ai-forth-experiment/forth.h

#ifndef FORTH_H
#define FORTH_H

#include <stdint.h>
#include <stddef.h>
#include <stdarg.h>

// Configuration (all hard limits removed)
#define MAX_NAME_LEN 31
#define FORTH_EOF (-1)

/* Platform interface -- must be provided by the target platform layer. */
void forth_putchar(char c);
int forth_getchar(void);
void forth_printf(const char* fmt, ...);
void forth_fflush(void);
void forth_panic(void);

/* Platform string-to-number (base 10 is required). */
int64_t forth_strtoll(const char* str, char** endptr, int base);

/* Portable character classification. */
static inline int forth_isspace(int c) {
    return c == ' ' || c == '\t' || c == '\n' || c == '\r' || c == '\f' || c == '\v';
}

/* Portable string operations. */
static inline size_t forth_strlen(const char* s) {
    size_t n = 0;
    while (s[n]) n++;
    return n;
}

static inline void forth_memcpy(void* dst, const void* src, size_t n) {
    char* d = dst;
    const char* s = src;
    while (n--) *d++ = *s++;
}

static inline int forth_strcmp(const char* a, const char* b) {
    while (*a && *a == *b) { a++; b++; }
    return (unsigned char)*a - (unsigned char)*b;
}

/* Static memory sizes (override before including forth.h) */
#ifndef FORTH_DATA_STACK_SIZE
#define FORTH_DATA_STACK_SIZE 128
#endif
#ifndef FORTH_RET_STACK_SIZE
#define FORTH_RET_STACK_SIZE 128
#endif
#ifndef FORTH_COMPILE_STACK_SIZE
#define FORTH_COMPILE_STACK_SIZE 64
#endif
#ifndef FORTH_COMPILE_BUF_SIZE
#define FORTH_COMPILE_BUF_SIZE 256
#endif
#ifndef FORTH_USER_MEM_CELLS
#define FORTH_USER_MEM_CELLS (1024 * 1024)
#endif
#ifndef FORTH_MAX_WORDS
#define FORTH_MAX_WORDS 256
#endif
#ifndef FORTH_MAX_WORD_BODY_CELLS
#define FORTH_MAX_WORD_BODY_CELLS 4096
#endif

#define F_IMMEDIATE (1 << 7)
#define F_HIDDEN    (1 << 6)

// Core types
typedef struct Word Word;
typedef union Cell {
    Word* word;
    int64_t num;
    void* ptr;        // used for return stack (holds Cell*)
} Cell;

struct Word {
    Word* prev;
    uint8_t flags;   // Bit7=immediate, Bit6=hidden, Bits0-5=name length
    char name[MAX_NAME_LEN + 1];
    void (*code)(Word*);
    Cell* body;      // points to the word’s body (allocated dynamically)
};

// Globals (all dynamic)
extern int64_t *data_stack;
extern int32_t data_sp;           // stack indices are small enough for int32_t
extern int32_t data_cap;

extern Cell *ret_stack;
extern int32_t rp;
extern int32_t ret_cap;
extern Cell* ip;                  // instruction pointer

extern Word* dict_head;

extern int state;
extern Cell *compile_buf;
extern int32_t compile_idx;
extern int32_t compile_cap;
extern char compiling_name[MAX_NAME_LEN + 1];

extern char* input_buf;           // line buffer (dynamic, managed by getline)
extern size_t input_buf_cap;
extern char* input_ptr;

extern int64_t *compile_stack;    // holds indices into compile_buf
extern int32_t compile_sp;
extern int32_t compile_stack_cap;

extern Cell *user_mem;
extern int64_t user_mem_size;     // in cells
extern Cell* here;

// Pointers to critical hidden/primitive words (set during init)
extern Word* w_exit;
extern Word* w_docolon;
extern Word* w_lit;
extern Word* w_branch;
extern Word* w_zbranch;
extern Word* w_dot_quote_inner;

// Entry point
void forth_run(void);

// Core function prototypes
void data_push(int64_t val);
int64_t data_pop(void);
void ret_push_ip(Cell* val);
Cell* ret_pop_ip(void);
void ret_push_num(int64_t val);
int64_t ret_pop_num(void);
void compile_push(int64_t idx);
int64_t compile_pop(void);

Word* add_primitive(const char* name, void (*code)(Word*), uint8_t flags);
Word* lookup_word(const char* name);
Word* lookup_word_internal(const char* name);

char* next_token(void);
void inner_interpreter(void);
void process_token(const char* token);
void outer_interpreter(void);

// Ensure compile_buf has at least 'needed' free cells (returns 0 on overflow)
int ensure_compile_cap(int32_t needed);

Word* forth_alloc_word(void);
Cell* forth_alloc_body(int32_t cells);

// Stack ops
void do_dup(Word* w);
void do_drop(Word* w);
void do_swap(Word* w);
void do_over(Word* w);
void do_rot(Word* w);
void do_minus_rot(Word* w);
void do_nip(Word* w);
void do_tuck(Word* w);

// Arithmetic
void do_add(Word* w);
void do_sub(Word* w);
void do_mul(Word* w);
void do_div(Word* w);
void do_mod(Word* w);
void do_slash_mod(Word* w);
void do_one_plus(Word* w);
void do_one_minus(Word* w);
void do_two_plus(Word* w);
void do_two_minus(Word* w);
void do_negate(Word* w);
void do_abs(Word* w);
void do_min(Word* w);
void do_max(Word* w);

// Logic
void do_and(Word* w);
void do_or(Word* w);
void do_xor(Word* w);
void do_invert(Word* w);
void do_lshift(Word* w);
void do_rshift(Word* w);

// Comparison
void do_eq(Word* w);
void do_neq(Word* w);
void do_lt(Word* w);
void do_gt(Word* w);
void do_lte(Word* w);
void do_gte(Word* w);
void do_zero_eq(Word* w);
void do_zero_lt(Word* w);
void do_zero_gt(Word* w);

// I/O
void do_dot(Word* w);
void do_cr(Word* w);
void do_emit(Word* w);
void do_key(Word* w);
void do_dot_quote(Word* w);
void do_dot_quote_inner(Word* w);
void do_words(Word* w);

// Memory
void do_fetch(Word* w);
void do_store(Word* w);
void do_plus_store(Word* w);
void do_cfetch(Word* w);
void do_cstore(Word* w);
void do_variable(Word* w);
void do_constant(Word* w);
void do_do_var(Word* w);
void do_do_const(Word* w);
void do_here(Word* w);
void do_allot(Word* w);
void do_comma(Word* w);

// Return stack
void do_to_r(Word* w);
void do_r_from(Word* w);
void do_r_fetch(Word* w);

// Control flow
void do_exit(Word* w);
void do_docolon(Word* w);
void do_lit(Word* w);
void do_colon(Word* w);
void do_semicolon(Word* w);
void do_branch(Word* w);
void do_zero_branch(Word* w);
void do_if(Word* w);
void do_else(Word* w);
void do_then(Word* w);
void do_begin(Word* w);
void do_until(Word* w);
void do_while(Word* w);
void do_repeat(Word* w);

/* Additional stack words */
void do_depth(Word* w);
void do_pick(Word* w);
void do_roll(Word* w);
void do_qdup(Word* w);
void do_2dup(Word* w);
void do_2drop(Word* w);
void do_2swap(Word* w);

#endif