Index	Manual	Download	Source	Macros	Embedded word Case ins.
	Sitemap	Links	Forum	Tests	Projects

string16.htm
Macros
Concat$
Compare$
DecodeUTF8
GetLength$

This file can be included to 16bit programs written in Euro Assembler.
It contains OS-independent macros for zero-terminated (ASCIIZ) string operations.

All functions expect direction flag on input be zero and they do not change it.

Similar macros with identical names for different program width are defined in string32.htm and string64.htm.

 string16 HEAD

↑ GetLength$ String

This macro returns the size of ASCIIZ string in bytes. The terminating NUL character is not counted.

Input

String is pointer to a zero terminated string of ANSI characters. It may also be a literal string.
ES= segment of the String.

Output

CX= size of the string in bytes without the terminating NUL.

GetLength$ %MACRO String
     %IF "%String" !== "CX"
       MOV CX,%String
     %ENDIF
     CALL GetLength$@RT::
       GetLength$@RT:: PROC1
          PUSH AX,DI
            MOV DI,CX
            XOR AX,AX
            XOR CX,CX
            CLD
            DEC CX
            REPNE SCASB
            NOT CX
            DEC CX
         POP DI,AX
         RET
       ENDPROC1 GetLength$@RT::
   %ENDMACRO GetLength$

↑ Concat$ Destination, Size=, Source1, Source2,,,

Macro will concatenate one or more zero-terminated source strings to a destination string.
All source strings must be addressable with DS. Destination must be addressable with ES.

Input

Destination is a pointer to an output buffer where the result of concatenation will be stored as zero-terminated byte string.
ES= is segment of Destination string.
Size= is the size allocated for the output destination buffer including the zero terminator. By default it is set to SIZE# %Destination.
Source* operands are pointers to the ASCIIZ strings which are being concatenated. The first one (Source1) may be identical with the destination, when we need to append something to an existing string.
DS= is segment of all Source strings.

Output

CF=0, Destination string is filled with concatenation of all Source strings. Registers are preserved.
CF=1 when the allocated Size= is not long enough. The output buffer Size is never exceeded.

Example

Concat$ FullName$,Path$,FileName$,=B".htm"

Concat$ %MACRO Destination, Source,,, Size=
   %IF %# < 2                                                                      ; >
     %ERROR ID=5930, 'Missing operand of macro "Concat$".'
     %EXITMACRO Concat$
   %ENDIF
   PUSH BP
     MOV BP,SP ; Store stack pointer.
     ArgNr %FOR %#..2, STEP= -1
       PUSHW %*{%ArgNr} ; All Source pointers, starting with the last.
     %ENDFOR ArgNr
     PUSHW %# - 1  ; Number of Source strings to concatenate.
     %IF "%Size" === ""
       PUSHW SIZE# %Destination
     %ELSE
        PUSHW %Size
     %ENDIF
     PUSHW %Destination
     CALL Concat$@RT::
     MOV SP,BP ; Restore stack.
   POP BP
Concat$@RT:: PROC1
     PUSHAW
       MOV BP,SP
       MOV DI,[BP+18] ; %Destination.
       MOV DX,[BP+20] ; %Size.
       MOV CX,[BP+22] ; Number of source strings.
       ADD DX,DI
       CLD
       DEC DX ; End of allocated Destination.
 .20:  MOV SI,[BP+24] ; Source pointer.
 .30:  LODSB
       CMP AL,0
       JE .40:
       CMP DI,DX
       CMC
       JC .80: ; If destination size overflowed.
       STOSB
       JMP .30:
 .40:  INC BP,BP ; The next Source pointer.
       LOOP .20:
 .80:  MOV AL,0
       STOSB
     POPAW
     RET
   ENDP1 Concat$@RT::
 %ENDMACRO Concat$

↑ Compare$ String1, String2

Compare two zero-terminated strings.

Input

String1 is pointer to the first zero-terminated string. SI is assumed when the 1st operand is empty.
DS= is segment of the first string.
String2 is pointer to the second zero-terminated string. DI is assumed when the 2nd operand is empty.
ES= is segment of the second string.

Output

ZF=1 if both string are identical,
ZF=0 otherwise.

Compare$ %MACRO String1, String2
   %IF "%String2" === ""
     PUSHW DI
   %ELSE
     PUSHW %String2
   %ENDIF
   %IF "%String1" === ""
     PUSHW SI
   %ELSE
     PUSHW %String1
   %ENDIF
   CALL Compare$@RT::
Compare$@RT:: PROC1
       PUSHAW
         MOV BP,SP
         SUB AX,AX
         MOV DI,[BP+20] ; %String2.
         MOV CX,-1
         MOV BX,DI
         CLD
         REPNE:SCASB    ; Search for the terminator of string ES:DI.
         SUB DI,BX      ; Size of String2 including the NUL.
         MOV DX,DI
         MOV DI,[BP+18] ; %String1$.
         MOV CX,-1
         MOV SI,DI
         PUSH ES,DS
          POP ES        ; Temporarily load ES from DS.
          REPNE:SCASB   ; Search for the terminator of string DS:SI.
         POP ES
         MOV CX,DI
         SUB CX,SI      ; Size of %String1 including the NUL.
         CMP CX,DX      ; Compare string sizes.
         JNE .90        ; If sizes do not match.
         MOV DI,BX      ; String1$.
         REPE CMPSB
   .90:POPAW
       RET 2*2
      ENDPROC1 Compare$@RT::
  %ENDMACRO Compare$

DecodeUTF8 Source, CallbackProc, Size=-1, Width=16

Macro DecodeUTF8 converts Source UTF-8 string to UTF-16 or UTF-32 string in 16bit CPU mode. It requires CPU 386 or higher.
Source string is either zero-terminated, or its Size= must be specified. Conversion stops at NUL byte, which is not converted to output. Input never reads beyond Source+Size.

If Byte Order Mark (BOM, 0xEF,0xBB,0xBF ) is detected at the beginning of the Source string, it is ignored.

Invalid UTF-8 sequence will send a replacement character 0xFFFD � to the output.

Byte order in output encoding is always LittleEndian, the same which is used in MS Windows WIDE functions.

If you want to produce UTF-16BE, perform XCHG AL,AH in CallbackProc.
If you want to produce UTF-32BE, perform BSWAP EAX in CallbackProc.
If you want to prefix the output string with BOM, store it to destination buffer before invoking DecodeUTF8.
If you don't like replacement characters (usually displayed as little squares �), filter them out in CallbackProc.

Documented

[UTF8], [UTF16], [UTF32]

Input

ES= is segment of UTF8-encoded string.
Source is offset of the first byte of UTF8-encoded string.
Size= -1 is the maximal possible size of source string in bytes. It may be left on default when the Source string is terminated with NUL byte (this NUL is not written to output).
If Size is not -1, exactly that many input bytes are decoded, including NUL bytes.
Width=16 or 32 specifies the output encoding UTF-16 or UTF-32, respectively.
CallbackProc is offset in CS segment of the procedure which stores one converted character.

CallbackProc

is called with register calling convention.
It is expected to store the UTF-16 or UTF-32 character obtained in EAX and return with CF=0.

Input

CF=DF=0
EAX= contains one converted character encoded in UTF-16 or UTF-32, which should be stored.
EAX may contain surrogate code 0x0000_D800..0x0000_DFFF when the input UTF-8 character belongs to Unicode supplementary planes (Emoji, Asian characters etc).
EAX may also contain the replacement 0x0000_FFFD when the input UTF-8 string is malformed.
BP= original value of BP on input to the macro. Usually it is the frame pointer of the function which expanded DecodeUTF8, thus arguments and local variables of the function can be used in CallbackProc.
DI= original value of DI on input to the macro. It may be used in STOS, incremented value of DI will be supplied on the next invocation of CallbackProc.
BX,CX,DX,SI should be considered undefined.

Output

CF=0 when a character from EAX was successfully stored to the destination buffer by CallbackProc and macro should continue with parsing the next UTF-8 characters.
CF=1 signalizes that the macro should cancel further conversion. CF propagates to the output of DecodeUTF8.
AX,BX,CX,DX,BP,SI,DI,ES may be changed in Callback procedure. The value of
DI will be saved and provided in the next call of CallbackProc, thus DI can be used in CallbackProc as an output pointer for decoded data.

Output

CF=0, CX= number of unprocessed bytes at the end of text (0..3) due to incompleteness of the last UTF-8 character in input text block. The caller of DecodeUTF8 should LSEEK the input file by CX bytes back before reading the next block of text.
DI= as returned from CallbackProc.
All other registers are preserved.

Error

CF=1 when CallbackProc returned CF to the macro.
CX=0, other registers are preserved.

Example

MOV DI,DestString ; It should be long enough for the decoded string. MOV AX,PARA# SourceString MOV ES,AX DecodeUTF8 SourceString, Store: Store:PROC1 ; Thanks to using PROC1 instead of PROC it doesn't need bypass by JMP. STOSW ; Store UTF-16 character from AX and advance DI to the next free room. RET ; Return to DecodeUTF-8 macro with CF=0. ENDPROC1 Store: SUB AX,AX STOSW ; Zero-terminate DestString.

DecodeUTF8 %MACRO Source, CallbackProc, Size=-1, Width=16
    %IF %Width != 16 && %Width != 32
      %ERROR ID=5932,'Macro "DecodeUTF8" requires Width=16 or Width=32.'
      %EXITMACRO DecodeUTF8
    %ENDIF
    PUSHW %Width, %Size, %CallbackProc, %Source
    CALL DecodeUTF8@RT::
DecodeUTF8@RT:: PROC1
    PUSHAW
    MOV BP,SP
    PUSH ES
      SUB CX,CX
      MOV [BP+12],CX  ; Initialize %ReturnCX to 0.
      MOV DI,[BP+18]  ; %Source.
      MOV CX,[BP+22]  ; %Size.
      MOV SI,DI
      MOV AX,CX
      INC AX
      JZ .Scan:       ; If Size=-1, AX=0 and the Source size will be scanned.
      ADD DI,CX       ; Otherwise use the explicit %Size.
      JNC .No0:
      MOV DI,-1
      JMP .No0:
.Scan:REPNE:SCASB
      JNE .No0:
      DEC DI            ; Omit the terminator from conversion.
.No0: ; Source string without NUL is now at ES:SI..ES:DI.
 BOM  %FOR 0xEF,0xBB,0xBF ; Little-Endian BOM (0xFEFF) encoded in UTF-8.
        CMP SI,DI
        JNB .NoBOM:
        LODSB [ES:SI]
        CMP AL,%BOM
        JNE .NoBOM:
      %ENDFOR BOM
      JMP .Start:     ; BOM was detected, SI is advanced just behind it.
.NoBOM:MOV SI,[BP+18] ; No BOM detected, restore SI to the start of Source again.
.Start:CMP SI,DI      ; Decode one UTF8 character from the string ES:SI..ES:DI.
      JNB .End:
      XOR EBX,EBX
      LODSB [ES:SI]
      MOV BL,AL
      NOT BL
      BSR CX,BX   ; Scan bits 7..0 of inverted first byte of 1,2,3,4 bytes long UTF-8 character.
      MOV BL,AL   ; First byte of 1,2,3,4 bytes long UTF-8 character (not inverted).
      MOV DL,0x7F ; Prepare mask for payload bits in the 1st UTF-8 byte.
      SUB CX,7    ; CX=7,5,4,3 change to CX=0,-2,-3,-4.
      JZ .Out:    ; Done when BX is codepoint 0..0x7F (7bit ASCII character).
      NEG CX      ; ECX=2,3,4 (number of bytes in UTF-8 character).
      SHR DL,CL   ; DL=0x1F,0x0F,0x07 is the payload mask of the first UTF-8 byte.
      AND BL,DL   ; EBX will accumulate payload bits of codepoint.
      CMP CL,2
      JB .Invalid:
      CMP CL,4
      JBE .Good:
.Invalid:
      MOV EAX,0xFFFD ; Invalid UTF-8 detected, output the replacement instead.
      JMP .NoSg:
.Good:DEC CX         ; CX=1, 2 or 3 continuation bytes 10xxxxxxb expected.
      MOV AX,SI      ; Check if there's that many input bytes left.
      ADD AX,CX
      CMP AX,DI
      JBE .Cont:
      DEC SI         ; Rollback, the last UTF-8 character is incomplete.
      SUB DI,SI      ; DI characters (1..3) were not decoded.
      MOV [BP+12],DI ; %ReturnCX.
      JMP .End:      ; CF=0.
.Cont:LODSB [ES:SI]  ; Continuation byte AL=10xxxxxxb expected.
      BTR AX,7       ; Reset the marker bit 7.
      JNC .Invalid:
      BTR AX,6
      JC .Invalid:
      SHL EBX,6      ; Make room in EBX for the next 6 bits.
      OR  BL,AL      ; Accumulate them.
      DEC CX
      JNZ .Cont:
.Out: MOV EAX,EBX ; EAX=EBX is now the decoded codepoint 0..0x10_FFFF.
      ; Check for overlong encodings. DL=0x7F,0x1F,0x0F,0x07 for 1,2,3,4 bytes in UTF-8 character.
      CMP EBX,0x01_0000   ; Codepoint 0x01_0000..0x10_FFFF should be encoded in 4 bytes.
      JAE .NoOverlong:
      CMP BX,0x00_0800    ; Codepoint 0x00_0800..0x00_FFFF should be encoded in 3 bytes.
      JB .2Bts:
      CMP DL,0x0F
      JE  .NoOverlong:
      JMP .Invalid:
.2Bts:CMP BX,0x00_0080    ; Codepoint 0x00_0080..0x00_07FF should be encoded in 2 bytes.
      JB .1Bts:
      CMP DL,0x1F
      JE .NoOverlong:
      JMP .Invalid:
.1Bts:CMP DL,0x7F         ; Codepoint 0x00_0000..0x00_007F should be encoded in 1 byte.
      JE .NoOverlong:
      TEST BX
      JNZ .Invalid:
      CMP DL,0x1F         ; Exception: codepoint 0 may be encoded in 1 or 2 bytes.
      JNE .Invalid:
.NoOverlong:
      SHR EBX,11          ; Check for surrogate codepoints.
      CMP BL,0x1B
      JE .Invalid:            ; Do not accept surrogates 0xD800..0xDFFF from input.
      TEST BX,0x3E0
      JZ .NoSg:           ; If codepoint EAX is below 0x0001_0000, surrogates do not apply.
      CMPW [BP+24],16     ; Output UTF %Width (16 or 32).
      JNE .NoSg:          ; UTF-32 does not need surrogates.
      SUB EAX,0x0001_0000 ; Codepoint EAX was not encodable in one UTF-16 character.
      MOV EBX,0x0000_03FF ; Use two surrogate Unichars.
      AND EBX,EAX
      SHR EAX,10
      ADD EBX,0x0000_DC00 ; EBX is now low surrogate.
      ADD EAX,0x0000_D800 ; EAX is now high surrogate.
      CALL .OutEAX:       ; High surrogate first.
      MOV EAX,EBX         ; Low surrogate.
      JC .End:            ; If aborted by CallbackProc.
.NoSg:CALL .OutEAX:       ; Low surrogate or BMP codepoint or UTF-32.
      JNC .Start:         ; Parse the next UTF-8 character from string ES:SI..ES:DI.
.End:POP ES
     POPAW
     RET 2*4
.OutEAX:PROC              ; Send EAX to callback. Preserves EBX,ES,SI,DI, updates %ReturnDI.
        PUSH EBX,ES,SI,DI
          MOV DI,[BP+0]   ; Restore %ReturnDI from DecodeUTF8@RT frame.
          PUSH BP
            CALL [BP+20]  ; %CallbackProc.
          POP BP
          MOV [BP+0],DI   ; Update %ReturnDI from DecodeUTF8@RT frame.
        POP DI,SI,ES,EBX
        RET
      ENDPROC .OutEAX:
   ENDP1 DecodeUTF8@RT::
 %ENDMACRO DecodeUTF8

  ENDHEAD string16