roza/lib/lexer.c

#include "lexer.h"
#include "lib/commons.h"

void lexer_init(lexer_t* lexer, char const* source, err_t* err)
{
  assert(lexer);
  lexer->source = strdup(source);
  lexer->cursor = 0;
  lexer->err = err;
  lexer->line = 1;
}

void lexer_free(lexer_t* lexer)
{
  assert(lexer);
  free(lexer->source);
  lexer->source = NULL;
}

node_t* lexer_try_new_next(lexer_t* lexer)
{
  assert(lexer);
  size_t len = strlen(lexer->source);

  // skip spaces
  {
    while (lexer->cursor < len
           && isspace(lexer->source[lexer->cursor]))
      {
        if (lexer->source[lexer->cursor] == '\n')
          {
            lexer->line++;
          }

        lexer->cursor++;
      }
  }

  // scan num
  {
    size_t cursor = lexer->cursor;

    str_t res_str;
    str_init(&res_str);

    if (cursor < len && lexer->source[cursor] == '-')
      {
        str_push(&res_str, lexer->source[cursor]);
        cursor++;
      }

    while (cursor < len
           && isdigit(lexer->source[cursor]))
      {
        str_push(&res_str, lexer->source[cursor]);
        cursor += 1;
      }

    if (cursor < len && lexer->source[cursor] == '.')
      {
        str_push(&res_str, lexer->source[cursor]);
        cursor++;

        while (cursor < len
               && isdigit(lexer->source[cursor]))
          {
            str_push(&res_str, lexer->source[cursor]);
            cursor += 1;
          }
      }

    if (res_str.size > 0
        && (cursor >= len || isspace(lexer->source[cursor])))
      {
        node_t* tok = malloc(sizeof(node_t));
        node_init(tok, NODE_NUM, res_str.data, lexer->line);
        str_free(&res_str);

        lexer->cursor = cursor;
        return tok;
      }

    str_free(&res_str);
  }

  if (lexer->cursor < len && lexer->err)
    {
      size_t const SZ = RZ_STR_LIMIT;
      char msg[SZ];
      snprintf(msg, SZ, "unexpected symbol '%c'", lexer->source[lexer->cursor]);
      err_error(lexer->err, msg, lexer->line);
    }

  return NULL;
}