forked from alibaba/paimon-cpp
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathmap_shared_shredding_batch_converter.cpp
More file actions
344 lines (309 loc) · 16.7 KB
/
Copy pathmap_shared_shredding_batch_converter.cpp
File metadata and controls
344 lines (309 loc) · 16.7 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
/*
* Copyright 2026-present Alibaba Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "paimon/common/data/shredding/map_shared_shredding_batch_converter.h"
#include <map>
#include <string>
#include <utility>
#include "arrow/array.h"
#include "arrow/builder.h"
#include "arrow/c/bridge.h"
#include "arrow/type.h"
#include "fmt/format.h"
#include "paimon/common/data/shredding/lru_map_shared_shredding_column_allocator.h"
#include "paimon/common/data/shredding/map_shared_shredding_context.h"
#include "paimon/common/data/shredding/map_shared_shredding_utils.h"
#include "paimon/common/data/shredding/map_shredding_defs.h"
#include "paimon/common/data/shredding/plain_map_shared_shredding_column_allocator.h"
#include "paimon/common/data/shredding/sequential_map_shared_shredding_column_allocator.h"
#include "paimon/common/utils/arrow/mem_utils.h"
#include "paimon/common/utils/arrow/status_utils.h"
#include "paimon/core/core_options.h"
namespace paimon {
/// Checks that a dynamic_cast result is not null, returning Status::Invalid on failure.
#define PAIMON_CHECK_NOT_NULL(ptr, msg) \
do { \
if (PAIMON_UNLIKELY((ptr) == nullptr)) { \
return Status::Invalid(msg); \
} \
} while (false)
namespace {
Result<std::unique_ptr<MapSharedShreddingColumnAllocator>> CreateMapSharedShreddingColumnAllocator(
int32_t num_columns, MapSharedShreddingColumnPlacementPolicy placement_policy) {
switch (placement_policy) {
case MapSharedShreddingColumnPlacementPolicy::PLAIN:
return std::make_unique<PlainMapSharedShreddingColumnAllocator>(num_columns);
case MapSharedShreddingColumnPlacementPolicy::SEQUENTIAL:
return std::make_unique<SequentialMapSharedShreddingColumnAllocator>(num_columns);
case MapSharedShreddingColumnPlacementPolicy::LRU:
return std::make_unique<LruMapSharedShreddingColumnAllocator>(num_columns);
}
return Status::Invalid("unknown shared-shredding column placement policy");
}
} // namespace
Result<std::shared_ptr<MapSharedShreddingBatchConverter>> MapSharedShreddingBatchConverter::Create(
const std::shared_ptr<arrow::Schema>& logical_schema,
const std::shared_ptr<MapSharedShreddingContext>& context, const CoreOptions& options,
const std::shared_ptr<MemoryPool>& pool) {
std::map<std::string, int32_t> field_to_num_columns = context->ComputeNextK();
PAIMON_ASSIGN_OR_RAISE(
std::shared_ptr<arrow::Schema> physical_schema,
MapSharedShreddingUtils::LogicalToPhysicalSchema(logical_schema, field_to_num_columns));
std::vector<ColumnContext> contexts;
std::vector<std::string> shredding_field_names;
contexts.reserve(field_to_num_columns.size());
shredding_field_names.reserve(field_to_num_columns.size());
// Iterate in schema field order (not map order) so that shredding_field_names_
// matches the order in which shredding columns appear in the schema.
// This is critical for the sequential matching logic in Convert().
for (int32_t i = 0; i < logical_schema->num_fields(); ++i) {
const std::string& name = logical_schema->field(i)->name();
auto it = field_to_num_columns.find(name);
if (it != field_to_num_columns.end()) {
int32_t num_columns = it->second;
PAIMON_ASSIGN_OR_RAISE(MapSharedShreddingColumnPlacementPolicy placement_policy,
options.GetMapSharedShreddingColumnPlacementPolicy(name));
PAIMON_ASSIGN_OR_RAISE(
std::unique_ptr<MapSharedShreddingColumnAllocator> allocator,
CreateMapSharedShreddingColumnAllocator(num_columns, placement_policy));
contexts.emplace_back(name, num_columns, std::move(allocator));
shredding_field_names.push_back(name);
}
}
return std::shared_ptr<MapSharedShreddingBatchConverter>(
new MapSharedShreddingBatchConverter(logical_schema, physical_schema, std::move(contexts),
std::move(shredding_field_names), pool));
}
MapSharedShreddingBatchConverter::MapSharedShreddingBatchConverter(
const std::shared_ptr<arrow::Schema>& logical_schema,
const std::shared_ptr<arrow::Schema>& physical_schema, std::vector<ColumnContext>&& contexts,
std::vector<std::string>&& shredding_field_names, const std::shared_ptr<MemoryPool>& pool)
: logical_schema_(logical_schema),
physical_schema_(physical_schema),
contexts_(std::move(contexts)),
shredding_field_names_(std::move(shredding_field_names)),
pool_(GetArrowPool(pool)) {}
const std::shared_ptr<arrow::Schema>& MapSharedShreddingBatchConverter::GetPhysicalSchema() const {
return physical_schema_;
}
Result<std::unique_ptr<ArrowArray>> MapSharedShreddingBatchConverter::Convert(
ArrowArray* logical_batch) {
std::shared_ptr<arrow::DataType> logical_type = arrow::struct_(logical_schema_->fields());
PAIMON_ASSIGN_OR_RAISE_FROM_ARROW(std::shared_ptr<arrow::Array> logical_array,
arrow::ImportArray(logical_batch, logical_type));
auto logical_struct = std::dynamic_pointer_cast<arrow::StructArray>(logical_array);
PAIMON_CHECK_NOT_NULL(logical_struct,
"MapSharedShreddingBatchConverter: input is not a StructArray");
int32_t num_fields = logical_schema_->num_fields();
arrow::ArrayVector physical_columns;
physical_columns.reserve(num_fields);
size_t context_idx = 0;
for (int32_t col = 0; col < num_fields; ++col) {
auto column = logical_struct->field(col);
const std::string& field_name = logical_schema_->field(col)->name();
if (context_idx < shredding_field_names_.size() &&
shredding_field_names_[context_idx] == field_name) {
auto physical_struct_type = physical_schema_->field(col)->type();
PAIMON_ASSIGN_OR_RAISE(
std::shared_ptr<arrow::Array> physical_column,
ConvertOneColumn(column, physical_struct_type, &contexts_[context_idx]));
physical_columns.push_back(std::move(physical_column));
++context_idx;
} else {
physical_columns.push_back(column);
}
}
PAIMON_ASSIGN_OR_RAISE_FROM_ARROW(
std::shared_ptr<arrow::Array> physical_struct,
arrow::StructArray::Make(physical_columns, physical_schema_->field_names()));
std::unique_ptr<ArrowArray> result = std::make_unique<ArrowArray>();
PAIMON_RETURN_NOT_OK_FROM_ARROW(arrow::ExportArray(*physical_struct, result.get()));
return result;
}
Result<std::shared_ptr<arrow::Array>> MapSharedShreddingBatchConverter::ConvertOneColumn(
const std::shared_ptr<arrow::Array>& map_column,
const std::shared_ptr<arrow::DataType>& physical_struct_type, ColumnContext* context) const {
auto map_array = std::dynamic_pointer_cast<arrow::MapArray>(map_column);
PAIMON_CHECK_NOT_NULL(map_array, "MapSharedShreddingBatchConverter: column is not a MapArray");
int64_t num_rows = map_array->length();
int32_t num_cols = context->num_columns;
auto keys_array = std::dynamic_pointer_cast<arrow::StringArray>(map_array->keys());
PAIMON_CHECK_NOT_NULL(keys_array,
"MapSharedShreddingBatchConverter: MAP keys are not StringArray");
auto values_array = map_array->items();
// Create StructBuilder from physical struct type — it owns all child builders.
PAIMON_ASSIGN_OR_RAISE_FROM_ARROW(std::unique_ptr<arrow::ArrayBuilder> struct_builder_base,
arrow::MakeBuilder(physical_struct_type, pool_.get()));
auto* struct_builder = dynamic_cast<arrow::StructBuilder*>(struct_builder_base.get());
PAIMON_CHECK_NOT_NULL(struct_builder,
"MapSharedShreddingBatchConverter: failed to create StructBuilder");
PAIMON_RETURN_NOT_OK_FROM_ARROW(struct_builder->Reserve(num_rows));
// Extract child builders: [field_mapping, col_0..K-1, overflow]
auto* field_mapping_builder =
dynamic_cast<arrow::ListBuilder*>(struct_builder->field_builder(0));
PAIMON_CHECK_NOT_NULL(field_mapping_builder,
"MapSharedShreddingBatchConverter: field_mapping is not a ListBuilder");
auto* field_mapping_value_builder =
dynamic_cast<arrow::Int32Builder*>(field_mapping_builder->value_builder());
PAIMON_CHECK_NOT_NULL(
field_mapping_value_builder,
"MapSharedShreddingBatchConverter: field_mapping value is not Int32Builder");
PAIMON_RETURN_NOT_OK_FROM_ARROW(field_mapping_builder->Reserve(num_rows));
PAIMON_RETURN_NOT_OK_FROM_ARROW(field_mapping_value_builder->Reserve(num_rows * num_cols));
std::vector<arrow::ArrayBuilder*> col_builders_raw;
col_builders_raw.reserve(num_cols);
for (int32_t c = 0; c < num_cols; ++c) {
arrow::ArrayBuilder* col_builder = struct_builder->field_builder(1 + c);
PAIMON_CHECK_NOT_NULL(col_builder, "MapSharedShreddingBatchConverter: col builder is null");
PAIMON_RETURN_NOT_OK_FROM_ARROW(col_builder->Reserve(num_rows));
col_builders_raw.push_back(col_builder);
}
int32_t overflow_field_idx = 1 + num_cols;
auto* overflow_builder =
dynamic_cast<arrow::MapBuilder*>(struct_builder->field_builder(overflow_field_idx));
PAIMON_CHECK_NOT_NULL(overflow_builder,
"MapSharedShreddingBatchConverter: overflow is not a MapBuilder");
auto* overflow_key_builder =
dynamic_cast<arrow::Int32Builder*>(overflow_builder->key_builder());
PAIMON_CHECK_NOT_NULL(overflow_key_builder,
"MapSharedShreddingBatchConverter: overflow key is not Int32Builder");
arrow::ArrayBuilder* overflow_value_builder = overflow_builder->item_builder();
PAIMON_CHECK_NOT_NULL(overflow_value_builder,
"MapSharedShreddingBatchConverter: overflow value builder is null");
PAIMON_RETURN_NOT_OK_FROM_ARROW(overflow_builder->Reserve(num_rows));
// Process each row
for (int64_t row = 0; row < num_rows; ++row) {
if (map_array->IsNull(row)) {
// StructBuilder::AppendNull() auto-appends empty values to all children.
PAIMON_RETURN_NOT_OK_FROM_ARROW(struct_builder->AppendNull());
continue;
}
PAIMON_RETURN_NOT_OK_FROM_ARROW(struct_builder->Append());
int64_t start = map_array->value_offset(row);
int64_t length = map_array->value_length(row);
// Extract field ids and build lookup map
std::vector<int32_t> field_ids;
std::unordered_map<int32_t, int64_t> field_id_to_value_index;
ExtractRowFields(keys_array, start, length, &context->dict, &field_ids,
&field_id_to_value_index);
// Allocate columns
RowAllocation allocation = context->allocator->AllocateRow(field_ids);
// Fill sub-columns
PAIMON_RETURN_NOT_OK(AppendFieldMapping(allocation, num_cols, field_mapping_builder,
field_mapping_value_builder));
PAIMON_RETURN_NOT_OK(AppendColumnValues(values_array, allocation, field_id_to_value_index,
num_cols, col_builders_raw));
PAIMON_RETURN_NOT_OK(AppendOverflow(values_array, allocation, field_id_to_value_index,
overflow_builder, overflow_key_builder,
overflow_value_builder));
}
// Finalize
std::shared_ptr<arrow::StructArray> result;
PAIMON_RETURN_NOT_OK_FROM_ARROW(struct_builder->Finish(&result));
return result;
}
void MapSharedShreddingBatchConverter::ExtractRowFields(
const std::shared_ptr<arrow::StringArray>& keys_array, int64_t start, int64_t length,
MapSharedShreddingFieldDict* dict, std::vector<int32_t>* field_ids_out,
std::unordered_map<int32_t, int64_t>* field_id_to_value_index_out) const {
field_ids_out->clear();
field_ids_out->reserve(length);
field_id_to_value_index_out->clear();
field_id_to_value_index_out->reserve(length);
for (int64_t j = 0; j < length; ++j) {
std::string key_str = keys_array->GetString(start + j);
int32_t field_id = dict->GetOrAssign(key_str);
field_ids_out->push_back(field_id);
(*field_id_to_value_index_out)[field_id] = start + j;
}
}
Status MapSharedShreddingBatchConverter::AppendFieldMapping(
const RowAllocation& allocation, int32_t num_cols, arrow::ListBuilder* list_builder,
arrow::Int32Builder* value_builder) const {
PAIMON_RETURN_NOT_OK_FROM_ARROW(list_builder->Append());
for (int32_t c = 0; c < num_cols; ++c) {
PAIMON_RETURN_NOT_OK_FROM_ARROW(value_builder->Append(allocation.col_to_field[c]));
}
return Status::OK();
}
Status MapSharedShreddingBatchConverter::AppendColumnValues(
const std::shared_ptr<arrow::Array>& values_array, const RowAllocation& allocation,
const std::unordered_map<int32_t, int64_t>& field_id_to_value_index, int32_t num_cols,
const std::vector<arrow::ArrayBuilder*>& col_builders) const {
for (int32_t c = 0; c < num_cols; ++c) {
int32_t assigned_field_id = allocation.col_to_field[c];
if (assigned_field_id == -1) {
PAIMON_RETURN_NOT_OK_FROM_ARROW(col_builders[c]->AppendNull());
} else {
auto it = field_id_to_value_index.find(assigned_field_id);
if (PAIMON_UNLIKELY(it == field_id_to_value_index.end())) {
return Status::Invalid(
fmt::format("MapSharedShreddingBatchConverter: field_id {} assigned to col {} "
"but not found in current row",
assigned_field_id, c));
}
PAIMON_RETURN_NOT_OK_FROM_ARROW(
col_builders[c]->AppendArraySlice(*values_array->data(), it->second, 1));
}
}
return Status::OK();
}
Status MapSharedShreddingBatchConverter::AppendOverflow(
const std::shared_ptr<arrow::Array>& values_array, const RowAllocation& allocation,
const std::unordered_map<int32_t, int64_t>& field_id_to_value_index,
arrow::MapBuilder* overflow_builder, arrow::Int32Builder* overflow_key_builder,
arrow::ArrayBuilder* overflow_value_builder) const {
if (allocation.overflow_fields.empty()) {
PAIMON_RETURN_NOT_OK_FROM_ARROW(overflow_builder->AppendNull());
return Status::OK();
}
PAIMON_RETURN_NOT_OK_FROM_ARROW(overflow_builder->Append());
for (int32_t overflow_field_id : allocation.overflow_fields) {
auto it = field_id_to_value_index.find(overflow_field_id);
if (PAIMON_UNLIKELY(it == field_id_to_value_index.end())) {
return Status::Invalid(fmt::format(
"MapSharedShreddingBatchConverter: overflow field_id {} not found in current row",
overflow_field_id));
}
PAIMON_RETURN_NOT_OK_FROM_ARROW(overflow_key_builder->Append(overflow_field_id));
PAIMON_RETURN_NOT_OK_FROM_ARROW(
overflow_value_builder->AppendArraySlice(*values_array->data(), it->second, 1));
}
return Status::OK();
}
Result<MapSharedShreddingFieldMeta> MapSharedShreddingBatchConverter::BuildFieldMeta(
const std::string& field_name) const {
for (const auto& context : contexts_) {
if (context.field_name == field_name) {
MapSharedShreddingFieldMeta meta;
meta.name_to_id = context.dict.GetNameToId();
// Convert set<int32_t> -> vector<int32_t> for field_to_columns
for (const auto& [field_id, col_set] : context.allocator->GetFieldToColumns()) {
meta.field_to_columns[field_id] =
std::vector<int32_t>(col_set.begin(), col_set.end());
}
meta.overflow_field_set = context.allocator->GetOverflowFieldSet();
meta.num_columns = context.num_columns;
meta.max_row_width = context.allocator->GetMaxRowWidth();
return meta;
}
}
return Status::Invalid(fmt::format(
"cannot find field_name '{}' in MapSharedShreddingBatchConverter contexts", field_name));
}
const std::vector<std::string>& MapSharedShreddingBatchConverter::GetShreddingColumnNames() const {
return shredding_field_names_;
}
} // namespace paimon