Method Parse

be/src/exec/json/json-parser.cc:88–136

template <class Scanner>
Status JsonParser<Scanner>::Parse(int max_rows, int* num_rows) {
  while (*num_rows < max_rows) {
    constexpr auto parse_flags =
        kParseNumbersAsStringsFlag | kParseStopWhenDoneFlag | kParseNanAndInfFlag;
    // Reads characters from the stream, parses them and publishes events to this
    // handler (JsonParser).
    reader_.Parse<parse_flags>(stream_, *this);

    if (UNLIKELY(reader_.HasParseError())) {
      if (reader_.GetParseErrorCode() == kParseErrorDocumentEmpty) {
        // When the parser encounters the first non-empty character as '\0' during once
        // parsing, it assumes the end of the stream and throws the error code
        // "kParseErrorDocumentEmpty". If the stream has indeed reached its end, we can
        // return normally. However, if a file corruption causes a '\0' to be inserted
        // between JSON objects, the stream hasn't actually ended, and we should
        // continue scanning.
        if (UNLIKELY(!stream_.Eos())) {
          DCHECK_EQ(stream_.Peek(), '\0');
          stream_.Take();
          continue;
        }
        DCHECK(IsTidy());
        return Status::OK();
      }
      // Call the scanner to handling error. If the error is successfully handled,
      // continue parsing. Since parsing have been interrupted and we may be stopped in
      // the middle of a JSON object, we need to move to the starting position of the
      // next object and reset the parser status before starting the next parse.
      // But there is a special case where the error code reported by the parser is
      // kParseErrorObjectMissCommaOrCurlyBracket, indicating that the current JSON
      // object is missing a closing curly bracket. In this case, we should already be
      // stopped at the end of this JSON object and there is no need to move to the
      // starting position of the next object. If MoveToNextJson() is still called in
      // this case, it is highly likely to cause us to miss a complete row.
      RETURN_IF_ERROR(scanner_->HandleError(reader_.GetParseErrorCode(),
          reader_.GetErrorOffset()));
      if (row_initialized_) FinishRow();
      if (reader_.GetParseErrorCode() != kParseErrorObjectMissCommaOrCurlyBracket) {
        MoveToNextJson();
      }
      ResetParser();
    }

    ++(*num_rows);
    if (UNLIKELY(scanner_->BreakParse())) break;
  }

  return Status::OK();
}

template <class Scanner>
Status JsonParser<Scanner>::CountJsonObjects(int max_rows, int* num_rows) {