| 70 | } |
| 71 | |
| 72 | void ProcessRecordBatch(std::shared_ptr<arrow::Schema> schema, |
| 73 | std::shared_ptr<arrow::RecordBatch> record_batch, |
| 74 | std::int64_t num_rows) { |
| 75 | // If you want to see what the result looks like without parsing the |
| 76 | // datatypes, use `record_batch->ToString()` for quick debugging. |
| 77 | // Note: you might need to adjust the formatting depending on how big the data |
| 78 | // in your table is. |
| 79 | for (std::int64_t row = 0; row < record_batch->num_rows(); ++row) { |
| 80 | std::cout << std::format("Row {}: ", row + num_rows); |
| 81 | |
| 82 | for (std::int64_t col = 0; col < record_batch->num_columns(); ++col) { |
| 83 | std::shared_ptr<arrow::Array> column = record_batch->column(col); |
| 84 | arrow::Result<std::shared_ptr<arrow::Scalar> > result = |
| 85 | column->GetScalar(row); |
| 86 | if (!result.ok()) { |
| 87 | std::cout << "Unable to parse scalar\n"; |
| 88 | throw result.status(); |
| 89 | } |
| 90 | |
| 91 | std::shared_ptr<arrow::Scalar> scalar = result.ValueOrDie(); |
| 92 | switch (scalar->type->id()) { |
| 93 | case arrow::Type::INT64: |
| 94 | std::cout |
| 95 | << std::left << std::setw(15) |
| 96 | << std::dynamic_pointer_cast<arrow::Int64Scalar>(scalar)->value |
| 97 | << " "; |
| 98 | break; |
| 99 | case arrow::Type::STRING: |
| 100 | std::cout |
| 101 | << std::left << std::setw(15) |
| 102 | << std::dynamic_pointer_cast<arrow::StringScalar>(scalar)->view() |
| 103 | << " "; |
| 104 | break; |
| 105 | // Depending on the table you are reading, you might need to add cases |
| 106 | // for other datatypes here. The schema will tell you what datatypes |
| 107 | // need to be handled. |
| 108 | default: |
| 109 | std::cout << std::left << std::setw(15) << "UNDEFINED "; |
| 110 | } |
| 111 | } |
| 112 | std::cout << "\n"; |
| 113 | } |
| 114 | } |
| 115 | |
| 116 | } // namespace |
| 117 | |