forth/src/fh_runtime.c

#include <string.h>
#include <errno.h>
#include <ctype.h>

#include "fh_error.h"
#include "fh_runtime.h"
#include "fh_builtins.h"
#include "fh_stack.h"
#include "fh_mem.h"
#include "fh_globals.h"
#include "fh_print.h"

struct fh_global_s fh_globals = {};

/** State names */
static const char *statenames[FH_STATE_MAX] = {
    [FH_STATE_INTERPRET] = "INTERPRET",
    [FH_STATE_COMPILE] = "COMPILE",
    [FH_STATE_SHUTDOWN] = "SHUTDOWN",
};

/** Sub-state names */
static const char *substatenames[FH_SUBSTATE_MAX] = {
    [FH_SUBSTATE_NONE] = "NONE",
    [FH_SUBSTATE_COLONNAME] = "COLONNAME",
    [FH_SUBSTATE_SQUOTE] = "SQUOTE",
    [FH_SUBSTATE_DOTQUOTE] = "DOTQUOTE",
    [FH_SUBSTATE_PARENCOMMENT] = "PARENCOMMENT",
    [FH_SUBSTATE_LINECOMMENT] = "LINECOMMENT",
    [FH_SUBSTATE_EXIT] = "EXIT",
    [FH_SUBSTATE_SEENAME] = "SEENAME",
};

/** Add a word to the dictionary. */
enum fh_error fh_add_word(const struct fh_word_s *w, struct fh_thread_s *fh)
{
  if (fh->dict_top == DICT_SIZE) {
    return FH_ERR_DICT_FULL;
  }
  memcpy(&fh->dict[fh->dict_top++], w, sizeof(struct fh_word_s));
  return FH_OK;
}

/** Log current runtime state */
static void showstate(const struct fh_thread_s *fh)
{
  if (fh->substate == 0) {
    LOG("state = %s", statenames[fh->state]);
  } else {
    LOG("state = %s.%s", statenames[fh->state], substatenames[fh->substate]);
  }
}

/** Set runtime state and sub-state */
void fh_setstate(struct fh_thread_s *fh, enum fh_state state, enum fh_substate substate)
{
  fh->state = state;
  fh->substate = substate;
  showstate(fh);
}

/** Set runtime sub-state (state is unchanged) */
void fh_setsubstate(struct fh_thread_s *fh, enum fh_substate substate)
{
  fh->substate = substate;
  showstate(fh);
}

/** Execute a user word */
enum fh_error w_user_word(struct fh_thread_s *fh, const struct fh_word_s *w0)
{
  enum fh_error rv;
  const struct fh_word_s *w;
  const struct fh_word_s *w2;

  w = w0;
  call:
  if (!w) { return FH_ERR_INTERNAL; }

  LOG("Run user word: %s", w->name);

  TRY(rs_push(fh, fh->execptr));
  fh->execptr = w->start;

  instr:;
  if (fh->state == FH_STATE_QUIT) {
    /* abort or quit was called, return to interactive mode */
    fh_setstate(fh, FH_STATE_INTERPRET, FH_SUBSTATE_NONE);
    return FH_OK;
  }
  // make sure it's aligned
  fh->execptr = WORDALIGNED(fh->execptr);
  const struct fh_instruction_s *instr = (const struct fh_instruction_s *) &fh->compile[fh->execptr];
  fh->execptr += INSTR_SIZE;

  uint32_t strl;
  uint32_t val;
  uint32_t addr = 0;
  switch (instr->kind) {
    case FH_INSTR_NUMBER:
      TRY(ds_push(fh, instr->data));
      goto instr;

    case FH_INSTR_WORD:
      w2 = &fh->dict[instr->data];
      if (w2->builtin) {
        LOG("Exec: builtin-word %s", w2->name);
        w2->handler(fh, w2);
        if (fh->substate == FH_SUBSTATE_EXIT) {
          fh_setsubstate(fh, 0);
          LOG("Exec: early return");
          TRY(rs_pop(fh, &fh->execptr));
          if (fh->execptr == MAGICADDR_INTERACTIVE) {
            goto end;
          }
        }
        goto instr;
      } else {
        LOG("Exec: user-word %s (CALL)", w2->name);
        w = &fh->dict[instr->data];
        goto call;
      }

    case FH_INSTR_JUMPZERO:
      TRY(ds_pop(fh, &val));
      if (0 == val) {
        fh->execptr = instr->data;
      }
      goto instr;

    case FH_INSTR_JUMP:
      fh->execptr = instr->data;
      goto instr;

    /* special case for strings stored in compile memory */
    case FH_INSTR_ALLOCSTR:
    case FH_INSTR_TYPESTR:
      strl = instr->data;
      if (instr->kind == FH_INSTR_ALLOCSTR) {
        TRY(fh_heap_reserve(fh, strl, &addr));
        fh_heap_copy_from_compile(fh, addr, fh->execptr, strl);
        LOG("Exec: alloc-str \"%.*s\"", strl, &fh->heap[addr]);
        TRY(ds_push(fh, addr));
        TRY(ds_push(fh, strl));
      } else {
        LOG("Exec: type-str \"%.*s\"", strl, &fh->compile[fh->execptr]);
        FHPRINT("%.*s", (int) strl, &fh->compile[fh->execptr]);
      }
      fh->execptr += strl;
      goto instr;

    case FH_INSTR_ENDWORD:
      LOG("Exec: word-end (RETURN)");
      TRY(rs_pop(fh, &fh->execptr));
      if (fh->execptr == MAGICADDR_INTERACTIVE) {
        goto end;
      }
      goto instr;
  }

  end:
  return FH_OK;
}


/** Initialize a runtime */
enum fh_error fh_init(struct fh_thread_s *fh)
{
  enum fh_error rv;

  /* Make sure we have a clean state */
  memset(fh, 0, sizeof(struct fh_thread_s));

  TRY(register_builtin_words(fh));

  fh->execptr = MAGICADDR_INTERACTIVE;
  fh->base = 10;
  return FH_OK;
}

/** Process a quoted string read from input */
static enum fh_error fh_handle_quoted_string(
    struct fh_thread_s *fh,
    const char *start,
    size_t len
)
{
  enum fh_error rv;
  uint32_t addr = 0;
  struct fh_instruction_s instr;

  if (fh->state == FH_STATE_INTERPRET) {
    switch (fh->substate) {
      case FH_SUBSTATE_SQUOTE:
        TRY(fh_heap_put(fh, start, len));
        TRY(ds_push(fh, addr));
        TRY(ds_push(fh, len));
        break;
      case FH_SUBSTATE_DOTQUOTE:
        FHPRINT("%.*s", (int) len, start);
        break;

      default:
        LOGE("Bad substate in interpret mode: %s", substatenames[fh->substate]);
    }
  } else {
    LOG("Compile a string");
    /* compile */
    if (fh->substate == FH_SUBSTATE_SQUOTE) {
      instr_init(&instr, FH_INSTR_ALLOCSTR, len);
    } else {
      instr_init(&instr, FH_INSTR_TYPESTR, len);
    }
    TRY(fh_compile_put(fh, &instr, INSTR_SIZE));
    TRY(fh_compile_put(fh, start, len));
  }
  return FH_OK;
}

/** Process a word read from input */
static enum fh_error fh_handle_word(
    struct fh_thread_s *fh,
    const char *start,
    const size_t wordlen
)
{
  if (wordlen >= MAX_NAME_LEN) {
    return FH_ERR_NAME_TOO_LONG;
  }

  /* First, try if it's a known word */
  // TODO we could use binary search if the dict was ordered
  struct fh_word_s *w = &fh->dict[0];
  struct fh_instruction_s instr;
  uint32_t cnt = 0;
  enum fh_error rv;
  while (w->handler) {
    if (0 == strncasecmp(start, w->name, wordlen) && w->name[wordlen] == 0) {
      // word found!
      if (fh->state == FH_STATE_COMPILE && !w->immediate) {
        LOG("Compile word call: %s", w->name);
        instr_init(&instr, FH_INSTR_WORD, cnt);
        TRY(fh_compile_put(fh, &instr, INSTR_SIZE));
      } else {
        /* interpret */
        LOG("Interpret word: %s", w->name);
        TRY(w->handler(fh, w));
      }
      return FH_OK;
    }
    w++;
    cnt++;
  }

  /* word not found, try parsing as number */
  errno = 0;
  char *endptr;
  int base = (int) fh->base;

  // prefix can override BASE - this is a syntax extension
  if (start[0] == '0') {
    if (start[1] == 'x') {
      base = 16;
    } else if (start[1] == 'b') {
      base = 2;
    } else if (start[1] == 'o') {
      base = 8;
    }
  }

  long v = strtol(start, &endptr, base); // XXX if base is 0, this will use auto-detection
  if (errno != 0 || (endptr - start) != wordlen) {
    LOGE("Unknown word and fail to parse as number: %.*s", (int) wordlen, start);
    return FH_ERR_UNKNOWN_WORD;
  }

  if (fh->state == FH_STATE_COMPILE) {
    LOG("Compile number: %ld", v);
    instr_init(&instr, FH_INSTR_NUMBER, (uint32_t) v);
    TRY(fh_compile_put(fh, &instr, INSTR_SIZE));
  } else {
    /* interpret */
    LOG("Interpret number: %ld", v);
    TRY(ds_push(fh, (uint32_t) v));
  }

  return FH_OK;
}

static void show_word(struct fh_thread_s *fh, const struct fh_word_s *w) {
  if (w->handler == w_user_word) {
    uint32_t execptr = w->start;

    instr:;
    // make sure it's aligned
    execptr = WORDALIGNED(execptr);
    FHPRINT("0x%08x: ", execptr);
    const struct fh_instruction_s *instr = (const struct fh_instruction_s *) &fh->compile[execptr];
    execptr += INSTR_SIZE;

    uint32_t strl;
    uint32_t wn;
    const struct fh_word_s *w2;
    switch (instr->kind) {
      case FH_INSTR_NUMBER:
        FHPRINT("Number(%d)\n", instr->data);
        goto instr;

      case FH_INSTR_WORD:
        wn = instr->data;
        w2 = &fh->dict[instr->data];
        FHPRINT("Call(%s, 0x%08x)\n", w2->name, instr->data);
        goto instr;

      case FH_INSTR_JUMPZERO:
        FHPRINT("JumpIfZero(0x%08x)\n", instr->data);
        goto instr;

      case FH_INSTR_JUMP:
        FHPRINT("Jump(0x%08x)\n", instr->data);
        goto instr;

        /* special case for strings stored in compile memory */
      case FH_INSTR_ALLOCSTR:
      case FH_INSTR_TYPESTR:
        strl = instr->data;
        if (instr->kind == FH_INSTR_ALLOCSTR) {
          FHPRINT("AllocStr(\"%.*s\")\n", strl, &fh->compile[execptr]);
          execptr += strl;
        } else {
          FHPRINT("PrintStr(\"%.*s\")\n", strl, &fh->compile[execptr]);
          execptr += strl;
        }
        goto instr;

      case FH_INSTR_ENDWORD:
        FHPRINT("END\n");
        return;
    }

  } else {
    FHPRINT("(builtin)");
  }
}

/** Decompile a word */
static enum fh_error fh_see_word(
    struct fh_thread_s *fh,
    const char *start,
    const size_t wordlen
)
{
  struct fh_word_s *w = &fh->dict[0];
  uint32_t cnt = 0;
  enum fh_error rv;
  while (w->handler) {
    if (0 == strncasecmp(start, w->name, wordlen) && w->name[wordlen] == 0) {
      // word found!
      show_word(fh, w);
      return FH_OK;
    }
    w++;
    cnt++;
  }
  return FH_ERR_UNKNOWN_WORD;
}

/** True if the character is CR or LF */
static inline bool isnl(char c)
{
  return c == '\n' || c == '\r';
}

/** Process a line read from input */
enum fh_error fh_process_line(struct fh_thread_s *fh, const char *linebuf)
{
  enum fh_error rv;
  const char *rp = linebuf;
  char c;

  if (!fh_globals.interactive) {
    LOGI("%s", linebuf);
  }

  while (0 != (c = *rp) && fh->state != FH_STATE_SHUTDOWN) {
    /* end on newline */
    if (isnl(c)) {
      goto done;
    }
    /* skip whitespace */
    if (isspace(c)) {
      rp++;
      continue;
    }

    char *end;
    size_t length;
    switch (fh->substate) {
      case FH_SUBSTATE_NONE:
      case FH_SUBSTATE_COLONNAME:
      case FH_SUBSTATE_SEENAME:
        /* try to read a word */
        end = strchr(rp, ' ');
        if (end) {
          length = end - rp; /* exclude the space */
        } else {
          length = strlen(rp);
        }

        if (fh->substate == FH_SUBSTATE_NONE) {
          /* eval a word */
          LOG("Handle \"%.*s\"", (int) length, rp);
          TRY(fh_handle_word(fh, rp, length));
        } else if (fh->substate == FH_SUBSTATE_COLONNAME) {
          /* new word's name is found */
          LOG("New word name = \"%.*s\"", (int) length, rp);
          strncpy(fh->dict[fh->dict_top].name, rp, length);
          fh_setsubstate(fh, FH_SUBSTATE_NONE);
        } else if (fh->substate == FH_SUBSTATE_SEENAME) {
          TRY(fh_see_word(fh, rp, length));
          fh_setsubstate(fh, FH_SUBSTATE_NONE);
        }

        if (end) {
          rp = end + 1;
        } else {
          goto done;
        }
        break;

      case FH_SUBSTATE_SQUOTE:
      case FH_SUBSTATE_DOTQUOTE:
        end = strchr(rp, '"');
        if (end) {
          length = end - rp;
          LOG("Quoted string: \"%.*s\"", (int) length, rp);
          TRY(fh_handle_quoted_string(fh, rp, length));
          fh_setsubstate(fh, FH_SUBSTATE_NONE);
          rp = end + 1;
        } else {
          /* no end. this is weird. */
          LOGE("Unterminated quoted string!");
          goto done;
        }
        break;

      case FH_SUBSTATE_PARENCOMMENT:
        end = strchr(rp, ')');
        if (end) {
          LOG("Discard inline comment");
          fh_setsubstate(fh, FH_SUBSTATE_NONE);
          rp = end + 1;
        } else {
          /* no end, discard all */
          LOGE("Unterminated parenthesis comment");
          goto done;
        }
        break;

      case FH_SUBSTATE_LINECOMMENT:
        LOG("Discard line comment");
        goto done; // just discard the rest

      default:
        LOGE("Bad substate %s", substatenames[fh->substate]);
    }
  }
  done:
  LOG("Line done.");
  return FH_OK;
}