2

我有兴趣了解如何在没有像 unix 系统调用这样的高级函数的帮助下编译/创建一种非常简单的语言(即 Brainfuck)。我想在一些依赖于 CPU 的低级程序集中为该语言编写一个编译器,以便我可以提供简单语言的源代码并最终得到一个二进制文件。不确定这是否清楚,但基本问题是如何在没有硬件中不存在的任何帮助的情况下将源代码转换为二进制文件。

编辑:更简洁的问题陈述......

给定:

-硬件(主板/CPU等)

没有给:

-UNIX/DOS

-C/FORTRAN/任何其他语言

我将如何实现像 Brainfuck 这样的简单语言?

我知道有更多实用的编译方法,但出于教育目的,我对此很感兴趣。

抱歉,如果这个问题是多余的或显而易见的——我不是计算机科学家,所以也许我只是不知道在网上找到解决问题的正确词汇。如果有人可以提供有关该主题的链接或文本,将不胜感激。

4

4 回答 4

1

查看维基百科上的描述,这不是一项艰巨的任务。我仍然可能会以您知道的某种语言开始,也许喜欢,也许不是。C是一个不错的选择。文件 I/O 是一个小型或大型项目,具体取决于平台等。稍后担心,在语言的“源”中编译。对于该来源中的每个角色,执行任务

> ++ptr;
< --ptr;
+ ++*ptr;
etc

然后将其转换为组装。您只需要一个寄存器来保存 ptr,几行 asm 来初始化数组/ram 并将寄存器/ptr 设置为开头。另一个寄存器遍历源代码。你只是在寻找 8 个字符,你可以 if-then-else 遍历这些字符,除非有一些模式可以让它们更容易处理。如果你愿意,你可以制作一个 256 字节的查找表,并将其用作该指令的处理程序的地址,或者将它们转换为 0-7 的整数,然后在跳转表中使用它。

那是一个解析器,不一定是编译器。我会用 C 或某种高级语言编写编译器,它接收一个字节数组,即程序,对于每条输出实现该指令的 asm 源代码的指令,输入、输出(使用 ARM asm)

add r0,#1

减号

ldr r1,[r0]
sub r1,#1
str r1,[r0]

r0 是 ptr 寄存器,而 r1 只是帮忙。

如果你真的反对使用像 printf 这样的调用,那么让这段代码的输出成为一个字节数组,这些字节是 asm 源的 ascii 输出每个字符 a、d、d、空格、r、0、逗号、#、 1、cr、lf等。在 asm 和一些高级语言中相当容易实现。如果您想直接转为二进制,则只需输出机器代码,甚至更容易。

将源字符串输入此编译器并将输出输入到稍后可以执行的某个文件中可能会进行系统调用。如果您在同一平台上运行,则可以避免输出成为文件,并且可以在某种意义上执行自修改代码,即您在某个地址构建机器代码,然后在完成解析后跳转到该地址执行。

写这个答案比用 C 或 asm 实现解决方案花费的时间要长很多倍。你遇到的具体困难是什么?

于 2012-05-08T04:02:11.760 回答
1

您可以很容易地将 Brainfuck 源代码编译到 DOS .COM 应用程序中(您还需要 NASM 或一些额外的代码来发出指令操作码和计算跳转)。下面是一个稍微修改过的 bf 解释器,变成了某种编译器:

// file: bfcompil.c

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

#define MAX_CODE_SIZE 30000

char code[MAX_CODE_SIZE];
char* pc = &code[0];
char* pcEnd = &code[0];

#define MAX_DATA_SIZE 30000

char data[MAX_DATA_SIZE] = { 0 };
char* pd = &data[0];

// Structures for quick bracket matching
unsigned brStack[MAX_CODE_SIZE];
unsigned brSptr = 0;
unsigned brMatch[MAX_CODE_SIZE];

int main(int argc, char** argv)
{
  FILE* f = NULL;
  int ch;

  if (argc != 2)
  {
    fprintf(stderr, "usage:\n  bfcompil <brainfuck-source-code-file>\n"
                    "bfcompil will output NASM-compilable source code for"
                    "a DOS program\n");
    return EXIT_FAILURE;
  }

  if ((f = fopen(argv[1], "rb")) == NULL)
  {
    fprintf(stderr, "can't open file \"%s\" for reading\n", argv[1]);
    return EXIT_FAILURE;
  }

  while ((ch = getc(f)) != EOF)
  {
    if (strchr(" \t\r\n", ch) != NULL) // skip white space
    {
      continue;
    }
    else if (strchr("><+-.,[]", ch) != NULL) // store valid commands
    {
      if (pcEnd >= &code[sizeof(code)])
      {
        fprintf(stderr, "too many commands in file \"%s\", expected at most "
                        "%u commands\n", argv[1], (unsigned)sizeof(code));
        fclose(f);
        return EXIT_FAILURE;
      }

      if (ch == '[')
      {
        brStack[brSptr++] = (unsigned)(pcEnd - &code[0]);
      }
      else if (ch == ']')
      {
        if (brSptr == 0)
        {
          fprintf(stderr, "unmatched ']' in file \"%s\"\n", argv[1]);
          fclose(f);
          return EXIT_FAILURE;
        }

        brSptr--;
        brMatch[brStack[brSptr]] = (unsigned)(pcEnd - &code[0]);
        brMatch[pcEnd - &code[0]] = brStack[brSptr];
      }

      *pcEnd++ = ch;
    }
    else // fail on invalid commands
    {
      fprintf(stderr, "unexpected character '%c' in file \"%s\", valid command "
                      "set is: \"><+-.,[]\"\n", ch, argv[1]);
      fclose(f);
      return EXIT_FAILURE;
    }
  }

  fclose(f);

  if (brSptr != 0)
  {
    fprintf(stderr, "unmatched '[' in file \"%s\"\n", argv[1]);
    return EXIT_FAILURE;
  }

  if (pcEnd == &code[0])
  {
    fprintf(stderr, "no commands found in file \"%s\"\n", argv[1]);
    return EXIT_FAILURE;
  }

  printf("; how to compile: nasm -f bin <input file with this code.asm> -o "
         "<output executable.com>\n\n"
         "org 0x100\n"
         "bits 16\n\n"
         "    mov     bx, data\n"
         "    mov     di, bx\n"
         "    mov     cx, 30000\n"
         "    xor     al, al\n"
         "    cld\n"
         "    rep     stosb\n\n"
         "    jmp     code\n\n"
         "print:\n"
         "    mov     ah, 2\n"
         "    cmp     byte [bx], 10\n"
         "    jne     lprint1\n"
         "    mov     dl, 13\n"
         "    int     0x21\n"
         "lprint1:\n"
         "    mov     dl, [bx]\n"
         "    int     0x21\n"
         "    ret\n\n"
#if 01
         // buffered input
         "input:\n"
         "    cmp     byte [kbdbuf+1], 0\n"
         "    jne     linput1\n"
         "    mov     ah, 0xa\n"
         "    mov     dx, kbdbuf\n"
         "    int     0x21\n"
         "    inc     byte [kbdbuf+1]\n"
         "linput1:\n"
         "    mov     al, [kbdbuf+2]\n"
         "    cmp     al, 13\n"
         "    jne     linput4\n"
         "    mov     al, 10\n"
         "linput4:\n"
         "    mov     [bx], al\n"
         "    mov     si, kbdbuf+3\n"
         "    mov     di, kbdbuf+2\n"
         "    xor     cx, cx\n"
         "    dec     byte [kbdbuf+1]\n"
         "    mov     cl, [kbdbuf+1]\n"
         "    jz      linput3\n"
         "linput2:\n"
         "    lodsb\n"
         "    stosb\n"
         "    loop    linput2\n"
         "linput3:\n"
         "    ret\n\n"
#else
         // unbuffered input
         "input:\n"
         "    mov     ah, 1\n"
         "    int     0x21\n"
         "    cmp     al, 13\n"
         "    jne     linput\n"
         "    mov     al, 10\n"
         "linput:\n"
         "    mov     [bx], al\n"
         "    ret\n\n"
#endif
         "code:\n\n");

  for (pc = &code[0]; pc < pcEnd; pc++)
  {
    switch (*pc)
    {
    case '>':
      printf("    inc     bx\n");
      break;
    case '<':
      printf("    dec     bx\n");
      break;
    case '+':
      printf("    inc     byte [bx]\n");
      break;
    case '-':
      printf("    dec     byte [bx]\n");
      break;
    case '.':
      printf("    call    print\n");
      break;
    case ',':
      printf("    call    input\n");
      break;
    case '[':
      printf("label%u:\n", (unsigned)(pc - &code[0]));
      printf("    cmp     byte [bx], 0\n");
      printf("    je      label%u\n", (unsigned)brMatch[pc - &code[0]]);
      break;
    case ']':
      printf("    jmp     label%u\n", brMatch[pc - &code[0]]);
      printf("label%u:\n", (unsigned)(pc - &code[0]));
      break;
    }
  }

  printf("\n    ret\n\n");
  printf("kbdbuf:\n"
         "    db      254\n"
         "    db      0\n"
         "    times   256 db 0\n\n");
  printf("data:\n");

  return EXIT_SUCCESS;
}

如果你喂它 hello world 程序:

++++++++++[>+++++++>++++++++++>+++>+<<<<-]>++.>+.+++++++..+++.>++.<<+++++++++++++++.>.+++.------.--------.>+.>.

它将产生可编译的汇编代码:

; how to compile: nasm -f bin <input file with this code.asm> -o <output executable.com>

org 0x100
bits 16

    mov     bx, data
    mov     di, bx
    mov     cx, 30000
    xor     al, al
    cld
    rep     stosb

    jmp     code

print:
    mov     ah, 2
    cmp     byte [bx], 10
    jne     lprint1
    mov     dl, 13
    int     0x21
lprint1:
    mov     dl, [bx]
    int     0x21
    ret

input:
    cmp     byte [kbdbuf+1], 0
    jne     linput1
    mov     ah, 0xa
    mov     dx, kbdbuf
    int     0x21
    inc     byte [kbdbuf+1]
linput1:
    mov     al, [kbdbuf+2]
    cmp     al, 13
    jne     linput4
    mov     al, 10
linput4:
    mov     [bx], al
    mov     si, kbdbuf+3
    mov     di, kbdbuf+2
    xor     cx, cx
    dec     byte [kbdbuf+1]
    mov     cl, [kbdbuf+1]
    jz      linput3
linput2:
    lodsb
    stosb
    loop    linput2
linput3:
    ret

code:

    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
label10:
    cmp     byte [bx], 0
    je      label41
    inc     bx
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     bx
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     bx
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     bx
    inc     byte [bx]
    dec     bx
    dec     bx
    dec     bx
    dec     bx
    dec     byte [bx]
    jmp     label10
label41:
    inc     bx
    inc     byte [bx]
    inc     byte [bx]
    call    print
    inc     bx
    inc     byte [bx]
    call    print
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    call    print
    call    print
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    call    print
    inc     bx
    inc     byte [bx]
    inc     byte [bx]
    call    print
    dec     bx
    dec     bx
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    call    print
    inc     bx
    call    print
    inc     byte [bx]
    inc     byte [bx]
    inc     byte [bx]
    call    print
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    call    print
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    dec     byte [bx]
    call    print
    inc     bx
    inc     byte [bx]
    call    print
    inc     bx
    call    print

    ret

kbdbuf:
    db      254
    db      0
    times   256 db 0

data:

如果您编译它,您将能够在 DOS、Windows 9x/XP(可能是 32 位 Vista/7)和 DosBox 中运行它。

不出所料,输出是:

Hello World!

更新:上述代码中基于 DOS 的输入和输出例程可以被直接访问屏幕缓冲区和键盘端口所取代。键盘代码还需要处理键盘中断。在 x86 PC 上做起来并不难。你真的可以为一种语言实现一个编译器,让它在没有操作系统的裸硬件上运行。

您还应该看看Forth这正是给定环境的那种语言。而且很容易实施。比 C 容易得多。比 Brainfuck 难,在某种程度上可以与组装相媲美。

更新 2:这是一个不使用任何 DOS 或 BIOS 功能的小型(约 1KB 大小)brainfuck 解释器:

; file: bfint.asm
; compile: nasm.exe -f bin bfint.asm -o bfint.com
; run in: DOS, DosBox or equivalent

bits 16
org 0x100

section .text

SCREEN_WIDTH   equ 80
SCREEN_HEIGHT  equ 25
SCAN_BUF_SIZE  equ 256

MAX_CODE_SIZE  equ 20000
MAX_DATA_SIZE  equ 30000

    cld

    ; set new keyboard (IRQ1) ISR
    push    byte 0
    pop     es
    cli                         ; update ISR address w/ ints disabled
    mov     word [es:9*4], Irq1Isr
    mov     [es:9*4+2], cs
    sti

    push    cs
    pop     es

Restart:

    call    ClearScreen
    mov     si, MsgHello
    call    PrintStr

    mov     word [CodeSize], 0
    mov     byte [EnterCount], 0

WaitForKey:
    call    GetKey

    ; Escape erases code
    cmp     ah, 1      ; Escape
    je      Restart

    ; Non-characters are ignored
    cmp     al, 0      ; non-character key
    je      WaitForKey

    ; Enter is "printed" but not stored, use for formatting
    cmp     al, 10     ; Enter
    je      KeyEnter
    mov     byte [EnterCount], 0

    ; Backspace deletes last character
    cmp     al, 8      ; Backspace
    je      KeyBackspace

    ; Space is printed but not stored, use for formatting
    cmp     al, " "    ; Space
    je      PrintOnly

    ; 0 runs a test program
    cmp     al, "0"
    je      TestProgram

    ; Other chracters are stored as code
    mov     bx, [CodeSize]
    cmp     bx, MAX_CODE_SIZE
    jae     ErrCodeTooBig
    mov     [Code + bx], al
    inc     word [CodeSize]
PrintOnly:
    call    PrintChar
    jmp     WaitForKey

ErrCodeTooBig:
    mov     si, MsgCodeTooBig
    call    PrintStr
    mov     word [CodeSize], 0
    jmp     WaitForKey

KeyEnter:
    call    PrintChar
    inc     byte [EnterCount]
    cmp     byte [EnterCount], 1
    je      WaitForKey
    mov     byte [EnterCount], 0
    call    Execute
    jmp     WaitForKey

KeyBackspace:
    call    PrintChar
    cmp     word [CodeSize], 0
    je      WaitForKey
    dec     word [CodeSize]
    jmp     WaitForKey

TestProgram:
    mov     si, TestCode
    mov     di, Code
    mov     cx, TestCodeEnd - TestCode
    mov     [CodeSize], cx
    rep     movsb
    call    Execute
    jmp     WaitForKey

Execute:
    mov     si, Code ; code start
    xor     bp, bp ; instruction index

    mov     di, Data ; data start
    mov     cx, MAX_DATA_SIZE
    xor     al, al
    rep     stosb
    sub     di, MAX_DATA_SIZE
    xor     bx, bx ; data index

ExecuteLoop:
    cmp     bp, [CodeSize]
    jae     ExecuteDone

    mov     al, [bp+si]
    cmp     al, ">"
    je      IncPtr
    cmp     al, "<"
    je      DecPtr
    cmp     al, "+"
    je      IncData
    cmp     al, "-"
    je      DecData
    cmp     al, "."
    je      PrintData
    cmp     al, ","
    je      InputData
    cmp     al, "["
    je      While
    cmp     al, "]"
    je      EndWhile

    mov     si, MsgInvalidChar
    call    PrintStr
    call    PrintChar
    mov     al, 10
    call    PrintChar
    jmp     ExecuteDone

IncPtr:
    inc     bx
    jmp     ExecuteContinue

DecPtr:
    dec     bx
    jmp     ExecuteContinue

IncData:
    inc     byte [bx+di]
    jmp     ExecuteContinue

DecData:
    dec     byte [bx+di]
    jmp     ExecuteContinue

PrintData:
    mov     al, [bx+di]
    call    PrintChar
    jmp     ExecuteContinue

InputData:
    call    GetKey
    or      al, al
    jz      InputData
    mov     [bx+di], al
    jmp     ExecuteContinue

While:
    cmp     byte [bx+di], 0
    jne     ExecuteContinue
    mov     ax, 1
    mov     dx, "[]"
    call    FindMatchingBracket
ExecuteContinue:
    inc     bp
    jmp     ExecuteLoop

EndWhile:
    mov     ax, -1
    mov     dx, "]["
    call    FindMatchingBracket
    jmp     ExecuteLoop

ExecuteDone:
    mov     word [CodeSize], 0
    mov     si, MsgCompleted
    jmp     PrintStr

FindMatchingBracket:
    xor     cx, cx
FindMatchingBracket1:
    cmp     byte [bp+si], dl
    jne     FindMatchingBracket2
    inc     cx
    jmp     FindMatchingBracket3
FindMatchingBracket2:
    cmp     byte [bp+si], dh
    jne     FindMatchingBracket3
    dec     cx
    jnz     FindMatchingBracket3
    ret
FindMatchingBracket3:
    add     bp, ax
    jmp     FindMatchingBracket1

; Inputs:
; AL = ASCII character code
PrintChar:
    ; assuming it's a color text mode (not monochrome or graphics)
    pusha
    push    es

    push    word 0xb800
    pop     es
    mov     bx, [CursorPos]

    cmp     al, 8
    je      PrintCharBackSpace

    cmp     al, 10
    je      PrintCharBackLF

    cmp     al, 13
    je      PrintCharBackCR

    mov     [es:bx], al
    call    AdvanceCursorPosition

    jmp     PrintCharDone

PrintCharBackSpace:
    ; move the cursor back and erase the last character
    or      bx, bx
    jz      PrintCharDone
    dec     bx
    dec     bx
    mov     word [es:bx], 0x0720
    jmp     PrintCharSetCursorPos

PrintCharBackLF:
    ; move the cursor to the beginning of the next line - '\n' behavior
    add     bx, SCREEN_WIDTH * 2
    cmp     bx, SCREEN_WIDTH * SCREEN_HEIGHT * 2
    jc      PrintCharBackCR
    sub     bx, SCREEN_WIDTH * 2
    call    ScrollUp

PrintCharBackCR:
    ; move the cursor to the beginning of the current line - '\r' behavior
    mov     ax, SCREEN_WIDTH * 2
    xchg    ax, bx
    xor     dx, dx
    div     bx
    mul     bx
    mov     bx, ax

PrintCharSetCursorPos:
    mov     [CursorPos], bx
    shr     bx, 1
    call    SetCursorPosition

PrintCharDone:
PopEsAllRet:
    pop     es
    popa
    ret

ClearScreen:
    ; assuming it's a color text mode (not monochrome or graphics)
    pusha
    push    es

    push    word 0xb800
    pop     es
    xor     di, di
    mov     cx, SCREEN_WIDTH * SCREEN_HEIGHT
    mov     ax, 0x0720 ; character = space, color = lightgray on black
    rep     stosw

    xor     bx, bx
    mov     [CursorPos], bx
    call    SetCursorPosition

    jmp     PopEsAllRet

ScrollUp:
    ; assuming it's a color text mode (not monochrome or graphics)
    pusha
    push    es
    push    ds

    push    word 0xb800
    pop     es
    push    es
    pop     ds
    mov     si, SCREEN_WIDTH * 2
    xor     di, di
    mov     cx, SCREEN_WIDTH * (SCREEN_HEIGHT - 1)
    rep     movsw

    mov     cx, SCREEN_WIDTH
    mov     ax, 0x0720 ; character = space, color = lightgray on black
    rep     stosw

    pop     ds
    jmp     PopEsAllRet

; Inputs:
; DS:SI = address of NUL-terminated ASCII string
PrintStr:
    pusha
PrintStr1:
    lodsb
    or      al, al
    jz      PrintStrDone
    call    PrintChar
    jmp     PrintStr1
PrintStrDone:
    popa
    ret

; Inputs:
; BX = Y * SCREEN_WIDTH + X
SetCursorPosition:
    ; assuming it's a color text mode (not monochrome or graphics)
    pusha

%if 0
    mov     dx, 0x3d4
    mov     al, 0x0f
    out     dx, al
    inc     dx
    mov     al, bl
    out     dx, al

    dec     dx
    mov     al, 0x0e
    out     dx, al
    inc     dx
    mov     al, bh
    out     dx, al
%else
    mov     dx, 0x3d4
    mov     al, 0x0f
    mov     ah, bl
    out     dx, ax

    dec     al
    mov     ah, bh
    out     dx, ax
%endif

    popa
    ret

AdvanceCursorPosition:
    ; assuming it's a color text mode (not monochrome or graphics)
    pusha

    mov     ax, [CursorPos]
    inc     ax
    inc     ax
    cmp     ax, SCREEN_WIDTH * SCREEN_HEIGHT * 2
    jc      AdvanceCursorPosition1

    sub     ax, SCREEN_WIDTH * 2
    call    ScrollUp

AdvanceCursorPosition1:
    mov     [CursorPos], ax
    shr     ax, 1
    xchg    ax, bx
    call    SetCursorPosition

    popa
    ret

; Outputs:
; AH = scan code
; AL = character
GetKey:
    push    bx
    push    si

GetKeyRepeat:
    mov     ax, [ScanWriteIdx]
    mov     si, [ScanReadIdx]
    sub     ax, si
    jz      GetKeyRepeat
    mov     bx, si
    mov     ax, [ScanBuf + bx + si]
    inc     si
    and     si, SCAN_BUF_SIZE - 1
    mov     [ScanReadIdx], si

    pop     si
    pop     bx
    ret

Irq1Isr:
    pusha
    push    ds

    push    cs
    pop     ds

    ; read keyboard scan code
    in      al, 0x60

    cmp     al, 0x2a ; Left Shift down
    jne     Irq1Isr1
    or      byte [Shift], 1
Irq1Isr1:
    cmp     al, 0x36 ; Right Shift down
    jne     Irq1Isr2
    or      byte [Shift], 2
Irq1Isr2:
    cmp     al, 0xaa ; Left Shift up
    jne     Irq1Isr3
    and     byte [Shift], ~1
Irq1Isr3:
    cmp     al, 0xb6 ; Right Shift up
    jne     Irq1Isr4
    and     byte [Shift], ~2
Irq1Isr4:

    test    al, 0x80
    jnz     Irq1IsrEois ; key released

    mov     ah, al
    cmp     al, 58
    jc      Irq1Isr5
    xor     al, al   ; don't translate non-character keys
    jmp     Irq1Isr7
Irq1Isr5:
    mov     bx, ScanToChar
    cmp     byte [Shift], 0
    je      Irq1Isr6
    add     bx, ScanToCharShift - ScanToChar
Irq1Isr6:
    xlatb

Irq1Isr7:
    mov     bx, [ScanWriteIdx]
    mov     di, bx
    mov     [ScanBuf + bx + di], ax
    inc     bx
    and     bx, SCAN_BUF_SIZE - 1
    mov     [ScanWriteIdx], bx

Irq1IsrEois:
%if 0
    ; send EOI to XT keyboard
    in      al, 0x61
    mov     ah, al
    or      al, 0x80
    out     0x61, al
    mov     al, ah
    out     0x61, al
%endif

    ; send EOI to master PIC
    mov     al, 0x20
    out     0x20, al

    pop     ds
    popa
    iret

ScanToChar:
    db      0 ; unused
    db      0 ; Escape
    db      "1234567890-="
    db      8 ; Backspace
    db      9 ; Tab
    db      "qwertyuiop[]"
    db      10 ; Enter
    db      0 ; Ctrl
    db      "asdfghjkl;'`"
    db      0 ; Left Shift
    db      "\zxcvbnm,./"
    db      0 ; Right Shift
    db      0 ; Print Screen
    db      0 ; Alt
    db      " " ; Space
ScanToCharShift:
    db      0 ; unused
    db      0 ; Escape
    db      "!@#$%^&*()_+"
    db      8 ; Backspace
    db      9 ; Tab
    db      "QWERTYUIOP{}"
    db      10 ; Enter
    db      0 ; Ctrl
    db      'ASDFGHJKL:"~'
    db      0 ; Left Shift
    db      "|ZXCVBNM<>?"
    db      0 ; Right Shift
    db      0 ; Print Screen
    db      0 ; Alt
    db      " " ; Space

MsgHello:
    db      "Brainfuck Interpreter", 10, 10
    db      "Press 0 to run test code OR", 10
    db      "Type your code.", 10
    db      "Use Esc to erase it all or Backspace to delete last character.", 10
    db      "Press Enter twice to run it.", 10, 10, 0

MsgCodeTooBig:
    db      10, "Code's too big", 10, 0

MsgCompleted:
    db      10, "Code's completed", 10, 0

MsgInvalidChar:
    db      10, "Invalid character: ", 0

Shift           db      0

CursorPos       dw      0

ScanReadIdx     dw      0
ScanWriteIdx    dw      0

EnterCount      db      0

CodeSize        dw      0

TestCode:
    ; Hello World!
    db "++++++++++[>+++++++>++++++++++>+++>+<<<<-]>++.>+.+++++++..+++.>++.<<+++++++++++++++.>.+++.------.--------.>+.>."
    ; Squares of 0 through 100
;    db "++++[>+++++<-]>[<+++++>-]+<+[>[>+>+<<-]++>>[<<+>>-]>>>[-]++>[-]+>>>+[[-]++++++>>>]<<<[[<++++++++<++>>-]+<.<[>----<-]<]<<[>>>>>[>>>[-]+++++++++<[>-<-]+++++++++>[-[<->-]+[<<<]]<[>+<-]>]<<-]<<-]"
    ; ROT13
;    db "+[,+[-[>+>+<<-]>[<+>-]+>>++++++++[<-------->-]<-[<[-]>>>+[<+<+>>-]<[>+<-]<[<++>>>+[<+<->>-]<[>+<-]]>[<]<]>>[-]<<<[[-]<[>>+>+<<<-]>>[<<+>>-]>>++++++++[<-------->-]<->>++++[<++++++++>-]<-<[>>>+<<[>+>[-]<<-]>[<+>-]>[<<<<<+>>>>++++[<++++++++>-]>-]<<-<-]>[<<<<[-]>>>>[<<<<->>>>-]]<<++++[<<++++++++>>-]<<-[>>+>+<<<-]>>[<<+>>-]+>>+++++[<----->-]<-[<[-]>>>+[<+<->>-]<[>+<-]<[<++>>>+[<+<+>>-]<[>+<-]]>[<]<]>>[-]<<<[[-]<<[>>+>+<<<-]>>[<<+>>-]+>------------[<[-]>>>+[<+<->>-]<[>+<-]<[<++>>>+[<+<+>>-]<[>+<-]]>[<]<]>>[-]<<<<<------------->>[[-]+++++[<<+++++>>-]<<+>>]<[>++++[<<++++++++>>-]<-]>]<[-]++++++++[<++++++++>-]<+>]<.[-]+>>+<]>[[-]<]<]"
TestCodeEnd:

section .bss

ScanBuf:
    resw SCAN_BUF_SIZE

Code:
    resb MAX_CODE_SIZE

Data:
    resb MAX_DATA_SIZE

如果您想将 DOS(作为宿主环境)和 NASM 排除在外,欢迎您手动对上述汇编代码进行编码,制作一张可启动软盘并启动它。

于 2012-05-08T11:52:18.140 回答
0

规范的编译器学习书籍是 Dragon Book,http://dragonbook.stanford.edu/

然而,它确实指向更多……复杂的语言。你可能不想谈论上下文无关的解析之类的东西(虽然我确实推荐那本书,它是双倍加难但真棒。)

考虑到这一点,您可能希望首先为一种非常简单的语言寻找解释器或编译器——也许是 Brainfuck 本身,也许是像 Scheme 实现这样更有用的东西。阅读、分析、了解它的作用。实现编译器使用的任何较低级别的库函数,调整其代码生成器以输出您想要定位的任何品牌的机器代码,然后您就完成了。

于 2012-05-08T03:26:06.473 回答
0

实际上,我也有一个类似的项目。您想要的是编写一个在裸硬件(没有任何操作系统)上运行的编译器。编译器就像其他所有程序一样只是一个程序,除了它将在您的情况下运行在裸硬件上。

您可能会考虑为您的编译器使用引导加载程序(引导加载程序不是操作系统)。引导加载程序通常将操作系统从磁盘加载到内存中以供执行,只是这次它加载的是编译器而不是操作系统。

那里有许多免费和开源的引导加载程序(谷歌为他们)。选择一个适合您的需求,或者如果您和我一样好奇/好奇,您甚至可以自己编写。

选择引导加载程序(或编写引导加载程序)需要您了解目标系统的引导方式——它通过 BIOS(旧)或 UEFI(新。它是 BIOS 的替代品)。随着您的编程语言成熟,您可以用新的编程语言编写自己的引导加载程序。

希望这可以帮助

于 2014-12-31T04:11:00.547 回答