Method ParseLine

cpp/src/arrow/csv/parser.cc:280–469

  template <typename SpecializedOptions, bool UseBulkFilter, typename ValueDescWriter,
            typename DataWriter, typename BulkFilter>
  Status ParseLine(ValueDescWriter* values_writer, DataWriter* parsed_writer,
                   const char* data, const char* data_end, bool is_final,
                   const char** out_data, const BulkFilter& bulk_filter) {
    int32_t num_cols = 0;
    char c;
    const auto start = data;

    DCHECK_GT(data_end, data);

    auto FinishField = [&]() { values_writer->FinishField(parsed_writer); };

    values_writer->BeginLine();
    parsed_writer->BeginLine();

    // The parsing state machine

    // Special case empty lines: do we start with a newline separator?
    c = *data;
    if (ARROW_PREDICT_FALSE(IsControlChar(c))) {
      if (c == '\r') {
        data++;
        if (data < data_end && *data == '\n') {
          data++;
        }
        goto EmptyLine;
      }
      if (c == '\n') {
        data++;
        goto EmptyLine;
      }
    }

  FieldStart:
    // At the start of a field
    if (*data == options_.delimiter) {
      // Empty cells are very common in some files, shortcut them
      values_writer->StartField(false /* quoted */);
      FinishField();
      ++data;
      ++num_cols;
      if (ARROW_PREDICT_FALSE(data == data_end)) {
        goto AbortLine;
      }
      goto FieldStart;
    }

    // Quoting is only recognized at start of field
    if (SpecializedOptions::quoting &&
        ARROW_PREDICT_FALSE(*data == options_.quote_char)) {
      ++data;
      values_writer->StartField(true /* quoted */);
      goto InQuotedField;
    } else {
      values_writer->StartField(false /* quoted */);
      goto InField;
    }

  InField: