| 1 | #include <DataStreams/FinishSortingBlockInputStream.h> |
| 2 | #include <DataStreams/MergeSortingBlockInputStream.h> |
| 3 | #include <DataStreams/processConstants.h> |
| 4 | |
| 5 | |
| 6 | namespace DB |
| 7 | { |
| 8 | |
| 9 | namespace ErrorCodes |
| 10 | { |
| 11 | extern const int LOGICAL_ERROR; |
| 12 | } |
| 13 | |
| 14 | static bool isPrefix(const SortDescription & pref_descr, const SortDescription & descr) |
| 15 | { |
| 16 | if (pref_descr.size() > descr.size()) |
| 17 | return false; |
| 18 | |
| 19 | for (size_t i = 0; i < pref_descr.size(); ++i) |
| 20 | if (pref_descr[i] != descr[i]) |
| 21 | return false; |
| 22 | |
| 23 | return true; |
| 24 | } |
| 25 | |
| 26 | FinishSortingBlockInputStream::FinishSortingBlockInputStream( |
| 27 | const BlockInputStreamPtr & input, const SortDescription & description_sorted_, |
| 28 | const SortDescription & description_to_sort_, |
| 29 | size_t max_merged_block_size_, UInt64 limit_) |
| 30 | : description_sorted(description_sorted_), description_to_sort(description_to_sort_), |
| 31 | max_merged_block_size(max_merged_block_size_), limit(limit_) |
| 32 | { |
| 33 | if (!isPrefix(description_sorted, description_to_sort)) |
| 34 | throw Exception("Can`t finish sorting. SortDescription of already sorted stream is not prefix of " |
| 35 | "SortDescription needed to sort" , ErrorCodes::LOGICAL_ERROR); |
| 36 | |
| 37 | children.push_back(input); |
| 38 | header = children.at(0)->getHeader(); |
| 39 | removeConstantsFromSortDescription(header, description_to_sort); |
| 40 | } |
| 41 | |
| 42 | |
| 43 | struct Less |
| 44 | { |
| 45 | const ColumnsWithSortDescriptions & left_columns; |
| 46 | const ColumnsWithSortDescriptions & right_columns; |
| 47 | |
| 48 | Less(const ColumnsWithSortDescriptions & left_columns_, const ColumnsWithSortDescriptions & right_columns_) : |
| 49 | left_columns(left_columns_), right_columns(right_columns_) {} |
| 50 | |
| 51 | bool operator() (size_t a, size_t b) const |
| 52 | { |
| 53 | for (auto it = left_columns.begin(), jt = right_columns.begin(); it != left_columns.end(); ++it, ++jt) |
| 54 | { |
| 55 | int res = it->description.direction * it->column->compareAt(a, b, *jt->column, it->description.nulls_direction); |
| 56 | if (res < 0) |
| 57 | return true; |
| 58 | else if (res > 0) |
| 59 | return false; |
| 60 | } |
| 61 | return false; |
| 62 | } |
| 63 | }; |
| 64 | |
| 65 | |
| 66 | Block FinishSortingBlockInputStream::readImpl() |
| 67 | { |
| 68 | if (limit && total_rows_processed >= limit) |
| 69 | return {}; |
| 70 | |
| 71 | Block res; |
| 72 | if (impl) |
| 73 | res = impl->read(); |
| 74 | |
| 75 | /// If res block is empty, we have finished sorting previous chunk of blocks. |
| 76 | if (!res) |
| 77 | { |
| 78 | if (end_of_stream) |
| 79 | return {}; |
| 80 | |
| 81 | blocks.clear(); |
| 82 | if (tail_block) |
| 83 | blocks.push_back(std::move(tail_block)); |
| 84 | |
| 85 | while (true) |
| 86 | { |
| 87 | Block block = children.back()->read(); |
| 88 | |
| 89 | /// End of input stream, but we can`t return immediately, we need to merge already read blocks. |
| 90 | /// Check it later, when get end of stream from impl. |
| 91 | if (!block) |
| 92 | { |
| 93 | end_of_stream = true; |
| 94 | break; |
| 95 | } |
| 96 | |
| 97 | // If there were only const columns in sort description, then there is no need to sort. |
| 98 | // Return the blocks as is. |
| 99 | if (description_to_sort.empty()) |
| 100 | return block; |
| 101 | |
| 102 | if (block.rows() == 0) |
| 103 | continue; |
| 104 | |
| 105 | |
| 106 | removeConstantsFromBlock(block); |
| 107 | |
| 108 | /// Find the position of last already read key in current block. |
| 109 | if (!blocks.empty()) |
| 110 | { |
| 111 | const Block & last_block = blocks.back(); |
| 112 | auto last_columns = getColumnsWithSortDescription(last_block, description_sorted); |
| 113 | auto current_columns = getColumnsWithSortDescription(block, description_sorted); |
| 114 | |
| 115 | Less less(last_columns, current_columns); |
| 116 | |
| 117 | size_t size = block.rows(); |
| 118 | IColumn::Permutation perm(size); |
| 119 | for (size_t i = 0; i < size; ++i) |
| 120 | perm[i] = i; |
| 121 | |
| 122 | auto it = std::upper_bound(perm.begin(), perm.end(), last_block.rows() - 1, less); |
| 123 | |
| 124 | /// We need to save tail of block, because next block may starts with the same key as in tail |
| 125 | /// and we should sort these rows in one chunk. |
| 126 | if (it != perm.end()) |
| 127 | { |
| 128 | size_t tail_pos = it - perm.begin(); |
| 129 | Block head_block = block.cloneEmpty(); |
| 130 | tail_block = block.cloneEmpty(); |
| 131 | |
| 132 | for (size_t i = 0; i < block.columns(); ++i) |
| 133 | { |
| 134 | head_block.getByPosition(i).column = block.getByPosition(i).column->cut(0, tail_pos); |
| 135 | tail_block.getByPosition(i).column = block.getByPosition(i).column->cut(tail_pos, block.rows() - tail_pos); |
| 136 | } |
| 137 | |
| 138 | if (head_block.rows()) |
| 139 | blocks.push_back(head_block); |
| 140 | |
| 141 | break; |
| 142 | } |
| 143 | } |
| 144 | |
| 145 | /// If we reach here, that means that current block is first in chunk |
| 146 | /// or it all consists of rows with the same key as tail of a previous block. |
| 147 | blocks.push_back(block); |
| 148 | } |
| 149 | |
| 150 | if (!blocks.empty()) |
| 151 | { |
| 152 | impl = std::make_unique<MergeSortingBlocksBlockInputStream>(blocks, description_to_sort, max_merged_block_size, limit); |
| 153 | res = impl->read(); |
| 154 | } |
| 155 | } |
| 156 | |
| 157 | if (res) |
| 158 | enrichBlockWithConstants(res, header); |
| 159 | |
| 160 | total_rows_processed += res.rows(); |
| 161 | |
| 162 | return res; |
| 163 | } |
| 164 | } |
| 165 | |