Coverage Report

Created: 2026-07-23 17:58

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
be/src/format_v2/parquet/parquet_profile.cpp
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
#include "format_v2/parquet/parquet_profile.h"
19
20
#include "format_v2/parquet/parquet_statistics.h"
21
#include "runtime/file_scan_profile.h"
22
23
namespace doris::format::parquet {
24
25
438
void ParquetProfile::init(RuntimeProfile* profile) {
26
438
    if (profile == nullptr) {
27
204
        return;
28
204
    }
29
30
234
    file_scan_profile::ensure_hierarchy(profile);
31
234
    static const char* parquet_profile = "ParquetReader";
32
234
    total_time =
33
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, parquet_profile, file_scan_profile::FILE_READER, 1);
34
35
    // Row-group counters are part of the long-standing ParquetReader profile contract. Keep them
36
    // below the format node so profile parsers and operators can attribute pruning to Parquet.
37
234
    filtered_row_groups = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowGroupsFiltered", TUnit::UNIT,
38
234
                                                       parquet_profile, 1);
39
234
    filtered_row_groups_by_min_max = ADD_CHILD_COUNTER_WITH_LEVEL(
40
234
            profile, "RowGroupsFilteredByMinMax", TUnit::UNIT, parquet_profile, 1);
41
234
    filtered_row_groups_by_dictionary = ADD_CHILD_COUNTER_WITH_LEVEL(
42
234
            profile, "RowGroupsFilteredByDictionary", TUnit::UNIT, parquet_profile, 1);
43
234
    filtered_row_groups_by_bloom_filter = ADD_CHILD_COUNTER_WITH_LEVEL(
44
234
            profile, "RowGroupsFilteredByBloomFilter", TUnit::UNIT, parquet_profile, 1);
45
234
    filtered_row_groups_by_page_index = ADD_CHILD_COUNTER_WITH_LEVEL(
46
234
            profile, "RowGroupsFilteredByPageIndex", TUnit::UNIT, parquet_profile, 1);
47
234
    to_read_row_groups = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowGroupsReadNum", TUnit::UNIT,
48
234
                                                      parquet_profile, 1);
49
234
    total_row_groups = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowGroupsTotalNum", TUnit::UNIT,
50
234
                                                    parquet_profile, 1);
51
234
    selected_row_ranges = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "SelectedRowRanges", TUnit::UNIT,
52
234
                                                       parquet_profile, 1);
53
234
    filtered_group_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FilteredRowsByGroup", TUnit::UNIT,
54
234
                                                       parquet_profile, 1);
55
234
    filtered_page_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FilteredRowsByPage", TUnit::UNIT,
56
234
                                                      parquet_profile, 1);
57
234
    pages_skipped_by_data_page_filter = ADD_CHILD_COUNTER_WITH_LEVEL(
58
234
            profile, "PagesSkippedByDataPageFilter", TUnit::UNIT, parquet_profile, 1);
59
234
    data_page_filter_skip_bytes = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DataPageFilterSkipBytes",
60
234
                                                               TUnit::BYTES, parquet_profile, 1);
61
234
    selected_rows =
62
234
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "SelectedRows", TUnit::UNIT, parquet_profile, 1);
63
    // Keep every Parquet scan metric below the format node: profile consumers extract that
64
    // subtree and otherwise silently lose this counter even though filtering happened.
65
234
    rows_filtered_by_conjunct = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowsFilteredByConjunct",
66
234
                                                             TUnit::UNIT, parquet_profile, 1);
67
234
    total_batches =
68
234
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "TotalBatches", TUnit::UNIT, parquet_profile, 1);
69
234
    dense_batches =
70
234
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DenseBatches", TUnit::UNIT, parquet_profile, 1);
71
234
    selected_batches = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "SelectedBatches", TUnit::UNIT,
72
234
                                                    parquet_profile, 1);
73
234
    empty_selection_batches = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "EmptySelectionBatches",
74
234
                                                           TUnit::UNIT, parquet_profile, 1);
75
234
    range_gap_skipped_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RangeGapSkippedRows",
76
234
                                                          TUnit::UNIT, parquet_profile, 1);
77
234
    reader_read_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ReaderReadRows", TUnit::UNIT,
78
234
                                                    parquet_profile, 1);
79
234
    reader_skip_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ReaderSkipRows", TUnit::UNIT,
80
234
                                                    parquet_profile, 1);
81
234
    reader_select_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ReaderSelectRows", TUnit::UNIT,
82
234
                                                      parquet_profile, 1);
83
234
    level_only_read_time =
84
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "LevelOnlyReadTime", parquet_profile, 1);
85
234
    level_only_skip_time =
86
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "LevelOnlySkipTime", parquet_profile, 1);
87
234
    materialization_time =
88
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "MaterializationTime", parquet_profile, 1);
89
234
    hybrid_selection_batches = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "HybridSelectionBatches",
90
234
                                                            TUnit::UNIT, parquet_profile, 1);
91
234
    hybrid_selection_ranges = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "HybridSelectionRanges",
92
234
                                                           TUnit::UNIT, parquet_profile, 1);
93
234
    hybrid_selection_null_fallback_batches = ADD_CHILD_COUNTER_WITH_LEVEL(
94
234
            profile, "HybridSelectionNullFallbackBatches", TUnit::UNIT, parquet_profile, 1);
95
234
    native_read_calls = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "NativeReadCalls", TUnit::UNIT,
96
234
                                                     parquet_profile, 1);
97
234
    native_page_fragments = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "NativePageFragments",
98
234
                                                         TUnit::UNIT, parquet_profile, 1);
99
234
    page_crossing_batches = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageCrossingBatches",
100
234
                                                         TUnit::UNIT, parquet_profile, 1);
101
234
    nested_batches =
102
234
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "NestedBatches", TUnit::UNIT, parquet_profile, 1);
103
234
    lazy_read_filtered_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FilteredRowsByLazyRead",
104
234
                                                           TUnit::UNIT, parquet_profile, 1);
105
    // Format-specific counters stay below ParquetReader so subtree consumers cannot silently
106
    // attribute them to a different file format initialized on the same RuntimeProfile.
107
234
    filtered_bytes = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FilteredBytes", TUnit::BYTES,
108
234
                                                  parquet_profile, 1);
109
234
    raw_rows_read =
110
234
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RawRowsRead", TUnit::UNIT, parquet_profile, 1);
111
234
    column_read_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "ColumnReadTime", parquet_profile, 1);
112
234
    parse_meta_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "ParseMetaTime", parquet_profile, 1);
113
234
    parse_footer_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "ParseFooterTime", parquet_profile, 1);
114
234
    file_reader_create_time =
115
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "FileReaderCreateTime", parquet_profile, 1);
116
234
    open_file_num =
117
234
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FileNum", TUnit::UNIT, parquet_profile, 1);
118
234
    page_index_read_calls = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageIndexReadCalls", TUnit::UNIT,
119
234
                                                         parquet_profile, 1);
120
234
    page_index_filter_time =
121
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageIndexFilterTime", parquet_profile, 1);
122
234
    read_page_index_time =
123
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageIndexReadTime", parquet_profile, 1);
124
234
    parse_page_index_time =
125
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageIndexParseTime", parquet_profile, 1);
126
234
    expr_zonemap_unusable = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ExprZoneMapUnusableEvals",
127
234
                                                         TUnit::UNIT, parquet_profile, 1);
128
234
    in_zonemap_point_check = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "InZoneMapPointCheckCount",
129
234
                                                          TUnit::UNIT, parquet_profile, 1);
130
234
    in_zonemap_range_only = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "InZoneMapRangeOnlyCount",
131
234
                                                         TUnit::UNIT, parquet_profile, 1);
132
234
    row_group_filter_time =
133
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "RowGroupFilterTime", parquet_profile, 1);
134
234
    file_footer_read_calls = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FileFooterReadCalls",
135
234
                                                          TUnit::UNIT, parquet_profile, 1);
136
234
    file_footer_hit_cache = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FileFooterHitCache", TUnit::UNIT,
137
234
                                                         parquet_profile, 1);
138
234
    decompress_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecompressTime", parquet_profile, 1);
139
234
    decompress_cnt = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DecompressCount", TUnit::UNIT,
140
234
                                                  parquet_profile, 1);
141
234
    page_read_counter =
142
234
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageReadCount", TUnit::UNIT, parquet_profile, 1);
143
234
    page_cache_write_counter = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageCacheWriteCount",
144
234
                                                            TUnit::UNIT, parquet_profile, 1);
145
234
    page_cache_compressed_write_counter = ADD_CHILD_COUNTER_WITH_LEVEL(
146
234
            profile, "PageCacheCompressedWriteCount", TUnit::UNIT, parquet_profile, 1);
147
234
    page_cache_decompressed_write_counter = ADD_CHILD_COUNTER_WITH_LEVEL(
148
234
            profile, "PageCacheDecompressedWriteCount", TUnit::UNIT, parquet_profile, 1);
149
234
    page_cache_hit_counter = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageCacheHitCount", TUnit::UNIT,
150
234
                                                          parquet_profile, 1);
151
234
    page_cache_missing_counter = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageCacheMissingCount",
152
234
                                                              TUnit::UNIT, parquet_profile, 1);
153
234
    page_cache_compressed_hit_counter = ADD_CHILD_COUNTER_WITH_LEVEL(
154
234
            profile, "PageCacheCompressedHitCount", TUnit::UNIT, parquet_profile, 1);
155
234
    page_cache_decompressed_hit_counter = ADD_CHILD_COUNTER_WITH_LEVEL(
156
234
            profile, "PageCacheDecompressedHitCount", TUnit::UNIT, parquet_profile, 1);
157
234
    decode_header_time =
158
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageHeaderDecodeTime", parquet_profile, 1);
159
234
    read_page_header_time =
160
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageHeaderReadTime", parquet_profile, 1);
161
234
    decode_value_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecodeValueTime", parquet_profile, 1);
162
234
    decode_dict_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecodeDictTime", parquet_profile, 1);
163
234
    decode_level_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecodeLevelTime", parquet_profile, 1);
164
234
    decode_null_map_time =
165
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecodeNullMapTime", parquet_profile, 1);
166
234
    skip_page_header_num = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "SkipPageHeaderNum", TUnit::UNIT,
167
234
                                                        parquet_profile, 1);
168
234
    parse_page_header_num = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ParsePageHeaderNum", TUnit::UNIT,
169
234
                                                         parquet_profile, 1);
170
234
    predicate_filter_time =
171
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PredicateFilterTime", parquet_profile, 1);
172
234
    predicate_compaction_time =
173
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PredicateCompactionTime", parquet_profile, 1);
174
234
    predicate_compaction_bytes = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PredicateCompactionBytes",
175
234
                                                              TUnit::BYTES, parquet_profile, 1);
176
234
    predicate_compaction_count = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PredicateCompactionCount",
177
234
                                                              TUnit::UNIT, parquet_profile, 1);
178
234
    plain_predicate_direct_batches = ADD_CHILD_COUNTER_WITH_LEVEL(
179
234
            profile, "PlainPredicateDirectBatches", TUnit::UNIT, parquet_profile, 1);
180
234
    plain_predicate_direct_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PlainPredicateDirectRows",
181
234
                                                               TUnit::UNIT, parquet_profile, 1);
182
234
    dict_filter_rewrite_time =
183
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DictFilterRewriteTime", parquet_profile, 1);
184
234
    dict_filter_expr_rewrite_time =
185
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DictFilterExprRewriteTime", parquet_profile, 1);
186
234
    dict_filter_read_dict_time =
187
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DictFilterReadDictTime", parquet_profile, 1);
188
234
    dict_filter_build_time =
189
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DictFilterBuildTime", parquet_profile, 1);
190
234
    dict_filter_candidate_columns = ADD_CHILD_COUNTER_WITH_LEVEL(
191
234
            profile, "DictFilterCandidateColumns", TUnit::UNIT, parquet_profile, 1);
192
234
    dict_filter_columns = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DictFilterColumns", TUnit::UNIT,
193
234
                                                       parquet_profile, 1);
194
234
    dict_filter_unsupported_columns = ADD_CHILD_COUNTER_WITH_LEVEL(
195
234
            profile, "DictFilterUnsupportedColumns", TUnit::UNIT, parquet_profile, 1);
196
234
    dict_filter_read_failures = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DictFilterReadFailures",
197
234
                                                             TUnit::UNIT, parquet_profile, 1);
198
234
    rows_filtered_by_dict_filter = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowsFilteredByDictFilter",
199
234
                                                                TUnit::UNIT, parquet_profile, 1);
200
234
    bloom_filter_read_time =
201
234
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "BloomFilterReadTime", parquet_profile, 1);
202
234
}
203
204
111
void ParquetProfile::update_pruning_stats(const ParquetPruningStats& pruning_stats) const {
205
111
    COUNTER_UPDATE(filtered_row_groups,
206
111
                   pruning_stats.total_row_groups - pruning_stats.selected_row_groups);
207
111
    COUNTER_UPDATE(filtered_row_groups_by_min_max, pruning_stats.filtered_row_groups_by_statistics);
208
111
    COUNTER_UPDATE(filtered_row_groups_by_dictionary,
209
111
                   pruning_stats.filtered_row_groups_by_dictionary);
210
111
    COUNTER_UPDATE(filtered_row_groups_by_bloom_filter,
211
111
                   pruning_stats.filtered_row_groups_by_bloom_filter);
212
111
    COUNTER_UPDATE(filtered_row_groups_by_page_index,
213
111
                   pruning_stats.filtered_row_groups_by_page_index);
214
111
    COUNTER_UPDATE(to_read_row_groups, pruning_stats.selected_row_groups);
215
111
    COUNTER_UPDATE(total_row_groups, pruning_stats.total_row_groups);
216
111
    COUNTER_UPDATE(selected_row_ranges, pruning_stats.selected_row_ranges);
217
111
    COUNTER_UPDATE(filtered_group_rows, pruning_stats.filtered_group_rows);
218
111
    COUNTER_UPDATE(filtered_bytes, pruning_stats.filtered_bytes);
219
111
    COUNTER_UPDATE(filtered_page_rows, pruning_stats.filtered_page_rows);
220
111
    COUNTER_UPDATE(page_index_read_calls, pruning_stats.page_index_read_calls);
221
111
    COUNTER_UPDATE(bloom_filter_read_time, pruning_stats.bloom_filter_read_time);
222
111
    COUNTER_UPDATE(row_group_filter_time, pruning_stats.row_group_filter_time);
223
111
    COUNTER_UPDATE(page_index_filter_time, pruning_stats.page_index_filter_time);
224
111
    COUNTER_UPDATE(read_page_index_time, pruning_stats.read_page_index_time);
225
111
    COUNTER_UPDATE(parse_page_index_time, pruning_stats.parse_page_index_time);
226
111
    COUNTER_UPDATE(expr_zonemap_unusable, pruning_stats.expr_zonemap_unusable_evals);
227
111
    COUNTER_UPDATE(in_zonemap_point_check, pruning_stats.in_zonemap_point_check_count);
228
111
    COUNTER_UPDATE(in_zonemap_range_only, pruning_stats.in_zonemap_range_only_count);
229
111
}
230
231
void ParquetProfile::update_deferred_pruning_stats(const ParquetPruningStats& pruning_stats,
232
127
                                                   bool selected) const {
233
127
    const int64_t filtered = selected ? 0 : 1;
234
127
    COUNTER_UPDATE(filtered_row_groups, filtered);
235
127
    COUNTER_UPDATE(filtered_row_groups_by_dictionary,
236
127
                   pruning_stats.filtered_row_groups_by_dictionary);
237
127
    COUNTER_UPDATE(filtered_row_groups_by_bloom_filter,
238
127
                   pruning_stats.filtered_row_groups_by_bloom_filter);
239
127
    COUNTER_UPDATE(filtered_row_groups_by_page_index,
240
127
                   pruning_stats.filtered_row_groups_by_page_index);
241
    // Initial footer planning counts every surviving candidate as readable. Lazy probes correct
242
    // that estimate only when a later dictionary, Bloom, or page-index check removes the group.
243
127
    COUNTER_UPDATE(to_read_row_groups, -filtered);
244
127
    COUNTER_UPDATE(selected_row_ranges, pruning_stats.selected_row_ranges);
245
127
    COUNTER_UPDATE(filtered_group_rows, pruning_stats.filtered_group_rows);
246
127
    COUNTER_UPDATE(filtered_bytes, pruning_stats.filtered_bytes);
247
127
    COUNTER_UPDATE(filtered_page_rows, pruning_stats.filtered_page_rows);
248
127
    COUNTER_UPDATE(page_index_read_calls, pruning_stats.page_index_read_calls);
249
127
    COUNTER_UPDATE(bloom_filter_read_time, pruning_stats.bloom_filter_read_time);
250
127
    COUNTER_UPDATE(row_group_filter_time, pruning_stats.row_group_filter_time);
251
127
    COUNTER_UPDATE(page_index_filter_time, pruning_stats.page_index_filter_time);
252
127
    COUNTER_UPDATE(read_page_index_time, pruning_stats.read_page_index_time);
253
127
    COUNTER_UPDATE(parse_page_index_time, pruning_stats.parse_page_index_time);
254
127
    COUNTER_UPDATE(expr_zonemap_unusable, pruning_stats.expr_zonemap_unusable_evals);
255
127
    COUNTER_UPDATE(in_zonemap_point_check, pruning_stats.in_zonemap_point_check_count);
256
127
    COUNTER_UPDATE(in_zonemap_range_only, pruning_stats.in_zonemap_range_only_count);
257
127
}
258
259
207
ParquetPageSkipProfile ParquetProfile::page_skip_profile() const {
260
207
    return {
261
207
            .skipped_pages = pages_skipped_by_data_page_filter,
262
207
            .skipped_bytes = data_page_filter_skip_bytes,
263
207
    };
264
207
}
265
266
449
ParquetColumnReaderProfile ParquetProfile::column_reader_profile() const {
267
449
    return {
268
449
            .reader_read_rows = reader_read_rows,
269
449
            .reader_skip_rows = reader_skip_rows,
270
449
            .reader_select_rows = reader_select_rows,
271
449
            .level_only_read_time = level_only_read_time,
272
449
            .level_only_skip_time = level_only_skip_time,
273
449
            .materialization_time = materialization_time,
274
449
            .hybrid_selection_batches = hybrid_selection_batches,
275
449
            .hybrid_selection_ranges = hybrid_selection_ranges,
276
449
            .hybrid_selection_null_fallback_batches = hybrid_selection_null_fallback_batches,
277
449
            .decompress_time = decompress_time,
278
449
            .decompress_count = decompress_cnt,
279
449
            .decode_header_time = decode_header_time,
280
449
            .decode_value_time = decode_value_time,
281
449
            .decode_dictionary_time = decode_dict_time,
282
449
            .decode_level_time = decode_level_time,
283
449
            .decode_null_map_time = decode_null_map_time,
284
449
            .page_index_read_calls = page_index_read_calls,
285
449
            .skip_page_header_count = skip_page_header_num,
286
449
            .parse_page_header_count = parse_page_header_num,
287
449
            .read_page_header_time = read_page_header_time,
288
449
            .page_read_count = page_read_counter,
289
449
            .page_cache_write_count = page_cache_write_counter,
290
449
            .page_cache_compressed_write_count = page_cache_compressed_write_counter,
291
449
            .page_cache_decompressed_write_count = page_cache_decompressed_write_counter,
292
449
            .page_cache_hit_count = page_cache_hit_counter,
293
449
            .page_cache_miss_count = page_cache_missing_counter,
294
449
            .page_cache_compressed_hit_count = page_cache_compressed_hit_counter,
295
449
            .page_cache_decompressed_hit_count = page_cache_decompressed_hit_counter,
296
449
            .native_read_calls = native_read_calls,
297
449
            .native_page_fragments = native_page_fragments,
298
449
            .page_crossing_batches = page_crossing_batches,
299
449
            .nested_batches = nested_batches,
300
449
    };
301
449
}
302
303
214
ParquetScanProfile ParquetProfile::scan_profile() const {
304
214
    return {
305
214
            .raw_rows_read = raw_rows_read,
306
214
            .selected_rows = selected_rows,
307
214
            .rows_filtered_by_conjunct = rows_filtered_by_conjunct,
308
214
            .lazy_read_filtered_rows = lazy_read_filtered_rows,
309
214
            .total_batches = total_batches,
310
214
            .dense_batches = dense_batches,
311
214
            .selected_batches = selected_batches,
312
214
            .empty_selection_batches = empty_selection_batches,
313
214
            .range_gap_skipped_rows = range_gap_skipped_rows,
314
214
            .column_read_time = column_read_time,
315
214
            .predicate_filter_time = predicate_filter_time,
316
214
            .predicate_compaction_time = predicate_compaction_time,
317
214
            .predicate_compaction_bytes = predicate_compaction_bytes,
318
214
            .predicate_compaction_count = predicate_compaction_count,
319
214
            .plain_predicate_direct_batches = plain_predicate_direct_batches,
320
214
            .plain_predicate_direct_rows = plain_predicate_direct_rows,
321
214
            .dict_filter_rewrite_time = dict_filter_rewrite_time,
322
214
            .dict_filter_expr_rewrite_time = dict_filter_expr_rewrite_time,
323
214
            .dict_filter_read_dict_time = dict_filter_read_dict_time,
324
214
            .dict_filter_build_time = dict_filter_build_time,
325
214
            .dict_filter_candidate_columns = dict_filter_candidate_columns,
326
214
            .dict_filter_columns = dict_filter_columns,
327
214
            .dict_filter_unsupported_columns = dict_filter_unsupported_columns,
328
214
            .dict_filter_read_failures = dict_filter_read_failures,
329
214
            .rows_filtered_by_dict_filter = rows_filtered_by_dict_filter,
330
214
            .column_reader_profile = column_reader_profile(),
331
214
    };
332
214
}
333
334
} // namespace doris::format::parquet