be/src/format/table/paimon_predicate_converter.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_predicate_converter.h" |
19 | | |
20 | | #include <algorithm> |
21 | | #include <cctype> |
22 | | #include <utility> |
23 | | |
24 | | #include "core/column/column_const.h" |
25 | | #include "core/column/column_nullable.h" |
26 | | #include "core/data_type/data_type.h" |
27 | | #include "core/data_type/data_type_nullable.h" |
28 | | #include "core/field.h" |
29 | | #include "core/types.h" |
30 | | #include "core/value/decimalv2_value.h" |
31 | | #include "core/value/timestamptz_value.h" |
32 | | #include "core/value/vdatetime_value.h" |
33 | | #include "exprs/vcompound_pred.h" |
34 | | #include "exprs/vdirect_in_predicate.h" |
35 | | #include "exprs/vectorized_fn_call.h" |
36 | | #include "exprs/vexpr.h" |
37 | | #include "exprs/vin_predicate.h" |
38 | | #include "exprs/vliteral.h" |
39 | | #include "exprs/vslot_ref.h" |
40 | | #include "paimon/data/decimal.h" |
41 | | #include "paimon/data/timestamp.h" |
42 | | #include "paimon/predicate/predicate_builder.h" |
43 | | #include "runtime/descriptors.h" |
44 | | #include "runtime/runtime_state.h" |
45 | | #include "util/timezone_utils.h" |
46 | | |
47 | | namespace doris { |
48 | | #include "common/compile_check_begin.h" |
49 | | |
50 | | PaimonPredicateConverter::PaimonPredicateConverter( |
51 | | const std::vector<SlotDescriptor*>& file_slot_descs, RuntimeState* state) |
52 | 0 | : _state(state) { |
53 | 0 | _field_index_by_name.reserve(file_slot_descs.size()); |
54 | 0 | for (size_t i = 0; i < file_slot_descs.size(); ++i) { |
55 | 0 | const auto& name = file_slot_descs[i]->col_name(); |
56 | 0 | auto normalized = _normalize_name(name); |
57 | 0 | if (_field_index_by_name.find(normalized) == _field_index_by_name.end()) { |
58 | 0 | _field_index_by_name.emplace(std::move(normalized), static_cast<int32_t>(i)); |
59 | 0 | } |
60 | 0 | } |
61 | |
|
62 | 0 | if (!TimezoneUtils::find_cctz_time_zone("GMT", _gmt_tz)) { |
63 | 0 | TimezoneUtils::find_cctz_time_zone(TimezoneUtils::default_time_zone, _gmt_tz); |
64 | 0 | } |
65 | 0 | } |
66 | | |
67 | | std::shared_ptr<paimon::Predicate> PaimonPredicateConverter::build( |
68 | 0 | const VExprContextSPtrs& conjuncts) { |
69 | 0 | std::vector<std::shared_ptr<paimon::Predicate>> predicates; |
70 | 0 | predicates.reserve(conjuncts.size()); |
71 | 0 | for (const auto& conjunct : conjuncts) { |
72 | 0 | if (!conjunct || !conjunct->root()) { |
73 | 0 | continue; |
74 | 0 | } |
75 | 0 | auto root = conjunct->root(); |
76 | 0 | if (root->is_rf_wrapper()) { |
77 | 0 | if (auto impl = root->get_impl()) { |
78 | 0 | root = impl; |
79 | 0 | } |
80 | 0 | } |
81 | 0 | auto predicate = _convert_expr(root); |
82 | 0 | if (predicate) { |
83 | 0 | predicates.emplace_back(std::move(predicate)); |
84 | 0 | } |
85 | 0 | } |
86 | |
|
87 | 0 | if (predicates.empty()) { |
88 | 0 | return nullptr; |
89 | 0 | } |
90 | 0 | if (predicates.size() == 1) { |
91 | 0 | return predicates.front(); |
92 | 0 | } |
93 | 0 | auto and_result = paimon::PredicateBuilder::And(predicates); |
94 | 0 | if (!and_result.ok()) { |
95 | 0 | return nullptr; |
96 | 0 | } |
97 | 0 | return std::move(and_result).value(); |
98 | 0 | } |
99 | | |
100 | 0 | std::shared_ptr<paimon::Predicate> PaimonPredicateConverter::_convert_expr(const VExprSPtr& expr) { |
101 | 0 | if (!expr) { |
102 | 0 | return nullptr; |
103 | 0 | } |
104 | | |
105 | 0 | auto uncast = VExpr::expr_without_cast(expr); |
106 | |
|
107 | 0 | if (auto* direct_in = dynamic_cast<VDirectInPredicate*>(uncast.get())) { |
108 | 0 | VExprSPtr in_expr; |
109 | 0 | if (direct_in->get_slot_in_expr(in_expr)) { |
110 | 0 | return _convert_in(in_expr); |
111 | 0 | } |
112 | 0 | return nullptr; |
113 | 0 | } |
114 | | |
115 | 0 | if (dynamic_cast<VInPredicate*>(uncast.get()) != nullptr) { |
116 | 0 | return _convert_in(uncast); |
117 | 0 | } |
118 | | |
119 | 0 | switch (uncast->op()) { |
120 | 0 | case TExprOpcode::COMPOUND_AND: |
121 | 0 | case TExprOpcode::COMPOUND_OR: |
122 | 0 | return _convert_compound(uncast); |
123 | 0 | case TExprOpcode::COMPOUND_NOT: |
124 | 0 | return nullptr; |
125 | 0 | case TExprOpcode::EQ: |
126 | 0 | case TExprOpcode::EQ_FOR_NULL: |
127 | 0 | case TExprOpcode::NE: |
128 | 0 | case TExprOpcode::GE: |
129 | 0 | case TExprOpcode::GT: |
130 | 0 | case TExprOpcode::LE: |
131 | 0 | case TExprOpcode::LT: |
132 | 0 | return _convert_binary(uncast); |
133 | 0 | default: |
134 | 0 | break; |
135 | 0 | } |
136 | | |
137 | 0 | if (auto* fn = dynamic_cast<VectorizedFnCall*>(uncast.get())) { |
138 | 0 | auto fn_name = _normalize_name(fn->function_name()); |
139 | 0 | if (fn_name == "is_null_pred" || fn_name == "is_not_null_pred") { |
140 | 0 | return _convert_is_null(uncast, fn_name); |
141 | 0 | } |
142 | 0 | if (fn_name == "like") { |
143 | 0 | return _convert_like_prefix(uncast); |
144 | 0 | } |
145 | 0 | } |
146 | | |
147 | 0 | return nullptr; |
148 | 0 | } |
149 | | |
150 | | std::shared_ptr<paimon::Predicate> PaimonPredicateConverter::_convert_compound( |
151 | 0 | const VExprSPtr& expr) { |
152 | 0 | if (!expr || expr->get_num_children() != 2) { |
153 | 0 | return nullptr; |
154 | 0 | } |
155 | 0 | auto left = _convert_expr(expr->get_child(0)); |
156 | 0 | if (!left) { |
157 | 0 | return nullptr; |
158 | 0 | } |
159 | 0 | auto right = _convert_expr(expr->get_child(1)); |
160 | 0 | if (!right) { |
161 | 0 | return nullptr; |
162 | 0 | } |
163 | | |
164 | 0 | if (expr->op() == TExprOpcode::COMPOUND_AND) { |
165 | 0 | auto and_result = paimon::PredicateBuilder::And({left, right}); |
166 | 0 | return and_result.ok() ? std::move(and_result).value() : nullptr; |
167 | 0 | } |
168 | 0 | if (expr->op() == TExprOpcode::COMPOUND_OR) { |
169 | 0 | auto or_result = paimon::PredicateBuilder::Or({left, right}); |
170 | 0 | return or_result.ok() ? std::move(or_result).value() : nullptr; |
171 | 0 | } |
172 | 0 | return nullptr; |
173 | 0 | } |
174 | | |
175 | 0 | std::shared_ptr<paimon::Predicate> PaimonPredicateConverter::_convert_in(const VExprSPtr& expr) { |
176 | 0 | auto* in_pred = dynamic_cast<VInPredicate*>(expr.get()); |
177 | 0 | if (!in_pred || expr->get_num_children() < 2) { |
178 | 0 | return nullptr; |
179 | 0 | } |
180 | 0 | auto field_meta = _resolve_field(expr->get_child(0)); |
181 | 0 | if (!field_meta) { |
182 | 0 | return nullptr; |
183 | 0 | } |
184 | | |
185 | 0 | std::vector<paimon::Literal> literals; |
186 | 0 | literals.reserve(expr->get_num_children() - 1); |
187 | 0 | for (uint16_t i = 1; i < expr->get_num_children(); ++i) { |
188 | 0 | auto literal = _convert_literal(expr->get_child(i), *field_meta->slot_desc, |
189 | 0 | field_meta->field_type); |
190 | 0 | if (!literal) { |
191 | 0 | return nullptr; |
192 | 0 | } |
193 | 0 | literals.emplace_back(std::move(*literal)); |
194 | 0 | } |
195 | | |
196 | 0 | if (literals.empty()) { |
197 | 0 | return nullptr; |
198 | 0 | } |
199 | 0 | if (in_pred->is_not_in()) { |
200 | 0 | return paimon::PredicateBuilder::NotIn(field_meta->index, field_meta->slot_desc->col_name(), |
201 | 0 | field_meta->field_type, literals); |
202 | 0 | } |
203 | 0 | return paimon::PredicateBuilder::In(field_meta->index, field_meta->slot_desc->col_name(), |
204 | 0 | field_meta->field_type, literals); |
205 | 0 | } |
206 | | |
207 | | std::shared_ptr<paimon::Predicate> PaimonPredicateConverter::_convert_binary( |
208 | 0 | const VExprSPtr& expr) { |
209 | 0 | if (!expr || expr->get_num_children() != 2) { |
210 | 0 | return nullptr; |
211 | 0 | } |
212 | 0 | auto field_meta = _resolve_field(expr->get_child(0)); |
213 | 0 | if (!field_meta) { |
214 | 0 | return nullptr; |
215 | 0 | } |
216 | | |
217 | 0 | if (expr->op() == TExprOpcode::EQ_FOR_NULL) { |
218 | 0 | return paimon::PredicateBuilder::IsNull( |
219 | 0 | field_meta->index, field_meta->slot_desc->col_name(), field_meta->field_type); |
220 | 0 | } |
221 | | |
222 | 0 | auto literal = |
223 | 0 | _convert_literal(expr->get_child(1), *field_meta->slot_desc, field_meta->field_type); |
224 | 0 | if (!literal) { |
225 | 0 | return nullptr; |
226 | 0 | } |
227 | | |
228 | 0 | switch (expr->op()) { |
229 | 0 | case TExprOpcode::EQ: |
230 | 0 | return paimon::PredicateBuilder::Equal(field_meta->index, field_meta->slot_desc->col_name(), |
231 | 0 | field_meta->field_type, *literal); |
232 | 0 | case TExprOpcode::NE: |
233 | 0 | return paimon::PredicateBuilder::NotEqual(field_meta->index, |
234 | 0 | field_meta->slot_desc->col_name(), |
235 | 0 | field_meta->field_type, *literal); |
236 | 0 | case TExprOpcode::GE: |
237 | 0 | return paimon::PredicateBuilder::GreaterOrEqual(field_meta->index, |
238 | 0 | field_meta->slot_desc->col_name(), |
239 | 0 | field_meta->field_type, *literal); |
240 | 0 | case TExprOpcode::GT: |
241 | 0 | return paimon::PredicateBuilder::GreaterThan(field_meta->index, |
242 | 0 | field_meta->slot_desc->col_name(), |
243 | 0 | field_meta->field_type, *literal); |
244 | 0 | case TExprOpcode::LE: |
245 | 0 | return paimon::PredicateBuilder::LessOrEqual(field_meta->index, |
246 | 0 | field_meta->slot_desc->col_name(), |
247 | 0 | field_meta->field_type, *literal); |
248 | 0 | case TExprOpcode::LT: |
249 | 0 | return paimon::PredicateBuilder::LessThan(field_meta->index, |
250 | 0 | field_meta->slot_desc->col_name(), |
251 | 0 | field_meta->field_type, *literal); |
252 | 0 | default: |
253 | 0 | break; |
254 | 0 | } |
255 | 0 | return nullptr; |
256 | 0 | } |
257 | | |
258 | | std::shared_ptr<paimon::Predicate> PaimonPredicateConverter::_convert_is_null( |
259 | 0 | const VExprSPtr& expr, const std::string& fn_name) { |
260 | 0 | if (!expr || expr->get_num_children() != 1) { |
261 | 0 | return nullptr; |
262 | 0 | } |
263 | 0 | auto field_meta = _resolve_field(expr->get_child(0)); |
264 | 0 | if (!field_meta) { |
265 | 0 | return nullptr; |
266 | 0 | } |
267 | 0 | if (fn_name == "is_not_null_pred") { |
268 | 0 | return paimon::PredicateBuilder::IsNotNull( |
269 | 0 | field_meta->index, field_meta->slot_desc->col_name(), field_meta->field_type); |
270 | 0 | } |
271 | 0 | return paimon::PredicateBuilder::IsNull(field_meta->index, field_meta->slot_desc->col_name(), |
272 | 0 | field_meta->field_type); |
273 | 0 | } |
274 | | |
275 | | std::shared_ptr<paimon::Predicate> PaimonPredicateConverter::_convert_like_prefix( |
276 | 0 | const VExprSPtr& expr) { |
277 | 0 | if (!expr || expr->get_num_children() != 2) { |
278 | 0 | return nullptr; |
279 | 0 | } |
280 | 0 | auto field_meta = _resolve_field(expr->get_child(0)); |
281 | 0 | if (!field_meta || field_meta->field_type != paimon::FieldType::STRING) { |
282 | 0 | return nullptr; |
283 | 0 | } |
284 | | |
285 | 0 | auto pattern_opt = _extract_string_literal(expr->get_child(1)); |
286 | 0 | if (!pattern_opt) { |
287 | 0 | return nullptr; |
288 | 0 | } |
289 | 0 | const std::string& pattern = *pattern_opt; |
290 | 0 | if (!pattern.empty() && pattern.front() == '%') { |
291 | 0 | return nullptr; |
292 | 0 | } |
293 | 0 | if (pattern.empty() || pattern.back() != '%') { |
294 | 0 | return nullptr; |
295 | 0 | } |
296 | | |
297 | 0 | std::string prefix = pattern.substr(0, pattern.size() - 1); |
298 | 0 | paimon::Literal lower_literal(paimon::FieldType::STRING, prefix.data(), prefix.size()); |
299 | 0 | auto lower_pred = paimon::PredicateBuilder::GreaterOrEqual( |
300 | 0 | field_meta->index, field_meta->slot_desc->col_name(), field_meta->field_type, |
301 | 0 | lower_literal); |
302 | |
|
303 | 0 | auto upper_prefix = _next_prefix(prefix); |
304 | 0 | if (!upper_prefix) { |
305 | 0 | return lower_pred; |
306 | 0 | } |
307 | | |
308 | 0 | paimon::Literal upper_literal(paimon::FieldType::STRING, upper_prefix->data(), |
309 | 0 | upper_prefix->size()); |
310 | 0 | auto upper_pred = |
311 | 0 | paimon::PredicateBuilder::LessThan(field_meta->index, field_meta->slot_desc->col_name(), |
312 | 0 | field_meta->field_type, upper_literal); |
313 | 0 | auto and_result = paimon::PredicateBuilder::And({lower_pred, upper_pred}); |
314 | 0 | return and_result.ok() ? std::move(and_result).value() : nullptr; |
315 | 0 | } |
316 | | |
317 | | std::optional<PaimonPredicateConverter::FieldMeta> PaimonPredicateConverter::_resolve_field( |
318 | 0 | const VExprSPtr& expr) const { |
319 | 0 | if (!_state || !expr) { |
320 | 0 | return std::nullopt; |
321 | 0 | } |
322 | 0 | auto slot_expr = VExpr::expr_without_cast(expr); |
323 | 0 | auto* slot_ref = dynamic_cast<VSlotRef*>(slot_expr.get()); |
324 | 0 | if (!slot_ref) { |
325 | 0 | return std::nullopt; |
326 | 0 | } |
327 | 0 | auto* slot_desc = _state->desc_tbl().get_slot_descriptor(slot_ref->slot_id()); |
328 | 0 | if (!slot_desc) { |
329 | 0 | return std::nullopt; |
330 | 0 | } |
331 | 0 | auto normalized = _normalize_name(slot_desc->col_name()); |
332 | 0 | auto it = _field_index_by_name.find(normalized); |
333 | 0 | if (it == _field_index_by_name.end()) { |
334 | 0 | return std::nullopt; |
335 | 0 | } |
336 | 0 | auto slot_type = slot_desc->type(); |
337 | 0 | auto field_type = |
338 | 0 | _to_paimon_field_type(slot_type->get_primitive_type(), slot_type->get_precision()); |
339 | 0 | if (!field_type) { |
340 | 0 | return std::nullopt; |
341 | 0 | } |
342 | 0 | return FieldMeta {it->second, *field_type, slot_desc}; |
343 | 0 | } |
344 | | |
345 | | std::optional<paimon::Literal> PaimonPredicateConverter::_convert_literal( |
346 | | const VExprSPtr& expr, const SlotDescriptor& slot_desc, |
347 | 0 | paimon::FieldType field_type) const { |
348 | 0 | auto literal_expr = VExpr::expr_without_cast(expr); |
349 | 0 | auto* literal = dynamic_cast<VLiteral*>(literal_expr.get()); |
350 | 0 | if (!literal) { |
351 | 0 | return std::nullopt; |
352 | 0 | } |
353 | | |
354 | 0 | auto literal_type = remove_nullable(literal->get_data_type()); |
355 | 0 | PrimitiveType literal_primitive = literal_type->get_primitive_type(); |
356 | 0 | PrimitiveType slot_primitive = slot_desc.type()->get_primitive_type(); |
357 | |
|
358 | 0 | ColumnPtr col = literal->get_column_ptr()->convert_to_full_column_if_const(); |
359 | 0 | if (const auto* nullable = check_and_get_column<ColumnNullable>(*col)) { |
360 | 0 | if (nullable->is_null_at(0)) { |
361 | 0 | return std::nullopt; |
362 | 0 | } |
363 | 0 | col = nullable->get_nested_column_ptr(); |
364 | 0 | } |
365 | | |
366 | 0 | Field field; |
367 | 0 | col->get(0, field); |
368 | |
|
369 | 0 | switch (slot_primitive) { |
370 | 0 | case TYPE_BOOLEAN: { |
371 | 0 | if (literal_primitive != TYPE_BOOLEAN) { |
372 | 0 | return std::nullopt; |
373 | 0 | } |
374 | 0 | return paimon::Literal(static_cast<bool>(field.get<TYPE_BOOLEAN>())); |
375 | 0 | } |
376 | 0 | case TYPE_TINYINT: |
377 | 0 | case TYPE_SMALLINT: |
378 | 0 | case TYPE_INT: |
379 | 0 | case TYPE_BIGINT: { |
380 | 0 | if (!_is_integer_type(literal_primitive)) { |
381 | 0 | return std::nullopt; |
382 | 0 | } |
383 | 0 | int64_t value = 0; |
384 | 0 | switch (literal_primitive) { |
385 | 0 | case TYPE_TINYINT: |
386 | 0 | value = field.get<TYPE_TINYINT>(); |
387 | 0 | break; |
388 | 0 | case TYPE_SMALLINT: |
389 | 0 | value = field.get<TYPE_SMALLINT>(); |
390 | 0 | break; |
391 | 0 | case TYPE_INT: |
392 | 0 | value = field.get<TYPE_INT>(); |
393 | 0 | break; |
394 | 0 | case TYPE_BIGINT: |
395 | 0 | value = field.get<TYPE_BIGINT>(); |
396 | 0 | break; |
397 | 0 | default: |
398 | 0 | return std::nullopt; |
399 | 0 | } |
400 | 0 | if (slot_primitive == TYPE_TINYINT) { |
401 | 0 | return paimon::Literal(static_cast<int8_t>(value)); |
402 | 0 | } |
403 | 0 | if (slot_primitive == TYPE_SMALLINT) { |
404 | 0 | return paimon::Literal(static_cast<int16_t>(value)); |
405 | 0 | } |
406 | 0 | if (slot_primitive == TYPE_INT) { |
407 | 0 | return paimon::Literal(static_cast<int32_t>(value)); |
408 | 0 | } |
409 | 0 | return paimon::Literal(static_cast<int64_t>(value)); |
410 | 0 | } |
411 | 0 | case TYPE_DOUBLE: { |
412 | 0 | if (literal_primitive != TYPE_DOUBLE && literal_primitive != TYPE_FLOAT) { |
413 | 0 | return std::nullopt; |
414 | 0 | } |
415 | 0 | double value = 0; |
416 | 0 | if (literal_primitive == TYPE_FLOAT) { |
417 | 0 | value = static_cast<double>(field.get<TYPE_FLOAT>()); |
418 | 0 | } else { |
419 | 0 | value = field.get<TYPE_DOUBLE>(); |
420 | 0 | } |
421 | 0 | return paimon::Literal(value); |
422 | 0 | } |
423 | 0 | case TYPE_DATE: |
424 | 0 | case TYPE_DATEV2: { |
425 | 0 | if (!_is_date_type(literal_primitive)) { |
426 | 0 | return std::nullopt; |
427 | 0 | } |
428 | 0 | int64_t seconds = 0; |
429 | 0 | if (literal_primitive == TYPE_DATE) { |
430 | 0 | const auto& dt = field.get<TYPE_DATE>(); |
431 | 0 | if (!dt.is_valid_date()) { |
432 | 0 | return std::nullopt; |
433 | 0 | } |
434 | 0 | dt.unix_timestamp(&seconds, _gmt_tz); |
435 | 0 | } else if (literal_primitive == TYPE_DATEV2) { |
436 | 0 | const auto& dt = field.get<TYPE_DATEV2>(); |
437 | 0 | if (!dt.is_valid_date()) { |
438 | 0 | return std::nullopt; |
439 | 0 | } |
440 | 0 | dt.unix_timestamp(&seconds, _gmt_tz); |
441 | 0 | } |
442 | 0 | int32_t days = _seconds_to_days(seconds); |
443 | 0 | return paimon::Literal(paimon::FieldType::DATE, days); |
444 | 0 | } |
445 | 0 | case TYPE_DATETIME: |
446 | 0 | case TYPE_DATETIMEV2: { |
447 | 0 | if (!_is_datetime_type(literal_primitive)) { |
448 | 0 | return std::nullopt; |
449 | 0 | } |
450 | 0 | if (literal_primitive == TYPE_DATETIME) { |
451 | 0 | const auto& dt = field.get<TYPE_DATETIME>(); |
452 | 0 | if (!dt.is_valid_date()) { |
453 | 0 | return std::nullopt; |
454 | 0 | } |
455 | 0 | int64_t seconds = 0; |
456 | 0 | dt.unix_timestamp(&seconds, _gmt_tz); |
457 | 0 | return paimon::Literal(paimon::Timestamp::FromEpochMillis(seconds * 1000)); |
458 | 0 | } |
459 | 0 | std::pair<int64_t, int64_t> ts; |
460 | 0 | const auto& dt = field.get<TYPE_DATETIMEV2>(); |
461 | 0 | if (!dt.is_valid_date()) { |
462 | 0 | return std::nullopt; |
463 | 0 | } |
464 | 0 | dt.unix_timestamp(&ts, _gmt_tz); |
465 | 0 | int64_t millis = ts.first * 1000 + ts.second / 1000; |
466 | 0 | return paimon::Literal(paimon::Timestamp::FromEpochMillis(millis)); |
467 | 0 | } |
468 | 0 | case TYPE_VARCHAR: |
469 | 0 | case TYPE_STRING: { |
470 | 0 | if (!_is_string_type(literal_primitive)) { |
471 | 0 | return std::nullopt; |
472 | 0 | } |
473 | 0 | const auto& value = field.get<TYPE_STRING>(); |
474 | 0 | return paimon::Literal(field_type, value.data(), value.size()); |
475 | 0 | } |
476 | 0 | case TYPE_DECIMALV2: |
477 | 0 | case TYPE_DECIMAL32: |
478 | 0 | case TYPE_DECIMAL64: |
479 | 0 | case TYPE_DECIMAL128I: |
480 | 0 | case TYPE_DECIMAL256: { |
481 | 0 | if (!_is_decimal_type(literal_primitive)) { |
482 | 0 | return std::nullopt; |
483 | 0 | } |
484 | 0 | int32_t precision = static_cast<int32_t>(literal_type->get_precision()); |
485 | 0 | int32_t scale = static_cast<int32_t>(literal_type->get_scale()); |
486 | 0 | if (precision <= 0 || precision > paimon::Decimal::MAX_PRECISION) { |
487 | 0 | return std::nullopt; |
488 | 0 | } |
489 | | |
490 | 0 | paimon::Decimal::int128_t value = 0; |
491 | 0 | switch (literal_primitive) { |
492 | 0 | case TYPE_DECIMALV2: { |
493 | 0 | const auto& dec = field.get<TYPE_DECIMALV2>(); |
494 | 0 | value = dec.value(); |
495 | 0 | break; |
496 | 0 | } |
497 | 0 | case TYPE_DECIMAL32: { |
498 | 0 | const auto& dec = field.get<TYPE_DECIMAL32>(); |
499 | 0 | value = dec.value; |
500 | 0 | break; |
501 | 0 | } |
502 | 0 | case TYPE_DECIMAL64: { |
503 | 0 | const auto& dec = field.get<TYPE_DECIMAL64>(); |
504 | 0 | value = dec.value; |
505 | 0 | break; |
506 | 0 | } |
507 | 0 | case TYPE_DECIMAL128I: { |
508 | 0 | const auto& dec = field.get<TYPE_DECIMAL128I>(); |
509 | 0 | value = dec.value; |
510 | 0 | break; |
511 | 0 | } |
512 | 0 | default: |
513 | 0 | return std::nullopt; |
514 | 0 | } |
515 | 0 | return paimon::Literal(paimon::Decimal(precision, scale, value)); |
516 | 0 | } |
517 | 0 | default: |
518 | 0 | break; |
519 | 0 | } |
520 | 0 | return std::nullopt; |
521 | 0 | } |
522 | | |
523 | | std::optional<std::string> PaimonPredicateConverter::_extract_string_literal( |
524 | 0 | const VExprSPtr& expr) const { |
525 | 0 | auto literal_expr = VExpr::expr_without_cast(expr); |
526 | 0 | auto* literal = dynamic_cast<VLiteral*>(literal_expr.get()); |
527 | 0 | if (!literal) { |
528 | 0 | return std::nullopt; |
529 | 0 | } |
530 | 0 | auto literal_type = remove_nullable(literal->get_data_type()); |
531 | 0 | PrimitiveType literal_primitive = literal_type->get_primitive_type(); |
532 | 0 | if (!_is_string_type(literal_primitive)) { |
533 | 0 | return std::nullopt; |
534 | 0 | } |
535 | | |
536 | 0 | ColumnPtr col = literal->get_column_ptr()->convert_to_full_column_if_const(); |
537 | 0 | if (const auto* nullable = check_and_get_column<ColumnNullable>(*col)) { |
538 | 0 | if (nullable->is_null_at(0)) { |
539 | 0 | return std::nullopt; |
540 | 0 | } |
541 | 0 | col = nullable->get_nested_column_ptr(); |
542 | 0 | } |
543 | 0 | Field field; |
544 | 0 | col->get(0, field); |
545 | 0 | const auto& value = field.get<TYPE_STRING>(); |
546 | 0 | return value; |
547 | 0 | } |
548 | | |
549 | 0 | std::string PaimonPredicateConverter::_normalize_name(std::string_view name) { |
550 | 0 | std::string out(name); |
551 | 0 | std::transform(out.begin(), out.end(), out.begin(), |
552 | 0 | [](unsigned char c) { return static_cast<char>(std::tolower(c)); }); |
553 | 0 | return out; |
554 | 0 | } |
555 | | |
556 | 0 | std::optional<std::string> PaimonPredicateConverter::_next_prefix(const std::string& prefix) { |
557 | 0 | if (prefix.empty()) { |
558 | 0 | return std::nullopt; |
559 | 0 | } |
560 | 0 | std::string upper = prefix; |
561 | 0 | for (int i = static_cast<int>(upper.size()) - 1; i >= 0; --i) { |
562 | 0 | auto c = static_cast<unsigned char>(upper[i]); |
563 | 0 | if (c != 0xFF) { |
564 | 0 | upper[i] = static_cast<char>(c + 1); |
565 | 0 | upper.resize(i + 1); |
566 | 0 | return upper; |
567 | 0 | } |
568 | 0 | } |
569 | 0 | return std::nullopt; |
570 | 0 | } |
571 | | |
572 | 0 | int32_t PaimonPredicateConverter::_seconds_to_days(int64_t seconds) { |
573 | 0 | static constexpr int64_t kSecondsPerDay = 24 * 60 * 60; |
574 | 0 | int64_t days = seconds / kSecondsPerDay; |
575 | 0 | if (seconds < 0 && seconds % kSecondsPerDay != 0) { |
576 | 0 | --days; |
577 | 0 | } |
578 | 0 | return static_cast<int32_t>(days); |
579 | 0 | } |
580 | | |
581 | 0 | bool PaimonPredicateConverter::_is_integer_type(PrimitiveType type) { |
582 | 0 | switch (type) { |
583 | 0 | case TYPE_TINYINT: |
584 | 0 | case TYPE_SMALLINT: |
585 | 0 | case TYPE_INT: |
586 | 0 | case TYPE_BIGINT: |
587 | 0 | return true; |
588 | 0 | default: |
589 | 0 | return false; |
590 | 0 | } |
591 | 0 | } |
592 | | |
593 | 0 | bool PaimonPredicateConverter::_is_string_type(PrimitiveType type) { |
594 | 0 | return type == TYPE_CHAR || type == TYPE_VARCHAR || type == TYPE_STRING; |
595 | 0 | } |
596 | | |
597 | 0 | bool PaimonPredicateConverter::_is_decimal_type(PrimitiveType type) { |
598 | 0 | switch (type) { |
599 | 0 | case TYPE_DECIMALV2: |
600 | 0 | case TYPE_DECIMAL32: |
601 | 0 | case TYPE_DECIMAL64: |
602 | 0 | case TYPE_DECIMAL128I: |
603 | 0 | case TYPE_DECIMAL256: |
604 | 0 | return true; |
605 | 0 | default: |
606 | 0 | return false; |
607 | 0 | } |
608 | 0 | } |
609 | | |
610 | 0 | bool PaimonPredicateConverter::_is_date_type(PrimitiveType type) { |
611 | 0 | return type == TYPE_DATE || type == TYPE_DATEV2; |
612 | 0 | } |
613 | | |
614 | 0 | bool PaimonPredicateConverter::_is_datetime_type(PrimitiveType type) { |
615 | 0 | return type == TYPE_DATETIME || type == TYPE_DATETIMEV2; |
616 | 0 | } |
617 | | |
618 | | std::optional<paimon::FieldType> PaimonPredicateConverter::_to_paimon_field_type( |
619 | 0 | PrimitiveType type, uint32_t precision) { |
620 | 0 | switch (type) { |
621 | 0 | case TYPE_BOOLEAN: |
622 | 0 | return paimon::FieldType::BOOLEAN; |
623 | 0 | case TYPE_TINYINT: |
624 | 0 | return paimon::FieldType::TINYINT; |
625 | 0 | case TYPE_SMALLINT: |
626 | 0 | return paimon::FieldType::SMALLINT; |
627 | 0 | case TYPE_INT: |
628 | 0 | return paimon::FieldType::INT; |
629 | 0 | case TYPE_BIGINT: |
630 | 0 | return paimon::FieldType::BIGINT; |
631 | 0 | case TYPE_DOUBLE: |
632 | 0 | return paimon::FieldType::DOUBLE; |
633 | 0 | case TYPE_VARCHAR: |
634 | 0 | case TYPE_STRING: |
635 | 0 | return paimon::FieldType::STRING; |
636 | 0 | case TYPE_DATE: |
637 | 0 | case TYPE_DATEV2: |
638 | 0 | return paimon::FieldType::DATE; |
639 | 0 | case TYPE_DATETIME: |
640 | 0 | case TYPE_DATETIMEV2: |
641 | 0 | return paimon::FieldType::TIMESTAMP; |
642 | 0 | case TYPE_DECIMALV2: |
643 | 0 | case TYPE_DECIMAL32: |
644 | 0 | case TYPE_DECIMAL64: |
645 | 0 | case TYPE_DECIMAL128I: |
646 | 0 | case TYPE_DECIMAL256: |
647 | 0 | if (precision > 0 && precision > paimon::Decimal::MAX_PRECISION) { |
648 | 0 | return std::nullopt; |
649 | 0 | } |
650 | 0 | return paimon::FieldType::DECIMAL; |
651 | 0 | case TYPE_FLOAT: |
652 | 0 | case TYPE_CHAR: |
653 | 0 | default: |
654 | 0 | return std::nullopt; |
655 | 0 | } |
656 | 0 | } |
657 | | |
658 | | #include "common/compile_check_end.h" |
659 | | } // namespace doris |