Function getWord

source/text_handler.cpp:97–144

Fetches the next word from the given text stream starting from the given position. On success, writes the decoded word into *word and updates position to the location past the returned word. A word ends at the next comment or whitespace. However, double-quoted strings remain intact, and a backslash always escapes the next character.

// A word ends at the next comment or whitespace.  However, double-quoted
// strings remain intact, and a backslash always escapes the next character.
spv_result_t getWord(spv_text text, spv_position position, std::string* word) {
  if (!text->str || !text->length) return SPV_ERROR_INVALID_TEXT;
  if (!position) return SPV_ERROR_INVALID_POINTER;

  const size_t start_index = position->index;

  bool quoting = false;
  bool escaping = false;

  // NOTE: Assumes first character is not white space!
  while (true) {
    if (position->index >= text->length) {
      word->assign(text->str + start_index, text->str + position->index);
      return SPV_SUCCESS;
    }
    const char ch = text->str[position->index];
    if (ch == '\\') {
      escaping = !escaping;
    } else {
      switch (ch) {
        case '"':
          if (!escaping) quoting = !quoting;
          break;
        case ' ':
        case ';':
        case ',':
        case '(':
        case ')':
        case '\t':
        case '\n':
        case '\r':
          if (escaping || quoting) break;
          word->assign(text->str + start_index, text->str + position->index);
          return SPV_SUCCESS;
        case '\0': {  // NOTE: End of word found!
          word->assign(text->str + start_index, text->str + position->index);
          return SPV_SUCCESS;
        }
        default:
          break;
      }
      escaping = false;
    }

    position->column++;
    position->index++;
  }
}

// Returns true if the characters in the text as position represent
// the start of an Opcode.