Coverage Report

Created: 2026-08-06 20:25

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
be/src/format/jni/jni_data_bridge.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 "jni_data_bridge.h"
19
20
#include <glog/logging.h>
21
22
#include <sstream>
23
#include <variant>
24
25
#include "core/block/block.h"
26
#include "core/column/column_array.h"
27
#include "core/column/column_map.h"
28
#include "core/column/column_nullable.h"
29
#include "core/column/column_string.h"
30
#include "core/column/column_struct.h"
31
#include "core/column/column_varbinary.h"
32
#include "core/data_type/data_type_array.h"
33
#include "core/data_type/data_type_map.h"
34
#include "core/data_type/data_type_nullable.h"
35
#include "core/data_type/data_type_struct.h"
36
#include "core/data_type/data_type_varbinary.h"
37
#include "core/data_type/define_primitive_type.h"
38
#include "core/data_type/primitive_type.h"
39
#include "core/types.h"
40
#include "core/value/decimalv2_value.h"
41
#include "util/string_util.h"
42
#include "util/url_coding.h"
43
44
namespace doris {
45
46
#define FOR_FIXED_LENGTH_TYPES(M)                                  \
47
0
    M(PrimitiveType::TYPE_TINYINT, ColumnInt8, Int8)               \
48
0
    M(PrimitiveType::TYPE_BOOLEAN, ColumnUInt8, UInt8)             \
49
0
    M(PrimitiveType::TYPE_SMALLINT, ColumnInt16, Int16)            \
50
0
    M(PrimitiveType::TYPE_INT, ColumnInt32, Int32)                 \
51
0
    M(PrimitiveType::TYPE_BIGINT, ColumnInt64, Int64)              \
52
0
    M(PrimitiveType::TYPE_LARGEINT, ColumnInt128, Int128)          \
53
0
    M(PrimitiveType::TYPE_FLOAT, ColumnFloat32, Float32)           \
54
0
    M(PrimitiveType::TYPE_DOUBLE, ColumnFloat64, Float64)          \
55
0
    M(PrimitiveType::TYPE_DECIMALV2, ColumnDecimal128V2, Int128)   \
56
0
    M(PrimitiveType::TYPE_DECIMAL128I, ColumnDecimal128V3, Int128) \
57
0
    M(PrimitiveType::TYPE_DECIMAL32, ColumnDecimal32, Int32)       \
58
0
    M(PrimitiveType::TYPE_DECIMAL64, ColumnDecimal64, Int64)       \
59
0
    M(PrimitiveType::TYPE_DATE, ColumnDate, Int64)                 \
60
0
    M(PrimitiveType::TYPE_DATEV2, ColumnDateV2, UInt32)            \
61
0
    M(PrimitiveType::TYPE_DATETIME, ColumnDateTime, Int64)         \
62
0
    M(PrimitiveType::TYPE_DATETIMEV2, ColumnDateTimeV2, UInt64)    \
63
0
    M(PrimitiveType::TYPE_TIMESTAMPTZ, ColumnTimeStampTz, UInt64)  \
64
0
    M(PrimitiveType::TYPE_IPV4, ColumnIPv4, IPv4)                  \
65
0
    M(PrimitiveType::TYPE_IPV6, ColumnIPv6, IPv6)
66
67
0
Status JniDataBridge::fill_block(Block* block, const ColumnNumbers& arguments, long table_address) {
68
0
    if (table_address == 0) {
69
0
        return Status::InternalError("table_address is 0");
70
0
    }
71
0
    TableMetaAddress table_meta(table_address);
72
0
    long num_rows = table_meta.next_meta_as_long();
73
0
    for (size_t i : arguments) {
74
0
        if (block->get_by_position(i).column.get() == nullptr) {
75
0
            auto return_type = block->get_data_type(i);
76
0
            bool result_nullable = return_type->is_nullable();
77
0
            ColumnUInt8::MutablePtr null_col = nullptr;
78
0
            if (result_nullable) {
79
0
                return_type = remove_nullable(return_type);
80
0
                null_col = ColumnUInt8::create();
81
0
            }
82
0
            auto res_col = return_type->create_column();
83
0
            if (result_nullable) {
84
0
                block->replace_by_position(
85
0
                        i, ColumnNullable::create(std::move(res_col), std::move(null_col)));
86
0
            } else {
87
0
                block->replace_by_position(i, std::move(res_col));
88
0
            }
89
0
        } else if (is_column_const(*(block->get_by_position(i).column))) {
90
0
            auto doris_column = block->get_by_position(i).column->convert_to_full_column_if_const();
91
0
            bool is_nullable = block->get_by_position(i).type->is_nullable();
92
0
            block->replace_by_position(i, is_nullable ? make_nullable(doris_column) : doris_column);
93
0
        }
94
0
        auto& column_with_type_and_name = block->get_by_position(i);
95
0
        auto& column_ptr = column_with_type_and_name.column;
96
0
        auto& column_type = column_with_type_and_name.type;
97
0
        RETURN_IF_ERROR(fill_column(table_meta, column_ptr, column_type, num_rows));
98
0
    }
99
0
    return Status::OK();
100
0
}
101
102
Status JniDataBridge::fill_column(TableMetaAddress& address, ColumnPtr& doris_column,
103
0
                                  const DataTypePtr& data_type, size_t num_rows) {
104
0
    auto logical_type = data_type->get_primitive_type();
105
0
    void* null_map_ptr = address.next_meta_as_ptr();
106
0
    if (null_map_ptr == nullptr) {
107
        // org.apache.doris.common.jni.vec.ColumnType.Type#UNSUPPORTED will set column address as 0
108
0
        return Status::InternalError("Unsupported type {} in java side", data_type->get_name());
109
0
    }
110
0
    auto mutable_doris_column = IColumn::mutate(std::move(doris_column));
111
0
    MutableColumnPtr data_column;
112
0
    if (auto* nullable_column = check_and_get_column<ColumnNullable>(mutable_doris_column.get())) {
113
0
        data_column = nullable_column->get_nested_column_ptr();
114
0
        NullMap& null_map = nullable_column->get_null_map_data();
115
0
        size_t origin_size = null_map.size();
116
0
        null_map.resize(origin_size + num_rows);
117
0
        memcpy(null_map.data() + origin_size, static_cast<bool*>(null_map_ptr), num_rows);
118
0
    } else {
119
0
        data_column = mutable_doris_column->get_ptr();
120
0
    }
121
    // Date and DateTime are deprecated and not supported.
122
0
    Status status = Status::OK();
123
0
    switch (logical_type) {
124
0
#define DISPATCH(TYPE_INDEX, COLUMN_TYPE, CPP_TYPE)                                             \
125
0
    case TYPE_INDEX: {                                                                          \
126
0
        auto* data = reinterpret_cast<CPP_TYPE*>(address.next_meta_as_ptr());                   \
127
0
        status = _fill_fixed_length_column<COLUMN_TYPE, CPP_TYPE>(data_column, data, num_rows); \
128
0
        break;                                                                                  \
129
0
    }
130
0
        FOR_FIXED_LENGTH_TYPES(DISPATCH)
131
0
#undef DISPATCH
132
0
    case PrimitiveType::TYPE_STRING:
133
0
        [[fallthrough]];
134
0
    case PrimitiveType::TYPE_CHAR:
135
0
        [[fallthrough]];
136
0
    case PrimitiveType::TYPE_VARCHAR:
137
0
        status = _fill_string_column(address, data_column, num_rows);
138
0
        break;
139
0
    case PrimitiveType::TYPE_ARRAY:
140
0
        status = _fill_array_column(address, data_column, data_type, num_rows);
141
0
        break;
142
0
    case PrimitiveType::TYPE_MAP:
143
0
        status = _fill_map_column(address, data_column, data_type, num_rows);
144
0
        break;
145
0
    case PrimitiveType::TYPE_STRUCT:
146
0
        status = _fill_struct_column(address, data_column, data_type, num_rows);
147
0
        break;
148
0
    case PrimitiveType::TYPE_VARBINARY:
149
0
        status = _fill_varbinary_column(address, data_column, num_rows);
150
0
        break;
151
0
    default:
152
0
        status = Status::InvalidArgument("Unsupported type {} in jni scanner",
153
0
                                         data_type->get_name());
154
0
        break;
155
0
    }
156
0
    doris_column = std::move(mutable_doris_column);
157
0
    return status;
158
0
}
159
160
Status JniDataBridge::_fill_varbinary_column(TableMetaAddress& address,
161
0
                                             MutableColumnPtr& doris_column, size_t num_rows) {
162
0
    auto* meta_base = reinterpret_cast<char*>(address.next_meta_as_ptr());
163
0
    auto& varbinary_col = assert_cast<ColumnVarbinary&>(*doris_column);
164
    // Java side writes per-row metadata as 16 bytes: [len: long][addr: long]
165
0
    for (size_t i = 0; i < num_rows; ++i) {
166
        // Read length (first 8 bytes)
167
0
        int64_t len = 0;
168
0
        memcpy(&len, meta_base + 16 * i, sizeof(len));
169
0
        if (len <= 0) {
170
0
            varbinary_col.insert_default();
171
0
        } else {
172
            // Read address (next 8 bytes)
173
0
            uint64_t addr_u = 0;
174
0
            memcpy(&addr_u, meta_base + 16 * i + 8, sizeof(addr_u));
175
0
            const char* src = reinterpret_cast<const char*>(addr_u);
176
0
            varbinary_col.insert_data(src, static_cast<size_t>(len));
177
0
        }
178
0
    }
179
0
    return Status::OK();
180
0
}
181
182
Status JniDataBridge::_fill_string_column(TableMetaAddress& address, MutableColumnPtr& doris_column,
183
0
                                          size_t num_rows) {
184
0
    auto& string_col = static_cast<ColumnString&>(*doris_column);
185
0
    ColumnString::Chars& string_chars = string_col.get_chars();
186
0
    ColumnString::Offsets& string_offsets = string_col.get_offsets();
187
0
    int* offsets = reinterpret_cast<int*>(address.next_meta_as_ptr());
188
0
    char* chars = reinterpret_cast<char*>(address.next_meta_as_ptr());
189
190
    // This judgment is necessary, otherwise the following statement `offsets[num_rows - 1]` out of bounds
191
    // What's more, This judgment must be placed after `address.next_meta_as_ptr()`
192
    // because `address.next_meta_as_ptr` will make `address._meta_index` plus 1
193
0
    if (num_rows == 0) {
194
0
        return Status::OK();
195
0
    }
196
197
0
    size_t origin_chars_size = string_chars.size();
198
0
    string_chars.resize(origin_chars_size + offsets[num_rows - 1]);
199
0
    memcpy(string_chars.data() + origin_chars_size, chars, offsets[num_rows - 1]);
200
201
0
    size_t origin_offsets_size = string_offsets.size();
202
0
    size_t start_offset = string_offsets[origin_offsets_size - 1];
203
0
    string_offsets.resize(origin_offsets_size + num_rows);
204
0
    for (size_t i = 0; i < num_rows; ++i) {
205
0
        string_offsets[origin_offsets_size + i] =
206
0
                static_cast<unsigned int>(offsets[i] + start_offset);
207
0
    }
208
0
    return Status::OK();
209
0
}
210
211
Status JniDataBridge::_fill_array_column(TableMetaAddress& address, MutableColumnPtr& doris_column,
212
0
                                         const DataTypePtr& data_type, size_t num_rows) {
213
0
    ColumnPtr& element_column = static_cast<ColumnArray&>(*doris_column).get_data_ptr();
214
0
    const DataTypePtr& element_type =
215
0
            (assert_cast<const DataTypeArray*>(remove_nullable(data_type).get()))
216
0
                    ->get_nested_type();
217
0
    ColumnArray::Offsets64& offsets_data = static_cast<ColumnArray&>(*doris_column).get_offsets();
218
219
0
    int64_t* offsets = reinterpret_cast<int64_t*>(address.next_meta_as_ptr());
220
0
    size_t origin_size = offsets_data.size();
221
0
    offsets_data.resize(origin_size + num_rows);
222
0
    size_t start_offset = offsets_data[origin_size - 1];
223
0
    for (size_t i = 0; i < num_rows; ++i) {
224
0
        offsets_data[origin_size + i] = offsets[i] + start_offset;
225
0
    }
226
227
0
    return fill_column(address, element_column, element_type,
228
0
                       offsets_data[origin_size + num_rows - 1] - start_offset);
229
0
}
230
231
Status JniDataBridge::_fill_map_column(TableMetaAddress& address, MutableColumnPtr& doris_column,
232
0
                                       const DataTypePtr& data_type, size_t num_rows) {
233
0
    auto& map = static_cast<ColumnMap&>(*doris_column);
234
0
    const DataTypePtr& key_type =
235
0
            reinterpret_cast<const DataTypeMap*>(remove_nullable(data_type).get())->get_key_type();
236
0
    const DataTypePtr& value_type =
237
0
            reinterpret_cast<const DataTypeMap*>(remove_nullable(data_type).get())
238
0
                    ->get_value_type();
239
0
    ColumnPtr& key_column = map.get_keys_ptr();
240
0
    ColumnPtr& value_column = map.get_values_ptr();
241
0
    ColumnArray::Offsets64& map_offsets = map.get_offsets();
242
243
0
    int64_t* offsets = reinterpret_cast<int64_t*>(address.next_meta_as_ptr());
244
0
    size_t origin_size = map_offsets.size();
245
0
    map_offsets.resize(origin_size + num_rows);
246
0
    size_t start_offset = map_offsets[origin_size - 1];
247
0
    for (size_t i = 0; i < num_rows; ++i) {
248
0
        map_offsets[origin_size + i] = offsets[i] + start_offset;
249
0
    }
250
251
0
    RETURN_IF_ERROR(fill_column(address, key_column, key_type,
252
0
                                map_offsets[origin_size + num_rows - 1] - start_offset));
253
0
    RETURN_IF_ERROR(fill_column(address, value_column, value_type,
254
0
                                map_offsets[origin_size + num_rows - 1] - start_offset));
255
0
    return Status::OK();
256
0
}
257
258
Status JniDataBridge::_fill_struct_column(TableMetaAddress& address, MutableColumnPtr& doris_column,
259
0
                                          const DataTypePtr& data_type, size_t num_rows) {
260
0
    auto& doris_struct = static_cast<ColumnStruct&>(*doris_column);
261
0
    const DataTypeStruct* doris_struct_type =
262
0
            reinterpret_cast<const DataTypeStruct*>(remove_nullable(data_type).get());
263
0
    for (int i = 0; i < doris_struct.tuple_size(); ++i) {
264
0
        ColumnPtr& struct_field = doris_struct.get_column_ptr(i);
265
0
        const DataTypePtr& field_type = doris_struct_type->get_element(i);
266
0
        RETURN_IF_ERROR(fill_column(address, struct_field, field_type, num_rows));
267
0
    }
268
0
    return Status::OK();
269
0
}
270
271
0
std::string JniDataBridge::get_jni_type(const DataTypePtr& data_type) {
272
0
    DataTypePtr type = remove_nullable(data_type);
273
0
    std::ostringstream buffer;
274
0
    switch (type->get_primitive_type()) {
275
0
    case TYPE_BOOLEAN:
276
0
        return "boolean";
277
0
    case TYPE_TINYINT:
278
0
        return "tinyint";
279
0
    case TYPE_SMALLINT:
280
0
        return "smallint";
281
0
    case TYPE_INT:
282
0
        return "int";
283
0
    case TYPE_BIGINT:
284
0
        return "bigint";
285
0
    case TYPE_LARGEINT:
286
0
        return "largeint";
287
0
    case TYPE_FLOAT:
288
0
        return "float";
289
0
    case TYPE_DOUBLE:
290
0
        return "double";
291
0
    case TYPE_IPV4:
292
0
        return "ipv4";
293
0
    case TYPE_IPV6:
294
0
        return "ipv6";
295
0
    case TYPE_VARCHAR:
296
0
        [[fallthrough]];
297
0
    case TYPE_CHAR:
298
0
        [[fallthrough]];
299
0
    case TYPE_STRING:
300
0
        return "string";
301
0
    case TYPE_DATE:
302
0
        return "datev1";
303
0
    case TYPE_DATEV2:
304
0
        return "datev2";
305
0
    case TYPE_DATETIME:
306
0
        return "datetimev1";
307
0
    case TYPE_DATETIMEV2:
308
0
        [[fallthrough]];
309
0
    case TYPE_TIMEV2: {
310
0
        buffer << "datetimev2(" << type->get_scale() << ")";
311
0
        return buffer.str();
312
0
    }
313
0
    case TYPE_TIMESTAMPTZ: {
314
0
        buffer << "timestamptz(" << type->get_scale() << ")";
315
0
        return buffer.str();
316
0
    }
317
0
    case TYPE_BINARY:
318
0
        return "binary";
319
0
    case TYPE_DECIMALV2: {
320
0
        buffer << "decimalv2(" << DecimalV2Value::PRECISION << "," << DecimalV2Value::SCALE << ")";
321
0
        return buffer.str();
322
0
    }
323
0
    case TYPE_DECIMAL32: {
324
0
        buffer << "decimal32(" << type->get_precision() << "," << type->get_scale() << ")";
325
0
        return buffer.str();
326
0
    }
327
0
    case TYPE_DECIMAL64: {
328
0
        buffer << "decimal64(" << type->get_precision() << "," << type->get_scale() << ")";
329
0
        return buffer.str();
330
0
    }
331
0
    case TYPE_DECIMAL128I: {
332
0
        buffer << "decimal128(" << type->get_precision() << "," << type->get_scale() << ")";
333
0
        return buffer.str();
334
0
    }
335
0
    case TYPE_STRUCT: {
336
0
        const DataTypeStruct* struct_type = reinterpret_cast<const DataTypeStruct*>(type.get());
337
0
        buffer << "struct<";
338
0
        for (int i = 0; i < struct_type->get_elements().size(); ++i) {
339
0
            if (i != 0) {
340
0
                buffer << ",";
341
0
            }
342
0
            buffer << struct_type->get_element_names()[i] << ":"
343
0
                   << get_jni_type(struct_type->get_element(i));
344
0
        }
345
0
        buffer << ">";
346
0
        return buffer.str();
347
0
    }
348
0
    case TYPE_ARRAY: {
349
0
        const DataTypeArray* array_type = reinterpret_cast<const DataTypeArray*>(type.get());
350
0
        buffer << "array<" << get_jni_type(array_type->get_nested_type()) << ">";
351
0
        return buffer.str();
352
0
    }
353
0
    case TYPE_MAP: {
354
0
        const DataTypeMap* map_type = reinterpret_cast<const DataTypeMap*>(type.get());
355
0
        buffer << "map<" << get_jni_type(map_type->get_key_type()) << ","
356
0
               << get_jni_type(map_type->get_value_type()) << ">";
357
0
        return buffer.str();
358
0
    }
359
0
    case TYPE_VARBINARY:
360
0
        return "varbinary";
361
    // bitmap, hll, quantile_state, jsonb are transferred as strings via JNI
362
0
    case TYPE_BITMAP:
363
0
        [[fallthrough]];
364
0
    case TYPE_HLL:
365
0
        [[fallthrough]];
366
0
    case TYPE_QUANTILE_STATE:
367
0
        [[fallthrough]];
368
0
    case TYPE_JSONB:
369
0
        return "string";
370
0
    default:
371
0
        return "unsupported";
372
0
    }
373
0
}
374
375
6
std::string JniDataBridge::get_jni_type_with_different_string(const DataTypePtr& data_type) {
376
6
    DataTypePtr type = remove_nullable(data_type);
377
6
    std::ostringstream buffer;
378
6
    switch (data_type->get_primitive_type()) {
379
0
    case TYPE_BOOLEAN:
380
0
        return "boolean";
381
0
    case TYPE_TINYINT:
382
0
        return "tinyint";
383
0
    case TYPE_SMALLINT:
384
0
        return "smallint";
385
0
    case TYPE_INT:
386
0
        return "int";
387
0
    case TYPE_BIGINT:
388
0
        return "bigint";
389
0
    case TYPE_LARGEINT:
390
0
        return "largeint";
391
0
    case TYPE_FLOAT:
392
0
        return "float";
393
0
    case TYPE_DOUBLE:
394
0
        return "double";
395
0
    case TYPE_IPV4:
396
0
        return "ipv4";
397
0
    case TYPE_IPV6:
398
0
        return "ipv6";
399
0
    case TYPE_VARCHAR: {
400
0
        buffer << "varchar("
401
0
               << assert_cast<const DataTypeString*>(remove_nullable(data_type).get())->len()
402
0
               << ")";
403
0
        return buffer.str();
404
0
    }
405
0
    case TYPE_DATE:
406
0
        return "datev1";
407
0
    case TYPE_DATEV2:
408
0
        return "datev2";
409
0
    case TYPE_DATETIME:
410
0
        return "datetimev1";
411
0
    case TYPE_DATETIMEV2:
412
0
        [[fallthrough]];
413
0
    case TYPE_TIMEV2: {
414
0
        buffer << "datetimev2(" << data_type->get_scale() << ")";
415
0
        return buffer.str();
416
0
    }
417
0
    case TYPE_TIMESTAMPTZ: {
418
0
        buffer << "timestamptz(" << data_type->get_scale() << ")";
419
0
        return buffer.str();
420
0
    }
421
0
    case TYPE_BINARY:
422
0
        return "binary";
423
0
    case TYPE_CHAR: {
424
0
        buffer << "char("
425
0
               << assert_cast<const DataTypeString*>(remove_nullable(data_type).get())->len()
426
0
               << ")";
427
0
        return buffer.str();
428
0
    }
429
6
    case TYPE_STRING:
430
6
        return "string";
431
0
    case TYPE_VARBINARY:
432
0
        buffer << "varbinary("
433
0
               << assert_cast<const DataTypeVarbinary*>(remove_nullable(data_type).get())->len()
434
0
               << ")";
435
0
        return buffer.str();
436
0
    case TYPE_DECIMALV2: {
437
0
        buffer << "decimalv2(" << DecimalV2Value::PRECISION << "," << DecimalV2Value::SCALE << ")";
438
0
        return buffer.str();
439
0
    }
440
0
    case TYPE_DECIMAL32: {
441
0
        buffer << "decimal32(" << data_type->get_precision() << "," << data_type->get_scale()
442
0
               << ")";
443
0
        return buffer.str();
444
0
    }
445
0
    case TYPE_DECIMAL64: {
446
0
        buffer << "decimal64(" << data_type->get_precision() << "," << data_type->get_scale()
447
0
               << ")";
448
0
        return buffer.str();
449
0
    }
450
0
    case TYPE_DECIMAL128I: {
451
0
        buffer << "decimal128(" << data_type->get_precision() << "," << data_type->get_scale()
452
0
               << ")";
453
0
        return buffer.str();
454
0
    }
455
0
    case TYPE_STRUCT: {
456
0
        const auto* type_struct =
457
0
                assert_cast<const DataTypeStruct*>(remove_nullable(data_type).get());
458
0
        buffer << "struct<";
459
0
        for (int i = 0; i < type_struct->get_elements().size(); ++i) {
460
0
            if (i != 0) {
461
0
                buffer << ",";
462
0
            }
463
0
            buffer << type_struct->get_element_name(i) << ":"
464
0
                   << get_jni_type_with_different_string(type_struct->get_element(i));
465
0
        }
466
0
        buffer << ">";
467
0
        return buffer.str();
468
0
    }
469
0
    case TYPE_ARRAY: {
470
0
        const auto* type_arr = assert_cast<const DataTypeArray*>(remove_nullable(data_type).get());
471
0
        buffer << "array<" << get_jni_type_with_different_string(type_arr->get_nested_type())
472
0
               << ">";
473
0
        return buffer.str();
474
0
    }
475
0
    case TYPE_MAP: {
476
0
        const auto* type_map = assert_cast<const DataTypeMap*>(remove_nullable(data_type).get());
477
0
        buffer << "map<" << get_jni_type_with_different_string(type_map->get_key_type()) << ","
478
0
               << get_jni_type_with_different_string(type_map->get_value_type()) << ">";
479
0
        return buffer.str();
480
0
    }
481
    // bitmap, hll, quantile_state, jsonb are transferred as strings via JNI
482
0
    case TYPE_BITMAP:
483
0
        [[fallthrough]];
484
0
    case TYPE_HLL:
485
0
        [[fallthrough]];
486
0
    case TYPE_QUANTILE_STATE:
487
0
        [[fallthrough]];
488
0
    case TYPE_JSONB:
489
0
        return "string";
490
0
    default:
491
0
        return "unsupported";
492
6
    }
493
6
}
494
495
9
std::string JniDataBridge::encode_schema_values(const std::vector<std::string>& values) {
496
9
    std::vector<std::string> encoded_values;
497
9
    encoded_values.reserve(values.size());
498
9
    for (const auto& value : values) {
499
6
        std::string encoded;
500
6
        base64_encode(value, &encoded);
501
        // Prefix every element so an empty Base64 token remains distinct from an empty list.
502
6
        encoded_values.emplace_back("$" + encoded);
503
6
    }
504
9
    return join(encoded_values, ",");
505
9
}
506
507
5
std::string JniDataBridge::get_jni_type_with_encoded_struct_fields(const DataTypePtr& data_type) {
508
5
    switch (data_type->get_primitive_type()) {
509
1
    case TYPE_STRUCT: {
510
1
        const auto* type_struct =
511
1
                assert_cast<const DataTypeStruct*>(remove_nullable(data_type).get());
512
1
        std::ostringstream buffer;
513
1
        buffer << "struct<";
514
4
        for (int i = 0; i < type_struct->get_elements().size(); ++i) {
515
3
            if (i != 0) {
516
2
                buffer << ",";
517
2
            }
518
3
            std::string encoded_name;
519
3
            base64_encode(type_struct->get_element_name(i), &encoded_name);
520
            // '$' versions the nested-name token and cannot collide with Base64 or grammar
521
            // delimiters; Java rejects an unversioned name in the encoded schema path.
522
3
            buffer << "$" << encoded_name << ":"
523
3
                   << get_jni_type_with_encoded_struct_fields(type_struct->get_element(i));
524
3
        }
525
1
        buffer << ">";
526
1
        return buffer.str();
527
0
    }
528
0
    case TYPE_ARRAY: {
529
0
        const auto* type_array =
530
0
                assert_cast<const DataTypeArray*>(remove_nullable(data_type).get());
531
0
        return "array<" + get_jni_type_with_encoded_struct_fields(type_array->get_nested_type()) +
532
0
               ">";
533
0
    }
534
0
    case TYPE_MAP: {
535
0
        const auto* type_map = assert_cast<const DataTypeMap*>(remove_nullable(data_type).get());
536
0
        return "map<" + get_jni_type_with_encoded_struct_fields(type_map->get_key_type()) + "," +
537
0
               get_jni_type_with_encoded_struct_fields(type_map->get_value_type()) + ">";
538
0
    }
539
4
    default:
540
4
        return get_jni_type_with_different_string(data_type);
541
5
    }
542
5
}
543
544
Status JniDataBridge::_fill_column_meta(const ColumnPtr& doris_column, const DataTypePtr& data_type,
545
0
                                        std::vector<long>& meta_data) {
546
0
    auto logical_type = data_type->get_primitive_type();
547
0
    const IColumn* column = nullptr;
548
    // insert const flag
549
0
    if (is_column_const(*doris_column)) {
550
0
        meta_data.emplace_back((long)1);
551
0
        const auto& const_column = assert_cast<const ColumnConst&>(*doris_column);
552
0
        column = &(const_column.get_data_column());
553
0
    } else {
554
0
        meta_data.emplace_back((long)0);
555
0
        column = &(*doris_column);
556
0
    }
557
558
    // insert null map address
559
0
    const IColumn* data_column = nullptr;
560
0
    if (const auto* nullable_column = check_and_get_column<ColumnNullable>(column)) {
561
0
        data_column = &(nullable_column->get_nested_column());
562
0
        const auto& null_map = nullable_column->get_null_map_data();
563
0
        meta_data.emplace_back((long)null_map.data());
564
0
    } else {
565
0
        meta_data.emplace_back(0);
566
0
        data_column = column;
567
0
    }
568
0
    switch (logical_type) {
569
0
#define DISPATCH(TYPE_INDEX, COLUMN_TYPE, CPP_TYPE)                                          \
570
0
    case TYPE_INDEX: {                                                                       \
571
0
        meta_data.emplace_back(_get_fixed_length_column_address<COLUMN_TYPE>(*data_column)); \
572
0
        break;                                                                               \
573
0
    }
574
0
        FOR_FIXED_LENGTH_TYPES(DISPATCH)
575
0
#undef DISPATCH
576
0
    case PrimitiveType::TYPE_STRING:
577
0
        [[fallthrough]];
578
0
    case PrimitiveType::TYPE_CHAR:
579
0
        [[fallthrough]];
580
0
    case PrimitiveType::TYPE_VARCHAR: {
581
0
        const auto& string_column = assert_cast<const ColumnString&>(*data_column);
582
        // insert offsets
583
0
        meta_data.emplace_back((long)string_column.get_offsets().data());
584
0
        meta_data.emplace_back((long)string_column.get_chars().data());
585
0
        break;
586
0
    }
587
0
    case PrimitiveType::TYPE_ARRAY: {
588
0
        const auto& element_column = assert_cast<const ColumnArray&>(*data_column).get_data_ptr();
589
0
        meta_data.emplace_back(
590
0
                (long)assert_cast<const ColumnArray&>(*data_column).get_offsets().data());
591
0
        const auto& element_type =
592
0
                (assert_cast<const DataTypeArray*>(remove_nullable(data_type).get()))
593
0
                        ->get_nested_type();
594
0
        RETURN_IF_ERROR(_fill_column_meta(element_column, element_type, meta_data));
595
0
        break;
596
0
    }
597
0
    case PrimitiveType::TYPE_STRUCT: {
598
0
        const auto& doris_struct = assert_cast<const ColumnStruct&>(*data_column);
599
0
        const auto* doris_struct_type =
600
0
                assert_cast<const DataTypeStruct*>(remove_nullable(data_type).get());
601
0
        for (int i = 0; i < doris_struct.tuple_size(); ++i) {
602
0
            const auto& struct_field = doris_struct.get_column_ptr(i);
603
0
            const auto& field_type = doris_struct_type->get_element(i);
604
0
            RETURN_IF_ERROR(_fill_column_meta(struct_field, field_type, meta_data));
605
0
        }
606
0
        break;
607
0
    }
608
0
    case PrimitiveType::TYPE_MAP: {
609
0
        const auto& map = assert_cast<const ColumnMap&>(*data_column);
610
0
        const auto& key_type =
611
0
                assert_cast<const DataTypeMap*>(remove_nullable(data_type).get())->get_key_type();
612
0
        const auto& value_type =
613
0
                assert_cast<const DataTypeMap*>(remove_nullable(data_type).get())->get_value_type();
614
0
        const auto& key_column = map.get_keys_ptr();
615
0
        const auto& value_column = map.get_values_ptr();
616
0
        meta_data.emplace_back((long)map.get_offsets().data());
617
0
        RETURN_IF_ERROR(_fill_column_meta(key_column, key_type, meta_data));
618
0
        RETURN_IF_ERROR(_fill_column_meta(value_column, value_type, meta_data));
619
0
        break;
620
0
    }
621
0
    case PrimitiveType::TYPE_VARBINARY: {
622
0
        const auto& varbinary_col = assert_cast<const ColumnVarbinary&>(*data_column);
623
0
        meta_data.emplace_back((long)varbinary_col.get_data().data());
624
0
        break;
625
0
    }
626
0
    default:
627
0
        return Status::InternalError("Unsupported type: {}", data_type->get_name());
628
0
    }
629
0
    return Status::OK();
630
0
}
631
632
0
Status JniDataBridge::to_java_table(Block* block, std::unique_ptr<long[]>& meta) {
633
0
    ColumnNumbers arguments;
634
0
    for (size_t i = 0; i < block->columns(); ++i) {
635
0
        arguments.emplace_back(i);
636
0
    }
637
0
    return to_java_table(block, block->rows(), arguments, meta);
638
0
}
639
640
Status JniDataBridge::to_java_table(Block* block, size_t num_rows, const ColumnNumbers& arguments,
641
0
                                    std::unique_ptr<long[]>& meta) {
642
0
    std::vector<long> meta_data;
643
    // insert number of rows
644
0
    meta_data.emplace_back(num_rows);
645
0
    for (size_t i : arguments) {
646
0
        auto& column_with_type_and_name = block->get_by_position(i);
647
0
        RETURN_IF_ERROR(_fill_column_meta(column_with_type_and_name.column,
648
0
                                          column_with_type_and_name.type, meta_data));
649
0
    }
650
651
0
    meta.reset(new long[meta_data.size()]);
652
0
    memcpy(meta.get(), &meta_data[0], meta_data.size() * 8);
653
0
    return Status::OK();
654
0
}
655
656
std::pair<std::string, std::string> JniDataBridge::parse_table_schema(
657
0
        Block* block, const ColumnNumbers& arguments, bool ignore_column_name) {
658
    // prepare table schema
659
0
    std::ostringstream required_fields;
660
0
    std::ostringstream columns_types;
661
0
    for (int i = 0; i < arguments.size(); ++i) {
662
0
        std::string type = JniDataBridge::get_jni_type(block->get_by_position(arguments[i]).type);
663
0
        if (i == 0) {
664
0
            if (ignore_column_name) {
665
0
                required_fields << "_col_" << arguments[i];
666
0
            } else {
667
0
                required_fields << block->get_by_position(arguments[i]).name;
668
0
            }
669
0
            columns_types << type;
670
0
        } else {
671
0
            if (ignore_column_name) {
672
0
                required_fields << ","
673
0
                                << "_col_" << arguments[i];
674
0
            } else {
675
0
                required_fields << "," << block->get_by_position(arguments[i]).name;
676
0
            }
677
0
            columns_types << "#" << type;
678
0
        }
679
0
    }
680
0
    return std::make_pair(required_fields.str(), columns_types.str());
681
0
}
682
683
0
std::pair<std::string, std::string> JniDataBridge::parse_table_schema(Block* block) {
684
0
    ColumnNumbers arguments;
685
0
    for (size_t i = 0; i < block->columns(); ++i) {
686
0
        arguments.emplace_back(i);
687
0
    }
688
0
    return parse_table_schema(block, arguments, true);
689
0
}
690
691
} // namespace doris