Coverage Report

Created: 2026-07-23 15:22

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
446
void ParquetProfile::init(RuntimeProfile* profile) {
26
446
    if (profile == nullptr) {
27
204
        return;
28
204
    }
29
30
242
    file_scan_profile::ensure_hierarchy(profile);
31
242
    static const char* parquet_profile = "ParquetReader";
32
242
    total_time =
33
242
            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
242
    filtered_row_groups = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowGroupsFiltered", TUnit::UNIT,
38
242
                                                       parquet_profile, 1);
39
242
    filtered_row_groups_by_min_max = ADD_CHILD_COUNTER_WITH_LEVEL(
40
242
            profile, "RowGroupsFilteredByMinMax", TUnit::UNIT, parquet_profile, 1);
41
242
    filtered_row_groups_by_dictionary = ADD_CHILD_COUNTER_WITH_LEVEL(
42
242
            profile, "RowGroupsFilteredByDictionary", TUnit::UNIT, parquet_profile, 1);
43
242
    filtered_row_groups_by_bloom_filter = ADD_CHILD_COUNTER_WITH_LEVEL(
44
242
            profile, "RowGroupsFilteredByBloomFilter", TUnit::UNIT, parquet_profile, 1);
45
242
    filtered_row_groups_by_page_index = ADD_CHILD_COUNTER_WITH_LEVEL(
46
242
            profile, "RowGroupsFilteredByPageIndex", TUnit::UNIT, parquet_profile, 1);
47
242
    to_read_row_groups = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowGroupsReadNum", TUnit::UNIT,
48
242
                                                      parquet_profile, 1);
49
242
    total_row_groups = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowGroupsTotalNum", TUnit::UNIT,
50
242
                                                    parquet_profile, 1);
51
242
    selected_row_ranges = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "SelectedRowRanges", TUnit::UNIT,
52
242
                                                       parquet_profile, 1);
53
242
    filtered_group_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FilteredRowsByGroup", TUnit::UNIT,
54
242
                                                       parquet_profile, 1);
55
242
    filtered_page_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FilteredRowsByPage", TUnit::UNIT,
56
242
                                                      parquet_profile, 1);
57
242
    pages_skipped_by_data_page_filter = ADD_CHILD_COUNTER_WITH_LEVEL(
58
242
            profile, "PagesSkippedByDataPageFilter", TUnit::UNIT, parquet_profile, 1);
59
242
    data_page_filter_skip_bytes = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DataPageFilterSkipBytes",
60
242
                                                               TUnit::BYTES, parquet_profile, 1);
61
242
    selected_rows =
62
242
            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
242
    rows_filtered_by_conjunct = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowsFilteredByConjunct",
66
242
                                                             TUnit::UNIT, parquet_profile, 1);
67
242
    total_batches =
68
242
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "TotalBatches", TUnit::UNIT, parquet_profile, 1);
69
242
    dense_batches =
70
242
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DenseBatches", TUnit::UNIT, parquet_profile, 1);
71
242
    selected_batches = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "SelectedBatches", TUnit::UNIT,
72
242
                                                    parquet_profile, 1);
73
242
    empty_selection_batches = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "EmptySelectionBatches",
74
242
                                                           TUnit::UNIT, parquet_profile, 1);
75
242
    range_gap_skipped_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RangeGapSkippedRows",
76
242
                                                          TUnit::UNIT, parquet_profile, 1);
77
242
    reader_read_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ReaderReadRows", TUnit::UNIT,
78
242
                                                    parquet_profile, 1);
79
242
    reader_skip_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ReaderSkipRows", TUnit::UNIT,
80
242
                                                    parquet_profile, 1);
81
242
    reader_select_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ReaderSelectRows", TUnit::UNIT,
82
242
                                                      parquet_profile, 1);
83
242
    level_only_read_time =
84
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "LevelOnlyReadTime", parquet_profile, 1);
85
242
    level_only_skip_time =
86
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "LevelOnlySkipTime", parquet_profile, 1);
87
242
    materialization_time =
88
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "MaterializationTime", parquet_profile, 1);
89
242
    hybrid_selection_batches = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "HybridSelectionBatches",
90
242
                                                            TUnit::UNIT, parquet_profile, 1);
91
242
    hybrid_selection_ranges = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "HybridSelectionRanges",
92
242
                                                           TUnit::UNIT, parquet_profile, 1);
93
242
    hybrid_selection_null_fallback_batches = ADD_CHILD_COUNTER_WITH_LEVEL(
94
242
            profile, "HybridSelectionNullFallbackBatches", TUnit::UNIT, parquet_profile, 1);
95
242
    native_read_calls = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "NativeReadCalls", TUnit::UNIT,
96
242
                                                     parquet_profile, 1);
97
242
    native_page_fragments = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "NativePageFragments",
98
242
                                                         TUnit::UNIT, parquet_profile, 1);
99
242
    page_crossing_batches = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageCrossingBatches",
100
242
                                                         TUnit::UNIT, parquet_profile, 1);
101
242
    nested_batches =
102
242
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "NestedBatches", TUnit::UNIT, parquet_profile, 1);
103
242
    lazy_read_filtered_rows = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FilteredRowsByLazyRead",
104
242
                                                           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
242
    filtered_bytes = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FilteredBytes", TUnit::BYTES,
108
242
                                                  parquet_profile, 1);
109
242
    raw_rows_read =
110
242
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RawRowsRead", TUnit::UNIT, parquet_profile, 1);
111
242
    column_read_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "ColumnReadTime", parquet_profile, 1);
112
242
    parse_meta_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "ParseMetaTime", parquet_profile, 1);
113
242
    parse_footer_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "ParseFooterTime", parquet_profile, 1);
114
242
    file_reader_create_time =
115
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "FileReaderCreateTime", parquet_profile, 1);
116
242
    open_file_num =
117
242
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FileNum", TUnit::UNIT, parquet_profile, 1);
118
242
    page_index_read_calls = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageIndexReadCalls", TUnit::UNIT,
119
242
                                                         parquet_profile, 1);
120
242
    page_index_filter_time =
121
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageIndexFilterTime", parquet_profile, 1);
122
242
    read_page_index_time =
123
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageIndexReadTime", parquet_profile, 1);
124
242
    parse_page_index_time =
125
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageIndexParseTime", parquet_profile, 1);
126
242
    expr_zonemap_unusable = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ExprZoneMapUnusableEvals",
127
242
                                                         TUnit::UNIT, parquet_profile, 1);
128
242
    in_zonemap_point_check = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "InZoneMapPointCheckCount",
129
242
                                                          TUnit::UNIT, parquet_profile, 1);
130
242
    in_zonemap_range_only = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "InZoneMapRangeOnlyCount",
131
242
                                                         TUnit::UNIT, parquet_profile, 1);
132
242
    row_group_filter_time =
133
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "RowGroupFilterTime", parquet_profile, 1);
134
242
    file_footer_read_calls = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FileFooterReadCalls",
135
242
                                                          TUnit::UNIT, parquet_profile, 1);
136
242
    file_footer_hit_cache = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "FileFooterHitCache", TUnit::UNIT,
137
242
                                                         parquet_profile, 1);
138
242
    decompress_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecompressTime", parquet_profile, 1);
139
242
    decompress_cnt = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DecompressCount", TUnit::UNIT,
140
242
                                                  parquet_profile, 1);
141
242
    page_read_counter =
142
242
            ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageReadCount", TUnit::UNIT, parquet_profile, 1);
143
242
    page_cache_write_counter = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageCacheWriteCount",
144
242
                                                            TUnit::UNIT, parquet_profile, 1);
145
242
    page_cache_compressed_write_counter = ADD_CHILD_COUNTER_WITH_LEVEL(
146
242
            profile, "PageCacheCompressedWriteCount", TUnit::UNIT, parquet_profile, 1);
147
242
    page_cache_decompressed_write_counter = ADD_CHILD_COUNTER_WITH_LEVEL(
148
242
            profile, "PageCacheDecompressedWriteCount", TUnit::UNIT, parquet_profile, 1);
149
242
    page_cache_hit_counter = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageCacheHitCount", TUnit::UNIT,
150
242
                                                          parquet_profile, 1);
151
242
    page_cache_missing_counter = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PageCacheMissingCount",
152
242
                                                              TUnit::UNIT, parquet_profile, 1);
153
242
    page_cache_compressed_hit_counter = ADD_CHILD_COUNTER_WITH_LEVEL(
154
242
            profile, "PageCacheCompressedHitCount", TUnit::UNIT, parquet_profile, 1);
155
242
    page_cache_decompressed_hit_counter = ADD_CHILD_COUNTER_WITH_LEVEL(
156
242
            profile, "PageCacheDecompressedHitCount", TUnit::UNIT, parquet_profile, 1);
157
242
    decode_header_time =
158
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageHeaderDecodeTime", parquet_profile, 1);
159
242
    read_page_header_time =
160
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PageHeaderReadTime", parquet_profile, 1);
161
242
    decode_value_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecodeValueTime", parquet_profile, 1);
162
242
    decode_dict_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecodeDictTime", parquet_profile, 1);
163
242
    decode_level_time = ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecodeLevelTime", parquet_profile, 1);
164
242
    decode_null_map_time =
165
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DecodeNullMapTime", parquet_profile, 1);
166
242
    skip_page_header_num = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "SkipPageHeaderNum", TUnit::UNIT,
167
242
                                                        parquet_profile, 1);
168
242
    parse_page_header_num = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "ParsePageHeaderNum", TUnit::UNIT,
169
242
                                                         parquet_profile, 1);
170
242
    predicate_filter_time =
171
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PredicateFilterTime", parquet_profile, 1);
172
242
    predicate_compaction_time =
173
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "PredicateCompactionTime", parquet_profile, 1);
174
242
    predicate_compaction_bytes = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PredicateCompactionBytes",
175
242
                                                              TUnit::BYTES, parquet_profile, 1);
176
242
    predicate_compaction_count = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "PredicateCompactionCount",
177
242
                                                              TUnit::UNIT, parquet_profile, 1);
178
242
    fixed_width_predicate_direct_batches = ADD_CHILD_COUNTER_WITH_LEVEL(
179
242
            profile, "FixedWidthPredicateDirectBatches", TUnit::UNIT, parquet_profile, 1);
180
242
    fixed_width_predicate_direct_rows = ADD_CHILD_COUNTER_WITH_LEVEL(
181
242
            profile, "FixedWidthPredicateDirectRows", TUnit::UNIT, parquet_profile, 1);
182
242
    dict_filter_rewrite_time =
183
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DictFilterRewriteTime", parquet_profile, 1);
184
242
    dict_filter_expr_rewrite_time =
185
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DictFilterExprRewriteTime", parquet_profile, 1);
186
242
    dict_filter_read_dict_time =
187
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DictFilterReadDictTime", parquet_profile, 1);
188
242
    dict_filter_build_time =
189
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "DictFilterBuildTime", parquet_profile, 1);
190
242
    dict_filter_candidate_columns = ADD_CHILD_COUNTER_WITH_LEVEL(
191
242
            profile, "DictFilterCandidateColumns", TUnit::UNIT, parquet_profile, 1);
192
242
    dict_filter_columns = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DictFilterColumns", TUnit::UNIT,
193
242
                                                       parquet_profile, 1);
194
242
    dict_filter_unsupported_columns = ADD_CHILD_COUNTER_WITH_LEVEL(
195
242
            profile, "DictFilterUnsupportedColumns", TUnit::UNIT, parquet_profile, 1);
196
242
    dict_filter_read_failures = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "DictFilterReadFailures",
197
242
                                                             TUnit::UNIT, parquet_profile, 1);
198
242
    rows_filtered_by_dict_filter = ADD_CHILD_COUNTER_WITH_LEVEL(profile, "RowsFilteredByDictFilter",
199
242
                                                                TUnit::UNIT, parquet_profile, 1);
200
242
    bloom_filter_read_time =
201
242
            ADD_CHILD_TIMER_WITH_LEVEL(profile, "BloomFilterReadTime", parquet_profile, 1);
202
242
}
203
204
115
void ParquetProfile::update_pruning_stats(const ParquetPruningStats& pruning_stats) const {
205
115
    COUNTER_UPDATE(filtered_row_groups,
206
115
                   pruning_stats.total_row_groups - pruning_stats.selected_row_groups);
207
115
    COUNTER_UPDATE(filtered_row_groups_by_min_max, pruning_stats.filtered_row_groups_by_statistics);
208
115
    COUNTER_UPDATE(filtered_row_groups_by_dictionary,
209
115
                   pruning_stats.filtered_row_groups_by_dictionary);
210
115
    COUNTER_UPDATE(filtered_row_groups_by_bloom_filter,
211
115
                   pruning_stats.filtered_row_groups_by_bloom_filter);
212
115
    COUNTER_UPDATE(filtered_row_groups_by_page_index,
213
115
                   pruning_stats.filtered_row_groups_by_page_index);
214
115
    COUNTER_UPDATE(to_read_row_groups, pruning_stats.selected_row_groups);
215
115
    COUNTER_UPDATE(total_row_groups, pruning_stats.total_row_groups);
216
115
    COUNTER_UPDATE(selected_row_ranges, pruning_stats.selected_row_ranges);
217
115
    COUNTER_UPDATE(filtered_group_rows, pruning_stats.filtered_group_rows);
218
115
    COUNTER_UPDATE(filtered_bytes, pruning_stats.filtered_bytes);
219
115
    COUNTER_UPDATE(filtered_page_rows, pruning_stats.filtered_page_rows);
220
115
    COUNTER_UPDATE(page_index_read_calls, pruning_stats.page_index_read_calls);
221
115
    COUNTER_UPDATE(bloom_filter_read_time, pruning_stats.bloom_filter_read_time);
222
115
    COUNTER_UPDATE(row_group_filter_time, pruning_stats.row_group_filter_time);
223
115
    COUNTER_UPDATE(page_index_filter_time, pruning_stats.page_index_filter_time);
224
115
    COUNTER_UPDATE(read_page_index_time, pruning_stats.read_page_index_time);
225
115
    COUNTER_UPDATE(parse_page_index_time, pruning_stats.parse_page_index_time);
226
115
    COUNTER_UPDATE(expr_zonemap_unusable, pruning_stats.expr_zonemap_unusable_evals);
227
115
    COUNTER_UPDATE(in_zonemap_point_check, pruning_stats.in_zonemap_point_check_count);
228
115
    COUNTER_UPDATE(in_zonemap_range_only, pruning_stats.in_zonemap_range_only_count);
229
115
}
230
231
void ParquetProfile::update_deferred_pruning_stats(const ParquetPruningStats& pruning_stats,
232
131
                                                   bool selected) const {
233
131
    const int64_t filtered = selected ? 0 : 1;
234
131
    COUNTER_UPDATE(filtered_row_groups, filtered);
235
131
    COUNTER_UPDATE(filtered_row_groups_by_dictionary,
236
131
                   pruning_stats.filtered_row_groups_by_dictionary);
237
131
    COUNTER_UPDATE(filtered_row_groups_by_bloom_filter,
238
131
                   pruning_stats.filtered_row_groups_by_bloom_filter);
239
131
    COUNTER_UPDATE(filtered_row_groups_by_page_index,
240
131
                   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
131
    COUNTER_UPDATE(to_read_row_groups, -filtered);
244
131
    COUNTER_UPDATE(selected_row_ranges, pruning_stats.selected_row_ranges);
245
131
    COUNTER_UPDATE(filtered_group_rows, pruning_stats.filtered_group_rows);
246
131
    COUNTER_UPDATE(filtered_bytes, pruning_stats.filtered_bytes);
247
131
    COUNTER_UPDATE(filtered_page_rows, pruning_stats.filtered_page_rows);
248
131
    COUNTER_UPDATE(page_index_read_calls, pruning_stats.page_index_read_calls);
249
131
    COUNTER_UPDATE(bloom_filter_read_time, pruning_stats.bloom_filter_read_time);
250
131
    COUNTER_UPDATE(row_group_filter_time, pruning_stats.row_group_filter_time);
251
131
    COUNTER_UPDATE(page_index_filter_time, pruning_stats.page_index_filter_time);
252
131
    COUNTER_UPDATE(read_page_index_time, pruning_stats.read_page_index_time);
253
131
    COUNTER_UPDATE(parse_page_index_time, pruning_stats.parse_page_index_time);
254
131
    COUNTER_UPDATE(expr_zonemap_unusable, pruning_stats.expr_zonemap_unusable_evals);
255
131
    COUNTER_UPDATE(in_zonemap_point_check, pruning_stats.in_zonemap_point_check_count);
256
131
    COUNTER_UPDATE(in_zonemap_range_only, pruning_stats.in_zonemap_range_only_count);
257
131
}
258
259
211
ParquetPageSkipProfile ParquetProfile::page_skip_profile() const {
260
211
    return {
261
211
            .skipped_pages = pages_skipped_by_data_page_filter,
262
211
            .skipped_bytes = data_page_filter_skip_bytes,
263
211
    };
264
211
}
265
266
457
ParquetColumnReaderProfile ParquetProfile::column_reader_profile() const {
267
457
    return {
268
457
            .reader_read_rows = reader_read_rows,
269
457
            .reader_skip_rows = reader_skip_rows,
270
457
            .reader_select_rows = reader_select_rows,
271
457
            .level_only_read_time = level_only_read_time,
272
457
            .level_only_skip_time = level_only_skip_time,
273
457
            .materialization_time = materialization_time,
274
457
            .hybrid_selection_batches = hybrid_selection_batches,
275
457
            .hybrid_selection_ranges = hybrid_selection_ranges,
276
457
            .hybrid_selection_null_fallback_batches = hybrid_selection_null_fallback_batches,
277
457
            .decompress_time = decompress_time,
278
457
            .decompress_count = decompress_cnt,
279
457
            .decode_header_time = decode_header_time,
280
457
            .decode_value_time = decode_value_time,
281
457
            .decode_dictionary_time = decode_dict_time,
282
457
            .decode_level_time = decode_level_time,
283
457
            .decode_null_map_time = decode_null_map_time,
284
457
            .page_index_read_calls = page_index_read_calls,
285
457
            .skip_page_header_count = skip_page_header_num,
286
457
            .parse_page_header_count = parse_page_header_num,
287
457
            .read_page_header_time = read_page_header_time,
288
457
            .page_read_count = page_read_counter,
289
457
            .page_cache_write_count = page_cache_write_counter,
290
457
            .page_cache_compressed_write_count = page_cache_compressed_write_counter,
291
457
            .page_cache_decompressed_write_count = page_cache_decompressed_write_counter,
292
457
            .page_cache_hit_count = page_cache_hit_counter,
293
457
            .page_cache_miss_count = page_cache_missing_counter,
294
457
            .page_cache_compressed_hit_count = page_cache_compressed_hit_counter,
295
457
            .page_cache_decompressed_hit_count = page_cache_decompressed_hit_counter,
296
457
            .native_read_calls = native_read_calls,
297
457
            .native_page_fragments = native_page_fragments,
298
457
            .page_crossing_batches = page_crossing_batches,
299
457
            .nested_batches = nested_batches,
300
457
    };
301
457
}
302
303
218
ParquetScanProfile ParquetProfile::scan_profile() const {
304
218
    return {
305
218
            .raw_rows_read = raw_rows_read,
306
218
            .selected_rows = selected_rows,
307
218
            .rows_filtered_by_conjunct = rows_filtered_by_conjunct,
308
218
            .lazy_read_filtered_rows = lazy_read_filtered_rows,
309
218
            .total_batches = total_batches,
310
218
            .dense_batches = dense_batches,
311
218
            .selected_batches = selected_batches,
312
218
            .empty_selection_batches = empty_selection_batches,
313
218
            .range_gap_skipped_rows = range_gap_skipped_rows,
314
218
            .column_read_time = column_read_time,
315
218
            .predicate_filter_time = predicate_filter_time,
316
218
            .predicate_compaction_time = predicate_compaction_time,
317
218
            .predicate_compaction_bytes = predicate_compaction_bytes,
318
218
            .predicate_compaction_count = predicate_compaction_count,
319
218
            .fixed_width_predicate_direct_batches = fixed_width_predicate_direct_batches,
320
218
            .fixed_width_predicate_direct_rows = fixed_width_predicate_direct_rows,
321
218
            .dict_filter_rewrite_time = dict_filter_rewrite_time,
322
218
            .dict_filter_expr_rewrite_time = dict_filter_expr_rewrite_time,
323
218
            .dict_filter_read_dict_time = dict_filter_read_dict_time,
324
218
            .dict_filter_build_time = dict_filter_build_time,
325
218
            .dict_filter_candidate_columns = dict_filter_candidate_columns,
326
218
            .dict_filter_columns = dict_filter_columns,
327
218
            .dict_filter_unsupported_columns = dict_filter_unsupported_columns,
328
218
            .dict_filter_read_failures = dict_filter_read_failures,
329
218
            .rows_filtered_by_dict_filter = rows_filtered_by_dict_filter,
330
218
            .column_reader_profile = column_reader_profile(),
331
218
    };
332
218
}
333
334
} // namespace doris::format::parquet