equi/equi.c

/*
 * Equi platform reference implementation
 *
 * Created in 2022 by Luxferre, released into public domain
 *
 * See equi.md file for the specification and manual
 *
 * @license Unlicense <https://unlicense.org>
 * @author Luxferre
 */

/* Standard includes with size optimizations for TCC */

#ifdef __TINYC__
#include <tcclib.h>
#define stderr 2
#else
#include <stdlib.h>
#include <stdio.h>
#endif

/*
 * Non-standard includes
 * To save space, we emulate necessary conio functions on non-6502 systems, not vice versa
 */

#ifdef __CC65__
#include <conio.h>
#else

#define cgetc() (getchar())
#define cputc(c) (putchar(c))
#endif

/* Definitions section */

#define EQUI_VER "0.0.1"

#define cerr(s) (fputs(s, stderr))
#define ushort unsigned short /* basic 16-bit integer */
#define uchar unsigned char /* basic 8-bit integer */
#define WS sizeof(ushort) /* Equi word size in bytes */
#define CLT_ENTRY_LEN 6u /* Amount of significant compiled word characters */
#define CLT_ENTRY_SIZE (CLT_ENTRY_LEN + WS) /* Full size in bytes taken by one CLT entry */
#define BS 8u /* Backspace character code */
#define CR 13u /* Character return code */

/* Configuration section (constants overridable at compile-time) */

/* Main and return stack size in bytes */
#ifndef STACK_SIZE
#define STACK_SIZE 256u
#endif

/* Literal stack size in bytes */
#ifndef LIT_STACK_SIZE
#define LIT_STACK_SIZE 32u
#endif

/* Command buffer size in bytes */
#ifndef GPD_AREA_SIZE
#define GPD_AREA_SIZE 6400u
#endif

/* Command buffer size in bytes */
#ifndef CMD_BUF_SIZE
#define CMD_BUF_SIZE 15600u
#endif

/* Maximum amount of CLT entries */
#ifndef CLT_ENTRIES_MAX
#define CLT_ENTRIES_MAX 512u
#endif

/* Some necessary constants and offsets derived from the above values */
#define STACK_SIZE_WORDS (STACK_SIZE / WS) /* Main and return stack size in words */

/*
 * Structures that describe Equi machine RAM using the above configuration
 */

struct CLTEntry {
  uchar name[CLT_ENTRY_LEN]; /* compiled word name */
  ushort loc;    /* compiled word location */
};

struct EquiRAM { 
  ushort main_stack[STACK_SIZE_WORDS];
  ushort return_stack[STACK_SIZE_WORDS];
  uchar literal_stack[LIT_STACK_SIZE];
  ushort pc; /* program counter */
  ushort msp; /* main stack pointer */
  ushort rsp; /* return stack pointer */
  uchar lsp; /* literal stack pointer */
  ushort cbp; /* compilation buffer pointer */
  ushort cltp; /* compilation lookup table pointer */
  unsigned int II:1; /* instruction ignore mode */
  unsigned int CM:1; /* compilation mode */
  unsigned int IM:1; /* interpretation mode */
  ushort gpd_start;
  ushort cmd_start;
  ushort cmd_size;
  ushort ibp; /* input buffer pointer */
  struct CLTEntry clt[CLT_ENTRIES_MAX]; /* compilation lookup table */
  uchar gpd[GPD_AREA_SIZE];
  uchar cmdbuf[CMD_BUF_SIZE];
};

/* Before running the main code, instantiate the machine RAM */
static struct EquiRAM ram;

/* Also create an alternative view of the same RAM area for direct offset-based access */
static uchar* flatram = (uchar *)&ram;

/* Error reporting codes */
enum EquiErrors {
  SUCCESS=0,
  STACK_OVERFLOW,
  STACK_UNDERFLOW,
  DIV_BY_ZERO,
  CLT_OVERFLOW,
  CMD_OVERFLOW,
  INVALID_INSTRUCTION,
  PORT_IO_ERROR
};

/* Error reporting method */
void trapout(errcode) {
  if(errcode > 0) {
    fprintf(stderr, "Error %d at 0x%x (instruction %c): ", errcode, ram.pc, ram.cmdbuf[ram.pc]);
    switch(errcode) {
      case STACK_OVERFLOW:
        cerr("Stack overflow");
        break;
      case STACK_UNDERFLOW:
        cerr("Stack underflow");
        break;
      case DIV_BY_ZERO:
        cerr("Division by zero");
        break;
      case CLT_OVERFLOW:
        cerr("Compilation lookup table full");
        break;
      case CMD_OVERFLOW:
        cerr("Command buffer full");
        break;
      case INVALID_INSTRUCTION:
        cerr("Invalid instruction");
        break;
      case PORT_IO_ERROR:
        cerr("Port I/O error");
        break;
    }
    exit(errcode);
  }
}

/* Equi instructions enum */

enum EquiInstructions {
  INS_IISTART='(',
  INS_IIEND=')',
  INS_CMSTART=':',
  INS_CMEND=';',
  INS_LITINT='#',
  INS_LITSTR='"',
  INS_LITCALL='\'',
  INS_RET='R',
  INS_M2R=']', /* main -> return */
  INS_R2M='[', /* return -> main */
  INS_LOAD='L',
  INS_STORE='S',
  INS_STOREBYTE='W',
  INS_DROP='!',
  INS_DUP='$',
  INS_SWAP='%',
  INS_ROT='@',
  INS_OVER='\\',
  INS_JUMP='J',
  INS_IF='I',
  INS_EXPOINT='X', /* locate execution point */
  INS_GT='>',
  INS_LT='<',
  INS_EQ='=',
  INS_ADD='+',
  INS_SUB='-',
  INS_MUL='*',
  INS_DIV='/',
  INS_NEG='N',
  INS_NOT='~',
  INS_AND='&',
  INS_OR='|',
  INS_XOR='^',
  INS_COUT='.',
  INS_NBKIN=',', /* non-blocking key input */
  INS_BKIN='?', /* blocking key input */
  INS_PORTIO='P',
  INS_PERSIST_WRITE='}',
  INS_PERSIST_READ='{',
  INS_QUIT='Q'
};

/* push a value onto the main stack */
void pushMain(ushort val) {
  ram.main_stack[ram.msp++] = val;
  if(ram.msp >= STACK_SIZE_WORDS)
    trapout(STACK_OVERFLOW);
}

/* push a value onto the return stack */
void pushRet(ushort val) {
  ram.return_stack[ram.rsp++] = val;
  if(ram.rsp >= STACK_SIZE_WORDS)
    trapout(STACK_OVERFLOW);
}

/* pop a value from the main stack */
ushort popMain() {
  if(ram.msp == 0)
    trapout(STACK_UNDERFLOW);
  else
    return ram.main_stack[--ram.msp];
}

/* pop a value from the return stack */
ushort popRet() {
  if(ram.rsp == 0)
    trapout(STACK_UNDERFLOW);
  else
    return ram.return_stack[--ram.rsp];
}

/* push a character to the literal stack, auto-discarding overflown items */
void pushLit(uchar c) {
  ram.literal_stack[ram.lsp++] = c;
  if(ram.lsp >= LIT_STACK_SIZE)
    ram.lsp = 0;
}

/* pop a character from the literal stack */
uchar popLit() {
  if(ram.lsp == 0)
    trapout(STACK_UNDERFLOW);
  else
    return ram.literal_stack[--ram.lsp];
}

/* check literal stack and push 4 top values as hex short literal if not empty and discard the rest of the stack */
void attemptPushNumericLiteral() {

}

/* Main interpreter loop */

void equi_main_loop() {
  uchar instr;
  while(1) { /* iterate over the instructions in the command buffer */
    instr = ram.cmdbuf[++ram.pc];
    /* first, check for II mode */
    if(ram.II) {
      if(instr == ')')
        ram.II = 0; /* unset instruction ignore mode flag */
      continue;
    }
    /* then, check for compilation mode */
    if(ram.CM) {
      if(instr == ';') { /* trigger word compilation logic as per the spec */
        ram.cmdbuf[ram.pc] = 'R'; /* in-place patch this instruction to R */
        
        ram.clt[ram.cltp].loc = ram.cbp; /* stored the compiled code location from the most recent CBP value */
        ++ram.cltp; /* increase the word */
        if(ram.cltp == CLT_ENTRIES_MAX) {
          trapout(CLT_OVERFLOW);
          break;
        }

        ram.CM = 0; /* unset compilation mode flag */
      }
      continue;
    }
    /* other than that, continue with interpretation mode */
    /* first, detect lowercase, underscore, digit and A to F */
    if((instr >= '0' && instr <= '9') || (instr >= 'A' && instr <= 'F') || instr == '_' || (instr >= 'a' && instr <= 'z')) {
      pushLit(instr);
      continue;
    }
    /* then trigger literal auto-push if applicable */
    if(instr != '"' && instr != '\'' && instr != '#' && instr != ':')
      attemptPushNumericLiteral();
    switch(instr) {

      default:
        goto brx;
    }
    continue;
    brx: break;
  }
  /* unset interpretation mode flag and exit */
  ram.IM = 0;
}

/* Equi VM entry point */

int main(int argc, char* argv[]) {
  /* _attempt_ to disable buffering for char input/output */
#if !defined __CC65__ && !defined __TINYC__
  setvbuf(stdin, NULL, _IONBF, 0);
  setvbuf(stdout, NULL, _IONBF, 0);
#endif
  uchar instr;
  /* initialize the RAM in the most standard way */
  ram.gpd_start = (uchar *)&ram.gpd - (uchar *)&ram.main_stack;
  ram.cmd_start = (uchar *)&ram.cmdbuf - (uchar *)&ram.main_stack;
  ram.cmd_size = CMD_BUF_SIZE;
  ram.II = ram.CM = ram.IM = 0; /* reset all flags */
  /* Start both execution and input buffering from the start of command buffer (-1 because we use prefix increment) */
  ram.pc = ram.ibp = ram.cmd_start-1;

  printf("Welcome to Equi v" EQUI_VER " by Luxferre, 2022\n\nCLT: 0x%X (%d bytes)\nGPD: 0x%X (%d bytes)\nCommand buffer: 0x%X (%d bytes)\nEqui ready\n\n",
      (uchar *)&ram.clt - (uchar *)&ram.main_stack,
      (uchar *)&ram.gpd - (uchar *)&ram.clt,
      ram.gpd_start,
      ram.cmd_start - ram.gpd_start,
      ram.cmd_start, ram.cmd_size);

  cputc('>');
  cputc(' ');

  while(1) { /* Now, we're in the command mode loop */
    instr = ram.cmdbuf[++ram.ibp] = cgetc(); /* Fetch the next instruction from the keyboard */
    if(instr == BS && ram.ibp > ram.cmd_start) { /* process the backspace */
#ifdef __CC65__
      cputc(instr); /* echo it */
#endif
      --ram.ibp; 
    } else if(instr == '(') { /* process II just to avoid quitting in command mode */
#ifdef __CC65__
      cputc(instr); /* echo it */
#endif
      ram.II = 1;
    } else if(instr == ')') { /* process II just to avoid quitting in command mode */
#ifdef __CC65__
      cputc(instr); /* echo it */
#endif
      ram.II = 0;
    } else if(instr == CR) { /* process carriage return */
      cputc(CR); /* echo it */
#ifdef __CC65__
      cputc(10); /* echo it */
#endif
      ram.IM = 1; /* set the mandatory interpretation mode flag */
      equi_main_loop(); /* and run the interpreter loop */
      cputc(CR); /* echo it */
#ifdef __CC65__
      cputc(10); /* echo it */
#endif
      cputc('>');
      cputc(' ');
    }
#ifdef __CC65__
    else cputc(instr); /* echo it */
#endif
  } /* command mode loop end */
  return 0;
}