be/src/storage/segment/column_writer.h
Line | Count | Source |
1 | | // Licensed to the Apache Software Foundation (ASF) under one |
2 | | // or more contributor license agreements. See the NOTICE file |
3 | | // distributed with this work for additional information |
4 | | // regarding copyright ownership. The ASF licenses this file |
5 | | // to you under the Apache License, Version 2.0 (the |
6 | | // "License"); you may not use this file except in compliance |
7 | | // with the License. You may obtain a copy of the License at |
8 | | // |
9 | | // http://www.apache.org/licenses/LICENSE-2.0 |
10 | | // |
11 | | // Unless required by applicable law or agreed to in writing, |
12 | | // software distributed under the License is distributed on an |
13 | | // "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY |
14 | | // KIND, either express or implied. See the License for the |
15 | | // specific language governing permissions and limitations |
16 | | // under the License. |
17 | | |
18 | | #pragma once |
19 | | |
20 | | #include <gen_cpp/segment_v2.pb.h> |
21 | | #include <stddef.h> |
22 | | #include <stdint.h> |
23 | | |
24 | | #include <algorithm> |
25 | | #include <memory> // for unique_ptr |
26 | | #include <ostream> |
27 | | #include <string> |
28 | | #include <unordered_map> |
29 | | #include <utility> |
30 | | #include <vector> |
31 | | |
32 | | #include "common/status.h" // for Status |
33 | | #include "core/column/column_variant.h" |
34 | | #include "storage/field.h" // for StorageField |
35 | | #include "storage/index/ann/ann_index_writer.h" |
36 | | #include "storage/index/bloom_filter/bloom_filter.h" |
37 | | #include "storage/index/inverted/inverted_index_writer.h" |
38 | | #include "storage/segment/common.h" |
39 | | #include "storage/segment/options.h" |
40 | | #include "storage/segment/variant/nested_group_provider.h" |
41 | | #include "storage/segment/variant/variant_statistics.h" |
42 | | #include "util/bitmap.h" // for BitmapChange |
43 | | #include "util/slice.h" // for OwnedSlice |
44 | | |
45 | | namespace doris { |
46 | | |
47 | | class BlockCompressionCodec; |
48 | | class TabletColumn; |
49 | | class TabletIndex; |
50 | | struct RowsetWriterContext; |
51 | | |
52 | | namespace io { |
53 | | class FileWriter; |
54 | | } |
55 | | |
56 | | namespace segment_v2 { |
57 | | |
58 | | struct ColumnWriterOptions { |
59 | | // input and output parameter: |
60 | | // - input: column_id/unique_id/type/length/encoding/compression/is_nullable members |
61 | | // - output: encoding/indexes/dict_page members |
62 | | ColumnMetaPB* meta = nullptr; |
63 | | size_t data_page_size = STORAGE_PAGE_SIZE_DEFAULT_VALUE; |
64 | | size_t dict_page_size = STORAGE_DICT_PAGE_SIZE_DEFAULT_VALUE; |
65 | | // store compressed page only when space saving is above the threshold. |
66 | | // space saving = 1 - compressed_size / uncompressed_size |
67 | | double compression_min_space_saving = 0.1; |
68 | | bool need_zone_map = false; |
69 | | bool need_bloom_filter = false; |
70 | | bool is_ngram_bf_index = false; |
71 | | bool need_inverted_index = false; |
72 | | bool need_ann_index = false; |
73 | | uint8_t gram_size; |
74 | | uint16_t gram_bf_size; |
75 | | BloomFilterOptions bf_options; |
76 | | std::vector<const TabletIndex*> inverted_indexes; |
77 | | IndexFileWriter* index_file_writer = nullptr; |
78 | | |
79 | | SegmentFooterPB* footer = nullptr; |
80 | | io::FileWriter* file_writer = nullptr; |
81 | | CompressionTypePB compression_type = UNKNOWN_COMPRESSION; |
82 | | RowsetWriterContext* rowset_ctx = nullptr; |
83 | | // For collect segment statistics for compaction |
84 | | std::vector<RowsetReaderSharedPtr> input_rs_readers; |
85 | | const TabletIndex* ann_index = nullptr; |
86 | | |
87 | | EncodingPreference encoding_preference {}; |
88 | | |
89 | 0 | std::string to_string() const { |
90 | 0 | std::stringstream ss; |
91 | 0 | ss << std::boolalpha << "meta=" << meta->DebugString() |
92 | 0 | << ", data_page_size=" << data_page_size << ", dict_page_size=" << dict_page_size |
93 | 0 | << ", compression_min_space_saving = " << compression_min_space_saving |
94 | 0 | << ", need_zone_map=" << need_zone_map << ", need_bloom_filter" << need_bloom_filter; |
95 | 0 | return ss.str(); |
96 | 0 | } |
97 | | }; |
98 | | |
99 | | class EncodingInfo; |
100 | | class NullBitmapBuilder; |
101 | | class OrdinalIndexWriter; |
102 | | class PageBuilder; |
103 | | class BloomFilterIndexWriter; |
104 | | class ZoneMapIndexWriter; |
105 | | class VariantColumnWriterImpl; |
106 | | class ColumnWriter; |
107 | | |
108 | | class ColumnWriter { |
109 | | public: |
110 | | static Status create(const ColumnWriterOptions& opts, const TabletColumn* column, |
111 | | io::FileWriter* file_writer, std::unique_ptr<ColumnWriter>* writer); |
112 | | static Status create_struct_writer(const ColumnWriterOptions& opts, const TabletColumn* column, |
113 | | io::FileWriter* file_writer, |
114 | | std::unique_ptr<ColumnWriter>* writer); |
115 | | static Status create_array_writer(const ColumnWriterOptions& opts, const TabletColumn* column, |
116 | | io::FileWriter* file_writer, |
117 | | std::unique_ptr<ColumnWriter>* writer); |
118 | | static Status create_map_writer(const ColumnWriterOptions& opts, const TabletColumn* column, |
119 | | io::FileWriter* file_writer, |
120 | | std::unique_ptr<ColumnWriter>* writer); |
121 | | |
122 | | static Status create_variant_writer(const ColumnWriterOptions& opts, const TabletColumn* column, |
123 | | io::FileWriter* file_writer, |
124 | | std::unique_ptr<ColumnWriter>* writer); |
125 | | |
126 | | static Status create_agg_state_writer(const ColumnWriterOptions& opts, |
127 | | const TabletColumn* column, io::FileWriter* file_writer, |
128 | | std::unique_ptr<ColumnWriter>* writer); |
129 | | |
130 | | explicit ColumnWriter(std::unique_ptr<StorageField> field, bool is_nullable, |
131 | | ColumnMetaPB* meta); |
132 | | |
133 | 1.04M | virtual ~ColumnWriter() = default; |
134 | | |
135 | | virtual Status init() = 0; |
136 | | |
137 | | template <typename CellType> |
138 | 651k | Status append(const CellType& cell) { |
139 | 651k | if (_is_nullable) { |
140 | 651k | uint8_t nullmap = 0; |
141 | 651k | BitmapChange(&nullmap, 0, cell.is_null()); |
142 | 651k | return append_nullable(&nullmap, cell.cell_ptr(), 1); |
143 | 651k | } else { |
144 | 4 | auto* cel_ptr = cell.cell_ptr(); |
145 | 4 | return append_data((const uint8_t**)&cel_ptr, 1); |
146 | 4 | } |
147 | 651k | } |
148 | | |
149 | | // Now we only support append one by one, we should support append |
150 | | // multi rows in one call |
151 | 0 | Status append(bool is_null, void* data) { |
152 | 0 | uint8_t nullmap = 0; |
153 | 0 | BitmapChange(&nullmap, 0, is_null); |
154 | 0 | return append_nullable(&nullmap, data, 1); |
155 | 0 | } |
156 | | |
157 | | Status append(const uint8_t* nullmap, const void* data, size_t num_rows); |
158 | | |
159 | | Status append_nullable(const uint8_t* nullmap, const void* data, size_t num_rows); |
160 | | |
161 | | // use only in vectorized load |
162 | | virtual Status append_nullable(const uint8_t* null_map, const uint8_t** data, size_t num_rows); |
163 | | |
164 | | virtual Status append_nulls(size_t num_rows) = 0; |
165 | | |
166 | | virtual Status finish_current_page() = 0; |
167 | | |
168 | | virtual uint64_t estimate_buffer_size() = 0; |
169 | | |
170 | | // finish append data |
171 | | virtual Status finish() = 0; |
172 | | |
173 | | // write all data into file |
174 | | virtual Status write_data() = 0; |
175 | | |
176 | | virtual Status write_ordinal_index() = 0; |
177 | | |
178 | | virtual Status write_zone_map() = 0; |
179 | | |
180 | | virtual Status write_inverted_index() = 0; |
181 | | |
182 | 674k | virtual Status write_ann_index() { return Status::OK(); } |
183 | | |
184 | | virtual Status write_bloom_filter_index() = 0; |
185 | | |
186 | | virtual ordinal_t get_next_rowid() const = 0; |
187 | | |
188 | | virtual uint64_t get_raw_data_bytes() const = 0; |
189 | | virtual uint64_t get_total_uncompressed_data_pages_bytes() const = 0; |
190 | | virtual uint64_t get_total_compressed_data_pages_bytes() const = 0; |
191 | | |
192 | | // used for append not null data. |
193 | | virtual Status append_data(const uint8_t** ptr, size_t num_rows) = 0; |
194 | | |
195 | 6.54M | bool is_nullable() const { return _is_nullable; } |
196 | | |
197 | 8.81M | StorageField* get_field() const { return _field.get(); } |
198 | | |
199 | 716k | ColumnMetaPB* get_column_meta() const { return _column_meta; } |
200 | | |
201 | | protected: |
202 | | DataTypePtr _data_type; |
203 | | |
204 | | private: |
205 | | std::unique_ptr<StorageField> _field; |
206 | | bool _is_nullable; |
207 | | ColumnMetaPB* _column_meta; |
208 | | std::vector<uint8_t> _null_bitmap; |
209 | | }; |
210 | | |
211 | | class FlushPageCallback { |
212 | | public: |
213 | 69.4k | virtual ~FlushPageCallback() = default; |
214 | 0 | virtual void put_extra_info_in_page(DataPageFooterPB* footer) {} |
215 | | }; |
216 | | |
217 | | // Encode one column's data into some memory slice. |
218 | | // Because some columns would be stored in a file, we should wait |
219 | | // until all columns has been finished, and then data can be written |
220 | | // to file |
221 | | class ScalarColumnWriter : public ColumnWriter { |
222 | | public: |
223 | | ScalarColumnWriter(const ColumnWriterOptions& opts, std::unique_ptr<StorageField> field, |
224 | | io::FileWriter* file_writer); |
225 | | |
226 | | ~ScalarColumnWriter() override; |
227 | | |
228 | | Status init() override; |
229 | | |
230 | | Status append_nulls(size_t num_rows) override; |
231 | | |
232 | | Status finish_current_page() override; |
233 | | |
234 | | uint64_t estimate_buffer_size() override; |
235 | | |
236 | | // finish append data |
237 | | Status finish() override; |
238 | | |
239 | | Status write_data() override; |
240 | | Status write_ordinal_index() override; |
241 | | Status write_zone_map() override; |
242 | | Status write_inverted_index() override; |
243 | | Status write_bloom_filter_index() override; |
244 | 175k | ordinal_t get_next_rowid() const override { return _next_rowid; } |
245 | | |
246 | 818k | uint64_t get_raw_data_bytes() const override { return _raw_data_bytes; } |
247 | | |
248 | 818k | uint64_t get_total_uncompressed_data_pages_bytes() const override { |
249 | 818k | return _total_uncompressed_data_pages_size; |
250 | 818k | } |
251 | | |
252 | 818k | uint64_t get_total_compressed_data_pages_bytes() const override { |
253 | 818k | return _total_compressed_data_pages_size; |
254 | 818k | } |
255 | | |
256 | 69.4k | void register_flush_page_callback(FlushPageCallback* flush_page_callback) { |
257 | 69.4k | _new_page_callback = flush_page_callback; |
258 | 69.4k | } |
259 | | Status append_data(const uint8_t** ptr, size_t num_rows) override; |
260 | | Status append_nullable(const uint8_t* null_map, const uint8_t** ptr, size_t num_rows) override; |
261 | | |
262 | | // used for append not null data. When page is full, will append data not reach num_rows. |
263 | | Status append_data_in_current_page(const uint8_t** ptr, size_t* num_written); |
264 | | |
265 | 4.44M | Status append_data_in_current_page(const uint8_t* ptr, size_t* num_written) { |
266 | 4.44M | RETURN_IF_CATCH_EXCEPTION( |
267 | 4.44M | { return _internal_append_data_in_current_page(ptr, num_written); }); |
268 | 4.44M | } |
269 | | friend class ArrayColumnWriter; |
270 | | friend class OffsetColumnWriter; |
271 | | |
272 | | private: |
273 | | Status _internal_append_data_in_current_page(const uint8_t* ptr, size_t* num_written); |
274 | | |
275 | | private: |
276 | | struct NullRun { |
277 | | bool is_null; |
278 | | uint32_t len; |
279 | | }; |
280 | | |
281 | | std::vector<NullRun> _null_run_buffer; |
282 | | std::unique_ptr<PageBuilder> _page_builder; |
283 | | |
284 | | std::unique_ptr<NullBitmapBuilder> _null_bitmap_builder; |
285 | | |
286 | | ColumnWriterOptions _opts; |
287 | | |
288 | | const EncodingInfo* _encoding_info = nullptr; |
289 | | |
290 | | ordinal_t _next_rowid = 0; |
291 | | |
292 | | // All Pages will be organized into a linked list |
293 | | struct Page { |
294 | | // the data vector may contain: |
295 | | // 1. one OwnedSlice if the page body is compressed |
296 | | // 2. one OwnedSlice if the page body is not compressed and doesn't have nullmap |
297 | | // 3. two OwnedSlice if the page body is not compressed and has nullmap |
298 | | // use vector for easier management for lifetime of OwnedSlice |
299 | | std::vector<OwnedSlice> data; |
300 | | PageFooterPB footer; |
301 | | }; |
302 | | |
303 | 1.00M | void _push_back_page(std::unique_ptr<Page> page) { |
304 | 1.91M | for (auto& data_slice : page->data) { |
305 | 1.91M | _data_size += data_slice.slice().size; |
306 | 1.91M | } |
307 | | // estimate (page footer + footer size + checksum) took 20 bytes |
308 | 1.00M | _data_size += 20; |
309 | | // add page to pages' tail |
310 | 1.00M | _pages.emplace_back(std::move(page)); |
311 | 1.00M | } |
312 | | |
313 | | Status _write_data_page(Page* page); |
314 | | |
315 | | private: |
316 | | io::FileWriter* _file_writer = nullptr; |
317 | | // total size of data page list |
318 | | uint64_t _data_size; |
319 | | |
320 | | uint64_t _raw_data_bytes {0}; |
321 | | uint64_t _total_uncompressed_data_pages_size {0}; |
322 | | uint64_t _total_compressed_data_pages_size {0}; |
323 | | |
324 | | // cached generated pages, |
325 | | std::vector<std::unique_ptr<Page>> _pages; |
326 | | ordinal_t _first_rowid = 0; |
327 | | |
328 | | BlockCompressionCodec* _compress_codec; |
329 | | |
330 | | std::unique_ptr<OrdinalIndexWriter> _ordinal_index_builder; |
331 | | std::unique_ptr<ZoneMapIndexWriter> _zone_map_index_builder; |
332 | | std::vector<std::unique_ptr<IndexColumnWriter>> _inverted_index_builders; |
333 | | std::unique_ptr<BloomFilterIndexWriter> _bloom_filter_index_builder; |
334 | | |
335 | | // call before flush data page. |
336 | | FlushPageCallback* _new_page_callback = nullptr; |
337 | | }; |
338 | | |
339 | | // offsetColumnWriter is used column which has offset column, like array, map. |
340 | | // column type is only uint64 and should response for whole column value [start, end], end will set |
341 | | // in footer.next_array_item_ordinal which in finish_cur_page() callback put_extra_info_in_page() |
342 | | class OffsetColumnWriter final : public ScalarColumnWriter, FlushPageCallback { |
343 | | public: |
344 | | OffsetColumnWriter(const ColumnWriterOptions& opts, std::unique_ptr<StorageField> field, |
345 | | io::FileWriter* file_writer); |
346 | | |
347 | | ~OffsetColumnWriter() override; |
348 | | |
349 | | Status init() override; |
350 | | |
351 | | Status append_data(const uint8_t** ptr, size_t num_rows) override; |
352 | | |
353 | | private: |
354 | | void put_extra_info_in_page(DataPageFooterPB* footer) override; |
355 | | |
356 | | uint64_t _next_offset; |
357 | | }; |
358 | | |
359 | | class StructColumnWriter final : public ColumnWriter { |
360 | | public: |
361 | | explicit StructColumnWriter(const ColumnWriterOptions& opts, |
362 | | std::unique_ptr<StorageField> field, |
363 | | ScalarColumnWriter* null_writer, |
364 | | std::vector<std::unique_ptr<ColumnWriter>>& sub_column_writers); |
365 | 2.30k | ~StructColumnWriter() override = default; |
366 | | |
367 | | Status init() override; |
368 | | |
369 | | Status append_nullable(const uint8_t* null_map, const uint8_t** data, size_t num_rows) override; |
370 | | Status append_data(const uint8_t** ptr, size_t num_rows) override; |
371 | | |
372 | | uint64_t estimate_buffer_size() override; |
373 | | |
374 | | Status finish() override; |
375 | | Status write_data() override; |
376 | | Status write_ordinal_index() override; |
377 | | Status append_nulls(size_t num_rows) override; |
378 | | |
379 | | Status finish_current_page() override; |
380 | | |
381 | 1.62k | Status write_zone_map() override { |
382 | 1.62k | if (_opts.need_zone_map) { |
383 | 0 | return Status::NotSupported("struct not support zone map"); |
384 | 0 | } |
385 | 1.62k | return Status::OK(); |
386 | 1.62k | } |
387 | | |
388 | | Status write_inverted_index() override; |
389 | 1.62k | Status write_bloom_filter_index() override { |
390 | 1.62k | if (_opts.need_bloom_filter) { |
391 | 0 | return Status::NotSupported("struct not support bloom filter index"); |
392 | 0 | } |
393 | 1.62k | return Status::OK(); |
394 | 1.62k | } |
395 | | |
396 | 2.69k | ordinal_t get_next_rowid() const override { return _sub_column_writers[0]->get_next_rowid(); } |
397 | | |
398 | 2.30k | uint64_t get_raw_data_bytes() const override { |
399 | 2.30k | return _get_total_data_pages_bytes(&ColumnWriter::get_raw_data_bytes); |
400 | 2.30k | } |
401 | | |
402 | 2.30k | uint64_t get_total_uncompressed_data_pages_bytes() const override { |
403 | 2.30k | return _get_total_data_pages_bytes(&ColumnWriter::get_total_uncompressed_data_pages_bytes); |
404 | 2.30k | } |
405 | | |
406 | 2.30k | uint64_t get_total_compressed_data_pages_bytes() const override { |
407 | 2.30k | return _get_total_data_pages_bytes(&ColumnWriter::get_total_compressed_data_pages_bytes); |
408 | 2.30k | } |
409 | | |
410 | | private: |
411 | | template <typename Func> |
412 | 6.91k | uint64_t _get_total_data_pages_bytes(Func func) const { |
413 | 6.91k | uint64_t size = is_nullable() ? std::invoke(func, _null_writer.get()) : 0; |
414 | 28.7k | for (const auto& writer : _sub_column_writers) { |
415 | 28.7k | size += std::invoke(func, writer.get()); |
416 | 28.7k | } |
417 | 6.91k | return size; |
418 | 6.91k | } |
419 | | |
420 | | private: |
421 | | size_t _num_sub_column_writers; |
422 | | std::unique_ptr<ScalarColumnWriter> _null_writer; |
423 | | std::vector<std::unique_ptr<ColumnWriter>> _sub_column_writers; |
424 | | ColumnWriterOptions _opts; |
425 | | }; |
426 | | |
427 | | class ArrayColumnWriter final : public ColumnWriter { |
428 | | public: |
429 | | explicit ArrayColumnWriter(const ColumnWriterOptions& opts, std::unique_ptr<StorageField> field, |
430 | | OffsetColumnWriter* offset_writer, ScalarColumnWriter* null_writer, |
431 | | std::unique_ptr<ColumnWriter> item_writer); |
432 | 45.7k | ~ArrayColumnWriter() override = default; |
433 | | |
434 | | Status init() override; |
435 | | |
436 | | Status append_data(const uint8_t** ptr, size_t num_rows) override; |
437 | | |
438 | | uint64_t estimate_buffer_size() override; |
439 | | |
440 | | Status finish() override; |
441 | | Status write_data() override; |
442 | | Status write_ordinal_index() override; |
443 | | Status append_nulls(size_t num_rows) override; |
444 | | Status append_nullable(const uint8_t* null_map, const uint8_t** ptr, size_t num_rows) override; |
445 | | |
446 | | Status finish_current_page() override; |
447 | | |
448 | 41.7k | Status write_zone_map() override { |
449 | 41.7k | if (_opts.need_zone_map) { |
450 | 0 | return Status::NotSupported("array not support zone map"); |
451 | 0 | } |
452 | 41.7k | return Status::OK(); |
453 | 41.7k | } |
454 | | |
455 | | Status write_inverted_index() override; |
456 | | Status write_ann_index() override; |
457 | 41.7k | Status write_bloom_filter_index() override { |
458 | 41.7k | if (_opts.need_bloom_filter) { |
459 | 0 | return Status::NotSupported("array not support bloom filter index"); |
460 | 0 | } |
461 | 41.7k | return Status::OK(); |
462 | 41.7k | } |
463 | 49.3k | ordinal_t get_next_rowid() const override { return _offset_writer->get_next_rowid(); } |
464 | | |
465 | 43.7k | uint64_t get_raw_data_bytes() const override { |
466 | 43.7k | return _get_total_data_pages_bytes(&ColumnWriter::get_raw_data_bytes); |
467 | 43.7k | } |
468 | | |
469 | 43.7k | uint64_t get_total_uncompressed_data_pages_bytes() const override { |
470 | 43.7k | return _get_total_data_pages_bytes(&ColumnWriter::get_total_uncompressed_data_pages_bytes); |
471 | 43.7k | } |
472 | | |
473 | 43.7k | uint64_t get_total_compressed_data_pages_bytes() const override { |
474 | 43.7k | return _get_total_data_pages_bytes(&ColumnWriter::get_total_compressed_data_pages_bytes); |
475 | 43.7k | } |
476 | | |
477 | | private: |
478 | | template <typename Func> |
479 | 131k | uint64_t _get_total_data_pages_bytes(Func func) const { |
480 | 131k | uint64_t size = std::invoke(func, _offset_writer.get()); |
481 | 131k | if (is_nullable()) { |
482 | 81.3k | size += std::invoke(func, _null_writer.get()); |
483 | 81.3k | } |
484 | 131k | size += std::invoke(func, _item_writer.get()); |
485 | 131k | return size; |
486 | 131k | } |
487 | | |
488 | | private: |
489 | | Status write_null_column(size_t num_rows, bool is_null); // 写入num_rows个null标记 |
490 | 45.2k | bool has_empty_items() const { return _item_writer->get_next_rowid() == 0; } |
491 | | |
492 | | private: |
493 | | std::unique_ptr<OffsetColumnWriter> _offset_writer; |
494 | | std::unique_ptr<ScalarColumnWriter> _null_writer; |
495 | | std::unique_ptr<ColumnWriter> _item_writer; |
496 | | std::unique_ptr<IndexColumnWriter> _inverted_index_writer; |
497 | | std::unique_ptr<AnnIndexColumnWriter> _ann_index_writer; |
498 | | ColumnWriterOptions _opts; |
499 | | }; |
500 | | |
501 | | class MapColumnWriter final : public ColumnWriter { |
502 | | public: |
503 | | explicit MapColumnWriter(const ColumnWriterOptions& opts, std::unique_ptr<StorageField> field, |
504 | | ScalarColumnWriter* null_writer, OffsetColumnWriter* offsets_writer, |
505 | | std::vector<std::unique_ptr<ColumnWriter>>& _kv_writers); |
506 | | |
507 | 23.6k | ~MapColumnWriter() override = default; |
508 | | |
509 | | Status init() override; |
510 | | |
511 | | Status append_data(const uint8_t** ptr, size_t num_rows) override; |
512 | | Status append_nullable(const uint8_t* null_map, const uint8_t** ptr, size_t num_rows) override; |
513 | | uint64_t estimate_buffer_size() override; |
514 | | |
515 | | Status finish() override; |
516 | | Status write_data() override; |
517 | | Status write_ordinal_index() override; |
518 | | Status write_inverted_index() override; |
519 | | Status append_nulls(size_t num_rows) override; |
520 | | |
521 | | Status finish_current_page() override; |
522 | | |
523 | 9.45k | Status write_zone_map() override { |
524 | 9.45k | if (_opts.need_zone_map) { |
525 | 0 | return Status::NotSupported("map not support zone map"); |
526 | 0 | } |
527 | 9.45k | return Status::OK(); |
528 | 9.45k | } |
529 | | |
530 | 9.45k | Status write_bloom_filter_index() override { |
531 | 9.45k | if (_opts.need_bloom_filter) { |
532 | 0 | return Status::NotSupported("map not support bloom filter index"); |
533 | 0 | } |
534 | 9.45k | return Status::OK(); |
535 | 9.45k | } |
536 | | |
537 | | // according key writer to get next rowid |
538 | 24.6k | ordinal_t get_next_rowid() const override { return _offsets_writer->get_next_rowid(); } |
539 | | |
540 | 10.2k | uint64_t get_raw_data_bytes() const override { |
541 | 10.2k | return _get_total_data_pages_bytes(&ColumnWriter::get_raw_data_bytes); |
542 | 10.2k | } |
543 | | |
544 | 10.2k | uint64_t get_total_uncompressed_data_pages_bytes() const override { |
545 | 10.2k | return _get_total_data_pages_bytes(&ColumnWriter::get_total_uncompressed_data_pages_bytes); |
546 | 10.2k | } |
547 | | |
548 | 10.2k | uint64_t get_total_compressed_data_pages_bytes() const override { |
549 | 10.2k | return _get_total_data_pages_bytes(&ColumnWriter::get_total_compressed_data_pages_bytes); |
550 | 10.2k | } |
551 | | |
552 | | private: |
553 | | template <typename Func> |
554 | 30.6k | uint64_t _get_total_data_pages_bytes(Func func) const { |
555 | 30.6k | uint64_t size = std::invoke(func, _offsets_writer.get()); |
556 | 30.6k | if (is_nullable()) { |
557 | 23.9k | size += std::invoke(func, _null_writer.get()); |
558 | 23.9k | } |
559 | 61.2k | for (const auto& writer : _kv_writers) { |
560 | 61.2k | size += std::invoke(func, writer.get()); |
561 | 61.2k | } |
562 | 30.6k | return size; |
563 | 30.6k | } |
564 | | |
565 | | private: |
566 | | std::vector<std::unique_ptr<ColumnWriter>> _kv_writers; |
567 | | // we need null writer to make sure a row is null or not |
568 | | std::unique_ptr<ScalarColumnWriter> _null_writer; |
569 | | std::unique_ptr<OffsetColumnWriter> _offsets_writer; |
570 | | std::unique_ptr<IndexColumnWriter> _index_builder; |
571 | | ColumnWriterOptions _opts; |
572 | | }; |
573 | | |
574 | | // used for compaction to write sub variant column |
575 | | class VariantSubcolumnWriter : public ColumnWriter { |
576 | | public: |
577 | | explicit VariantSubcolumnWriter(const ColumnWriterOptions& opts, const TabletColumn* column, |
578 | | std::unique_ptr<StorageField> field); |
579 | | |
580 | 22 | ~VariantSubcolumnWriter() override = default; |
581 | | |
582 | | Status init() override; |
583 | | |
584 | | Status append_data(const uint8_t** ptr, size_t num_rows) override; |
585 | | |
586 | | uint64_t estimate_buffer_size() override; |
587 | | |
588 | | Status finish() override; |
589 | | Status write_data() override; |
590 | | Status write_ordinal_index() override; |
591 | | |
592 | | Status write_zone_map() override; |
593 | | |
594 | | Status write_inverted_index() override; |
595 | | Status write_bloom_filter_index() override; |
596 | 0 | ordinal_t get_next_rowid() const override { return _next_rowid; } |
597 | | |
598 | 21 | uint64_t get_raw_data_bytes() const override { |
599 | 21 | return 0; // TODO |
600 | 21 | } |
601 | | |
602 | 21 | uint64_t get_total_uncompressed_data_pages_bytes() const override { |
603 | 21 | return 0; // TODO |
604 | 21 | } |
605 | | |
606 | 21 | uint64_t get_total_compressed_data_pages_bytes() const override { |
607 | 21 | return 0; // TODO |
608 | 21 | } |
609 | | |
610 | 0 | Status append_nulls(size_t num_rows) override { |
611 | 0 | return Status::NotSupported("variant writer can not append_nulls"); |
612 | 0 | } |
613 | | Status append_nullable(const uint8_t* null_map, const uint8_t** ptr, size_t num_rows) override; |
614 | | |
615 | 0 | Status finish_current_page() override { |
616 | 0 | return Status::NotSupported("variant writer has no data, can not finish_current_page"); |
617 | 0 | } |
618 | | |
619 | 0 | size_t get_non_null_size() const { return none_null_size; } |
620 | | |
621 | | Status finalize(); |
622 | | |
623 | | private: |
624 | | bool is_finalized() const; |
625 | | bool _is_finalized = false; |
626 | | ordinal_t _next_rowid = 0; |
627 | | size_t none_null_size = 0; |
628 | | ColumnVariant::MutablePtr _column; |
629 | | const TabletColumn* _tablet_column = nullptr; |
630 | | ColumnWriterOptions _opts; |
631 | | std::unique_ptr<ColumnWriter> _writer; |
632 | | TabletIndexes _indexes; |
633 | | |
634 | | std::unique_ptr<NestedGroupWriteProvider> _nested_group_provider; |
635 | | VariantStatistics _statistics; |
636 | | }; |
637 | | |
638 | | class VariantColumnWriter : public ColumnWriter { |
639 | | public: |
640 | | explicit VariantColumnWriter(const ColumnWriterOptions& opts, const TabletColumn* column, |
641 | | std::unique_ptr<StorageField> field); |
642 | | |
643 | 6.02k | ~VariantColumnWriter() override = default; |
644 | | |
645 | | Status init() override; |
646 | | |
647 | | Status append_data(const uint8_t** ptr, size_t num_rows) override; |
648 | | |
649 | | uint64_t estimate_buffer_size() override; |
650 | | |
651 | | Status finish() override; |
652 | | Status write_data() override; |
653 | | Status write_ordinal_index() override; |
654 | | |
655 | | Status write_zone_map() override; |
656 | | |
657 | | Status write_inverted_index() override; |
658 | | Status write_bloom_filter_index() override; |
659 | 16 | ordinal_t get_next_rowid() const override { return _next_rowid; } |
660 | | |
661 | 6.00k | uint64_t get_raw_data_bytes() const override { |
662 | 6.00k | return 0; // TODO |
663 | 6.00k | } |
664 | | |
665 | 6.00k | uint64_t get_total_uncompressed_data_pages_bytes() const override { |
666 | 6.00k | return 0; // TODO |
667 | 6.00k | } |
668 | | |
669 | 6.00k | uint64_t get_total_compressed_data_pages_bytes() const override { |
670 | 6.00k | return 0; // TODO |
671 | 6.00k | } |
672 | | |
673 | 0 | Status append_nulls(size_t num_rows) override { |
674 | 0 | return Status::NotSupported("variant writer can not append_nulls"); |
675 | 0 | } |
676 | | Status append_nullable(const uint8_t* null_map, const uint8_t** ptr, size_t num_rows) override; |
677 | | |
678 | 0 | Status finish_current_page() override { |
679 | 0 | return Status::NotSupported("variant writer has no data, can not finish_current_page"); |
680 | 0 | } |
681 | | |
682 | | private: |
683 | | std::unique_ptr<VariantColumnWriterImpl> _impl; |
684 | | ordinal_t _next_rowid = 0; |
685 | | }; |
686 | | |
687 | | } // namespace segment_v2 |
688 | | } // namespace doris |