numeral do
  def encode(0), do: ''
  def encode(x) when x >= 1000, do: [?M | encode(x - 1000)]
  def encode(x) when x >= 100,  do: digit(div(x,100), ?C, ?D, ?M) ++ encode(rem(x,100))
  def encode(x) when x >= 10,   do: digit(div(x,10), ?X, ?L, ?C) ++ encode(rem(x,10))
  def encode(x) when x >= 1,    do: digit(x, ?I, ?V, ?X)
  
  defp digit(1, x, _, _), do: [x]
  defp digit(2, x, _, _), do: [x, x]
  defp digit(3, x, _, _), do: [x, x, x]
  defp digit(4, x, y, _), do: [x, y]
  defp digit(5, _, y, _), do: [y]
  defp digit(6, x, y, _), do: [y, x]
  defp digit(7, x, y, _), do: [y, x, x]
  defp digit(8, x, y, _), do: [y, x, x, x]
  defp digit(9, x, _, z), do: [x, z]
end

Another:

Translation of: Ruby
defmodule Roman_numeral do
  @symbols [ {1000, 'M'}, {900, 'CM'}, {500, 'D'}, {400, 'CD'}, {100, 'C'}, {90, 'XC'},
             {50, 'L'}, {40, 'XL'}, {10, 'X'}, {9, 'IX'}, {5, 'V'}, {4, 'IV'}, {1, 'I'} ]
  def encode(num) do
    {roman,_} = Enum.reduce(@symbols, {[], num}, fn {divisor, letter}, {memo, n} ->
                  {memo ++ List.duplicate(letter, div(n, divisor)), rem(n, divisor)}
                end)
    Enum.join(roman)
  end
end

Test:

Enum.each([1990, 2008, 1666], fn n ->
  IO.puts "#{n}: #{Roman_numeral.encode(n)}"
end)
Output:
1990: MCMXC
2008: MMVIII
1666: MDCLXVI

Emacs Lisp

(defun ar2ro (AN)
  "Translate from arabic number AN to roman number.
   For example, (ar2ro 1666) returns (M D C L X V I)."
  (cond
   ((>= AN 1000) (cons 'M (ar2ro (- AN 1000))))
   ((>= AN 900) (cons 'C (cons 'M (ar2ro (- AN 900)))))
   ((>= AN 500) (cons 'D (ar2ro (- AN 500))))
   ((>= AN 400) (cons 'C (cons 'D (ar2ro (- AN 400)))))
   ((>= AN 100) (cons 'C (ar2ro (- AN 100))))
   ((>= AN 90) (cons 'X (cons 'C (ar2ro (- AN 90)))))
   ((>= AN 50) (cons 'L (ar2ro (- AN 50))))
   ((>= AN 40) (cons 'X (cons 'L (ar2ro (- AN 40)))))
   ((>= AN 10) (cons 'X (ar2ro (- AN 10))))
   ((>= AN 5) (cons 'V (ar2ro (- AN 5))))
   ((>= AN 4) (cons 'I (cons 'V (ar2ro (- AN 4)))))
   ((>= AN 1) (cons 'I (ar2ro (- AN 1))))
   ((= AN 0) nil)))

Erlang

Translation of: OCaml
-module(roman).
-export([to_roman/1]).

to_roman(0) -> [];
to_roman(X) when X >= 1000 -> [$M | to_roman(X - 1000)];
to_roman(X) when X >= 100 ->
    digit(X div 100, $C, $D, $M) ++ to_roman(X rem 100);
to_roman(X) when X >= 10 ->
    digit(X div 10, $X, $L, $C) ++ to_roman(X rem 10);
to_roman(X) when X >= 1 -> digit(X, $I, $V, $X).

digit(1, X, _, _) -> [X];
digit(2, X, _, _) -> [X, X];
digit(3, X, _, _) -> [X, X, X];
digit(4, X, Y, _) -> [X, Y];
digit(5, _, Y, _) -> [Y];
digit(6, X, Y, _) -> [Y, X];
digit(7, X, Y, _) -> [Y, X, X];
digit(8, X, Y, _) -> [Y, X, X, X];
digit(9, X, _, Z) -> [X, Z].

sample:

1> c(roman).            
{ok,roman}
2> roman:to_roman(1999).
"MCMXCIX"
3> roman:to_roman(25).  
"XXV"
4> roman:to_roman(944).
"CMXLIV"

Alternative:

-module( roman_numerals ).

-export( [encode_from_integer/1]).

-record( encode_acc, {n, romans=""} ).

encode_from_integer( N ) when N > 0 ->
        #encode_acc{romans=Romans} = lists:foldl( fun encode_from_integer/2, #encode_acc{n=N}, map() ),
        Romans.


encode_from_integer( _Map, #encode_acc{n=0}=Acc ) -> Acc;
encode_from_integer( {_Roman, Value}, #encode_acc{n=N}=Acc ) when N < Value -> Acc;
encode_from_integer( {Roman, Value}, #encode_acc{n=N, romans=Romans} ) ->
        Times = N div Value,
        New_roman = lists:flatten( lists:duplicate(Times, Roman) ),
        #encode_acc{n=N - (Times * Value), romans=Romans ++ New_roman}.

map() -> [{"M",1000}, {"CM",900}, {"D",500}, {"CD",400}, {"C",100}, {"XC",90}, {"L",50}, {"XL",40}, {"X",10}, {"IX",9}, {"V",5}, {"IV",4}, {"I\
",1}].
Output:
36> roman_numerals:encode_from_integer( 1990 ).
"MCMXC"
37> roman_numerals:encode_from_integer( 2008 ).
"MMVIII"
38> roman_numerals:encode_from_integer( 1666 ).
"MDCLXVI"

ERRE

PROGRAM ARAB2ROMAN

DIM ARABIC%[12],ROMAN$[12]

PROCEDURE TOROMAN(VALUE->ANS$)
LOCAL RESULT$
    FOR I%=0 TO 12 DO
        WHILE VALUE>=ARABIC%[I%] DO
            RESULT$+=ROMAN$[I%]
            VALUE-=ARABIC%[I%]
        END WHILE
    END FOR
    ANS$=RESULT$
END PROCEDURE

BEGIN
!
!Testing
!
   ARABIC%[]=(1000,900,500,400,100,90,50,40,10,9,5,4,1)
   ROMAN$[]=("M","CM","D","CD","C","XC","L","XL","X","IX","V","IV","I")
   TOROMAN(2009->ANS$) PRINT("2009 = ";ANS$)
   TOROMAN(1666->ANS$) PRINT("1666 = ";ANS$)
   TOROMAN(3888->ANS$) PRINT("3888 = ";ANS$)
END PROGRAM

Euphoria

Translation of: BASIC
constant arabic = {1000, 900, 500, 400, 100, 90, 50,  40,  10,  9,  5,   4,  1 }
constant roman  = {"M", "CM", "D","CD", "C","XC","L","XL","X","IX","V","IV","I"}

function toRoman(integer val)
    sequence result
    result = ""
    for i = 1 to 13 do
        while val >= arabic[i] do
            result &= roman[i]
            val -= arabic[i]
        end while
    end for
    return result
end function

printf(1,"%d = %s\n",{2009,toRoman(2009)})
printf(1,"%d = %s\n",{1666,toRoman(1666)})
printf(1,"%d = %s\n",{3888,toRoman(3888)})
Output:
 2009 = MMIX
 1666 = MDCLXVI
 3888 = MMMDCCCLXXXVIII

Excel

Excel can encode numbers in Roman forms in 5 successively concise forms. These can be indicated from 0 to 4. Type in a cell:

=ROMAN(2013,0)

It becomes:

MMXIII

F#

let digit x y z = function
    1 -> x
  | 2 -> x + x
  | 3 -> x + x + x
  | 4 -> x + y
  | 5 -> y
  | 6 -> y + x
  | 7 -> y + x + x
  | 8 -> y + x + x + x
  | 9 -> x + z
  | _ -> failwith "invalid call to digit"
 
let rec to_roman acc = function
    | x when x >= 1000 -> to_roman (acc + "M") (x - 1000)
    | x when x >= 100 -> to_roman (acc + digit "C" "D" "M" (x / 100)) (x % 100)
    | x when x >= 10 -> to_roman (acc + digit "X" "L" "C" (x / 10)) (x % 10)
    | x when x > 0 -> acc + digit "I" "V" "X" x
    | 0 -> acc
    | _ -> failwith "invalid call to_roman (negative input)"

let roman n = to_roman "" n

[<EntryPoint>]
let main args =
    [1990; 2008; 1666]
    |> List.map (fun n -> roman n)
    |> List.iter (printfn "%s")
    0
Output:
MCMXC
MMVIII
MDCLXVI

Factor

A roman numeral library ships with Factor.

USE: roman
( scratchpad ) 3333 >roman .
"mmmcccxxxiii"

Parts of the implementation:

CONSTANT: roman-digits
    { "m" "cm" "d" "cd" "c" "xc" "l" "xl" "x" "ix" "v" "iv" "i" }

CONSTANT: roman-values
    { 1000 900 500 400 100 90 50 40 10 9 5 4 1 }

ERROR: roman-range-error n ;

: roman-range-check ( n -- n )
    dup 1 10000 between? [ roman-range-error ] unless ;

: >roman ( n -- str )
    roman-range-check
    roman-values roman-digits [
        [ /mod swap ] dip <repetition> concat
    ] 2map "" concat-as nip ;

FALSE

^$." "
[$999>][1000- "M"]#
 $899> [ 900-"CM"]?
 $499> [ 500- "D"]?
 $399> [ 400-"CD"]?
[$ 99>][ 100- "C"]#
 $ 89> [  90-"XC"]?
 $ 49> [  50- "L"]?
 $ 39> [  40-"XL"]?
[$  9>][  10- "X"]#
 $  8> [   9-"IX"]?
 $  4> [   5- "V"]?
 $  3> [   4-"IV"]?
[$    ][   1- "I"]#%

Fan

**
** converts a number to its roman numeral representation
**
class RomanNumerals
{

  private Str digit(Str x, Str y, Str z, Int i)
  {
    switch (i)
    {
      case 1: return x
      case 2: return x+x
      case 3: return x+x+x
      case 4: return x+y
      case 5: return y
      case 6: return y+x
      case 7: return y+x+x
      case 8: return y+x+x+x
      case 9: return x+z
    }
    return ""
  }

  Str toRoman(Int i)
  {
    if (i>=1000) { return "M" + toRoman(i-1000) }
    if (i>=100) { return digit("C", "D", "M", i/100) + toRoman(i%100) }
    if (i>=10) { return digit("X", "L", "C", i/10) + toRoman(i%10) }
    if (i>=1) { return digit("I", "V", "X", i) }
    return ""
  }

  Void main()
  {
    2000.times |i| { echo("$i = ${toRoman(i)}") }
  }

}

Forth

: vector create ( n -- ) 0 do , loop  does>  ( n -- ) swap cells + @ execute ; 
\ these are ( numerals -- numerals )
: ,I  dup c@ C, ;  : ,V  dup 1 + c@ C, ;  : ,X  dup 2 + c@ C, ; 

\ these are ( numerals -- )
:noname  ,I ,X     drop ;   :noname  ,V ,I ,I ,I  drop ;   :noname  ,V ,I ,I  drop ;  
:noname  ,V ,I     drop ;   :noname  ,V           drop ;   :noname  ,I ,V     drop ;
:noname  ,I ,I ,I  drop ;   :noname  ,I ,I        drop ;   :noname  ,I        drop ;  
' drop ( 0 : no output )  10 vector ,digit 
	 
: roman-rec ( numerals n -- )  10 /mod dup if >r over 2 + r> recurse else drop then ,digit ; 
: roman ( n -- c-addr u )  
  dup 0 4000 within 0= abort" EX LIMITO!" 
  HERE SWAP  s" IVXLCDM" drop swap roman-rec  HERE OVER - ;

1999 roman type     \ MCMXCIX 
  25 roman type     \ XXV
 944 roman type     \ CMXLIV

Alternative implementation

create romans 0 , 1 , 5 , 21 , 9 , 2 , 6 , 22 , 86 , 13 ,
  does> swap cells + @ ;

: roman-digit                          ( a1 n1 a2 n2 -- a3)
  drop >r romans
  begin dup while tuck 4 mod 1- chars r@ + c@ over c! char+ swap 4 / repeat
  r> drop drop
;

: (split) swap >r /mod r> swap ;

: >roman                               ( n1 a -- a n2)
  tuck 1000 (split) s" M  " roman-digit 100 (split) s" CDM" roman-digit
  10 (split) s" XLC" roman-digit 1 (split) s" IVX" roman-digit nip over -
;

create (roman) 16 chars allot

1999 (roman) >roman type cr

Fortran

Works with: Fortran version 90 and later
program roman_numerals

  implicit none

  write (*, '(a)') roman (2009)
  write (*, '(a)') roman (1666)
  write (*, '(a)') roman (3888)

contains

function roman (n) result (r)

  implicit none
  integer, intent (in) :: n
  integer, parameter   :: d_max = 13
  integer              :: d
  integer              :: m
  integer              :: m_div
  character (32)       :: r
  integer,        dimension (d_max), parameter :: d_dec = &
    & (/1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1/)
  character (32), dimension (d_max), parameter :: d_rom = &
    & (/'M ', 'CM', 'D ', 'CD', 'C ', 'XC', 'L ', 'XL', 'X ', 'IX', 'V ', 'IV', 'I '/)

  r = ''
  m = n
  do d = 1, d_max
    m_div = m / d_dec (d)
    r = trim (r) // repeat (trim (d_rom (d)), m_div)
    m = m - d_dec (d) * m_div
  end do

end function roman

end program roman_numerals
Output:
  MMIX
  MDCLXVI
  MMMDCCCLXXXVIII

Go

For fluff, the unicode overbar is recognized as a factor of 1000, as described in WP.

If you see boxes in the code below, those are supposed to be the Unicode combining overline (U+0305) and look like IVXLCDM. Or, if you see overstruck combinations of letters, that's a different font rendering problem. (If you need roman numerals > 3999 reliably, it might best to stick to chiseling them in stone...)

package main

import "fmt"

var (
    m0 = []string{"", "I", "II", "III", "IV", "V", "VI", "VII", "VIII", "IX"}
    m1 = []string{"", "X", "XX", "XXX", "XL", "L", "LX", "LXX", "LXXX", "XC"}
    m2 = []string{"", "C", "CC", "CCC", "CD", "D", "DC", "DCC", "DCCC", "CM"}
    m3 = []string{"", "M", "MM", "MMM", "I̅V̅",
        "V̅", "V̅I̅", "V̅I̅I̅", "V̅I̅I̅I̅", "I̅X̅"}
    m4 = []string{"", "X̅", "X̅X̅", "X̅X̅X̅", "X̅L̅",
        "L̅", "L̅X̅", "L̅X̅X̅", "L̅X̅X̅X̅", "X̅C̅"}
    m5 = []string{"", "C̅", "C̅C̅", "C̅C̅C̅", "C̅D̅",
        "D̅", "D̅C̅", "D̅C̅C̅", "D̅C̅C̅C̅", "C̅M̅"}
    m6 = []string{"", "M̅", "M̅M̅", "M̅M̅M̅"}
)

func formatRoman(n int) (string, bool) {
    if n < 1 || n >= 4e6 {
        return "", false
    }
    // this is efficient in Go.  the seven operands are evaluated,
    // then a single allocation is made of the exact size needed for the result.
    return m6[n/1e6] + m5[n%1e6/1e5] + m4[n%1e5/1e4] + m3[n%1e4/1e3] +
        m2[n%1e3/1e2] + m1[n%100/10] + m0[n%10],
        true
}

func main() {
    // show three numbers mentioned in task descriptions
    for _, n := range []int{1990, 2008, 1666} {
        r, ok := formatRoman(n)
        if ok {
            fmt.Println(n, "==", r)
        } else {
            fmt.Println(n, "not representable")
        }
    }
}
Output:
1990 == MCMXC
2008 == MMVIII
1666 == MDCLXVI

Golo

#!/usr/bin/env golosh
----
This module takes a decimal integer and converts it to a Roman numeral.
----
module Romannumeralsencode

augment java.lang.Integer {

  function digits = |this| {

    var remaining = this
    let digits = vector[]
    while remaining > 0 {
      digits: prepend(remaining % 10)
      remaining = remaining / 10
    }
    return digits
  }

  ----
  123: digitsWithPowers() will return [[1, 2], [2, 1], [3, 0]]
  ----
  function digitsWithPowers = |this| -> vector[
    [ this: digits(): get(i), (this: digits(): size() - 1) - i ] for (var i = 0, i < this: digits(): size(), i = i + 1)
  ]

  function encode = |this| {

    require(this > 0, "the integer must be positive!")

    let romanPattern = |digit, powerOf10| -> match {
      when digit == 1 then i
      when digit == 2 then i + i
      when digit == 3 then i + i + i
      when digit == 4 then i + v
      when digit == 5 then v
      when digit == 6 then v + i
      when digit == 7 then v + i + i
      when digit == 8 then v + i + i + i
      when digit == 9 then i + x
      otherwise ""
    } with {
      i, v, x = [
        [ "I", "V", "X" ],
        [ "X", "L", "C" ],
        [ "C", "D", "M" ],
        [ "M", "?", "?" ]
      ]: get(powerOf10)
    }

    return vector[ romanPattern(digit, power) foreach digit, power in this: digitsWithPowers() ]: join("")
  }
}

function main = |args| {
  println("1990 == MCMXC? " + (1990: encode() == "MCMXC"))
  println("2008 == MMVIII? " + (2008: encode() == "MMVIII"))
  println("1666 == MDCLXVI? " + (1666: encode() == "MDCLXVI"))
}

Groovy

symbols = [ 1:'I', 4:'IV', 5:'V', 9:'IX', 10:'X', 40:'XL', 50:'L', 90:'XC', 100:'C', 400:'CD', 500:'D', 900:'CM', 1000:'M' ]

def roman(arabic) {
    def result = ""
    symbols.keySet().sort().reverse().each { 
        while (arabic >= it) {
            arabic-=it
            result+=symbols[it]
        }
    }
    return result
}
assert roman(1) == 'I'
assert roman(2) == 'II'
assert roman(4) == 'IV'
assert roman(8) == 'VIII'
assert roman(16) == 'XVI'
assert roman(32) == 'XXXII'
assert roman(25) == 'XXV'
assert roman(64) == 'LXIV'
assert roman(128) == 'CXXVIII'
assert roman(256) == 'CCLVI'
assert roman(512) == 'DXII'
assert roman(954) == 'CMLIV'
assert roman(1024) == 'MXXIV'
assert roman(1666) == 'MDCLXVI'
assert roman(1990) == 'MCMXC'
assert roman(2008) == 'MMVIII'

Haskell

With an explicit decimal digit representation list:

digit :: Char -> Char -> Char -> Integer -> String
digit x y z k =
  [[x], [x, x], [x, x, x], [x, y], [y], [y, x], [y, x, x], [y, x, x, x], [x, z]] !!
  (fromInteger k - 1)

toRoman :: Integer -> String
toRoman 0 = ""
toRoman x
  | x < 0 = error "Negative roman numeral"
toRoman x
  | x >= 1000 = 'M' : toRoman (x - 1000)
toRoman x
  | x >= 100 = digit 'C' 'D' 'M' q ++ toRoman r
  where
    (q, r) = x `divMod` 100
toRoman x
  | x >= 10 = digit 'X' 'L' 'C' q ++ toRoman r
  where
    (q, r) = x `divMod` 10
toRoman x = digit 'I' 'V' 'X' x

main :: IO ()
main = print $ toRoman <$> [1999, 25, 944]
Output:
["MCMXCIX","XXV","CMXLIV"]

or, defining romanFromInt in terms of mapAccumL

import Data.Bifunctor (first)
import Data.List (mapAccumL)
import Data.Tuple (swap)

roman :: Int -> String
roman =
  romanFromInt $
  zip
    [1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1]
    (words "M CM D CD C XC L XL X IX V IV I")

romanFromInt :: [(Int, String)] -> Int -> String
romanFromInt nks n = concat . snd $ mapAccumL go n nks
  where
    go a (v, s) = swap $ first ((>> s) . enumFromTo 1) $ quotRem a v

main :: IO ()
main = (putStrLn . unlines) (roman <$> [1666, 1990, 2008, 2016, 2018])
Output:
MDCLXVI
MCMXC
MMVIII
MMXVI
MMXVIII

With the Roman patterns abstracted, and in a simple logic programming idiom:

module Main where

------------------------
--  ENCODER FUNCTION  --
------------------------

romanDigits = "IVXLCDM"

--  Meaning and indices of the romanDigits sequence:
--
--    magnitude |  1 5  | index
--   -----------|-------|-------
--        0     |  I V  |  0 1   
--        1     |  X L  |  2 3   
--        2     |  C D  |  4 5   
--        3     |  M    |  6     
--
--  romanPatterns are index offsets into romanDigits,
--  from an index base of 2 * magnitude.

romanPattern 0 = []      -- empty string
romanPattern 1 = [0]     -- I or X or C or M
romanPattern 2 = [0,0]   -- II or XX...
romanPattern 3 = [0,0,0] -- III...
romanPattern 4 = [0,1]   -- IV...
romanPattern 5 = [1]     -- ...
romanPattern 6 = [1,0]
romanPattern 7 = [1,0,0]
romanPattern 8 = [1,0,0,0]
romanPattern 9 = [0,2]

encodeValue 0 _ = ""
encodeValue value magnitude = encodeValue rest (magnitude + 1) ++ digits
  where
    low = rem value 10 -- least significant digit (encoded now)
    rest = div value 10 -- the other digits (to be encoded next)
    indices = map addBase (romanPattern low)
    addBase i = i + (2 * magnitude)
    digits = map pickDigit indices
    pickDigit i = romanDigits!!i

encode value = encodeValue value 0

------------------
--  TEST SUITE  --
------------------

main = do
  test "MCMXC" 1990
  test "MMVIII" 2008
  test "MDCLXVI" 1666

test expected value = putStrLn ((show value) ++ " = " ++ roman ++ remark)
  where
    roman = encode value
    remark =
      " (" ++
      (if roman == expected then "PASS"
       else ("FAIL, expected " ++ (show expected))) ++ ")"
Output:
1990 = MCMXC (PASS)
2008 = MMVIII (PASS)
1666 = MDCLXVI (PASS)

HicEst

CHARACTER Roman*20

CALL RomanNumeral(1990, Roman) ! MCMXC
CALL RomanNumeral(2008, Roman) ! MMVIII
CALL RomanNumeral(1666, Roman) ! MDCLXVI

END

SUBROUTINE RomanNumeral( arabic, roman)
  CHARACTER roman
  DIMENSION ddec(13)
  DATA      ddec/1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1/

  roman = ' '
  todo = arabic
  DO d = 1, 13
    DO rep = 1, todo / ddec(d)
      roman = TRIM(roman) // TRIM(CHAR(d, 13, "M  CM D  CD C  XC L  XL X  OX V  IV I  "))
      todo = todo - ddec(d)
    ENDDO
  ENDDO
END

Hoon

Library file (e.g. /lib/rhonda.hoon):

|%
++  parse
  |=  t=tape  ^-  @ud
  =.  t  (cass t)
  =|  result=@ud
  |-
  ?~  t  result
  ?~  t.t  (add result (from-numeral i.t))
  =+  [a=(from-numeral i.t) b=(from-numeral i.t.t)]
  ?:  (gte a b)  $(result (add result a), t t.t)
  $(result (sub (add result b) a), t t.t.t)
++  yield
  |=  n=@ud  ^-  tape
  =|  result=tape
  =/  values  to-numeral
  |-
  ?~  values  result
  ?:  (gte n -.i.values)
    $(result (weld result +.i.values), n (sub n -.i.values))
  $(values t.values)
++  from-numeral
  |=  c=@t  ^-  @ud
  ?:  =(c 'i')  1
  ?:  =(c 'v')  5
  ?:  =(c 'x')  10
  ?:  =(c 'l')  50
  ?:  =(c 'c')  100
  ?:  =(c 'd')  500
  ?:  =(c 'm')  1.000
  !!
++  to-numeral
  ^-  (list [@ud tape])
  :*
    [1.000 "m"]
    [900 "cm"]
    [500 "d"]
    [400 "cd"]
    [100 "c"]
    [90 "xc"]
    [50 "l"]
    [40 "xl"]
    [10 "x"]
    [9 "ix"]
    [5 "v"]
    [4 "iv"]
    [1 "i"]
    ~
  ==
--

Script file ("generator") (e.g. /gen/roman.hoon):

/+  *roman
:-  %say
|=  [* [x=$%([%from-roman tape] [%to-roman @ud]) ~] ~]
:-  %noun
^-  tape
?-  -.x
  %from-roman  "{<(parse +.x)>}"
  %to-roman  (yield +.x)
==

Icon and Unicon

link numbers   # commas, roman

procedure main(arglist)
every x := !arglist do
   write(commas(x), " -> ",roman(x)|"*** can't convert to Roman numerals ***")
end

numbers.icn provides roman as seen below and is based upon a James Gimple SNOBOL4 function.

procedure roman(n)		#: convert integer to Roman numeral
   local arabic, result
   static equiv

   initial equiv := ["","I","II","III","IV","V","VI","VII","VIII","IX"]

   integer(n) > 0 | fail
   result := ""
   every arabic := !n do
      result := map(result,"IVXLCDM","XLCDM**") || equiv[arabic + 1]
   if find("*",result) then fail else return result
end
Output:
#roman.exe  3 4 8 49 2010 1666 3000 3999 4000 

3 -> III
4 -> IV
8 -> VIII
49 -> XLIX
2,010 -> MMX
1,666 -> MDCLXVI
3,999 -> MMMCMXCIX
4,000 -> *** can't convert to Roman numerals ***

Intercal

INTERCAL outputs numbers as Roman numerals by default, so this is surprisingly trivial for a language that generally tries to make things as difficult as possible. Although you do still have to input the numbers as spelled out digitwise in all caps.

       PLEASE WRITE IN .1
           DO READ OUT .1
           DO GIVE UP
Output:
$ ./roman
ONE SIX SIX SIX
       
MDCLXVI

Io

Translation of: C#
Roman := Object clone do (
    nums := list(1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1)
    rum := list("M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I")
    
    numeral := method(number,
        result := ""
        for(i, 0, nums size,
            if(number == 0, break)
            while(number >= nums at(i),
                number = number - nums at(i)
                result = result .. rum at(i)
            )
        )
        return result
    )
)

Roman numeral(1666) println

J

rfd obtains Roman numerals from decimals.

R1000=. ;L:1 ,{ <@(<;._1);._2]0 :0
  C CC CCC CD D DC DCC DCCC CM
  X XX XXX XL L LX LXX LXXX XC
  I II III IV V VI VII VIII IX
)

rfd=: ('M' $~ <.@%&1000) , R1000 {::~ 1000&|

Explanation: R1000's definition contains rows representing each of 10 different digits in the 100s, 10s and 1s column (the first entry in each row is blank -- each entry is preceded by a space). R1000 itself represents the first 1000 roman numerals (the cartesian product of these three rows of roman numeral "digits" which is constructed so that they are in numeric order. And the first entry -- zero -- is just blank). To convert a number to its roman numeral representation, we will separate the number into the integer part after dividing by 1000 (that's the number of 'M's we need) and the remainder after dividing by 1000 (which will be an index into R1000).

For example:
   rfd 1234
MCCXXXIV
   rfd 567
DLXVII
   rfd 89
LXXXIX

Derived from the J Wiki. Further examples of use will be found there.

Java

Translation of: Ada

The conversion function throws an IllegalArgumentException for non-positive numbers, since Java does not have unsigned primitives.

Works with: Java version 1.5+
public class RN {

    enum Numeral {
        I(1), IV(4), V(5), IX(9), X(10), XL(40), L(50), XC(90), C(100), CD(400), D(500), CM(900), M(1000);
        int weight;

        Numeral(int weight) {
            this.weight = weight;
        }
    };

    public static String roman(long n) {
        
        if( n <= 0) {
            throw new IllegalArgumentException();
        }
        
        StringBuilder buf = new StringBuilder();

        final Numeral[] values = Numeral.values();
        for (int i = values.length - 1; i >= 0; i--) {
            while (n >= values[i].weight) {
                buf.append(values[i]);
                n -= values[i].weight;
            }
        }
        return buf.toString();
    }

    public static void test(long n) {
        System.out.println(n + " = " + roman(n));
    }

    public static void main(String[] args) {
        test(1999);
        test(25);
        test(944);
        test(0);
    }

}
Output:
1999 = MCMXCIX
25 = XXV
944 = CMXLIV
Exception in thread "main" java.lang.IllegalArgumentException
	at RN.roman(RN.java:15)
	at RN.test(RN.java:31)
	at RN.main(RN.java:38)
Works with: Java version 1.8+
import java.util.Set;
import java.util.EnumSet;
import java.util.Collections;
import java.util.stream.Collectors;
import java.util.stream.LongStream;

public interface RomanNumerals {
  public enum Numeral {
    M(1000), CM(900), D(500), CD(400), C(100), XC(90), L(50), XL(40), X(10), IX(9), V(5), IV(4), I(1);

    public final long weight;

    private static final Set<Numeral> SET = Collections.unmodifiableSet(EnumSet.allOf(Numeral.class));

    private Numeral(long weight) {
      this.weight = weight;
    }

    public static Numeral getLargest(long weight) {
      return SET.stream()
        .filter(numeral -> weight >= numeral.weight)
        .findFirst()
        .orElse(I)
      ;
    }
  };

  public static String encode(long n) {
    return LongStream.iterate(n, l -> l - Numeral.getLargest(l).weight)
      .limit(Numeral.values().length)
      .filter(l -> l > 0)
      .mapToObj(Numeral::getLargest)
      .map(String::valueOf)
      .collect(Collectors.joining())
    ;
  }

  public static long decode(String roman) {
    long result =  new StringBuilder(roman.toUpperCase()).reverse().chars()
      .mapToObj(c -> Character.toString((char) c))
      .map(numeral -> Enum.valueOf(Numeral.class, numeral))
      .mapToLong(numeral -> numeral.weight)
      .reduce(0, (a, b) -> a + (a <= b ? b : -b))
    ;
    if (roman.charAt(0) == roman.charAt(1)) {
      result += 2 * Enum.valueOf(Numeral.class, roman.substring(0, 1)).weight;
    }
    return result;
  }

  public static void test(long n) {
    System.out.println(n + " = " + encode(n));
    System.out.println(encode(n) + " = " + decode(encode(n)));
  }

  public static void main(String[] args) {
    LongStream.of(1999, 25, 944).forEach(RomanNumerals::test);
  }
}
Output:
1999 = MCMXCIX
MCMXCIX = 1999
25 = XXV
XXV = 25
944 = CMXLIV
CMXLIV = 944

JavaScript

ES5

Iteration

Translation of: Tcl
var roman = {
    map: [
        1000, 'M', 900, 'CM', 500, 'D', 400, 'CD', 100, 'C', 90, 'XC',
        50, 'L', 40, 'XL', 10, 'X', 9, 'IX', 5, 'V', 4, 'IV', 1, 'I',
    ],
    int_to_roman: function(n) {
        var value = '';
        for (var idx = 0; n > 0 && idx < this.map.length; idx += 2) {
            while (n >= this.map[idx]) {
                value += this.map[idx + 1];
                n -= this.map[idx];
            }
        }
        return value;
    }
}

roman.int_to_roman(1999); // "MCMXCIX"

Functional composition

(function () {
    'use strict';


    // If the Roman is a string, pass any delimiters through

    // (Int | String) -> String
    function romanTranscription(a) {
        if (typeof a === 'string') {
            var ps = a.split(/\d+/),
                dlm = ps.length > 1 ? ps[1] : undefined;

            return (dlm ? a.split(dlm)
                    .map(function (x) {
                        return Number(x);
                    }) : [a])
                .map(roman)
                .join(dlm);
        } else return roman(a);
    }

    // roman :: Int -> String
    function roman(n) {
        return [[1000, "M"], [900, "CM"], [500, "D"], [400, "CD"], [100,
        "C"], [90, "XC"], [50, "L"], [40, "XL"], [10, "X"], [9,
        "IX"], [5, "V"], [4, "IV"], [1, "I"]]
            .reduce(function (a, lstPair) {
                var m = a.remainder,
                    v = lstPair[0];

                return (v > m ? a : {
                    remainder: m % v,
                    roman: a.roman + Array(
                            Math.floor(m / v) + 1
                        )
                        .join(lstPair[1])
                });
            }, {
                remainder: n,
                roman: ''
            }).roman;   
    }

    // TEST

    return [2016, 1990, 2008, "14.09.2015", 2000, 1666].map(
        romanTranscription);

})();
Output:
["MMXVI", "MCMXC", "MMVIII", "XIV.IX.MMXV", "MM", "MDCLXVI"]

ES6

Functional

Translation of: Haskell

(mapAccumL version)

(() => {
    "use strict";

    // -------------- ROMAN INTEGER STRINGS --------------

    // roman :: Int -> String
    const roman = n =>
        mapAccumL(residue =>
            ([k, v]) => second(
                q => 0 < q ? (
                    k.repeat(q)
                ) : ""
            )(remQuot(residue)(v))
        )(n)(
            zip([
                "M", "CM", "D", "CD", "C", "XC",
                "L", "XL", "X", "IX", "V", "IV", "I"
            ])([
                1000, 900, 500, 400, 100, 90,
                50, 40, 10, 9, 5, 4, 1
            ])
        )[1]
        .join("");


    // ---------------------- TEST -----------------------
    // main :: IO ()
    const main = () => (
        [2016, 1990, 2008, 2000, 2020, 1666].map(roman)
    ).join("\n");


    // ---------------- GENERIC FUNCTIONS ----------------

    // mapAccumL :: (acc -> x -> (acc, y)) -> acc ->
    // [x] -> (acc, [y])
    const mapAccumL = f =>
        // A tuple of an accumulation and a list
        // obtained by a combined map and fold,
        // with accumulation from left to right.
        acc => xs => [...xs].reduce(
            (a, x) => {
                const tpl = f(a[0])(x);

                return [
                    tpl[0],
                    a[1].concat(tpl[1])
                ];
            },
            [acc, []]
        );


    // remQuot :: Int -> Int -> (Int, Int)
    const remQuot = m =>
        n => [m % n, Math.trunc(m / n)];


    // second :: (a -> b) -> ((c, a) -> (c, b))
    const second = f =>
        // A function over a simple value lifted
        // to a function over a tuple.
        // f (a, b) -> (a, f(b))
        xy => [xy[0], f(xy[1])];


    // zip :: [a] -> [b] -> [(a, b)]
    const zip = xs =>
        // The paired members of xs and ys, up to
        // the length of the shorter of the two lists.
        ys => Array.from({
            length: Math.min(xs.length, ys.length)
        }, (_, i) => [xs[i], ys[i]]);


    // MAIN --
    return main();
})();
Output:
MDCLXVI
MCMXC
MMVIII
MMXVI
MMXVIII
MMXX

Declarative

function toRoman(num) {
  return 'I'
    .repeat(num)
    .replace(/IIIII/g, 'V')
    .replace(/VV/g, 'X')
    .replace(/XXXXX/g, 'L')
    .replace(/LL/g, 'C')
    .replace(/CCCCC/g, 'D')
    .replace(/DD/g, 'M')
    .replace(/VIIII/g, 'IX')
    .replace(/LXXXX/g, 'XC')
    .replace(/XXXX/g, 'XL')
    .replace(/DCCCC/g, 'CM')
    .replace(/CCCC/g, 'CD')
    .replace(/IIII/g, 'IV');
}

console.log(toRoman(1666));
Output:
MDCLXVI

jq

Works with: jq

Works with gojq, the Go implementation of jq

The "easy-to-code" version is presented first, followed by the "orders of magnitude" version. Both versions work for positive integers up to and including 399,999, but note that the Unicode glyphs for 50,000 and 100,000 are not supported in many environments.

The test cases and output are identical for both versions and are therefore not repeated.

Easy-to-code version

def to_roman_numeral:
  def romans:
      [100000, "\u2188"],
      [90000,  "ↂ\u2188"],
      [50000,  "\u2187"],
      [40000,  "ↂ\u2187"],
      [10000,  "ↂ"],
      [9000,  "Mↂ"],
      [5000,   "ↁ"],
      [4000,  "Mↁ"],
      [1000,   "M"],
      [900,   "CM"],
      [500,    "D"],
      [400,   "CD"],
      [100,    "C"],
      [90,    "XC"],
      [50,     "L"],
      [40,    "XL"],
      [10,     "X"],
      [9,     "IX"],
      [5,      "V"],
      [4,     "IV"],
      [1,      "I"] ;
  
  if . < 1 or . > 399999
  then "to_roman_numeral: \(.) is out of range" | error
  else reduce romans as [$i, $r] ({n: .};
        until (.n < $i;
          .res += $r
          | .n = .n - $i ) )
  | .res
  end ;

Test Cases

def testcases: [1668, 1990, 2008, 2020, 4444, 5000, 8999, 39999, 89999, 399999];

"Decimal => Roman:",
 (testcases[]
  | "   \(.) => \(to_roman_numeral)" )
Output:
Decimal => Roman:
   1668 => MDCLXVIII
   1990 => MCMXC
   2008 => MMVIII
   2020 => MMXX
   4444 => MↁCDXLIV
   5000 => ↁ
   8999 => ↁMMMCMXCIX
   39999 => ↂↂↂMↂCMXCIX
   89999 => ↇↂↂↂMↂCMXCIX
   399999 => ↈↈↈↂↈMↂCMXCIX

"Orders of Magnitude" version

Translated from Julia extended to 399,999

def digits: tostring | explode | map( [.]|implode|tonumber);
# Non-negative integer to Roman numeral up to 399,999
def to_roman_numeral:
  if . < 1 or . > 399999
  then "to_roman_numeral: \(.) is out of range" | error
  else [["I", "X", "C", "M", "ↂ", "\u2188"], ["V", "L", "D", "ↁ", "\u2187"]] as $DR
  | (digits|reverse) as $digits
  | reduce range(0;$digits|length) as $omag ({rnum: ""};
     $digits[$omag] as $d
     | if   $d == 0 then .omr = ""
       elif $d <  4 then .omr = $DR[0][$omag] * $d
       elif $d == 4 then .omr = $DR[0][$omag] + $DR[1][$omag]
       elif $d == 5 then .omr = $DR[1][$omag]
       elif $d <  9 then .omr = $DR[1][$omag] + ($DR[0][$omag] * ($d - 5))
       else .omr = $DR[0][$omag] + $DR[0][$omag+1]
       end
     | .rnum = .omr + .rnum )
  | .rnum
  end;

Jsish

This covers both Encode (toRoman) and Decode (fromRoman).

/* Roman numerals, in Jsish */
var Roman = {
    ord: ['M', 'CM', 'D', 'CD', 'C', 'XC', 'L', 'XL', 'X', 'IX', 'V', 'IV', 'I'],
    val: [1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1],

    fromRoman: function(roman:string):number {
        var n = 0;
        var re = /IV|IX|I|V|XC|XL|X|L|CD|CM|C|D|M/g;
        var matches = roman.match(re);
        if (!matches) return NaN;
        for (var hit of matches) n += this.val[this.ord.indexOf(hit)];
        return n;
    },

    toRoman: function(n:number):string {
        var roman = '';
        var idx = 0;
        while (n > 0) {
            while (n >= this.val[idx]) {
                roman += this.ord[idx];
                n -= this.val[idx];
            }
            idx++;
        }
        return roman;
    }
};

provide('Roman', 1);

if (Interp.conf('unitTest')) {
;    Roman.fromRoman('VIII');
;    Roman.fromRoman('MMMDIV');
;    Roman.fromRoman('CDIV');
;    Roman.fromRoman('MDCLXVI');
;    Roman.fromRoman('not');

;    Roman.toRoman(8);
;    Roman.toRoman(3504);
;    Roman.toRoman(404);
;    Roman.toRoman(1666);
}

/*
=!EXPECTSTART!=
Roman.fromRoman('VIII') ==> 8
Roman.fromRoman('MMMDIV') ==> 3504
Roman.fromRoman('CDIV') ==> 404
Roman.fromRoman('MDCLXVI') ==> 1666
Roman.fromRoman('not') ==> NaN
Roman.toRoman(8) ==> VIII
Roman.toRoman(3504) ==> MMMDIV
Roman.toRoman(404) ==> CDIV
Roman.toRoman(1666) ==> MDCLXVI
=!EXPECTEND!=
*/
Output:
prompt$ jsish -u Roman.jsi
[PASS] Roman.jsi

Julia

using Printf

function romanencode(n::Integer)
    if n < 1 || n > 4999 throw(DomainError()) end

    DR = [["I", "X", "C", "M"] ["V", "L", "D", "MMM"]]
    rnum = ""
    for (omag, d) in enumerate(digits(n))
        if d == 0
            omr = ""
        elseif d <  4
            omr = DR[omag, 1] ^ d
        elseif d == 4
            omr = DR[omag, 1] * DR[omag, 2]
        elseif d == 5
            omr = DR[omag, 2]
        elseif d <  9
            omr = DR[omag, 2] * DR[omag, 1] ^ (d - 5)
        else
            omr = DR[omag, 1] * DR[omag + 1, 1]
        end
        rnum = omr * rnum
    end
    return rnum
end

testcases = [1990, 2008, 1668]
append!(testcases, rand(1:4999, 12))
testcases = unique(testcases)

println("Test romanencode, arabic => roman:")
for n in testcases
    @printf("%-4i => %s\n", n, romanencode(n))
end
Output:
Test romanencode, arabic => roman:
1990 => MCMXC
2008 => MMVIII
1668 => MDCLXVIII
2928 => MMCMXXVIII
129  => CXXIX
4217 => MMMMCCXVII
1503 => MDIII
2125 => MMCXXV
1489 => MCDLXXXIX
3677 => MMMDCLXXVII
1465 => MCDLXV
1421 => MCDXXI
1642 => MDCXLII
572  => DLXXII
3714 => MMMDCCXIV

Kotlin

val romanNumerals = mapOf(
    1000 to "M",
    900 to "CM",
    500 to "D",
    400 to "CD",
    100 to "C",
    90 to "XC",
    50 to "L",
    40 to "XL",
    10 to "X",
    9 to "IX",
    5 to "V",
    4 to "IV",
    1 to "I"
)

fun encode(number: Int): String? {
    if (number > 5000 || number < 1) {
        return null
    }
    var num = number
    var result = ""
    for ((multiple, numeral) in romanNumerals.entries) {
        while (num >= multiple) {
            num -= multiple
            result += numeral
        }
    }
    return result
}

fun main(args: Array<String>) {
    println(encode(1990))
    println(encode(1666))
    println(encode(2008))
}
Output:
MCMXC
MDCLXVI
MMVIII

Alternatively:

fun Int.toRomanNumeral(): String {
    fun digit(k: Int, unit: String, five: String, ten: String): String {
        return when (k) {
            in 1..3 -> unit.repeat(k)
            4 -> unit + five
            in 5..8 -> five + unit.repeat(k - 5)
            9 -> unit + ten
            else -> throw IllegalArgumentException("$k not in range 1..9")
        }
    }
    return when (this) {
        0 -> ""
        in 1..9 -> digit(this, "I", "V", "X")
        in 10..99 -> digit(this / 10, "X", "L", "C") + (this % 10).toRomanNumeral()
        in 100..999 -> digit(this / 100, "C", "D", "M") + (this % 100).toRomanNumeral()
        in 1000..3999 -> "M" + (this - 1000).toRomanNumeral()
        else -> throw IllegalArgumentException("${this} not in range 0..3999")
    }
}

Lasso

define br => '\r'
// encode roman
define encodeRoman(num::integer)::string => {
	local(ref = array('M'=1000, 'CM'=900, 'D'=500, 'CD'=400, 'C'=100, 'XC'=90, 'L'=50, 'XL'=40, 'X'=10, 'IX'=9, 'V'=5, 'IV'=4, 'I'=1))
	local(out = string)
	with i in #ref do => {
		while(#num >= #i->second) => {
			#out->append(#i->first)
			#num -= #i->second
		}
	}
	return #out
}

'1990 in roman is '+encodeRoman(1990)
br
'2008 in roman is '+encodeRoman(2008)
br
'1666 in roman is '+encodeRoman(1666)

LaTeX

The macro \Roman is defined for uppercase roman numeral, accepting as argument a name of an existing counter.

\documentclass{minimal}
\newcounter{currentyear}
\setcounter{currentyear}{\year}
\begin{document}
Anno Domini \Roman{currentyear}
\end{document}

LiveCode

function toRoman intNum
    local roman,numArabic
    put "M,CM,D,CD,C,XC,L,XL,X,IX,V,IV,I" into romans
    put "1000,900,500,400,100,90,50,40,10,9,5,4,1" into arabics
    put intNum into numArabic
    repeat with n = 1 to the number of items of romans
        put numArabic div item n of arabics into nums
        if nums > 0 then
            put repeatChar(item n of romans,nums) after roman
            add -(nums * item n of arabics) to numArabic
        end if
    end repeat
return roman
end toRoman

function repeatChar c n
    local cc
    repeat n times
        put c after cc
    end repeat
    return cc
end repeatChar

Examples

toRoman(2009) -- MMIX
toRoman(1666) -- MDCLXVI
toRoman(1984) -- MCMLXXXIV
toRoman(3888) -- MMMDCCCLXXXVIII

make "roman.rules [
  [1000 M] [900 CM] [500 D] [400 CD]
  [ 100 C] [ 90 XC] [ 50 L] [ 40 XL]
  [  10 X] [  9 IX] [  5 V] [  4 IV]
  [   1 I]
]

to roman :n [:rules :roman.rules] [:acc "||]
  if empty? :rules [output :acc]
  if :n < first first :rules [output (roman :n bf :rules :acc)]
  output (roman :n - first first :rules  :rules  word :acc last first :rules)
end
Works with: UCB Logo
make "patterns [[?] [? ?] [? ? ?] [? ?2] [?2] [?2 ?] [?2 ? ?] [?2 ? ? ?] [? ?3]]

to digit :d :numerals
  if :d = 0 [output "||]
  output apply (sentence "\( "word (item :d :patterns) "\)) :numerals
end
to digits :n :numerals
  output word ifelse :n < 10 ["||] [digits int :n/10 bf bf :numerals] ~
              digit modulo :n 10 :numerals
end
to roman :n
  if or :n < 0 :n >= 4000 [output [EX MODVS!]]
  output digits :n [I V X L C D M]
end

print roman 1999  ; MCMXCIX 
print roman 25    ; XXV
print roman 944   ; CMXLIV

LOLCODE

HAI 1.2
I HAS A Romunz ITZ A BUKKIT
Romunz HAS A SRS  0 ITZ "M"
Romunz HAS A SRS  1 ITZ "CM"
Romunz HAS A SRS  2 ITZ "D"
Romunz HAS A SRS  3 ITZ "CD"
Romunz HAS A SRS  4 ITZ "C"
Romunz HAS A SRS  5 ITZ "XC"
Romunz HAS A SRS  6 ITZ "L"
Romunz HAS A SRS  7 ITZ "XL"
Romunz HAS A SRS  8 ITZ "X"
Romunz HAS A SRS  9 ITZ "IX"
Romunz HAS A SRS 10 ITZ "V"
Romunz HAS A SRS 11 ITZ "IV"
Romunz HAS A SRS 12 ITZ "I"

I HAS A Valuez ITZ A BUKKIT
Valuez HAS A SRS  0 ITZ 1000
Valuez HAS A SRS  1 ITZ  900
Valuez HAS A SRS  2 ITZ  500
Valuez HAS A SRS  3 ITZ  400
Valuez HAS A SRS  4 ITZ  100
Valuez HAS A SRS  5 ITZ   90
Valuez HAS A SRS  6 ITZ   50
Valuez HAS A SRS  7 ITZ   40
Valuez HAS A SRS  8 ITZ   10
Valuez HAS A SRS  9 ITZ    9
Valuez HAS A SRS 10 ITZ    5
Valuez HAS A SRS 11 ITZ    4
Valuez HAS A SRS 12 ITZ    1

HOW IZ I Romunize YR Num
  I HAS A Result ITZ ""
  IM IN YR Outer UPPIN YR Dummy TIL BOTH SAEM Num AN 0
     IM IN YR Inner UPPIN YR Index TIL BOTH SAEM Index AN 13
       BOTH SAEM Num AN BIGGR OF Num AN Valuez'Z SRS Index, O RLY?
         YA RLY
            Num R DIFF OF Num AN Valuez'Z SRS Index
            Result R SMOOSH Result Romunz'Z SRS Index MKAY
            GTFO
        OIC
      IM OUTTA YR Inner
    IM OUTTA YR Outer
    FOUND YR Result
IF U SAY SO

VISIBLE SMOOSH 2009 " = " I IZ Romunize YR 2009 MKAY MKAY
VISIBLE SMOOSH 1666 " = " I IZ Romunize YR 1666 MKAY MKAY
VISIBLE SMOOSH 3888 " = " I IZ Romunize YR 3888 MKAY MKAY
KTHXBYE
Output:
2009 = MMIX
1666 = MDCLXVI
3888 = MMMDCCCLXXXVIII

LotusScript

Function toRoman(value) As String
	Dim arabic(12) As Integer
	Dim roman(12) As String
	
	arabic(0) = 1000
	arabic(1) = 900
	arabic(2) = 500
	arabic(3) = 400
	arabic(4) = 100
	arabic(5) = 90
	arabic(6) = 50
	arabic(7) = 40
	arabic(8) = 10
	arabic(9) = 9
	arabic(10) = 5
	arabic(11) = 4
	arabic(12) = 1
	
	roman(0) = "M"
	roman(1) = "CM"
	roman(2) = "D"
	roman(3) = "CD"
	roman(4) = "C"
	roman(5) = "XC"
	roman(6) = "L"
	roman(7) = "XL"
	roman(8) = "X"
	roman(9) = "IX"
	roman(10) = "V"
	roman(11) = "IV"
	roman(12) = "I"
	
	Dim i As Integer, result As String
	
	For i = 0 To 12
		Do While value >= arabic(i)
			result = result + roman(i)
			value = value - arabic(i)
		Loop
	Next i
	
	toRoman = result
End Function

Lua

romans = {
{1000, "M"},
{900, "CM"}, {500, "D"}, {400, "CD"}, {100, "C"},
{90, "XC"}, {50, "L"}, {40, "XL"}, {10, "X"},
{9, "IX"}, {5, "V"}, {4, "IV"}, {1, "I"} }

k = io.read() + 0
for _, v in ipairs(romans) do --note that this is -not- ipairs.
  val, let = unpack(v)
  while k >= val do
    k = k - val
	io.write(let)
  end
end
print()

M4

define(`roman',`ifelse(eval($1>=1000),1,`M`'roman(eval($1-1000))',
`ifelse(eval($1>=900),1,`CM`'roman(eval($1-900))',
`ifelse(eval($1>=500),1,`D`'roman(eval($1-500))',
`ifelse(eval($1>=100),1,`C`'roman(eval($1-100))',
`ifelse(eval($1>=90),1,`XC`'roman(eval($1-90))',
`ifelse(eval($1>=50),1,`L`'roman(eval($1-50))',
`ifelse(eval($1>=40),1,`XL`'roman(eval($1-40))',
`ifelse(eval($1>=10),1,`X`'roman(eval($1-10))',
`ifelse(eval($1>=9),1,`IX`'roman(eval($1-9))',
`ifelse(eval($1>=5),1,`V`'roman(eval($1-5))',
`ifelse(eval($1>=4),1,`IV`'roman(eval($1-4))',
`ifelse(eval($1>=1),1,`I`'roman(eval($1-1))'
)')')')')')')')')')')')')dnl
dnl
roman(3675)
Output:
MMMDCLXXV

Maple

> for n in [ 1666, 1990, 2008 ] do printf( "%d\t%s\n", n, convert( n, 'roman' ) ) end:            
1666    MDCLXVI
1990    MCMXC
2008    MMVIII

Mathematica/Wolfram Language

RomanNumeral is a built-in function in the Wolfram language. Examples:

RomanNumeral[4]
RomanNumeral[99]
RomanNumeral[1337]
RomanNumeral[1666]
RomanNumeral[6889]

gives back:

IV
XCIX
MCCCXXXVII
MDCLXVI
MMMMMMDCCCLXXXIX

Mercury

The non-ceremonial work in this program starts at the function to_roman/1. Unusually for Mercury the function is semi-deterministic. This is because some of the helper functions it calls are also semi-deterministic and the determinism subsystem propagates the status upward. (There are ways to stop it from doing this, but it would distract from the actual Roman numeral conversion process so the semi-determinism has been left in.)

to_roman/1 is just a string of chained function calls. The number is passed in as a string (and the main/2 predicate ensures that it is *only* digits!) is converted into a list of characters. This list is then reversed and the Roman numeral version is built from it. This resulting character list is then converted back into a string and returned.

build_roman/1 takes the lead character off the list (reversed numerals) and then recursively calls itself. It uses the promote/2 predicate to multiply the ensuing Roman numerals (if any) by an order of magnitude and converts the single remaining digit to the appropriate list of Roman numerals. To clarify, if it's passed the number "123" (encoded by this point as ['3', '2', '1']) the following transpires:

  • The '3' is removed and build_roman/1 is now called with ['2', '1'].
    • The '2' is removed and the function is recursively called with ['1'].
      • The '1' is removed and the function is recursively called with [] (the empty list)..
        • The function returns [].
      • The [] has its (non-existent) digits promoted and then gets ['I'] appended (1 converts to ['I'] via digit_to_roman/1).
    • The ['I'] has its (single) digit promoted and is converted to ['X'] and then gets ['I','I'] appended from the 2's conversion. The resulting list is now ['X','I','I'] (or 12).
  • The ['X','I','I'] has all of its digits promoted, yielding ['C','X','X'] before getting ['I','I','I'] appended. The resulting list is now ['C','X','X','I','I','I'] which is converted into the string "CXXIII" back up in to_roman/1.

It is possible for this to be implemented differently even keeping the same algorithm. For example the map module from the standard library could be used for looking up conversions and promotions instead of having digit_to_roman/1 and promote. This would require, however, either passing around the conversion tables constantly (bulking up the parameter lists of all functions and predicates) or creating said conversion tables each time at point of use (slowing down the implementation greatly).

Now the semi-determinism of the functions involved is a little bit of a problem. In the main/2 predicate you can see one means of dealing with it. main/2 *must* be deterministic (or cc_multi, but this is equivalent for this discussion). There can be *no* failure in a called function or predicate … unless that failure is explicitly handled somehow. In this implementation the failure is handled in the foldl/4's provided higher-order predicate lambda. The call to to_roman/1 is called within a conditional and both the success (true) and failure (false) branches are handled. This makes the passed-in predicate lambda deterministic, even though the implementation functions and predicates are semi-deterministic.

But why are they semi-deterministic? Well, this has to do with the type system. It doesn't permit sub-typing, so when the type of a predicate is, say pred(char, char) (as is the case for promote/2), the underlying implementation *must* handle *all* values that a type char could possibly hold. It is trivial to see that our code does not. This means that, in theory, it is possible that promote/2 (or digit_to_roman/1) could be passed a value which cannot be processed, thus triggering a false result, and thus being semi-deterministic.

roman.m

:- module roman.

:- interface.

:- import_module io.

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

:- implementation.

:- import_module char, int, list, string.

main(!IO) :-
    command_line_arguments(Args, !IO),
    filter(is_all_digits, Args, CleanArgs),
    foldl((pred(Arg::in, !.IO::di, !:IO::uo) is det :-
               ( Roman = to_roman(Arg) ->
                     format("%s => %s", [s(Arg), s(Roman)], !IO), nl(!IO)
               ;     format("%s cannot be converted.", [s(Arg)], !IO), nl(!IO) )
          ), CleanArgs, !IO).

:- func to_roman(string::in) = (string::out) is semidet.
to_roman(Number) = from_char_list(build_roman(reverse(to_char_list(Number)))).

:- func build_roman(list(char)) = list(char).
:- mode build_roman(in)         = out is semidet.
build_roman([]) = [].
build_roman([D|R]) = Roman :-
    map(promote, build_roman(R), Interim),
    Roman = Interim ++ digit_to_roman(D).

:- func digit_to_roman(char) = list(char).
:- mode digit_to_roman(in)   = out is semidet.
digit_to_roman('0') = [].
digit_to_roman('1') = ['I'].
digit_to_roman('2') = ['I','I'].
digit_to_roman('3') = ['I','I','I'].
digit_to_roman('4') = ['I','V'].
digit_to_roman('5') = ['V'].
digit_to_roman('6') = ['V','I'].
digit_to_roman('7') = ['V','I','I'].
digit_to_roman('8') = ['V','I','I','I'].
digit_to_roman('9') = ['I','X'].

:- pred promote(char::in, char::out) is semidet.
promote('I', 'X').
promote('V', 'L').
promote('X', 'C').
promote('L', 'D').
promote('C', 'M').

:- end_module roman.
Output:
 $ '''mmc roman && ./roman 1 8 27 64 125 216 343 512 729 1000 1331 1728 2197 2744 3375'''
 ''1 => I''
 ''8 => VIII''
 ''27 => XXVII''
 ''64 => LXIV''
 ''125 => CXXV''
 ''216 => CCXVI''
 ''343 => CCCXLIII''
 ''512 => DXII''
 ''729 => DCCXXIX''
 ''1000 => M''
 ''1331 => MCCCXXXI''
 ''1728 => MDCCXXVIII''
 ''2197 => MMCXCVII''
 ''2744 => MMDCCXLIV''
 ''3375 => MMMCCCLXXV''

roman2.m

Another implementation using an algorithm inspired by the Erlang implementation could look like this:

:- module roman2.

:- interface.

:- import_module io.

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

:- implementation.

:- import_module char, int, list, string.

main(!IO) :-
    command_line_arguments(Args, !IO),
    filter_map(to_int, Args, CleanArgs),
    foldl((pred(Arg::in, !.IO::di, !:IO::uo) is det :-
               ( Roman = to_roman(Arg) ->
                     format("%i => %s", 
                            [i(Arg), s(from_char_list(Roman))], !IO), 
                     nl(!IO)
               ;     format("%i cannot be converted.", [i(Arg)], !IO), nl(!IO) )
          ), CleanArgs, !IO).

:- func to_roman(int) = list(char).
:- mode to_roman(in)  = out is semidet.
to_roman(N) = ( N >= 1000 -> 
                    ['M'] ++ to_roman(N - 1000)
              ;( N >= 100 -> 
                     digit(N / 100, 'C', 'D', 'M') ++ to_roman(N rem 100)
               ;( N >= 10 ->
                      digit(N / 10, 'X', 'L', 'C') ++ to_roman(N rem 10)
                ;( N >= 1 ->
                       digit(N, 'I', 'V', 'X')
                 ; [] ) ) ) ).

:- func digit(int, char, char, char) = list(char).
:- mode digit(in,  in,   in,   in)   = out is semidet.
digit(1, X, _, _) = [X].
digit(2, X, _, _) = [X, X].
digit(3, X, _, _) = [X, X, X].
digit(4, X, Y, _) = [X, Y].
digit(5, _, Y, _) = [Y].
digit(6, X, Y, _) = [Y, X].
digit(7, X, Y, _) = [Y, X, X].
digit(8, X, Y, _) = [Y, X, X, X].
digit(9, X, _, Z) = [X, Z].

:- end_module roman2.

This implementation calculates the value of the thousands, then the hundreds, then the tens, then the ones. In each case it uses the digit/4 function and some tricks with unification to build the appropriate list of characters for the digit and multiplier being targeted.

Its output is identical to that of the previous version.

Miranda

main :: [sys_message]
main = [ Stdout (show n ++ ": " ++ toroman n ++ "\n")
       | n <- [1990, 2008, 1666, 2023]]

toroman :: num->[char]
toroman 0 = ""
toroman n = d ++ toroman (n - v)
            where digits = [("M",1000),("CM",900),("D",500),("CD",400),
                            ("C",100),("XC",90),("L",50),("XL",40),
                            ("X",10),("IX",9),("V",5),("IV",4),
                            ("I",1)]
                  (d, v) = hd [(d,v) | (d,v) <- digits; v <= n]
Output:
1990: MCMXC
2008: MMVIII
1666: MDCLXVI
2023: MMXXIII

Modula-2

Translation of: DWScript
Works with: ADW Modula-2 version any (Compile with the linker option Console Application).
MODULE RomanNumeralsEncode;

FROM Strings IMPORT
  Append;
FROM STextIO IMPORT
  WriteString, WriteLn;

CONST
  MaxChars = 15;
  (* 3888 or MMMDCCCLXXXVIII (15 chars) is the longest string properly encoded
     with these symbols. *)

TYPE
  TRomanNumeral = ARRAY [0 .. MaxChars - 1] OF CHAR;

PROCEDURE ToRoman(AValue: CARDINAL; VAR OUT Destination: ARRAY OF CHAR);
TYPE
  TRomanSymbols = ARRAY [0 .. 1] OF CHAR;
  TWeights = ARRAY [0 .. 12] OF CARDINAL;
  TSymbols = ARRAY [0 .. 12] OF TRomanSymbols;
CONST
  Weights = TWeights {1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1};
  Symbols = TSymbols {"M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX",
    "V", "IV", "I"};
VAR
  I: CARDINAL;
BEGIN
  Destination := "";
  I := 0;
  WHILE (I <= HIGH(Weights)) AND (AValue > 0) DO
    WHILE AValue >= Weights[I] DO
      Append(Symbols[I], Destination);
      AValue := AValue - Weights[I]
    END;
    INC(I);
  END;
END ToRoman;

VAR
  Numeral: TRomanNumeral;

BEGIN
  ToRoman(1990, Numeral); WriteString(Numeral); WriteLn; (* MCMXC *)
  ToRoman(2018, Numeral); WriteString(Numeral); WriteLn; (* MMXVIII *)
  ToRoman(3888, Numeral); WriteString(Numeral); WriteLn; (* MMMDCCCLXXXVIII *)
END RomanNumeralsEncode.
Output:
MCMXC
MMXVIII
MMMDCCCLXXXVIII

MUMPS

TOROMAN(INPUT)
 ;Converts INPUT into a Roman numeral. INPUT must be an integer between 1 and 3999
 ;OUTPUT is the string to return
 ;I is a loop variable
 ;CURRVAL is the current value in the loop
 QUIT:($FIND(INPUT,".")>1)!(INPUT<=0)!(INPUT>3999) "Invalid input"
 NEW OUTPUT,I,CURRVAL
 SET OUTPUT="",CURRVAL=INPUT
 SET:$DATA(ROMANNUM)=0 ROMANNUM="I^IV^V^IX^X^XL^L^XC^C^CD^D^CM^M"
 SET:$DATA(ROMANVAL)=0 ROMANVAL="1^4^5^9^10^40^50^90^100^400^500^900^1000"
 FOR I=$LENGTH(ROMANVAL,"^"):-1:1 DO
 .FOR  Q:CURRVAL<$PIECE(ROMANVAL,"^",I)  SET OUTPUT=OUTPUT_$PIECE(ROMANNUM,"^",I),CURRVAL=CURRVAL-$PIECE(ROMANVAL,"^",I)
 KILL I,CURRVAL
 QUIT OUTPUT
Output:
USER>W $$ROMAN^ROSETTA(1666)
MDCLXVI
USER>W $$TOROMAN^ROSETTA(2010)
MMX
USER>W $$TOROMAN^ROSETTA(949)
CMXLIX
USER>W $$TOROMAN^ROSETTA(949.24)
Invalid input
USER>W $$TOROMAN^ROSETTA(-949)
Invalid input

Another variant

TOROMAN(n)
    ;return empty string if input parameter 'n' is not in 1-3999
    Quit:(n'?1.4N)!(n'<4000)!'n ""
    New r  Set r=""
    New p  Set p=$Length(n)
    New j,x
    For j=1:1:p  Do
    . Set x=$Piece("~I~II~III~IV~V~VI~VII~VIII~IX","~",$Extract(n,j)+1)
    . Set x=$Translate(x,"IVX",$Piece("IVX~XLC~CDM~M","~",p-j+1))
    . Set r=r_x
    Quit r

Nim

Translation of: Python
import strutils

const nums = [(1000, "M"), (900, "CM"), (500, "D"), (400, "CD"), (100, "C"), (90, "XC"),
              (50, "L"), (40, "XL"), (10, "X"), (9, "IX"), (5, "V"), (4, "IV"), (1, "I")]

proc toRoman(n: Positive): string =
  var n = n.int
  for (a, r) in nums:
    result.add(repeat(r, n div a))
    n = n mod a

for i in [1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
          11, 12, 13, 14, 15, 16, 17, 18, 19, 20,
          25, 30, 40, 50, 60, 69, 70, 80, 90, 99,
          100, 200, 300, 400, 500, 600, 666, 700, 800, 900,
          1000, 1009, 1444, 1666, 1945, 1997, 1999,
          2000, 2008, 2010, 2011, 2500, 3000, 3999]:
  echo ($i).align(4), ": ", i.toRoman
Output:
   1: I
   2: II
   3: III
   4: IV
   5: V
   6: VI
   7: VII
   8: VIII
   9: IX
  10: X
  11: XI
  12: XII
  13: XIII
  14: XIV
  15: XV
  16: XVI
  17: XVII
  18: XVIII
  19: XIX
  20: XX
  25: XXV
  30: XXX
  40: XL
  50: L
  60: LX
  69: LXIX
  70: LXX
  80: LXXX
  90: XC
  99: XCIX
 100: C
 200: CC
 300: CCC
 400: CD
 500: D
 600: DC
 666: DCLXVI
 700: DCC
 800: DCCC
 900: CM
1000: M
1009: MIX
1444: MCDXLIV
1666: MDCLXVI
1945: MCMXLV
1997: MCMXCVII
1999: MCMXCIX
2000: MM
2008: MMVIII
2010: MMX
2011: MMXI
2500: MMD
3000: MMM
3999: MMMCMXCIX

Objeck

Translation of: C sharp
bundle Default {
  class Roman {
    nums: static : Int[];
    rum : static : String[];
  
    function : Init() ~ Nil {
      nums := [1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1];
      rum := ["M", "CM", "D", "CD", "C", "XC", "L", "XL", "X", "IX", "V", "IV", "I"];
    }

    function : native : ToRoman(number : Int) ~ String {
      result := "";

      for(i :=0; i < nums->Size(); i += 1;) {
        while(number >= nums[i]) {
          result->Append(rum[i]);
          number -= nums[i];
        };
      };

      return result;
    }

    function : Main(args : String[]) ~ Nil {
      Init();

      ToRoman(1999)->PrintLine();
      ToRoman(25)->PrintLine();
      ToRoman(944)->PrintLine();
    }
  }
}

OCaml

With an explicit decimal digit representation list:

let digit x y z = function
    1 -> [x]
  | 2 -> [x;x]
  | 3 -> [x;x;x]
  | 4 -> [x;y]
  | 5 -> [y]
  | 6 -> [y;x]
  | 7 -> [y;x;x]
  | 8 -> [y;x;x;x]
  | 9 -> [x;z]

let rec to_roman x =
  if x = 0 then []
  else if x < 0 then
    invalid_arg "Negative roman numeral"
  else if x >= 1000 then
    'M' :: to_roman (x - 1000)
  else if x >= 100 then
    digit 'C' 'D' 'M' (x / 100) @ to_roman (x mod 100)
  else if x >= 10 then
    digit 'X' 'L' 'C' (x / 10) @ to_roman (x mod 10)
  else
    digit 'I' 'V' 'X' x
Output:
# to_roman 1999;;
- : char list = ['M'; 'C'; 'M'; 'X'; 'C'; 'I'; 'X']
# to_roman 25;;
- : char list = ['X'; 'X'; 'V']
# to_roman 944;;
- : char list = ['C'; 'M'; 'X'; 'L'; 'I'; 'V']

Oforth

[ [1000,"M"], [900,"CM"], [500,"D"], [400,"CD"], [100,"C"], [90,"XC"], [50,"L"], [40,"XL"], [10,"X"], [9,"IX"], [5,"V"], [4,"IV"], [1,"I"] ] const: Romans

: roman(n)
| r |
   StringBuffer new
   Romans forEach: r [ while(r first n <=) [ r second << n r first - ->n ] ] ;

OpenEdge/Progress

FUNCTION encodeRoman RETURNS CHAR (
   i_i AS INT
):

   DEF VAR cresult   AS CHAR.
   DEF VAR croman    AS CHAR EXTENT 7 INIT [  "M", "D", "C", "L", "X", "V", "I" ].
   DEF VAR idecimal  AS INT  EXTENT 7 INIT [ 1000, 500, 100,  50,  10,   5,   1 ].
   DEF VAR ipos      AS INT  INIT 1.
   
   DO WHILE i_i > 0:

      IF i_i - idecimal[ ipos ] >= 0 THEN
         ASSIGN
            cresult  =  cresult + croman[ ipos ]
            i_i      =  i_i - idecimal[ ipos ]
            .
      ELSE IF ipos < EXTENT( croman ) - 1 AND i_i - ( idecimal[ ipos ] - idecimal[ ipos + 2 ] ) >= 0 THEN
         ASSIGN
            cresult  =  cresult + croman[ ipos + 2 ] + croman[ ipos ]
            i_i      =  i_i - ( idecimal[ ipos ] - idecimal[ ipos + 2 ] )
            ipos     =  ipos + 1
            .
      ELSE
         ipos = ipos + 1.
   END.

   RETURN cresult.

END FUNCTION. /* encodeRoman */

MESSAGE
   1990 encodeRoman( 1990 ) SKIP
   2008 encodeRoman( 2008 ) SKIP
   2000 encodeRoman( 2000 ) SKIP
   1666 encodeRoman( 1666 ) SKIP
VIEW-AS ALERT-BOX.
Output:
---------------------------
Message (Press HELP to view stack trace)
---------------------------
1990 MCMXC 
2008 MMVIII 
2000 MM 
1666 MDCLXVI 
---------------------------
OK   Help   
---------------------------

Oz

Translation of: Haskell
declare
  fun {Digit X Y Z K}
     unit([X] [X X] [X X X] [X Y] [Y] [Y X] [Y X X] [Y X X X] [X Z])
     .K
  end

  fun {ToRoman X}
     if     X == 0    then ""
     elseif X < 0     then raise toRoman(negativeInput X) end
     elseif X >= 1000 then "M"#{ToRoman X-1000}
     elseif X >= 100  then {Digit &C &D &M  X div 100}#{ToRoman X mod 100}
     elseif X >= 10   then {Digit &X &L &C  X div 10}#{ToRoman X mod 10}
     else                  {Digit &I &V &X  X}
     end
  end
in
  {ForAll {Map [1999 25 944] ToRoman} System.showInfo}

PARI/GP

Old-style Roman numerals

oldRoman(n)={
  while(n>999999,
    n-=1000000;
    print1("((((I))))")
  );
  if(n>499999,
    n-=500000;
    print1("I))))")
  );
  while(n>99999,
    n-=100000;
    print1("(((I)))")
  );
  if(n>49999,
    n-=50000;
    print1("I)))")
  );
  while(n>9999,
    n-=10000;
    print1("((I))")
  );
  if(n>4999,
    n-=5000;
    print1("I))")
  );
  while(n>999,
    n-=1000;
    print1("(I)")
  );
  if(n>499,
    n-=500;
    print1("I)")
  );
  while(n>99,
    n-=100;
    print1("C")
  );
  if(n>49,
    n-=50;
    print1("L");
  );
  while(n>9,
    n-=10;
    print1("X")
  );
  if(n>4,
    n-=5;
    print1("V");
  );
  while(n,
    n--;
    print1("I")
  );
  print()
};

This simple version of medieval Roman numerals does not handle large numbers.

medievalRoman(n)={
  while(n>999,
    n-=1000;
    print1("M")
  );
  if(n>899,
    n-=900;
    print1("CM")
  );
  if(n>499,
    n-=500;
    print1("D")
  );
  if(n>399,
    n-=400;
    print1("CD")
  );
  while(n>99,
    n-=100;
    print1("C")
  );
  if(n>89,
    n-=90;
    print1("XC")
  );
  if(n>49,
    n-=50;
    print1("L")
  );
  if(n>39,
    n-=40;
    print1("XL")
  );
  while(n>9,
    n-=10;
    print1("X")
  );
  if(n>8,
    n-=9;
    print1("IX")
  );
  if(n>4,
    n-=5;
    print1("V")
  );
  if(n>3,
    n-=4;
    print1("IV")
  );
  while(n,
    n--;
    print1("I")
  );
  print()
};

Pascal

See Delphi

Peloton

Roman numbers are built in to Peloton as a particular form of national number. However, for the sake of the task the _RO opcode has been defined.

<@ DEFUDOLITLIT>_RO|__Transformer|<@ DEFKEYPAR>__NationalNumericID|2</@><@ LETRESCS%NNMPAR>...|1</@></@>

<@ ENU$$DLSTLITLIT>1990,2008,1,2,64,124,1666,10001|,|
<@ SAYELTLST>...</@> is <@ SAY_ROELTLSTLIT>...|RomanLowerUnicode</@> <@ SAY_ROELTLSTLIT>...|RomanUpperUnicode</@> <@ SAY_ROELTLSTLIT>...|RomanASCII</@>
</@>

Same code in padded-out, variable-length English dialect

<# DEFINE USERDEFINEDOPCODE LITERAL LITERAL>_RO|__Transformer|<# DEFINE KEYWORD PARAMETER>__NationalNumericID|2</#><# LET RESULT CAST NATIONALNUMBER PARAMETER>...|1</#></#>

<# ENUMERATION LAMBDASPECIFIEDDELMITER LIST LITERAL LITERAL>1990,2008,1,2,64,124,1666,10001|,|
<# SAY ELEMENT LIST>...</#> is <# SAY _RO ELEMENT LIST LITERAL>...|RomanLowerUnicode</#> <# SAY _RO ELEMENT LIST LITERAL>...|RomanUpperUnicode</#> <# SAY _RO ELEMENT LIST LITERAL>...|RomanASCII</#>
</#>
Output:
Notice here the three different ways of representing the results.

For reasons for notational differences, see wp:Roman_numerals#Alternate_forms

1990 is ⅿⅽⅿⅹⅽ ⅯⅭⅯⅩⅭ MCMXC
2008 is ⅿⅿⅷ ⅯⅯⅧ MMVIII
1 is ⅰ Ⅰ I
2 is ⅱ Ⅱ II
64 is ⅼⅹⅳ ⅬⅩⅣ LXIV
124 is ⅽⅹⅹⅳ ⅭⅩⅩⅣ CXXIV
1666 is ⅿⅾⅽⅼⅹⅵ ⅯⅮⅭⅬⅩⅥ MDCLXVI
10001 is ⅿⅿⅿⅿⅿⅿⅿⅿⅿⅿⅰ ↂⅠ MMMMMMMMMMI

Perl

Simple program

Simple, fast, produces same output as the Math::Roman module and the Raku example, less crazy than writing a Latin program, and doesn't require experimental modules like the Raku translation.

my @symbols = ( [1000, 'M'], [900, 'CM'], [500, 'D'], [400, 'CD'], [100, 'C'], [90, 'XC'], [50, 'L'], [40, 'XL'], [10, 'X'], [9, 'IX'], [5, 'V'], [4, 'IV'], [1, 'I']  );

sub roman {
  my($n, $r) = (shift, '');
  ($r, $n) = ('-', -$n) if $n < 0;  # Optional handling of negative input
  foreach my $s (@symbols) {
    my($arabic, $roman) = @$s;
    ($r, $n) = ($r .= $roman x int($n/$arabic),  $n % $arabic)
       if $n >= $arabic;
  }
  $r;
}

say roman($_) for 1..2012;

Using a module

use Math::Roman qw/roman/;
say roman($_) for 1..2012'

Ported version of Raku

use List::MoreUtils qw( natatime );

my %symbols = (
    1 => "I", 5 => "V", 10 => "X", 50 => "L", 100 => "C",
    500 => "D", 1_000 => "M"
);

my @subtractors = (
        1_000, 100,  500, 100,  100, 10,  50, 10,  10, 1,  5, 1,  1, 0
);

sub roman {
    return '' if 0 == (my $n = shift);
    my $iter = natatime 2, @subtractors;
    while( my ($cut, $minus) = $iter->() ) {
        $n >= $cut
            and return $symbols{$cut} . roman($n - $cut);
        $n >= $cut - $minus
            and return $symbols{$minus} . roman($n + $minus);
    }
};

print roman($_) . "\n" for 1..2012;

Phix

with javascript_semantics
function toRoman(integer v)
    sequence roman  = {"M", "CM", "D","CD", "C","XC","L","XL","X","IX","V","IV","I"},
             decml  = {1000, 900, 500, 400, 100, 90, 50,  40,  10,  9,  5,   4,  1 }
    string res = ""
    integer val = v
    for i=1 to length(roman) do
        while val>=decml[i] do
            res &= roman[i]
            val -= decml[i]
        end while
    end for
    return {v,res}  -- (for output)
end function

?apply({1990,2008,1666},toRoman)
Output:
{{1990,"MCMXC"},{2008,"MMVIII"},{1666,"MDCLXVI"}}

cheating slightly

with javascript_semantics
requires("1.0.5")
function toRoman(integer n)
    return {n,sprintf("%R",n)}
end function

same output (builtins\VM\pprntfN.e/toRoman() is somewhat more obfuscated and faster than the above)

Phixmonti

include ..\Utilitys.pmt

def romanEnc   /# n -- s #/
    var number
    "" var res
    ( ( 1000 "M" ) ( 900 "CM" ) ( 500 "D" ) ( 400 "CD" ) ( 100 "C" ) ( 90 "XC" )
      ( 50 "L" ) ( 40 "XL" ) ( 10 "X" ) ( 9 "IX" ) ( 5 "V" ) ( 4 "IV" ) ( 1 "I" ) )
    
    len for
        get 1 get
        number over / int
        number rot mod var number
        swap 2 get rot dup if
            for drop res over chain var res endfor 
        else
            drop
        endif
        drop drop
    endfor
    drop
    res
enddef

1968 romanEnc print
Translation of: Lua
def romanEnc   /# n -- s #/
    var k   
    ( ( 1000 "M" ) ( 900 "CM" ) ( 500 "D" ) ( 400 "CD" ) ( 100 "C" ) ( 90 "XC" )
      ( 50 "L" ) ( 40 "XL" ) ( 10 "X" ) ( 9 "IX" ) ( 5 "V" ) ( 4 "IV" ) ( 1 "I" ) )

    len for
        get 2 get var let 1 get var val drop
        k val >=     
        while
            k val - var k 
            let print
            k val >=
        endwhile
    endfor
    drop nl
enddef

1968 romanEnc

Without vars

def romanEnc   /# n -- s #/
    >ps
    ( ( 1000 "M" ) ( 900 "CM" ) ( 500 "D" ) ( 400 "CD" ) ( 100 "C" ) ( 90 "XC" )
      ( 50 "L" ) ( 40 "XL" ) ( 10 "X" ) ( 9 "IX" ) ( 5 "V" ) ( 4 "IV" ) ( 1 "I" ) )

    len for
        get 2 get swap 1 get nip
        tps over >=     
        while
            ps> over - >ps 
            over print
            tps over >=
        endwhile
        drop drop
    endfor
    ps> drop drop nl
enddef

1968 romanEnc

PHP

Works with: PHP version 4+ tested in 5.2.12
/**
 * int2roman
 * Convert any positive value of a 32-bit signed integer to its modern roman 
 * numeral representation. Numerals within parentheses are multiplied by 
 * 1000. ie. M == 1 000, (M) == 1 000 000, ((M)) == 1 000 000 000
 * 
 * @param number - an integer between 1 and 2147483647
 * @return roman numeral representation of number
 */
function int2roman($number)
{
	if (!is_int($number) || $number < 1) return false; // ignore negative numbers and zero
	
	$integers = array(900, 500,  400, 100,   90,  50,   40,  10,    9,   5,    4,   1);
	$numerals = array('CM', 'D', 'CD', 'C', 'XC', 'L', 'XL', 'X', 'IX', 'V', 'IV', 'I');
	$major = intval($number / 1000) * 1000;
	$minor = $number - $major;
	$numeral = $leastSig = '';
	
	for ($i = 0; $i < sizeof($integers); $i++) {
		while ($minor >= $integers[$i]) {
			$leastSig .= $numerals[$i];
			$minor  -= $integers[$i];
		}
	}
	
	if ($number >= 1000 && $number < 40000) {
		if ($major >= 10000) {
			$numeral .= '(';
			while ($major >= 10000) {
				$numeral .= 'X';
				$major -= 10000;
			}
			$numeral .= ')';
		}
		if ($major == 9000) {
			$numeral .= 'M(X)';
			return $numeral . $leastSig;
		}
		if ($major == 4000) {
			$numeral .= 'M(V)';
			return $numeral . $leastSig;
		}
		if ($major >= 5000) {
			$numeral .= '(V)';
			$major -= 5000;
		}
		while ($major >= 1000) {
			$numeral .= 'M';
			$major -= 1000;
		}
	}
	
	if ($number >= 40000) {
		$major = $major/1000;
		$numeral .= '(' . int2roman($major) . ')';
	}
	
	return $numeral . $leastSig;
}

Picat

go => 
   List = [455,999,1990,1999,2000,2001,2008,2009,2010,2011,2012,1666,3456,3888,4000],
   foreach(Val in List)
      printf("%4d: %w\n", Val, roman_encode(Val))
   end,
   nl.

roman_encode(Val) = Res => 
  if Val <= 0 then
     Res := -1
  else
     Arabic = [1000, 900, 500, 400,  100, 90,  50, 40,  10,  9,  5,  4,   1],
     Roman  = ["M", "CM", "D", "CD", "C", "XC","L","XL","X","IX","V","IV","I"],
     Res = "",
     foreach(I in 1..Arabic.length)
        while(Val >= Arabic[I]) 
           Res := Res ++ Roman[I],
           Val := Val - Arabic[I]
        end
     end
  end.
Output:
 455: CDLV
 999: CMXCIX
1990: MCMXC
1999: MCMXCIX
2000: MM
2001: MMI
2008: MMVIII
2009: MMIX
2010: MMX
2011: MMXI
2012: MMXII
1666: MDCLXVI
3456: MMMCDLVI
3888: MMMDCCCLXXXVIII
4000: MMMM

Longest numeral

Which number encodes to the longest Roman numerals in the interval 1..4000:

go2 =>
  All = [Len=I=roman_encode(I) : I in 1..4000,E=roman_encode(I), Len=E.len].sort_down,
  println(All[1..2]),
  nl.
Output:
[15 = 3888 = MMMDCCCLXXXVIII,14 = 3887 = MMMDCCCLXXXVII]

PicoLisp

(de roman (N)
   (pack
      (make
         (mapc
            '((C D)
               (while (>= N D)
                  (dec 'N D)
                  (link C) ) )
            '(M CM D CD C XC L XL X IX V IV I)
            (1000 900 500 400 100 90 50 40 10 9 5 4 1) ) ) ) )
Output:
: (roman 1009)
-> "MIX"

: (roman 1666)
-> "MDCLXVI"

Pike

import String;
int main(){
   write(int2roman(2009) + "\n");
   write(int2roman(1666) + "\n");
   write(int2roman(1337) + "\n");
}

PL/I

/* From Wiki Fortran */
roman: procedure (n) returns(character (32) varying);
   declare n fixed binary nonassignable;
   declare (d, m) fixed binary;
   declare (r, m_div) character (32) varying;
   declare d_dec(13) fixed binary static initial
      (1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1);
   declare d_rom(13) character (2) varying static initial
      ('M', 'CM', 'D', 'CD', 'C', 'XC', 'L',
       'XL', 'X', 'IX', 'V', 'IV', 'I');
   r = '';
   m = n;
   do d = 1 to 13;
      m_div = m / d_dec (d);
      r = r || copy (d_rom (d), m_div);
      m = m - d_dec (d) * m_div;
   end;
   return (r);
end roman;

Results:

   11                   XI 
   1990                 MCMXC 
   2008                 MMVIII 
   1666                 MDCLXVI 
   1999                 MCMXCIX 

PL/SQL

/*****************************************************************
 * $Author: Atanas Kebedjiev $
 *****************************************************************
 * Encoding an Arabic numeral to a Roman in the range 1..3999 is much simpler as Oracle provides the conversion formats.
 * Please see also the SQL solution for the same task.
 */

CREATE OR REPLACE
FUNCTION rencode(an IN NUMBER) 
  RETURN VARCHAR2 
IS
BEGIN
  RETURN to_char(an, 'RN');
END rencode;

BEGIN

    DBMS_OUTPUT.PUT_LINE ('2012 = ' || rencode('2012'));     -- MMXII
    DBMS_OUTPUT.PUT_LINE ('1951 = ' || rencode('1951'));     -- MCMLI
    DBMS_OUTPUT.PUT_LINE ('1987 = ' || rencode('1987'));     -- MCMLXXXVII
    DBMS_OUTPUT.PUT_LINE ('1666 = ' || rencode('1666'));     -- MDCLXVI
    DBMS_OUTPUT.PUT_LINE ('1999 = ' || rencode('1999'));     -- MCMXCIX

END;

Plain TeX

TeX has its own way to convert a number into roman numeral, but it produces lowercase letters; the following macro (and usage example), produce uppercase roman numeral.

\def\upperroman#1{\uppercase\expandafter{\romannumeral#1}}
Anno Domini \upperroman{\year}
\bye

PowerShell

Filter ToRoman {
	$output = ''
	
	if ($_ -ge 4000) {
		throw 'Number too high'
	}
	
	$current = 1000
	$subtractor = 'M'
	$whole = $False
	$decimal = $_
	'C','D','X','L','I','V',' ' `
	| %{
		$divisor = $current
		if ($whole = !$whole) {
			$current /= 10
			$subtractor = $_ + $subtractor[0]
			$_ = $subtractor[1]
		}
		else {
			$divisor *= 5 
			$subtractor = $subtractor[0] + $_
		}
		
		$multiple = [Math]::floor($decimal / $divisor)
		if ($multiple) {
			$output += [string]$_ * $multiple
			$decimal %= $divisor
		}
		if ($decimal -ge ($divisor -= $current)) {
			$output += $subtractor
			$decimal -= $divisor
		}
	}
	
	$output
}
19,4,0,2479,3001 | ToRoman
Output:
XIX
IV

MMCDLXXIX
MMMI             

Prolog

Works with: SWI-Prolog
Library: clpfd

Library clpfd assures that the program works in both managements : Roman towards Arabic and Arabic towards Roman.

:- use_module(library(clpfd)).

roman :-
	LA =  [    _       , 2010,    _, 1449,         _],
	LR =  ['MDCCLXXXIX',  _  , 'CX',    _, 'MDCLXVI'],
	maplist(roman,   LA, LR),
	maplist(my_print,LA, LR).


roman(A, R) :-
	A #> 0,
	roman(A, [u, t, h, th], LR, []),
	label([A]),
	parse_Roman(CR, LR, []),
	atom_chars(R, CR).

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% using DCG
 
roman(0, []) --> [].

roman(N, [H | T]) -->
	{N1 #= N / 10,
	 N2 #= N mod 10},
	roman(N1, T),
	unity(N2, H).

unity(1, u) --> ['I'].
unity(1, t) --> ['X'].
unity(1, h) --> ['C'].
unity(1, th)--> ['M'].

unity(4, u) --> ['IV'].
unity(4, t) --> ['XL'].
unity(4, h) --> ['CD'].
unity(4, th)--> ['MMMM'].

unity(5, u) --> ['V'].
unity(5, t) --> ['L'].
unity(5, h) --> ['D'].
unity(5, th)--> ['MMMMM'].

unity(9, u) --> ['IX'].
unity(9, t) --> ['XC'].
unity(9, h) --> ['CM'].
unity(9, th)--> ['MMMMMMMMM'].

unity(0, _) --> [].


unity(V, U)-->
	{V #> 5,
	V1 #= V - 5},
	unity(5, U),
	unity(V1, U).

unity(V, U) -->
	{V #> 1, V #< 4,
	V1 #= V-1},
	unity(1, U),
	unity(V1, U).

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Extraction of roman "lexeme"
parse_Roman(['C','M'|T]) -->
	['CM'],
	parse_Roman(T).

parse_Roman(['C','D'|T]) -->
	['CD'],
	parse_Roman(T).

parse_Roman(['X','C'| T]) -->
	['XC'],
	parse_Roman(T).


parse_Roman(['X','L'| T]) -->
	['XL'],
	parse_Roman(T).


parse_Roman(['I','X'| T]) -->
	['IX'],
	parse_Roman(T).


parse_Roman(['I','V'| T]) -->
	['IV'],
	parse_Roman(T).

parse_Roman([H | T]) -->
	[H],
	parse_Roman(T).


parse_Roman([]) -->
	[].

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
my_print(A, R) :-
	format('~w in roman is ~w~n', [A, R]).
Output:
 ?- roman.
1789 in roman is MDCCLXXXIX
2010 in roman is MMX
110 in roman is CX
1449 in roman is MCDXLIX
1666 in roman is MDCLXVI
true .

Python

Pythonic

import roman
print(roman.toRoman(2022))

Minimalistic structuralism

def toRoman(n):
    res=''		#converts int to str(Roman numeral)
    reg=n		#using the numerals (M,D,C,L,X,V,I)
    if reg<4000:#no more than three repetitions
        while reg>=1000:	#thousands up to MMM
            res+='M'		#MAX is MMMCMXCIX
            reg-=1000		
        if reg>=900:		#nine hundreds in 900-999
            res+='CM'
            reg-=900
        if reg>=500:		#five hudreds in 500-899
            res+='D'
            reg-=500
        if reg>=400:		#four hundreds in 400-499
            res+='CD'
            reg-=400
        while reg>=100:		#hundreds in 100-399
            res+='C'
            reg-=100
        if reg>=90:			#nine tens in 90-99
            res+='XC'
            reg-=90
        if reg>=50:			#five Tens in 50-89
            res+='L'
            reg-=50
        if reg>=40:
            res+='XL'		#four Tens
            reg-=40
        while reg>=10:
            res+="X"		#tens
            reg-=10
        if reg>=9:
            res+='IX'		#nine Units
            reg-=9
        if reg>=5:
            res+='V'		#five Units
            reg-=5
        if reg>=4:
            res+='IV'		#four Units
            reg-=4
        while reg>0:		#three or less Units
            res+='I'
            reg-=1
    return res

Imperative

  1. Version for Python 2
roman =        "MDCLXVmdclxvi"; # UPPERCASE for thousands #
adjust_roman = "CCXXmmccxxii";
arabic =       (1000000, 500000, 100000, 50000, 10000, 5000, 1000, 500, 100, 50, 10, 5, 1);
adjust_arabic = (100000, 100000,  10000, 10000,  1000, 1000,  100, 100,  10, 10,  1, 1, 0);

def arabic_to_roman(dclxvi):
  org = dclxvi; # 666 #
  out = "";
  for scale,arabic_scale  in enumerate(arabic): 
    if org == 0: break
    multiples = org / arabic_scale;
    org -= arabic_scale * multiples;
    out += roman[scale] * multiples;
    if org >= -adjust_arabic[scale] + arabic_scale: 
      org -= -adjust_arabic[scale] + arabic_scale;
      out +=  adjust_roman[scale] +  roman[scale]
  return out
 
if __name__ == "__main__": 
  test = (1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,25,30,40,50,60,69,70,
     80,90,99,100,200,300,400,500,600,666,700,800,900,1000,1009,1444,1666,1945,1997,1999,
     2000,2008,2500,3000,4000,4999,5000,6666,10000,50000,100000,500000,1000000);
  for val in test: 
    print '%d - %s'%(val, arabic_to_roman(val))
An alternative which uses the divmod() function
romanDgts= 'ivxlcdmVXLCDM_'

def ToRoman(num):
   namoR = ''
   if num >=4000000:
      print 'Too Big -'
      return '-----'
   for rdix in range(0, len(romanDgts), 2):
      if num==0: break
      num,r = divmod(num,10)
      v,r = divmod(r, 5)
      if r==4:
         namoR += romanDgts[rdix+1+v] + romanDgts[rdix]
      else:
         namoR += r*romanDgts[rdix] + (romanDgts[rdix+1] if(v==1) else '')
   return namoR[-1::-1]

It is more Pythonic to use zip to iterate over two lists together:

anums = [1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1]
rnums = "M CM D CD C XC L XL X IX V IV I".split()

def to_roman(x):
    ret = []
    for a,r in zip(anums, rnums):
        n,x = divmod(x,a)
        ret.append(r*n)
    return ''.join(ret)
        
if __name__ == "__main__":
    test = (1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,25,30,40,
            50,60,69,70,80,90,99,100,200,300,400,500,600,666,700,800,900,
            1000,1009,1444,1666,1945,1997,1999,2000,2008,2010,2011,2500,
            3000,3999)
    
    for val in test:
        print '%d - %s'%(val, to_roman(val))
  1. Version for Python 3
def arabic_to_roman(dclxvi):
#===========================
  '''Convert an integer from the decimal notation to the Roman notation'''
  org = dclxvi; # 666 #
  out = "";
  for scale, arabic_scale  in enumerate(arabic):
    if org == 0: break
    multiples = org // arabic_scale;
    org -= arabic_scale * multiples;
    out += roman[scale] * multiples;
    if (org >= -adjust_arabic[scale] + arabic_scale):
      org -= -adjust_arabic[scale] + arabic_scale;
      out +=  adjust_roman[scale] +  roman[scale]
  return out

if __name__ == "__main__": 
  test = (1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,25,30,40,50,60,69,70,
     80,90,99,100,200,300,400,500,600,666,700,800,900,1000,1009,1444,1666,1945,1997,1999,
     2000,2008,2500,3000,4000,4999,5000,6666,10000,50000,100000,500000,1000000);
  
  for val in test: 
    print("%8d %s" %(val, arabic_to_roman(val)))

Declarative

Less readable, but a 'one liner':

rnl = [ { '4' : 'MMMM', '3' : 'MMM', '2' : 'MM', '1' : 'M', '0' : '' }, { '9' : 'CM', '8' : 'DCCC', '7' : 'DCC',
          '6' : 'DC', '5' : 'D', '4' : 'CD', '3' : 'CCC', '2' : 'CC', '1' : 'C', '0' : '' }, { '9' : 'XC',
          '8' : 'LXXX', '7' : 'LXX', '6' : 'LX', '5' : 'L', '4' : 'XL', '3' : 'XXX', '2' : 'XX', '1' : 'X',
          '0' : '' }, { '9' : 'IX', '8' : 'VIII', '7' : 'VII', '6' : 'VI', '5' : 'V', '4' : 'IV', '3' : 'III',
          '2' : 'II', '1' : 'I', '0' : '' }]
# Option 1
def number2romannumeral(n):
    return ''.join([rnl[x][y] for x, y in zip(range(4), str(n).zfill(4)) if n < 5000 and n > -1])
# Option 2
def number2romannumeral(n):
    return reduce(lambda x, y: x + y, map(lambda x, y: rnl[x][y], range(4), str(n).zfill(4))) if -1 < n < 5000 else None


Or, defining roman in terms of mapAccumL:

Works with: Python version 3
Translation of: Haskell
'''Encoding Roman Numerals'''

from functools import reduce
from itertools import chain


# romanFromInt ::  Int -> String
def romanFromInt(n):
    '''A string of Roman numerals encoding an integer.'''
    def go(a, ms):
        m, s = ms
        q, r = divmod(a, m)
        return (r, s * q)

    return concat(snd(mapAccumL(go)(n)(
        zip([
            1000, 900, 500, 400, 100, 90, 50,
            40, 10, 9, 5, 4, 1
        ], [
            'M', 'CM', 'D', 'CD', 'C', 'XC', 'L',
            'XL', 'X', 'IX', 'V', 'IV', 'I'
        ])
    )))


# ------------------------- TEST -------------------------
# main :: IO ()
def main():
    '''Sample of years'''
    for s in [
            romanFromInt(x) for x in [
                1666, 1990, 2008, 2016, 2018, 2020
            ]
    ]:
        print(s)


# ------------------ GENERIC FUNCTIONS -------------------

# concat :: [[a]] -> [a]
# concat :: [String] -> String
def concat(xxs):
    '''The concatenation of all the elements in a list.'''
    xs = list(chain.from_iterable(xxs))
    unit = '' if isinstance(xs, str) else []
    return unit if not xs else (
        ''.join(xs) if isinstance(xs[0], str) else xs
    )


# mapAccumL :: (acc -> x -> (acc, y)) -> acc -> [x] -> (acc, [y])
def mapAccumL(f):
    '''A tuple of an accumulation and a list derived by a
       combined map and fold,
       with accumulation from left to right.'''
    def go(a, x):
        tpl = f(a[0], x)
        return (tpl[0], a[1] + [tpl[1]])
    return lambda acc: lambda xs: (
        reduce(go, xs, (acc, []))
    )


# snd :: (a, b) -> b
def snd(tpl):
    '''Second component of a tuple.'''
    return tpl[1]


# MAIN ---
if __name__ == '__main__':
    main()
Output:
MDCLXVI
MCMXC
MMVIII
MMXVI
MMXVIII
MMXX

Quackery

Pasting epitomised.

  [ $ "" 
    swap 1000 /mod $ "M" rot of rot swap join swap
    dup   900 < not if [ 900 - dip [ $ "CM" join ] ]
    dup   500 < not if [ 500 - dip [ $  "D" join ] ]
    dup   400 < not if [ 400 - dip [ $ "CD" join ] ]
          100 /mod $ "C" rot of rot swap join swap
    dup    90 < not if [  90 - dip [ $ "XC" join ] ]
    dup    50 < not if [  50 - dip [ $  "L" join ] ]
    dup    40 < not if [  40 - dip [ $ "XL" join ] ]
           10 /mod $ "X" rot of rot swap join swap
    dup     9 < not if [   9 - dip [ $ "IX" join ] ]
    dup     5 < not if [   5 - dip [ $  "V" join ] ]
    dup     4 < not if [   4 - dip [ $ "IV" join ] ]
    $ "I" swap of join ] 
                              is ->roman ( n --> $ )

  1990 dup echo say " = " ->roman echo$ cr
  2008 dup echo say " = " ->roman echo$ cr
  1666 dup echo say " = " ->roman echo$ cr
Output:
1990 = MCMXC
2008 = MMVIII
1666 = MDCLXVI

R

R has a built-in function, as.roman, for conversion to Roman numerals. The implementation details are found in utils:::.numeric2roman (see previous link), and utils:::.roman2numeric, for conversion back to Arabic decimals.

as.roman(1666)   # MDCLXVI

Since the object as.roman creates is just an integer vector with a class, you can do arithmetic with Roman numerals:

as.roman(1666) + 334   # MM

Racket

Straight recursion:

#lang racket
(define (encode/roman number)
  (cond ((>= number 1000) (string-append "M" (encode/roman (- number 1000))))
        ((>= number 900) (string-append "CM" (encode/roman (- number 900))))
        ((>= number 500) (string-append "D" (encode/roman (- number 500))))
        ((>= number 400) (string-append "CD" (encode/roman (- number 400))))
        ((>= number 100) (string-append "C" (encode/roman (- number 100))))
        ((>= number 90) (string-append "XC" (encode/roman (- number 90))))
        ((>= number 50) (string-append "L" (encode/roman (- number 50))))
        ((>= number 40) (string-append "XL" (encode/roman (- number 40))))
        ((>= number 10) (string-append "X" (encode/roman (- number 10))))
        ((>= number 9) (string-append "IX" (encode/roman (- number 9))))
        ((>= number 5) (string-append "V" (encode/roman (- number 5))))
        ((>= number 4) (string-append "IV" (encode/roman (- number 4))))
        ((>= number 1) (string-append "I" (encode/roman (- number 1))))
        (else "")))

Using for/fold and quotient/remainder to remove repetition:

#lang racket
(define (number->list n)
  (for/fold ([result null])
    ([decimal '(1000 900 500 400 100 90 50 40 10 9  5 4  1)]
     [roman   '(M    CM  D   CD  C   XC L  XL X  IX V IV I)])
    #:break (= n 0)
    (let-values ([(q r) (quotient/remainder n decimal)])
      (set! n r)
      (append result (make-list q roman)))))

(define (encode/roman number)
  (string-join (map symbol->string (number->list number)) "")) 

(for ([n '(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 25 30 40 
           50 60 69 70 80 90 99 100 200 300 400 500 600 666 700 800 900 
           1000 1009 1444 1666 1945 1997 1999 2000 2008 2010 2011 2500 
           3000 3999)])
  (printf "~a ~a\n" n (encode/roman n)))

Raku

(formerly Perl 6)

my %symbols =
    1 => "I", 5 => "V", 10 => "X", 50 => "L", 100 => "C",
    500 => "D", 1_000 => "M";

my @subtractors =
    1_000, 100,  500, 100,  100, 10,  50, 10,  10, 1,  5, 1,  1, 0;

multi sub roman (0) { '' }
multi sub roman (Int $n) {
    for @subtractors -> $cut, $minus {
        $n >= $cut
            and return %symbols{$cut} ~ roman($n - $cut);
        $n >= $cut - $minus
            and return %symbols{$minus} ~ roman($n + $minus);
     }
}

# Sample usage

for 1 .. 2_010 -> $x {
    say roman($x);
}

Red

Straight iterative solution:

Red []

table: [1000 M 900 CM 500 D 400 CD 100 C 90 XC 50 L 40 XL 10 X 9 IX 5 V 4 IV 1 I]

to-Roman: function [n [integer!] return: [string!]][
    out: copy ""
    foreach [a r] table [while [n >= a][append out r n: n - a]]
    out
]

foreach number [40 33 1888 2016][print [number ":" to-Roman number]]

Straight recursive solution:

Red []

table: [1000 M 900 CM 500 D 400 CD 100 C 90 XC 50 L 40 XL 10 X 9 IX 5 V 4 IV 1 I]

to-Roman: func [n [integer!] return: [string!]][
    case [
        tail? table [table: head table copy ""]
        table/1 > n [table: skip table 2 to-Roman n]
        'else       [append copy form table/2 to-Roman n - table/1]
    ]
]

foreach number [40 33 1888 2016][print [number ":" to-Roman number]]

This solution builds, using metaprogramming, a `case` table, that relies on recursion to convert every digit.

Red []

to-Roman: function [n [integer!]] reduce [
    'case collect [
        foreach [a r] [1000 M 900 CM 500 D 400 CD 100 C 90 XC 50 L 40 XL 10 X 9 IX 5 V 4 IV 1 I][
            keep compose/deep [n >= (a) [append copy (form r) any [to-Roman n - (a) copy ""]]]
        ]	
    ]
]

foreach number [40 33 1888 2016][print [number ":" to-Roman number]]

Retro

This is a port of the Forth code; but returns a string rather than displaying the roman numerals. It only handles numbers between 1 and 3999.

: vector ( ...n"- )
  here [ &, times ] dip : .data ` swap ` + ` @ ` do ` ; ;
: .I  dup     @ ^buffer'add ;
: .V  dup 1 + @ ^buffer'add ;
: .X  dup 2 + @ ^buffer'add ;
 
[ .I .X       drop ]
[ .V .I .I .I drop ]
[ .V .I .I    drop ]
[ .V .I       drop ]
[ .V          drop ]
[ .I .V       drop ]
[ .I .I .I    drop ]
[ .I .I       drop ]
[ .I          drop ]
&drop
10 vector .digit
 
: record ( an- )
  10 /mod dup [ [ over 2 + ] dip record ] &drop if .digit ;
: toRoman   ( n-a )
  here ^buffer'set
  dup 1 3999 within 0 =
  [ "EX LIMITO!\n" ] [ "IVXLCDM" swap record here ] if ;

REXX

version 1

roman: procedure
arg number

/* handle only 1 to 3999, else return ? */
if number >= 4000 | number <= 0 then return "?"

romans = "   M  CM   D  CD   C  XC  L  XL  X IX  V IV  I"
arabic = "1000 900 500 400 100  90 50  40 10  9  5  4  1"

result = ""
do i = 1 to words(romans)
  do while number >= word(arabic,i)
    result = result || word(romans,i)
    number = number - word(arabic,i)
  end
end
return result

version 2

This version of a REXX program allows almost any non-negative decimal integer.

Most people think that the Romans had no word for "zero".   The Roman numeral system has no need for a
zero   placeholder,   so there was no name for it   (just as we have no name for a   "¶"   in the middle of our
numbers ─── as we don't have that possibility).   However, the Romans did have a name for zero (or nothing).
In fact the Romans had several names for zero   (see the REXX code),   as does modern English.   In American
English, many words can be used for   0:     zero, nothing, naught, bupkis, zilch, goose-egg, nebbish, squat, nil,
crapola, what-Patty-shot-at, nineteen (only in cribbage), love (in tennis), etc.

Also, this REXX version supports large numbers (with parentheses and deep parentheses).

(This REXX code was ripped out of my general routine that also supported versions for Attic, ancient Roman,
and modern Roman numerals.)

The general REXX code is bulkier than most at it deals with   any   non-negative decimal number,   and more
boilerplate code is in the general REXX code to handle the above versions.

/*REXX program converts (Arabic) non─negative decimal integers (≥0) ───► Roman numerals.*/
numeric digits 10000                             /*decimal digs can be higher if wanted.*/
parse arg #                                      /*obtain optional integers from the CL.*/
@er= "argument isn't a non-negative integer: "   /*literal used when issuing error msg. */
if #=''  then                                    /*Nothing specified?  Then generate #s.*/
    do
                                                  do j= 0  by  11  to  111; #=# j;     end
    #=# 49;                                       do k=88  by 100  to 1200; #=# k;     end
    #=# 1000 2000 3000 4000 5000 6000;            do m=88  by 200  to 1200; #=# m;     end
    #=# 1304 1405 1506 1607 1708 1809 1910 2011;  do p= 4          to   50; #=# 10**p; end
    end                                          /*finished with generation of numbers. */

  do i=1  for words(#);         x=word(#, i)     /*convert each of the numbers───►Roman.*/
  if \datatype(x, 'W') | x<0  then say "***error***"  @er  x     /*¬ whole #?  negative?*/
  say  right(x, 55)     dec2rom(x)
  end   /*i*/
exit                                             /*stick a fork in it,  we're all done. */
/*──────────────────────────────────────────────────────────────────────────────────────*/
dec2rom: procedure;   parse arg n,#              /*obtain the number, assign # to a null*/
         n=space(translate(n/1, , ','),  0)      /*remove commas from normalized integer*/
         nulla= 'ZEPHIRUM NULLAE NULLA NIHIL'    /*Roman words for "nothing" or "none". */
         if n==0  then return word(nulla, 1)     /*return a Roman word for  "zero".     */
         maxnp=(length(n)-1)%3                   /*find max(+1) # of parenthesis to use.*/
         highPos=(maxnp+1)*3                     /*highest position of number.          */
         nn=reverse( right(n, highPos, 0) )      /*digits for Arabic──►Roman conversion.*/
                       do j=highPos  to 1  by -3
                       _=substr(nn, j,   1);  select     /*════════════════════hundreds.*/
                                              when _==9  then hx='CM'
                                              when _>=5  then hx='D'copies("C", _-5)
                                              when _==4  then hx='CD'
                                              otherwise       hx=   copies('C', _)
                                              end  /*select hundreds*/
                       _=substr(nn, j-1, 1);  select     /*════════════════════════tens.*/
                                              when _==9  then tx='XC'
                                              when _>=5  then tx='L'copies("X", _-5)
                                              when _==4  then tx='XL'
                                              otherwise       tx=   copies('X', _)
                                              end  /*select tens*/
                       _=substr(nn, j-2, 1);  select     /*═══════════════════════units.*/
                                              when _==9  then ux='IX'
                                              when _>=5  then ux='V'copies("I", _-5)
                                              when _==4  then ux='IV'
                                              otherwise       ux=   copies('I', _)
                                              end  /*select units*/
                       $=hx || tx || ux
                       if $\==''  then #=# || copies("(", (j-1)%3)$ ||copies(')', (j-1)%3)
                       end   /*j*/
         if pos('(I',#)\==0  then do i=1  for 4           /*special case: M,MM,MMM,MMMM.*/
                                  if i==4  then _ = '(IV)'
                                           else _ = '('copies("I", i)')'
                                  if pos(_, #)\==0  then #=changestr(_, #, copies('M', i))
                                  end   /*i*/
         return #

Some older REXXes don't have a   changestr   BIF,   so one is included here   ──►   CHANGESTR.REX.

output   when using the default (internal) input):

                                                      0 ZEPHIRUM
                                                     11 XI
                                                     22 XXII
                                                     33 XXXIII
                                                     44 XLIV
                                                     55 LV
                                                     66 LXVI
                                                     77 LXXVII
                                                     88 LXXXVIII
                                                     99 XCIX
                                                    110 CX
                                                     49 XLIX
                                                     88 LXXXVIII
                                                    188 CLXXXVIII
                                                    288 CCLXXXVIII
                                                    388 CCCLXXXVIII
                                                    488 CDLXXXVIII
                                                    588 DLXXXVIII
                                                    688 DCLXXXVIII
                                                    788 DCCLXXXVIII
                                                    888 DCCCLXXXVIII
                                                    988 CMLXXXVIII
                                                   1088 MLXXXVIII
                                                   1188 MCLXXXVIII
                                                   1000 M
                                                   2000 MM
                                                   3000 MMM
                                                   4000 MMMM
                                                   5000 (V)
                                                   6000 (VI)
                                                     88 LXXXVIII
                                                    288 CCLXXXVIII
                                                    488 CDLXXXVIII
                                                    688 DCLXXXVIII
                                                    888 DCCCLXXXVIII
                                                   1088 MLXXXVIII
                                                   1304 MCCCIV
                                                   1405 MCDV
                                                   1506 MDVI
                                                   1607 MDCVII
                                                   1708 MDCCVIII
                                                   1809 MDCCCIX
                                                   1910 MCMX
                                                   2011 MMXI
                                                  10000 (X)
                                                 100000 (C)
                                                1000000 (M)
                                               10000000 ((X))
                                              100000000 ((C))
                                             1000000000 ((M))
                                            10000000000 (((X)))
                                           100000000000 (((C)))
                                          1000000000000 (((M)))
                                         10000000000000 ((((X))))
                                        100000000000000 ((((C))))
                                       1000000000000000 ((((M))))
                                      10000000000000000 (((((X)))))
                                     100000000000000000 (((((C)))))
                                    1000000000000000000 (((((M)))))
                                   10000000000000000000 ((((((X))))))
                                  100000000000000000000 ((((((C))))))
                                 1000000000000000000000 ((((((M))))))
                                10000000000000000000000 (((((((X)))))))
                               100000000000000000000000 (((((((C)))))))
                              1000000000000000000000000 (((((((M)))))))
                             10000000000000000000000000 ((((((((X))))))))
                            100000000000000000000000000 ((((((((C))))))))
                           1000000000000000000000000000 ((((((((M))))))))
                          10000000000000000000000000000 (((((((((X)))))))))
                         100000000000000000000000000000 (((((((((C)))))))))
                        1000000000000000000000000000000 (((((((((M)))))))))
                       10000000000000000000000000000000 ((((((((((X))))))))))
                      100000000000000000000000000000000 ((((((((((C))))))))))
                     1000000000000000000000000000000000 ((((((((((M))))))))))
                    10000000000000000000000000000000000 (((((((((((X)))))))))))
                   100000000000000000000000000000000000 (((((((((((C)))))))))))
                  1000000000000000000000000000000000000 (((((((((((M)))))))))))
                 10000000000000000000000000000000000000 ((((((((((((X))))))))))))
                100000000000000000000000000000000000000 ((((((((((((C))))))))))))
               1000000000000000000000000000000000000000 ((((((((((((M))))))))))))
              10000000000000000000000000000000000000000 (((((((((((((X)))))))))))))
             100000000000000000000000000000000000000000 (((((((((((((C)))))))))))))
            1000000000000000000000000000000000000000000 (((((((((((((M)))))))))))))
           10000000000000000000000000000000000000000000 ((((((((((((((X))))))))))))))
          100000000000000000000000000000000000000000000 ((((((((((((((C))))))))))))))
         1000000000000000000000000000000000000000000000 ((((((((((((((M))))))))))))))
        10000000000000000000000000000000000000000000000 (((((((((((((((X)))))))))))))))
       100000000000000000000000000000000000000000000000 (((((((((((((((C)))))))))))))))
      1000000000000000000000000000000000000000000000000 (((((((((((((((M)))))))))))))))
     10000000000000000000000000000000000000000000000000 ((((((((((((((((X))))))))))))))))
    100000000000000000000000000000000000000000000000000 ((((((((((((((((C))))))))))))))))

Ring

arabic = [1000, 900, 500, 400, 100, 90, 50,  40,  10,  9,  5,   4,  1]
roman  = ["M", "CM", "D", "CD", "C" ,"XC", "L", "XL" ,"X", "IX", "V", "IV", "I"]

see "2009 = " + toRoman(2009) + nl
see "1666 = " + toRoman(1666) + nl
see "3888 = " + toRoman(3888) + nl
 
func toRoman val
     result = ""
     for i = 1 to 13 
         while val >= arabic[i] 
               result = result + roman[i]
               val = val - arabic[i]
         end
      next
      return result

RPL

Translation of: Python
Works with: Halcyon Calc version 4.2.7
RPL code Comment
≪ { 1000 900 500 400 100 90 50 40 10 9 5 4 1 }
   { "M" "CM" "D" "CD" "C" "XC" "L" "XL" "X" "IX"
     "V" "IV" "I" } → divs rdig
   IF DUP 5000 < THEN
     "" SWAP 1 13 FOR j
        divs j GET MOD LAST / IP ROT SWAP
        WHILE DUP REPEAT 
           rdig j GET ROT SWAP + SWAP 1 - END
        DROP SWAP
     NEXT 
  END DROP 
≫ '→ROM' STO
→ROM ( n -- "ROMAN" )
store tables

if n < 5000 then
   scan divisors
   x,y = divmod(n, divisor)
   if x > 0 then
       add related digit x times
   n = y
   
clean stack

Alternate version

RPL code Comment
IF DUP 5000 < THEN 
    { "IIIVIIIX"  "XXXLXXXC" "CCCDCCCM" }
    { 11 21 31 43 44 54 64 74 87 88 } → rom args
    ≪ "" SWAP 
       1 3 FOR j 
          10 MOD LAST / IP 
          IF SWAP THEN 
             args LAST GET 10 MOD LAST / IP 
             rom j GET ROT ROT SUB ROT + SWAP END 
       NEXTWHILE DUP REPEAT 1 - "M" ROT + SWAP END
    DROP END 
≫ '→ROM' STO
→ROM ( n -- "M..CXVI" ) 
collapsed Roman digits
10 arguments to extract Roman digits
initialize stack
process units to hundreds
   divmod(n,10)
   if last digit ≠ 0 then
      get extraction arguments
      extract Roman digit

add thousands if any
clean stack
       

Ruby

Roman numeral generation was used as an example for demonstrating Test Driven Development in Ruby. The solution came to be:

Symbols = { 1=>'I', 5=>'V', 10=>'X', 50=>'L', 100=>'C', 500=>'D', 1000=>'M' }
Subtractors = [ [1000, 100], [500, 100], [100, 10], [50, 10], [10, 1], [5, 1], [1, 0] ]

def roman(num)
  return Symbols[num]  if Symbols.has_key?(num)
  Subtractors.each do |cutPoint, subtractor| 
    return roman(cutPoint) + roman(num - cutPoint)      if num >  cutPoint
    return roman(subtractor) + roman(num + subtractor)  if num >= cutPoint - subtractor and num < cutPoint
  end
end

[1990, 2008, 1666].each do |i|
  puts "%4d => %s" % [i, roman(i)]
end
Output:
1990 => MCMXC
2008 => MMVIII
1666 => MDCLXVI

Another shorter version if we don't consider calculating the substractors:

Symbols = [ [1000, 'M'], [900, 'CM'], [500, 'D'], [400, 'CD'], [100, 'C'], [90, 'XC'], [50, 'L'], [40, 'XL'], [10, 'X'], [9, 'IX'], [5, 'V'], [4, 'IV'], [1, 'I'] ]

def arabic_to_roman(arabic)
  return '' if arabic.zero?
  Symbols.each { |arabic_rep, roman_rep| return roman_rep + arabic_to_roman(arabic - arabic_rep) if arabic >= arabic_rep }
end

Yet another way to solve it in terms of reduce

Symbols = [ [1000, 'M'], [900, 'CM'], [500, 'D'], [400, 'CD'], [100, 'C'], [90, 'XC'], [50, 'L'], [40, 'XL'], [10, 'X'], [9, 'IX'], [5, 'V'], [4, 'IV'], [1, 'I'] ]

def to_roman(num)
    Symbols.reduce "" do |memo, (divisor, letter)|
        div, num = num.divmod(divisor)
        memo + letter * div
    end
end

Rust

struct RomanNumeral {
    symbol: &'static str,
    value: u32
}

const NUMERALS: [RomanNumeral; 13] = [
    RomanNumeral {symbol: "M",  value: 1000},
    RomanNumeral {symbol: "CM", value: 900},
    RomanNumeral {symbol: "D",  value: 500},
    RomanNumeral {symbol: "CD", value: 400},
    RomanNumeral {symbol: "C",  value: 100},
    RomanNumeral {symbol: "XC", value: 90},
    RomanNumeral {symbol: "L",  value: 50},
    RomanNumeral {symbol: "XL", value: 40},
    RomanNumeral {symbol: "X",  value: 10},
    RomanNumeral {symbol: "IX", value: 9},
    RomanNumeral {symbol: "V",  value: 5},
    RomanNumeral {symbol: "IV", value: 4},
    RomanNumeral {symbol: "I",  value: 1}
];

fn to_roman(mut number: u32) -> String {
    let mut min_numeral = String::new();
    for numeral in NUMERALS.iter() {
        while numeral.value <= number {
            min_numeral = min_numeral + numeral.symbol;
            number -= numeral.value;
        }
    }
    min_numeral
}

fn main() {
    let nums = [2014, 1999, 25, 1666, 3888];
    for &n in nums.iter() {
        // 4 is minimum printing width, for alignment
        println!("{:2$} = {}", n, to_roman(n), 4);
    }
}
Output:
2014 = MMXIV
1999 = MCMXCIX
  25 = XXV
1666 = MDCLXVI
3888 = MMMDCCCLXXXVIII

Scala

Works with: Scala version 2.8
val romanDigits = Map(
  1 -> "I", 5 -> "V", 
  10 -> "X", 50 -> "L", 
  100 -> "C", 500 -> "D", 
  1000 -> "M", 
  4 -> "IV", 9 -> "IX", 
  40 -> "XL", 90 -> "XC", 
  400 -> "CD", 900 -> "CM")
val romanDigitsKeys = romanDigits.keysIterator.toList sortBy (x => -x)
def toRoman(n: Int): String = romanDigitsKeys find (_ >= n) match {
  case Some(key) => romanDigits(key) + toRoman(n - key)
  case None => ""
}
Output:
scala> List(1990, 2008, 1666) map toRoman
res55: List[String] = List(MCMXC, MMVIII, MDCLXVI)

Using foldLeft

def toRoman( v:Int ) : String = {
  val romanNumerals = List(1000->"M",900->"CM",500->"D",400->"CD",100->"C",90->"XC",
                           50->"L",40->"XL",10->"X",9->"IX",5->"V",4->"IV",1->"I")	
                            
  var n = v                          
  romanNumerals.foldLeft(""){(s,t) => {val c = n/t._1; n = n-t._1*c;  s + (t._2 * c) } }
}
  
// A small test
def test( arabic:Int ) = println( arabic + " => " + toRoman( arabic ) )
  
test(1990)
test(2008)
test(1666)

Different code-style

def toRoman(num: Int): String = {
  case class RomanUnit(value: Int, token: String)
  val romanNumerals = List(
    RomanUnit(1000, "M"),
    RomanUnit(900, "CM"),
    RomanUnit(500, "D"),
    RomanUnit(400, "CD"),
    RomanUnit(100, "C"),
    RomanUnit(90, "XC"),
    RomanUnit(50, "L"),
    RomanUnit(40, "XL"),
    RomanUnit(10, "X"),
    RomanUnit(9, "IX"),
    RomanUnit(5, "V"),
    RomanUnit(4, "IV"),
    RomanUnit(1, "I"))

  var remainingNumber = num
  romanNumerals.foldLeft("") { (outputStr, romanUnit) =>
    {
      val times = remainingNumber / romanUnit.value
      remainingNumber -= romanUnit.value * times
      outputStr + (romanUnit.token * times)
    }
  }
}
Output:
1990 => MCMXC
2008 => MMVIII
1666 => MDCLXVI

Scheme

This uses format directives supported in Chez Scheme since v6.9b; YMMV.

(define (to-roman n)
  (format "~@r" n))

This is a general example using Chicken Scheme.

(define roman-decimal
  '(("M"  . 1000)
    ("CM" . 900)
    ("D"  . 500)
    ("CD" . 400)
    ("C"  . 100)
    ("XC" .  90)
    ("L"  .  50)
    ("XL" .  40)
    ("X"  .  10)
    ("IX" .   9)
    ("V"  .   5)
    ("IV" .   4)
    ("I"  .   1)))

(define (to-roman value)
  (apply string-append
         (let loop ((v value)
                    (decode roman-decimal))
           (let ((r (caar decode))
                 (d (cdar decode)))
             (cond
              ((= v 0) '())
              ((>= v d) (cons r (loop (- v d) decode)))
              (else (loop v (cdr decode))))))))


(let loop ((n '(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 25 30 40 
                  50 60 69 70 80 90 99 100 200 300 400 500 600 666 700 800 900 
                  1000 1009 1444 1666 1945 1997 1999 2000 2008 2010 2011 2500 
                  3000 3999)))
  (unless (null? n)
    (printf "~a ~a\n" (car n) (to-roman (car n)))
    (loop (cdr n))))

Seed7

The following program writes the numbers between 1 and 3999 as roman numerals. The wrinum.s7i library contains the function str(ROMAN,), which writes a roman numeral to a string.

$ include "seed7_05.s7i";
  include "stdio.s7i";
  include "wrinum.s7i";

const proc: main is func
  local
    var integer: number is 0;
  begin
    for number range 1 to 3999 do
      writeln(str(ROMAN, number));
    end for;
  end func;

Original source [1].

SenseTalk

function RomanNumeralsEncode number
	put [
		(1, "I"),
		(4, "IV"),
		(5, "V"),
		(9, "IX"),
		(10, "X"),
		(40, "XL"),
		(50, "L"),
		(90, "XC"),
		(100, "C"),
		(400, "CD"),
		(500, "D"),
		(900, "CM"),
		(1000, "M")
	] into values
	
	
	repeat for index = each item of (the number of items in values)..1
		put item index of values into pair
		repeat while number is greater than or equal to the first item of pair
			put the second item of pair after numerals
			subtract the first item of pair from number
		end repeat
	end repeat
	return numerals
end RomanNumeralsEncode
repeat for each item in [
	1990,
	2008,
	1666	
]
	put RomanNumeralsEncode(it)
end repeat
Output:
MCMXC
MMVIII
MDCLXVI

SETL

examples := [2008, 1666, 1990];

for example in examples loop
    print( roman_numeral(example) );
end loop;

proc roman_numeral( n );
    R := [[1000, 'M'], [900, 'CM'], [500, 'D'], [400, 'CD'], [100, 'C'], [90, 'XC'], [50, 'L'], [40, 'XL'], [10, 'X'], [9, 'IX'], [5, 'V'], [4, 'IV'], [1, 'I']];
    roman := '';
    for numeral in R loop
        while n >= numeral(1) loop
            n := n - numeral(1);
            roman := roman + numeral(2);
        end loop;
    end loop;
    return roman;
end;
Output:
MMVIII
MDCLXVI
MCMXC

Shen

(define encodeGlyphs
  ACC 0 _ -> ACC
  ACC N [Glyph Value | Rest] -> (encodeGlyphs (@s ACC Glyph) (- N Value) [Glyph Value | Rest]) where (>= N Value)
  ACC N [Glyph Value | Rest] -> (encodeGlyphs ACC N Rest)
)

(define encodeRoman
  N -> (encodeGlyphs "" N ["M" 1000 "CM" 900 "D" 500 "CD" 400 "C" 100 "XC" 90 "L" 50 "XL" 40 "X" 10 "IX" 9 "V" 5 "IV" 4 "I" 1])
)
Output:
(4-) (encodeRoman 1990)
"MCMXC"

(5-) (encodeRoman 2008)
"MMVIII"

(6-) (encodeRoman 1666)
"MDCLXVI"

Sidef

Translation of: ActionScript
func arabic2roman(num, roman='') {
    static lookup = [
        :M:1000, :CM:900, :D:500,
        :CD:400, :C:100,  :XC:90,
        :L:50,   :XL:40,  :X:10,
        :IX:9,   :V:5,    :IV:4,
        :I:1
    ];
    lookup.each { |pair|
        while (num >= pair.second) {
            roman += pair.first;
            num -= pair.second;
        }
    }
    return roman;
}
say("1990 in roman is " + arabic2roman(1990));
say("2008 in roman is " + arabic2roman(2008));
say("1666 in roman is " + arabic2roman(1666));
Output:
1990 in roman is MCMXC
2008 in roman is MMVIII
1666 in roman is MDCLXVI

Simula

BEGIN

    TEXT PROCEDURE TOROMAN(N); INTEGER N;
    BEGIN
        PROCEDURE P(WEIGHT,LIT); INTEGER WEIGHT; TEXT LIT;
        BEGIN
            WHILE N >= WEIGHT DO
            BEGIN
                T :- T & LIT;
                N := N - WEIGHT;
            END WHILE;
        END P;
        TEXT T; T :- NOTEXT;
        P( 1000, "M"  );
        P(  900, "CM" );
        P(  500, "D"  );
        P(  400, "CD" );
        P(  100, "C"  );
        P(   90, "XC" );
        P(   50, "L"  );
        P(   40, "XL" );
        P(   10, "X"  );
        P(    9, "IX" );
        P(    5, "V"  );
        P(    4, "IV" );
        P(    1, "I"  );
        TOROMAN :- T;
    END TOROMAN;

    INTEGER Y;
    FOR Y := 1990, 2008, 1666 DO
    BEGIN
        OUTTEXT("YEAR ");
        OUTINT(Y, 4);
        OUTTEXT(" => ");
        OUTTEXT(TOROMAN(Y));
        OUTIMAGE;
    END FOR;

END PROGRAM;
Output:
YEAR 1990 => MCMXC
YEAR 2008 => MMVIII
YEAR 1666 => MDCLXVI

Smalltalk

Works with: Smalltalk/X

in ST/X, integers already know how to print themselves as roman number:

2013 printRomanOn:Stdout naive:false
Output:
MMXIII

the implementation is:

printRomanOn:aStream naive:naive
    "print the receiver as roman number to the argument, aStream.
     The naive argument controls if the conversion is
     correct (i.e. subtracting prefix notation for 4,9,40,90, etc.),
     or naive (i.e. print 4 as IIII and 9 as VIIII); also called simple.
     The naive version is often used for page numbers in documents."

    |restValue spec|

    restValue := self.
    restValue > 0 ifFalse:[self error:'negative roman'].

    naive ifTrue:[
        spec := #(
                " value string repeat "
                   1000 'M'    true
                    500 'D'    false
                    100 'C'    true
                     50 'L'    false
                     10 'X'    true
                      5 'V'    false
                      1 'I'    true
                 ).
    ] ifFalse:[
        spec := #(
                " value string repeat "
                   1000 'M'    true
                    900 'CM'   false
                    500 'D'    false
                    400 'CD'   false
                    100 'C'    true
                     90 'XC'   false
                     50 'L'    false
                     40 'XL'   false
                     10 'X'    true
                      9 'IX'   false
                      5 'V'    false
                      4 'IV'   false
                      1 'I'    true
                 ).
    ].

    spec
        inGroupsOf:3
        do:[:rValue :rString :repeatFlag |

            [
                (restValue >= rValue) ifTrue:[
                    aStream nextPutAll:rString.
                    restValue := restValue - rValue.
                ].
            ] doWhile:[ repeatFlag and:[ restValue >= rValue] ].
        ].

SNOBOL4

Adapted from Catspaw SNOBOL Tutorial, Chapter 6

* ROMAN(N) - Convert integer N to Roman numeral form.
*
*  N must be positive and less than 4000.
*
*  An asterisk appears in the result if N >= 4000.
*
*  The function fails if N is not an integer.

	DEFINE('ROMAN(N)UNITS')              :(ROMAN_END)

*  Get rightmost digit to UNITS and remove it from N.
*  Return null result if argument is null.
ROMAN	N RPOS(1) LEN(1) . UNITS =           :F(RETURN)

*  Search for digit, replace with its Roman form.
*  Return failing if not a digit.
	'0,1I,2II,3III,4IV,5V,6VI,7VII,8VIII,9IX,'  UNITS 
+	BREAK(',') . UNITS                 :F(FRETURN)

*  Convert rest of N and multiply by 10.  Propagate a
*  failure return from recursive call back to caller.
	ROMAN = REPLACE(ROMAN(N), 'IVXLCDM', 'XLCDM**') 
+	UNITS            :S(RETURN) F(FRETURN)
ROMAN_END

*	Testing
	OUTPUT = "1999 = " ROMAN(1999)
	OUTPUT = "  24 = " ROMAN(24)
	OUTPUT = " 944 = " ROMAN(944)

END
Output:
1999 = MCMXCIX
  24 = XXIV
 944 = CMXLIV

Here's a non-recursive version, and a Roman-to-Arabic converter to boot.

*       # Arabic to Roman
        define('roman(n)s,ch,val,str') :(roman_end)
roman   roman = ge(n,4000) n :s(return)
        s = 'M1000 CM900 D500 CD400 C100 XC90 L50 XL40 X10 IX9 V5 IV4 I1 '
rom1    s span(&ucase) . ch break(' ') . val span(' ') = :f(rom2)
        str = str dupl(ch,(n / val))
        n = remdr(n,val) :(rom1)
rom2    roman = str :(return)
roman_end        

*       # Roman to Arabic
        define('arabic(n)s,ch,val,sum,x') :(arabic_end)
arabic  s = 'M1000 D500 C100 L50 X10 V5 I1 '
        n = reverse(n)
arab1   n len(1) . ch = :f(arab2)
        s ch break(' ') . val
        val = lt(val,x) (-1 * val)
        sum = sum + val; x = val :(arab1)
arab2   arabic = sum :(return)
arabic_end        

*       # Test and display
        tstr = '2010 1999 1492 1066 476 '
tloop   tstr break(' ') . year span(' ') = :f(out)
        r = roman(year)
        rstr = rstr year '=' r ' ' 
        astr = astr r '=' arabic(r) ' ' :(tloop)
out     output = rstr; output = astr
end
Output:
2010=MMX 1999=MCMXCIX 1492=MCDXCII 1066=MLXVI 476=CDLXXVI
MMX=2010 MCMXCIX=1999 MCDXCII=1492 MLXVI=1066 CDLXXVI=476

SPL

a2r(a)=
  r = ""
  n = [["M","CM","D","CD","C","XC","L","XL","X","IX","V","IV","I"],[1000,900,500,400,100,90,50,40,10,9,5,4,1]]
  > i, 1..13
    > a!<n[i,2]
      r += n[i,1]
      a -= n[i,2]
    <
  <
  <= r
.

t = [1990,2008,1666]
> i, 1..#.size(t,1)
  #.output(t[i]," = ",a2r(t[i]))
<
Output:
1990 = MCMXC
2008 = MMVIII
1666 = MDCLXVI

SQL

-- 
-- This only works under Oracle and has the limitation of 1 to 3999


SQL> select to_char(1666, 'RN') urcoman, to_char(1666, 'rn') lcroman from dual;

URCOMAN         LCROMAN
--------------- ---------------
        MDCLXVI         mdclxvi

Swift

func ator(var n: Int) -> String {

    var result = ""
    
    for (value, letter) in
       [( 1000,    "M"),
        (  900,   "CM"),
        (  500,    "D"),
        (  400,   "CD"),
        (  100,    "C"),
        (   90,   "XC"),
        (   50,    "L"),
        (   40,   "XL"),
        (   10,    "X"),
        (    9,   "IX"),
        (    5,    "V"),
        (    4,   "IV"),
        (    1,    "I")]
    {
        while n >= value {
            result += letter
            n   -= value
        }
    }
    return result
}

Sample call:

Works with: Swift version 1.x
println(ator(1666)) // MDCLXVI
Works with: Swift version 2.0
print(ator(1666)) // MDCLXVI
Output:
MDCLXVI 

Tailspin

def digits: [(M:1000"1"), (CM:900"1"), (D:500"1"), (CD:400"1"), (C:100"1"), (XC:90"1"), (L:50"1"), (XL:40"1"), (X:10"1"), (IX:9"1"), (V:5"1"), (IV:4"1"), (I:1"1")];
templates encodeRoman
  @: 1;
  '$ -> ($)"1" -> #;' !
  when <$digits($@)::value..> do
    $digits($@)::key !
    $ - $digits($@)::value -> #
  when <1"1"..> do
    @:$@ + 1;
    $ -> #
end encodeRoman

1990 -> encodeRoman -> !OUT::write
'
' -> !OUT::write
2008 -> encodeRoman -> !OUT::write
'
' -> !OUT::write
1666 -> encodeRoman -> !OUT::write
Output:
MCMXC
MMVIII
MDCLXVI

Tcl

proc to_roman {i} {
    set map {1000 M 900 CM 500 D 400 CD 100 C 90 XC 50 L 40 XL 10 X 9 IX 5 V 4 IV 1 I}
    foreach {value roman} $map {
        while {$i >= $value} {
            append res $roman
            incr i -$value
        }
    }
    return $res
}

TUSCRIPT

$$ MODE TUSCRIPT
LOOP arab_number="1990'2008'1666"
roman_number = ENCODE (arab_number,ROMAN)
PRINT "Arabic number ",arab_number, " equals ", roman_number
ENDLOOP
Output:
Arabic number 1990 equals MCMXC
Arabic number 2008 equals MMVIII
Arabic number 1666 equals MDCLXVI 

TypeScript

Translation of: DWScript

Weights and symbols in tuples.

// Roman numerals/Encode

const weightsSymbols: [number, string][] = 
  [[1000, 'M'], [900, 'CM'], [500, 'D'], [400, 'CD'], [100, 'C'], [90, 'XC'], 
  [50, 'L'], [40, 'XL'], [10, 'X'], [9, 'IX'], [5, 'V'], [4, 'IV'], [1, 'I']];
// 3888 or MMMDCCCLXXXVIII (15 chars) is the longest string properly encoded
// with these symbols.

function toRoman(n: number): string {
  var roman = ""; // Result
  for (i = 0; i <= 12 && n > 0; i++) {
    var w = weightsSymbols[i][0];
    while (n >= w) {
      roman += weightsSymbols[i][1];
      n -= w;
    }
  }
  return roman;
}

console.log(toRoman(1990)); // MCMXC
console.log(toRoman(2022)); // MMXXII
console.log(toRoman(3888)); // MMMDCCCLXXXVIII
Output:
MCMXC
MMXXII
MMMDCCCLXXXVIII

UNIX Shell

Translation of: Tcl
Works with: bash
roman() {
    local values=( 1000 900 500 400 100 90 50 40 10 9 5 4 1 )
    local roman=(
        [1000]=M [900]=CM [500]=D [400]=CD 
         [100]=C  [90]=XC  [50]=L  [40]=XL 
          [10]=X   [9]=IX   [5]=V   [4]=IV   
           [1]=I
    )
    local nvmber=""
    local num=$1
    for value in ${values[@]}; do
        while (( num >= value )); do
            nvmber+=${roman[value]}
            ((num -= value))
        done
    done
    echo $nvmber
}

for test in 1999 24 944 1666 2008; do
    printf "%d = %s\n" $test $(roman $test)
done
Output:
1999 = MCMXCIX
24 = XXIV
944 = CMXLIV
1666 = MDCLXVI
2008 = MMVIII

Ursala

The algorithm is to implement the subtractive principle by string substitution only after constucting the numeral from successive remainders. The order among the substitutions matters. For example, occurrences of DCCCC must be replaced by CM before any occurrences of CCCC are replaced by CD. The substitution operator (%=) is helpful here.

#import nat

roman = 

-+
   'IIII'%='IV'+ 'VIIII'%='IX'+ 'XXXX'%='XL'+ 'LXXXX'%='XC'+ 'CCCC'%='CD'+ 'DCCCC'%='CM',
   ~&plrDlSPSL/'MDCLXVI'+ iota*+ +^|(^|C/~&,\/division)@rlX=>~&iNC <1000,500,100,50,10,5>+-

This test program applies the function to each member of a list of numbers.

#show+

test = roman* <1990,2008,1,2,64,124,1666,10001>
Output:
MCMXC
MMVIII
I
II
LXIV
CXXIV
MDCLXVI
MMMMMMMMMMI

Vala

Translation of: D
string to_roman(int n) 
  requires (n > 0 && n < 5000)
{
  const int[] weights = {1000, 900, 500, 400, 100, 90,
                         50, 40, 10, 9, 5, 4, 1};
  const string[] symbols = {"M","CM","D","CD","C","XC","L",
                            "XL","X","IX","V","IV","I"};
  
  var roman = "", count = 0;
  foreach (var w in weights) {
    while (n >= w) {
      roman += symbols[count];
      n -= w;
    }
    if (n == 0)
      break;
    count++;
  }
  return roman;
}

void main() {
  print("%s\n", to_roman(455));
  print("%s\n", to_roman(3456));
  print("%s\n", to_roman(2488));
}
Output:
CDLV
MMMCDLVI
MMCDLXXXVIII

VBA

Private Function roman(n As Integer) As String
    roman = WorksheetFunction.roman(n)
End Function
Public Sub main()
    s = [{10, 2016, 800, 2769, 1666, 476, 1453}]
    For Each x In s
        Debug.Print roman(CInt(x)); " ";
    Next x
End Sub
Output:
X MMXVI DCCC MMDCCLXIX MDCLXVI CDLXXVI MCDLIII 

Vedit macro language

// Main program for testing the function
//
do {
    #1 = Get_Num("Number to convert: ", STATLINE)
    Call("NUM_TO_ROMAN")
    Num_Type(#1, NOCR) Message(" = ") Reg_Type(1) Type_Newline
} while (Reg_Size(1))
Return
 
// Convert numeric value into Roman number
//  #1 = number to convert; on return: T-reg(1) = Roman number
//
:NUM_TO_ROMAN:
    Reg_Empty(1)                        // @1 = Results (Roman number)
    if (#1 < 1) { Return }              // non-positive numbers return empty string
 
    Buf_Switch(Buf_Free)
    Ins_Text("M1000,CM900,D500,CD400,C100,XC90,L50,XL40,X10,IX9,V5,IV4,I1")
 
    BOF
    #2 = #1
    Repeat(ALL) {
        Search("|A|[|A]", ADVANCE+ERRBREAK)         // get next item from conversion list
        Reg_Copy_Block(20, CP-Chars_Matched, CP)    // @20 = Letter(s) to be inserted
        #11 = Num_Eval()                            // #11 = magnitude (1000...1)
        while (#2 >= #11) {
            Reg_Set(1, @20, APPEND)
            #2 -= #11
        }
    }
    Buf_Quit(OK)
Return
Output:
    4 = IV
   12 = XII
 1666 = MDCLXVI
 1990 = MCMXC
 2011 = MMXI

V (Vlang)

const numerals = {1000:"M", 900:"CM", 500:"D", 400:"CD", 100:"C",
 90:"XC", 50:"L", 40: "XL", 10:"X", 9:"IX", 5:"V", 4:"IV", 1:"I"}

fn main() {
    println(encode(1990))
    println(encode(2008))
    println(encode(1666))
}

fn encode(number int) string {
    mut num := number
    mut result := ""
    if number < 1 || number > 5000 {return result}
    for digit, roman in numerals {
        for num >= digit {
            num -= digit
            result += roman
        }
    }
    return result
}
Output:
MCMXC
MMVIII
MDCLXVI

Wren

Translation of: Kotlin
var romans = [
    [1000, "M"],
    [900, "CM"],
    [500,  "D"],
    [400, "CD"],
    [100,  "C"],
    [90,  "XC"],
    [50,   "L"],
    [40,  "XL"],
    [10,   "X"],
    [9,   "IX"],
    [5,    "V"],
    [4,   "IV"],
    [1,    "I"]
]

var encode = Fn.new { |n|
    if (n > 5000 || n < 1) return null
    var res = ""
    for (r in romans) {
        while (n >= r[0]) {
            n = n - r[0]
            res = res + r[1]
        }
    }
    return res
}

System.print(encode.call(1990))
System.print(encode.call(1666))
System.print(encode.call(2008))
System.print(encode.call(2020))
Output:
MCMXC
MDCLXVI
MMVIII
MMXX

XLISP

(defun roman (n)
    (define roman-numerals '((1000 "m") (900 "cm") (500 "d") (400 "cd") (100 "c") (90 "xc") (50 "l") (40 "xl") (10 "x") (9 "ix") (5 "v") (4 "iv") (1 "i")))
    (defun romanize (arabic-numeral numerals roman-numeral)
        (if (= arabic-numeral 0)
            roman-numeral
            (if (>= arabic-numeral (caar numerals))
                (romanize (- arabic-numeral (caar numerals)) numerals (string-append roman-numeral (cadar numerals)))
                (romanize arabic-numeral (cdr numerals) roman-numeral))))
    (romanize n roman-numerals ""))

; test the function:
(display (mapcar roman '(10 2016 800 2769 1666 476 1453)))
Output:
(x mmxvi dccc mmdcclxix mdclxvi cdlxxvi mcdliii)

XPL0

proc    Rom(N, A, B, C);        \Display 1..9 in Roman numerals
int     N, A, B, C, I;
[case N of
9:      [ChOut(0, C);  ChOut(0, A)];                            \XI
8,7,6,5:[ChOut(0, B);  for I:= 1 to rem(N/5) do ChOut(0, C)];   \V
4:      [ChOut(0, C);  ChOut(0, B)]                             \IV
other   for I:= 1 to N do ChOut(0, C);                          \I
];

proc    Roman(N);               \Display N in Roman numerals
int     N, Q;
[Q:= N/1000;  N:= rem(0);       \0..3999
Rom(Q, ^?, ^?, ^M);
Q:= N/100;  N:= rem(0);         \0..999
Rom(Q, ^M, ^D, ^C);
Q:= N/10;  N:= rem(0);          \0..99
Rom(Q, ^C, ^L, ^X);
Rom(N, ^X, ^V, ^I);             \0..9
];

int     Tbl, I;
[Tbl:= [1990, 2008, 1666, 0, 1, 3999, 2020, 1234];
for I:= 0 to 7 do
        [IntOut(0, Tbl(I));  Text(0, ". ");  Roman(Tbl(I));  CrLf(0)];
]
Output:
1990.   MCMXC
2008.   MMVIII
1666.   MDCLXVI
0.      
1.      I
3999.   MMMCMXCIX
2020.   MMXX
1234.   MCCXXXIV

XSLT

<xsl:stylesheet  version="1.0"    xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
    <xsl:template match="/data/number">
        <xsl:call-template name="for">
               <xsl:with-param name="stop">13</xsl:with-param>
        	<xsl:with-param name="value"><xsl:value-of select="@value"></xsl:value-of></xsl:with-param>
        </xsl:call-template>
    </xsl:template>
    
    <xsl:template name="for">
      <xsl:param name="start">1</xsl:param>
      <xsl:param name="stop">1</xsl:param>
      <xsl:param name="step">1</xsl:param>
      <xsl:param name="value">1</xsl:param>
      <xsl:text/>
      <xsl:choose>
      <xsl:when test="($value &gt; /data/roman
/numeral[@pos=$start]/@value or $value = /data/roman
/numeral[@pos=$start]/@value) ">
          <xsl:value-of select="/data/roman
/numeral[@pos=$start]/@letter"/>
          <xsl:call-template name="for">
          <xsl:with-param name="stop">
            <xsl:value-of select="$stop"/>
          </xsl:with-param>
          <xsl:with-param name="start">
            <xsl:value-of select="$start"/>
          </xsl:with-param>
          <xsl:with-param name="value">
          	<xsl:value-of select="$value - /data/roman/numeral[@pos=$start]/@value"/>
          </xsl:with-param>
        </xsl:call-template>
      </xsl:when>
      <xsl:otherwise>
        <xsl:if test="$start &lt; $stop">
        <xsl:call-template name="for">
          <xsl:with-param name="stop">
            <xsl:value-of select="$stop"/>
          </xsl:with-param>
          <xsl:with-param name="start">
            <xsl:value-of select="$start + $step"/>
          </xsl:with-param>
          <xsl:with-param name="value">
          	<xsl:value-of select="$value"/>
          </xsl:with-param>
        </xsl:call-template>
        </xsl:if>
      </xsl:otherwise>
      </xsl:choose>
    </xsl:template>
</xsl:stylesheet>

zkl

var [const] romans = L(
   L("M", 1000), L("CM", 900), L("D",  500), L("CD", 400), L("C",  100),
   L("XC",  90), L("L",   50), L("XL",  40), L("X",   10), L("IX",   9),
   L("V",    5), L("IV",   4), L("I",    1));
fcn toRoman(i){		// convert int to a roman number
   reg text = "";
   foreach R,N in (romans){ text += R*(i/N); i = i%N; }
   return(text);
}
toRoman(1990) //-->"MCMXC"
toRoman(2008) //-->"MMVIII"
toRoman(1666) //-->"MDCLXVI"

Zoea

program: decimal_roman
  input: 12
  output: 'XII'

Zoea Visual

Roman numerals encode

Zsh

Based on the python solution.

function printroman () {
  local -a conv
  local number=$1 div rom num out
  conv=(I 1 IV 4 V 5 IX 9 X 10 XL 40 L 50 XC 90 C 100 CD 400 D 500 CM 900 M 1000)
  for num rom in ${(Oa)conv}; do
    (( div = number / num, number = number % num ))
    while (( div-- > 0 )); do
      out+=$rom
    done
  done
  echo $out
}