xref: /aosp_15_r20/external/libtextclassifier/native/annotator/vocab/vocab-level-table.cc (revision 993b0882672172b81d12fad7a7ac0c3e5c824a12)
1*993b0882SAndroid Build Coastguard Worker /*
2*993b0882SAndroid Build Coastguard Worker  * Copyright (C) 2018 The Android Open Source Project
3*993b0882SAndroid Build Coastguard Worker  *
4*993b0882SAndroid Build Coastguard Worker  * Licensed under the Apache License, Version 2.0 (the "License");
5*993b0882SAndroid Build Coastguard Worker  * you may not use this file except in compliance with the License.
6*993b0882SAndroid Build Coastguard Worker  * You may obtain a copy of the License at
7*993b0882SAndroid Build Coastguard Worker  *
8*993b0882SAndroid Build Coastguard Worker  *      http://www.apache.org/licenses/LICENSE-2.0
9*993b0882SAndroid Build Coastguard Worker  *
10*993b0882SAndroid Build Coastguard Worker  * Unless required by applicable law or agreed to in writing, software
11*993b0882SAndroid Build Coastguard Worker  * distributed under the License is distributed on an "AS IS" BASIS,
12*993b0882SAndroid Build Coastguard Worker  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13*993b0882SAndroid Build Coastguard Worker  * See the License for the specific language governing permissions and
14*993b0882SAndroid Build Coastguard Worker  * limitations under the License.
15*993b0882SAndroid Build Coastguard Worker  */
16*993b0882SAndroid Build Coastguard Worker 
17*993b0882SAndroid Build Coastguard Worker #include "annotator/vocab/vocab-level-table.h"
18*993b0882SAndroid Build Coastguard Worker 
19*993b0882SAndroid Build Coastguard Worker #include <cstddef>
20*993b0882SAndroid Build Coastguard Worker #include <memory>
21*993b0882SAndroid Build Coastguard Worker 
22*993b0882SAndroid Build Coastguard Worker #include "annotator/model_generated.h"
23*993b0882SAndroid Build Coastguard Worker #include "utils/base/endian.h"
24*993b0882SAndroid Build Coastguard Worker #include "utils/container/bit-vector.h"
25*993b0882SAndroid Build Coastguard Worker #include "utils/optional.h"
26*993b0882SAndroid Build Coastguard Worker #include "marisa/trie.h"
27*993b0882SAndroid Build Coastguard Worker 
28*993b0882SAndroid Build Coastguard Worker namespace libtextclassifier3 {
29*993b0882SAndroid Build Coastguard Worker 
Create(const VocabModel * model)30*993b0882SAndroid Build Coastguard Worker std::unique_ptr<VocabLevelTable> VocabLevelTable::Create(
31*993b0882SAndroid Build Coastguard Worker     const VocabModel* model) {
32*993b0882SAndroid Build Coastguard Worker   if (!LittleEndian::IsLittleEndian()) {
33*993b0882SAndroid Build Coastguard Worker     // TODO(tonymak) Consider making this work on a big endian device.
34*993b0882SAndroid Build Coastguard Worker     TC3_LOG(ERROR)
35*993b0882SAndroid Build Coastguard Worker         << "VocabLevelTable is only working on a little endian device.";
36*993b0882SAndroid Build Coastguard Worker     return nullptr;
37*993b0882SAndroid Build Coastguard Worker   }
38*993b0882SAndroid Build Coastguard Worker   const flatbuffers::Vector<uint8_t>* trie_data = model->vocab_trie();
39*993b0882SAndroid Build Coastguard Worker   if (trie_data == nullptr) {
40*993b0882SAndroid Build Coastguard Worker     TC3_LOG(ERROR) << "vocab_trie is missing from the model file.";
41*993b0882SAndroid Build Coastguard Worker     return nullptr;
42*993b0882SAndroid Build Coastguard Worker   }
43*993b0882SAndroid Build Coastguard Worker   std::unique_ptr<marisa::Trie> vocab_trie(new marisa::Trie);
44*993b0882SAndroid Build Coastguard Worker   vocab_trie->map(trie_data->data(), trie_data->size());
45*993b0882SAndroid Build Coastguard Worker 
46*993b0882SAndroid Build Coastguard Worker   return std::unique_ptr<VocabLevelTable>(new VocabLevelTable(
47*993b0882SAndroid Build Coastguard Worker       model, std::move(vocab_trie), BitVector(model->beginner_level()),
48*993b0882SAndroid Build Coastguard Worker       BitVector(model->do_not_trigger_in_upper_case())));
49*993b0882SAndroid Build Coastguard Worker }
50*993b0882SAndroid Build Coastguard Worker 
VocabLevelTable(const VocabModel * model,std::unique_ptr<marisa::Trie> vocab_trie,const BitVector beginner_level,const BitVector do_not_trigger_in_upper_case)51*993b0882SAndroid Build Coastguard Worker VocabLevelTable::VocabLevelTable(const VocabModel* model,
52*993b0882SAndroid Build Coastguard Worker                                  std::unique_ptr<marisa::Trie> vocab_trie,
53*993b0882SAndroid Build Coastguard Worker                                  const BitVector beginner_level,
54*993b0882SAndroid Build Coastguard Worker                                  const BitVector do_not_trigger_in_upper_case)
55*993b0882SAndroid Build Coastguard Worker     : model_(model),
56*993b0882SAndroid Build Coastguard Worker       vocab_trie_(std::move(vocab_trie)),
57*993b0882SAndroid Build Coastguard Worker       beginner_level_(beginner_level),
58*993b0882SAndroid Build Coastguard Worker       do_not_trigger_in_upper_case_(do_not_trigger_in_upper_case) {}
59*993b0882SAndroid Build Coastguard Worker 
Lookup(const std::string & vocab) const60*993b0882SAndroid Build Coastguard Worker Optional<LookupResult> VocabLevelTable::Lookup(const std::string& vocab) const {
61*993b0882SAndroid Build Coastguard Worker   marisa::Agent agent;
62*993b0882SAndroid Build Coastguard Worker   agent.set_query(vocab.data(), vocab.size());
63*993b0882SAndroid Build Coastguard Worker   if (vocab_trie_->lookup(agent)) {
64*993b0882SAndroid Build Coastguard Worker     const int vector_idx = agent.key().id();
65*993b0882SAndroid Build Coastguard Worker     return Optional<LookupResult>({beginner_level_[vector_idx],
66*993b0882SAndroid Build Coastguard Worker                                    do_not_trigger_in_upper_case_[vector_idx]});
67*993b0882SAndroid Build Coastguard Worker   }
68*993b0882SAndroid Build Coastguard Worker   return Optional<LookupResult>();
69*993b0882SAndroid Build Coastguard Worker }
70*993b0882SAndroid Build Coastguard Worker }  // namespace libtextclassifier3
71