roza/lib/lexer.c

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

#define RZ_KEYWORD(KW, NODE, VAL)                                       \
  {                                                                     \
    node_t* kw = lexer_try_new_keyword(lexer, KW, NODE, VAL, 1);        \
    if (kw) { lexer_skip_spaces(lexer); return kw; }                    \
  }

#define RZ_TEXT(KW, NODE, VAL)                                       \
  {                                                                     \
    node_t* kw = lexer_try_new_keyword(lexer, KW, NODE, VAL, 0);        \
    if (kw) { lexer_skip_spaces(lexer); return kw; }                    \
  }

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);

  lexer_skip_spaces(lexer);

  // Text
  // ====
  RZ_TEXT("==", NODE_EQ, 0);
  RZ_TEXT("!=", NODE_NE, 0);

  // Keywords
  // ========
  RZ_KEYWORD("true", NODE_BOOL, 1);
  RZ_KEYWORD("false", NODE_BOOL, 1);
  RZ_KEYWORD("assert", NODE_ASSERT, 0);

  // scan str
  {
    node_t* node = lexer_try_new_str(lexer);

    if (node)
      {
        lexer_skip_spaces(lexer);
        return node;
      }
  }

  // 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;

        lexer_skip_spaces(lexer);
        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_fatal(lexer->err, msg, lexer->line);
    }

  return NULL;
}

NodeType lexer_peek(lexer_t* lexer, int lookahead)
{
  assert(lexer);
  size_t cursor = lexer->cursor;
  int line = lexer->line;

  NodeType type = -1;

  for (int i=0; i<lookahead; i++)
    {
      node_t* node = lexer_try_new_next(lexer);

      if (node)
        {
          type = node->type;
          node_free(node);
          free(node);
        }
      else
        {
          break;
        }
    }

  lexer->cursor = cursor;
  lexer->line = line;

  return type;
}

void lexer_skip_spaces(lexer_t* lexer)
{
  assert(lexer);
  size_t len = strlen(lexer->source);

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

      lexer->cursor++;
    }

}

node_t* lexer_try_new_keyword(lexer_t* lexer, char* kw,
                              NodeType type, int has_value,
                              int is_kw)
{
  assert(lexer);
  assert(kw);

  size_t len = strlen(kw);
  size_t cursor = lexer->cursor;

  if (cursor + len <= strlen(lexer->source))
    {
      int ok = 1;

      for (size_t i=cursor; i<cursor + len; i++)
        {
          if (lexer->source[i] != kw[i - cursor])
            {
              ok = 0;
              break;
            }
        }

      if (ok)
        {
          int next_idx = lexer->cursor + len;

          if (next_idx < strlen(lexer->source)
              && (is_kw && !lexer_is_sep(lexer, next_idx)))
            {
              return NULL;
            }

          node_t* node = malloc(sizeof(node_t));
          node_init(node, type, has_value ? (char*) kw : "", lexer->line);

          lexer->cursor += len;
          return node;
        }
    }

  return NULL;
}

node_t* lexer_try_new_str(lexer_t* lexer)
{
  assert(lexer);

  ssize_t cursor = lexer->cursor;
  ssize_t len = strlen(lexer->source);

  str_t res_str;
  str_init(&res_str);

  if (cursor >= len || lexer->source[cursor] != '"')
    {
      str_free(&res_str);
      return NULL;
    }

  cursor++;

  while (cursor < len
         && lexer->source[cursor] != '"')
    {
      if (lexer->source[cursor] == '\\')
        {
          if (cursor + 1 < len)
            {
              switch (lexer->source[cursor + 1])
                {
                case '"': {
                  str_push(&res_str, '"');
                  cursor += 2;
                } break;

                case 'n': {
                  str_push(&res_str, '\n');
                  cursor += 2;
                } break;

                case 't': {
                  str_push(&res_str, '\t');
                  cursor += 2;
                } break;

                case 'r': {
                  str_push(&res_str, '\r');
                  cursor += 2;
                } break;

                default: cursor++; break;
                }
            }
        }
      else
        {
          str_push(&res_str, lexer->source[cursor]);
          cursor++;
        }
    }

  if (cursor >= len || lexer->source[cursor] != '"')
    {
      str_free(&res_str);
      return NULL;
    }

  cursor++;

  node_t* tok = malloc(sizeof(node_t));
  node_init(tok, NODE_STR, res_str.data, lexer->line);
  str_free(&res_str);

  lexer->cursor = cursor;

  return tok;
}

int lexer_is_sep(lexer_t* lexer, size_t idx)
{
  assert(lexer);

  if (idx >= strlen(lexer->source))
    {
      return 1;
    }

  char c = lexer->source[idx];

  if (isspace(c))
    {
      return 1;
    }

  return c == '='
    || c == '!';
}