| 1 | #include "duckdb/execution/physical_plan_generator.hpp" |
| 2 | #include "duckdb/execution/operator/persistent/physical_copy_to_file.hpp" |
| 3 | #include "duckdb/execution/operator/persistent/physical_batch_copy_to_file.hpp" |
| 4 | #include "duckdb/execution/operator/persistent/physical_fixed_batch_copy.hpp" |
| 5 | #include "duckdb/planner/operator/logical_copy_to_file.hpp" |
| 6 | |
| 7 | namespace duckdb { |
| 8 | |
| 9 | unique_ptr<PhysicalOperator> PhysicalPlanGenerator::CreatePlan(LogicalCopyToFile &op) { |
| 10 | auto plan = CreatePlan(op&: *op.children[0]); |
| 11 | bool preserve_insertion_order = PhysicalPlanGenerator::PreserveInsertionOrder(context, plan&: *plan); |
| 12 | bool supports_batch_index = PhysicalPlanGenerator::UseBatchIndex(context, plan&: *plan); |
| 13 | auto &fs = FileSystem::GetFileSystem(context); |
| 14 | op.file_path = fs.ExpandPath(path: op.file_path); |
| 15 | if (op.use_tmp_file) { |
| 16 | op.file_path += ".tmp" ; |
| 17 | } |
| 18 | if (op.per_thread_output || op.partition_output || !op.partition_columns.empty() || op.overwrite_or_ignore) { |
| 19 | // hive-partitioning/per-thread output does not care about insertion order, and does not support batch indexes |
| 20 | preserve_insertion_order = false; |
| 21 | supports_batch_index = false; |
| 22 | } |
| 23 | auto mode = CopyFunctionExecutionMode::REGULAR_COPY_TO_FILE; |
| 24 | if (op.function.execution_mode) { |
| 25 | mode = op.function.execution_mode(preserve_insertion_order, supports_batch_index); |
| 26 | } |
| 27 | if (mode == CopyFunctionExecutionMode::BATCH_COPY_TO_FILE) { |
| 28 | if (!supports_batch_index) { |
| 29 | throw InternalException("BATCH_COPY_TO_FILE can only be used if batch indexes are supported" ); |
| 30 | } |
| 31 | // batched copy to file |
| 32 | if (op.function.desired_batch_size) { |
| 33 | auto copy = make_uniq<PhysicalFixedBatchCopy>(args&: op.types, args&: op.function, args: std::move(op.bind_data), |
| 34 | args&: op.estimated_cardinality); |
| 35 | copy->file_path = op.file_path; |
| 36 | copy->use_tmp_file = op.use_tmp_file; |
| 37 | copy->children.push_back(x: std::move(plan)); |
| 38 | return std::move(copy); |
| 39 | } else { |
| 40 | auto copy = make_uniq<PhysicalBatchCopyToFile>(args&: op.types, args&: op.function, args: std::move(op.bind_data), |
| 41 | args&: op.estimated_cardinality); |
| 42 | copy->file_path = op.file_path; |
| 43 | copy->use_tmp_file = op.use_tmp_file; |
| 44 | copy->children.push_back(x: std::move(plan)); |
| 45 | return std::move(copy); |
| 46 | } |
| 47 | } |
| 48 | // COPY from select statement to file |
| 49 | auto copy = make_uniq<PhysicalCopyToFile>(args&: op.types, args&: op.function, args: std::move(op.bind_data), args&: op.estimated_cardinality); |
| 50 | copy->file_path = op.file_path; |
| 51 | copy->use_tmp_file = op.use_tmp_file; |
| 52 | copy->overwrite_or_ignore = op.overwrite_or_ignore; |
| 53 | copy->filename_pattern = op.filename_pattern; |
| 54 | copy->per_thread_output = op.per_thread_output; |
| 55 | copy->partition_output = op.partition_output; |
| 56 | copy->partition_columns = op.partition_columns; |
| 57 | copy->names = op.names; |
| 58 | copy->expected_types = op.expected_types; |
| 59 | copy->parallel = mode == CopyFunctionExecutionMode::PARALLEL_COPY_TO_FILE; |
| 60 | |
| 61 | copy->children.push_back(x: std::move(plan)); |
| 62 | return std::move(copy); |
| 63 | } |
| 64 | |
| 65 | } // namespace duckdb |
| 66 | |