Coverage Report

Created: 2026-06-08 08:15

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
be/src/format/table/paimon_reader.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/table/paimon_reader.h"
19
20
#include <vector>
21
22
#include "common/status.h"
23
#include "format/table/deletion_vector_reader.h"
24
#include "runtime/runtime_state.h"
25
26
namespace doris {
27
28
// ============================================================================
29
// PaimonOrcReader
30
// ============================================================================
31
2.08k
void PaimonOrcReader::_init_paimon_profile() {
32
2.08k
    static const char* paimon_profile = "PaimonProfile";
33
2.08k
    ADD_TIMER(get_profile(), paimon_profile);
34
2.08k
    _paimon_profile.num_delete_rows =
35
2.08k
            ADD_CHILD_COUNTER(get_profile(), "NumDeleteRows", TUnit::UNIT, paimon_profile);
36
2.08k
    _paimon_profile.delete_files_read_time =
37
2.08k
            ADD_CHILD_TIMER(get_profile(), "DeleteFileReadTime", paimon_profile);
38
2.08k
    _paimon_profile.parse_deletion_vector_time =
39
2.08k
            ADD_CHILD_TIMER(get_profile(), "ParseDeletionVectorTime", paimon_profile);
40
2.08k
}
41
42
2.08k
Status PaimonOrcReader::on_before_init_reader(ReaderInitContext* ctx) {
43
2.08k
    _column_descs = ctx->column_descs;
44
2.08k
    _fill_col_name_to_block_idx = ctx->col_name_to_block_idx;
45
2.08k
    RETURN_IF_ERROR(_extract_partition_values(*ctx->range, ctx->tuple_descriptor,
46
2.08k
                                              _fill_partition_values,
47
2.08k
                                              &_fill_partition_value_is_null));
48
2.08k
    const orc::Type* orc_type_ptr = nullptr;
49
2.08k
    RETURN_IF_ERROR(get_file_type(&orc_type_ptr));
50
51
2.08k
    RETURN_IF_ERROR(gen_table_info_node_by_field_id(
52
2.08k
            get_scan_params(), get_scan_range().table_format_params.paimon_params.schema_id,
53
2.08k
            get_tuple_descriptor(), orc_type_ptr));
54
2.08k
    ctx->table_info_node = table_info_node_ptr;
55
56
12.4k
    for (const auto& desc : *ctx->column_descs) {
57
12.4k
        if (desc.category == ColumnCategory::REGULAR ||
58
12.4k
            desc.category == ColumnCategory::GENERATED) {
59
12.2k
            ctx->column_names.push_back(desc.name);
60
12.2k
        }
61
12.4k
    }
62
2.08k
    return Status::OK();
63
2.08k
}
64
65
2.08k
Status PaimonOrcReader::on_after_init_reader(ReaderInitContext* /*ctx*/) {
66
2.08k
    return _init_deletion_vector();
67
2.08k
}
68
69
2.08k
Status PaimonOrcReader::_init_deletion_vector() {
70
2.08k
    const auto& table_desc = get_scan_range().table_format_params.paimon_params;
71
2.08k
    if (!table_desc.__isset.deletion_file) {
72
2.06k
        return Status::OK();
73
2.06k
    }
74
75
    // Cannot do count push down if there are delete files
76
20
    if (!get_scan_range().table_format_params.paimon_params.__isset.row_count) {
77
20
        set_push_down_agg_type(TPushAggOp::NONE);
78
20
    }
79
20
    const auto& deletion_file = table_desc.deletion_file;
80
81
20
    Status create_status = Status::OK();
82
83
20
    std::string key;
84
20
    key.resize(deletion_file.path.size() + sizeof(deletion_file.offset));
85
20
    memcpy(key.data(), deletion_file.path.data(), deletion_file.path.size());
86
20
    memcpy(key.data() + deletion_file.path.size(), &deletion_file.offset,
87
20
           sizeof(deletion_file.offset));
88
89
20
    SCOPED_TIMER(_paimon_profile.delete_files_read_time);
90
20
    using DeleteRows = std::vector<int64_t>;
91
20
    _delete_rows = _kv_cache->get<DeleteRows>(key, [&]() -> DeleteRows* {
92
20
        auto* delete_rows = new DeleteRows;
93
94
20
        TFileRangeDesc delete_range;
95
20
        delete_range.__set_fs_name(get_scan_range().fs_name);
96
20
        delete_range.path = deletion_file.path;
97
20
        delete_range.start_offset = deletion_file.offset;
98
20
        delete_range.size = deletion_file.length + 4;
99
20
        delete_range.file_size = -1;
100
101
20
        DeletionVectorReader dv_reader(get_state(), get_profile(), get_scan_params(), delete_range,
102
20
                                       get_io_ctx());
103
20
        create_status = dv_reader.open();
104
20
        if (!create_status.ok()) [[unlikely]] {
105
0
            return nullptr;
106
0
        }
107
108
20
        size_t bytes_read = deletion_file.length + 4;
109
20
        std::vector<char> buffer(bytes_read);
110
20
        create_status = dv_reader.read_at(deletion_file.offset, {buffer.data(), bytes_read});
111
20
        if (!create_status.ok()) [[unlikely]] {
112
0
            return nullptr;
113
0
        }
114
115
20
        const char* buf = buffer.data();
116
20
        uint32_t actual_length;
117
20
        std::memcpy(reinterpret_cast<char*>(&actual_length), buf, 4);
118
20
        std::reverse(reinterpret_cast<char*>(&actual_length),
119
20
                     reinterpret_cast<char*>(&actual_length) + 4);
120
20
        buf += 4;
121
20
        if (actual_length != bytes_read - 4) [[unlikely]] {
122
0
            create_status = Status::RuntimeError(
123
0
                    "DeletionVector deserialize error: length not match, "
124
0
                    "actual length: {}, expect length: {}",
125
0
                    actual_length, bytes_read - 4);
126
0
            return nullptr;
127
0
        }
128
20
        uint32_t magic_number;
129
20
        std::memcpy(reinterpret_cast<char*>(&magic_number), buf, 4);
130
20
        std::reverse(reinterpret_cast<char*>(&magic_number),
131
20
                     reinterpret_cast<char*>(&magic_number) + 4);
132
20
        buf += 4;
133
20
        const static uint32_t MAGIC_NUMBER = 1581511376;
134
20
        if (magic_number != MAGIC_NUMBER) [[unlikely]] {
135
0
            create_status = Status::RuntimeError(
136
0
                    "DeletionVector deserialize error: invalid magic number {}", magic_number);
137
0
            return nullptr;
138
0
        }
139
140
20
        roaring::Roaring roaring_bitmap;
141
20
        SCOPED_TIMER(_paimon_profile.parse_deletion_vector_time);
142
20
        try {
143
20
            roaring_bitmap = roaring::Roaring::readSafe(buf, bytes_read - 4);
144
20
        } catch (const std::runtime_error& e) {
145
0
            create_status = Status::RuntimeError(
146
0
                    "DeletionVector deserialize error: failed to deserialize roaring bitmap, {}",
147
0
                    e.what());
148
0
            return nullptr;
149
0
        }
150
20
        delete_rows->reserve(roaring_bitmap.cardinality());
151
40
        for (auto it = roaring_bitmap.begin(); it != roaring_bitmap.end(); it++) {
152
20
            delete_rows->push_back(*it);
153
20
        }
154
20
        COUNTER_UPDATE(_paimon_profile.num_delete_rows, delete_rows->size());
155
20
        return delete_rows;
156
20
    });
157
20
    RETURN_IF_ERROR(create_status);
158
20
    if (!_delete_rows->empty()) [[likely]] {
159
20
        set_position_delete_rowids(_delete_rows);
160
20
    }
161
20
    return Status::OK();
162
20
}
163
164
// ============================================================================
165
// PaimonParquetReader
166
// ============================================================================
167
2.44k
void PaimonParquetReader::_init_paimon_profile() {
168
2.44k
    static const char* paimon_profile = "PaimonProfile";
169
2.44k
    ADD_TIMER(get_profile(), paimon_profile);
170
2.44k
    _paimon_profile.num_delete_rows =
171
2.44k
            ADD_CHILD_COUNTER(get_profile(), "NumDeleteRows", TUnit::UNIT, paimon_profile);
172
2.44k
    _paimon_profile.delete_files_read_time =
173
2.44k
            ADD_CHILD_TIMER(get_profile(), "DeleteFileReadTime", paimon_profile);
174
2.44k
    _paimon_profile.parse_deletion_vector_time =
175
2.44k
            ADD_CHILD_TIMER(get_profile(), "ParseDeletionVectorTime", paimon_profile);
176
2.44k
}
177
178
2.43k
Status PaimonParquetReader::on_before_init_reader(ReaderInitContext* ctx) {
179
2.43k
    _column_descs = ctx->column_descs;
180
2.43k
    _fill_col_name_to_block_idx = ctx->col_name_to_block_idx;
181
2.43k
    RETURN_IF_ERROR(_extract_partition_values(*ctx->range, ctx->tuple_descriptor,
182
2.43k
                                              _fill_partition_values,
183
2.43k
                                              &_fill_partition_value_is_null));
184
2.43k
    const FieldDescriptor* field_desc = nullptr;
185
2.43k
    RETURN_IF_ERROR(get_file_metadata_schema(&field_desc));
186
2.43k
    DCHECK(field_desc != nullptr);
187
188
2.43k
    RETURN_IF_ERROR(gen_table_info_node_by_field_id(
189
2.43k
            get_scan_params(), get_scan_range().table_format_params.paimon_params.schema_id,
190
2.43k
            get_tuple_descriptor(), *field_desc));
191
2.43k
    ctx->table_info_node = table_info_node_ptr;
192
193
7.25k
    for (const auto& desc : *ctx->column_descs) {
194
7.25k
        if (desc.category == ColumnCategory::REGULAR ||
195
7.25k
            desc.category == ColumnCategory::GENERATED) {
196
5.93k
            ctx->column_names.push_back(desc.name);
197
5.93k
        }
198
7.25k
    }
199
2.43k
    return Status::OK();
200
2.43k
}
201
202
2.44k
Status PaimonParquetReader::on_after_init_reader(ReaderInitContext* /*ctx*/) {
203
2.44k
    return _init_deletion_vector();
204
2.44k
}
205
206
2.44k
Status PaimonParquetReader::_init_deletion_vector() {
207
2.44k
    const auto& table_desc = get_scan_range().table_format_params.paimon_params;
208
2.44k
    if (!table_desc.__isset.deletion_file) {
209
2.41k
        return Status::OK();
210
2.41k
    }
211
212
28
    if (!get_scan_range().table_format_params.paimon_params.__isset.row_count) {
213
28
        set_push_down_agg_type(TPushAggOp::NONE);
214
28
    }
215
28
    const auto& deletion_file = table_desc.deletion_file;
216
217
28
    Status create_status = Status::OK();
218
219
28
    std::string key;
220
28
    key.resize(deletion_file.path.size() + sizeof(deletion_file.offset));
221
28
    memcpy(key.data(), deletion_file.path.data(), deletion_file.path.size());
222
28
    memcpy(key.data() + deletion_file.path.size(), &deletion_file.offset,
223
28
           sizeof(deletion_file.offset));
224
225
28
    SCOPED_TIMER(_paimon_profile.delete_files_read_time);
226
28
    using DeleteRows = std::vector<int64_t>;
227
28
    _delete_rows = _kv_cache->get<DeleteRows>(key, [&]() -> DeleteRows* {
228
28
        auto* delete_rows = new DeleteRows;
229
230
28
        TFileRangeDesc delete_range;
231
28
        delete_range.__set_fs_name(get_scan_range().fs_name);
232
28
        delete_range.path = deletion_file.path;
233
28
        delete_range.start_offset = deletion_file.offset;
234
28
        delete_range.size = deletion_file.length + 4;
235
28
        delete_range.file_size = -1;
236
237
28
        DeletionVectorReader dv_reader(get_state(), get_profile(), get_scan_params(), delete_range,
238
28
                                       get_io_ctx());
239
28
        create_status = dv_reader.open();
240
28
        if (!create_status.ok()) [[unlikely]] {
241
0
            return nullptr;
242
0
        }
243
244
28
        size_t bytes_read = deletion_file.length + 4;
245
28
        std::vector<char> buffer(bytes_read);
246
28
        create_status = dv_reader.read_at(deletion_file.offset, {buffer.data(), bytes_read});
247
28
        if (!create_status.ok()) [[unlikely]] {
248
0
            return nullptr;
249
0
        }
250
251
28
        const char* buf = buffer.data();
252
28
        uint32_t actual_length;
253
28
        std::memcpy(reinterpret_cast<char*>(&actual_length), buf, 4);
254
28
        std::reverse(reinterpret_cast<char*>(&actual_length),
255
28
                     reinterpret_cast<char*>(&actual_length) + 4);
256
28
        buf += 4;
257
28
        if (actual_length != bytes_read - 4) [[unlikely]] {
258
0
            create_status = Status::RuntimeError(
259
0
                    "DeletionVector deserialize error: length not match, "
260
0
                    "actual length: {}, expect length: {}",
261
0
                    actual_length, bytes_read - 4);
262
0
            return nullptr;
263
0
        }
264
28
        uint32_t magic_number;
265
28
        std::memcpy(reinterpret_cast<char*>(&magic_number), buf, 4);
266
28
        std::reverse(reinterpret_cast<char*>(&magic_number),
267
28
                     reinterpret_cast<char*>(&magic_number) + 4);
268
28
        buf += 4;
269
28
        const static uint32_t MAGIC_NUMBER = 1581511376;
270
28
        if (magic_number != MAGIC_NUMBER) [[unlikely]] {
271
0
            create_status = Status::RuntimeError(
272
0
                    "DeletionVector deserialize error: invalid magic number {}", magic_number);
273
0
            return nullptr;
274
0
        }
275
276
28
        roaring::Roaring roaring_bitmap;
277
28
        SCOPED_TIMER(_paimon_profile.parse_deletion_vector_time);
278
28
        try {
279
28
            roaring_bitmap = roaring::Roaring::readSafe(buf, bytes_read - 4);
280
28
        } catch (const std::runtime_error& e) {
281
0
            create_status = Status::RuntimeError(
282
0
                    "DeletionVector deserialize error: failed to deserialize roaring bitmap, {}",
283
0
                    e.what());
284
0
            return nullptr;
285
0
        }
286
28
        delete_rows->reserve(roaring_bitmap.cardinality());
287
56
        for (auto it = roaring_bitmap.begin(); it != roaring_bitmap.end(); it++) {
288
28
            delete_rows->push_back(*it);
289
28
        }
290
28
        COUNTER_UPDATE(_paimon_profile.num_delete_rows, delete_rows->size());
291
28
        return delete_rows;
292
28
    });
293
28
    RETURN_IF_ERROR(create_status);
294
28
    if (!_delete_rows->empty()) [[likely]] {
295
28
        ParquetReader::set_delete_rows(_delete_rows);
296
28
    }
297
28
    return Status::OK();
298
28
}
299
300
} // namespace doris