Method Make

r/src/altrep.cpp:452–518

  }

  static SEXP Make(const std::shared_ptr<ChunkedArray>& chunked_array) {
    // only dealing with dictionaries of strings
    if (internal::checked_cast<const DictionaryArray&>(*chunked_array->chunk(0))
            .dictionary()
            ->type_id() != Type::STRING) {
      return R_NilValue;
    }

    bool need_unification = DictionaryChunkArrayNeedUnification(chunked_array);

    std::shared_ptr<Array> dictionary;
    SEXP pointer_arrays_transpose;

    if (need_unification) {
      const auto& arr_type =
          internal::checked_cast<const DictionaryType&>(*chunked_array->type());
      std::unique_ptr<arrow::DictionaryUnifier> unifier_ =
          ValueOrStop(DictionaryUnifier::Make(arr_type.value_type()));

      int n_arrays = chunked_array->num_chunks();
      BufferVector arrays_transpose(n_arrays);

      for (int i = 0; i < n_arrays; i++) {
        const auto& dict_i =
            *internal::checked_cast<const DictionaryArray&>(*chunked_array->chunk(i))
                 .dictionary();
        StopIfNotOk(unifier_->Unify(dict_i, &arrays_transpose[i]));
      }

      std::shared_ptr<DataType> out_type;
      StopIfNotOk(unifier_->GetResult(&out_type, &dictionary));

      Pointer<BufferVector> ptr(
          new std::shared_ptr<BufferVector>(new BufferVector(arrays_transpose)));
      pointer_arrays_transpose = PROTECT(ptr);
    } else {
      // just use the first one
      const auto& dict_array =
          internal::checked_cast<const DictionaryArray&>(*chunked_array->chunk(0));
      dictionary = dict_array.dictionary();

      pointer_arrays_transpose = PROTECT(R_NilValue);
    }

    // the chunked array as data1
    SEXP data1 =
        PROTECT(external_pointer<ArrowAltrepData>(new ArrowAltrepData(chunked_array)));

    // a pairlist with the representation in the first node
    SEXP data2 = PROTECT(Rf_list2(R_NilValue,  // representation, empty at first
                                  pointer_arrays_transpose));

    SEXP alt = PROTECT(R_new_altrep(class_t, data1, data2));
    MARK_NOT_MUTABLE(alt);

    // set factor attributes
    Rf_setAttrib(alt, R_LevelsSymbol, Array__as_vector(dictionary));