| 1 | #include "duckdb/storage/table/scan_state.hpp" |
| 2 | #include "duckdb/storage/table/row_group.hpp" |
| 3 | #include "duckdb/storage/table/column_segment.hpp" |
| 4 | #include "duckdb/transaction/duck_transaction.hpp" |
| 5 | #include "duckdb/storage/table/column_data.hpp" |
| 6 | #include "duckdb/storage/table/row_group_collection.hpp" |
| 7 | #include "duckdb/storage/table/row_group_segment_tree.hpp" |
| 8 | |
| 9 | namespace duckdb { |
| 10 | |
| 11 | void TableScanState::Initialize(vector<column_t> column_ids, TableFilterSet *table_filters) { |
| 12 | this->column_ids = std::move(column_ids); |
| 13 | this->table_filters = table_filters; |
| 14 | if (table_filters) { |
| 15 | D_ASSERT(table_filters->filters.size() > 0); |
| 16 | this->adaptive_filter = make_uniq<AdaptiveFilter>(args&: table_filters); |
| 17 | } |
| 18 | } |
| 19 | |
| 20 | const vector<column_t> &TableScanState::GetColumnIds() { |
| 21 | D_ASSERT(!column_ids.empty()); |
| 22 | return column_ids; |
| 23 | } |
| 24 | |
| 25 | TableFilterSet *TableScanState::GetFilters() { |
| 26 | D_ASSERT(!table_filters || adaptive_filter.get()); |
| 27 | return table_filters; |
| 28 | } |
| 29 | |
| 30 | AdaptiveFilter *TableScanState::GetAdaptiveFilter() { |
| 31 | return adaptive_filter.get(); |
| 32 | } |
| 33 | |
| 34 | void ColumnScanState::NextInternal(idx_t count) { |
| 35 | if (!current) { |
| 36 | //! There is no column segment |
| 37 | return; |
| 38 | } |
| 39 | row_index += count; |
| 40 | while (row_index >= current->start + current->count) { |
| 41 | current = segment_tree->GetNextSegment(segment: current); |
| 42 | initialized = false; |
| 43 | segment_checked = false; |
| 44 | if (!current) { |
| 45 | break; |
| 46 | } |
| 47 | } |
| 48 | D_ASSERT(!current || (row_index >= current->start && row_index < current->start + current->count)); |
| 49 | } |
| 50 | |
| 51 | void ColumnScanState::Next(idx_t count) { |
| 52 | NextInternal(count); |
| 53 | for (auto &child_state : child_states) { |
| 54 | child_state.Next(count); |
| 55 | } |
| 56 | } |
| 57 | |
| 58 | const vector<storage_t> &CollectionScanState::GetColumnIds() { |
| 59 | return parent.GetColumnIds(); |
| 60 | } |
| 61 | |
| 62 | TableFilterSet *CollectionScanState::GetFilters() { |
| 63 | return parent.GetFilters(); |
| 64 | } |
| 65 | |
| 66 | AdaptiveFilter *CollectionScanState::GetAdaptiveFilter() { |
| 67 | return parent.GetAdaptiveFilter(); |
| 68 | } |
| 69 | |
| 70 | ParallelCollectionScanState::ParallelCollectionScanState() |
| 71 | : collection(nullptr), current_row_group(nullptr), processed_rows(0) { |
| 72 | } |
| 73 | |
| 74 | CollectionScanState::CollectionScanState(TableScanState &parent_p) |
| 75 | : row_group(nullptr), vector_index(0), max_row_group_row(0), row_groups(nullptr), max_row(0), batch_index(0), |
| 76 | parent(parent_p) { |
| 77 | } |
| 78 | |
| 79 | bool CollectionScanState::Scan(DuckTransaction &transaction, DataChunk &result) { |
| 80 | while (row_group) { |
| 81 | row_group->Scan(transaction, state&: *this, result); |
| 82 | if (result.size() > 0) { |
| 83 | return true; |
| 84 | } else if (max_row <= row_group->start + row_group->count) { |
| 85 | row_group = nullptr; |
| 86 | return false; |
| 87 | } else { |
| 88 | do { |
| 89 | row_group = row_groups->GetNextSegment(segment: row_group); |
| 90 | if (row_group) { |
| 91 | if (row_group->start >= max_row) { |
| 92 | row_group = nullptr; |
| 93 | break; |
| 94 | } |
| 95 | bool scan_row_group = row_group->InitializeScan(state&: *this); |
| 96 | if (scan_row_group) { |
| 97 | // scan this row group |
| 98 | break; |
| 99 | } |
| 100 | } |
| 101 | } while (row_group); |
| 102 | } |
| 103 | } |
| 104 | return false; |
| 105 | } |
| 106 | |
| 107 | bool CollectionScanState::ScanCommitted(DataChunk &result, SegmentLock &l, TableScanType type) { |
| 108 | while (row_group) { |
| 109 | row_group->ScanCommitted(state&: *this, result, type); |
| 110 | if (result.size() > 0) { |
| 111 | return true; |
| 112 | } else { |
| 113 | row_group = row_groups->GetNextSegment(l, segment: row_group); |
| 114 | if (row_group) { |
| 115 | row_group->InitializeScan(state&: *this); |
| 116 | } |
| 117 | } |
| 118 | } |
| 119 | return false; |
| 120 | } |
| 121 | |
| 122 | bool CollectionScanState::ScanCommitted(DataChunk &result, TableScanType type) { |
| 123 | while (row_group) { |
| 124 | row_group->ScanCommitted(state&: *this, result, type); |
| 125 | if (result.size() > 0) { |
| 126 | return true; |
| 127 | } else { |
| 128 | row_group = row_groups->GetNextSegment(segment: row_group); |
| 129 | if (row_group) { |
| 130 | row_group->InitializeScan(state&: *this); |
| 131 | } |
| 132 | } |
| 133 | } |
| 134 | return false; |
| 135 | } |
| 136 | |
| 137 | } // namespace duckdb |
| 138 | |