1bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi/* 2bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * Copyright (C) 2017 The Android Open Source Project 3bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * 4bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * Licensed under the Apache License, Version 2.0 (the "License"); 5bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * you may not use this file except in compliance with the License. 6bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * You may obtain a copy of the License at 7bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * 8bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * http://www.apache.org/licenses/LICENSE-2.0 9bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * 10bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * Unless required by applicable law or agreed to in writing, software 11bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * distributed under the License is distributed on an "AS IS" BASIS, 12bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * See the License for the specific language governing permissions and 14bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi * limitations under the License. 15bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi */ 16bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 17bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#ifndef LIBTEXTCLASSIFIER_COMMON_FEATURE_DESCRIPTORS_H_ 18bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#define LIBTEXTCLASSIFIER_COMMON_FEATURE_DESCRIPTORS_H_ 19bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 20bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#include <memory> 21bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#include <string> 22bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#include <vector> 23bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 24bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#include "util/base/integral_types.h" 25bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#include "util/base/logging.h" 26bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#include "util/base/macros.h" 27bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 28bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifinamespace libtextclassifier { 29bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifinamespace nlp_core { 30bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 31bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi// Named feature parameter. 32bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharificlass Parameter { 33bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi public: 34bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi Parameter() {} 35bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 36bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi void set_name(const std::string &name) { name_ = name; } 37bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi const std::string &name() const { return name_; } 38bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 39bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi void set_value(const std::string &value) { value_ = value; } 40bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi const std::string &value() const { return value_; } 41bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 42bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi private: 43bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi std::string name_; 44bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi std::string value_; 45bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi}; 46bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 47bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi// Descriptor for a feature function. Used to store the results of parsing one 48bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi// feature function. 49bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharificlass FeatureFunctionDescriptor { 50bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi public: 51bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi FeatureFunctionDescriptor() {} 52bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 53bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // Accessors for the feature function type. The function type is the string 54bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // that the feature extractor code is registered under. 55bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi void set_type(const std::string &type) { type_ = type; } 56bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi bool has_type() const { return !type_.empty(); } 57bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi const std::string &type() const { return type_; } 58bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 59bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // Accessors for the feature function name. The function name (if available) 60bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // is used for some log messages. Otherwise, a more precise, but also more 61bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // verbose name based on the feature specification is used. 62bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi void set_name(const std::string &name) { name_ = name; } 63bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi bool has_name() const { return !name_.empty(); } 64bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi const std::string &name() { return name_; } 65bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 66bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // Accessors for the default (name-less) parameter. 67bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi void set_argument(int32 argument) { argument_ = argument; } 68bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi bool has_argument() const { 69bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // If argument has not been specified, clients should treat it as 0. This 70bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // makes the test below correct, without having a separate has_argument_ 71bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // bool field. 72bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return argument_ != 0; 73bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 74bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi int32 argument() const { return argument_; } 75bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 76bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // Accessors for the named parameters. 77bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi Parameter *add_parameter() { 78bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi parameters_.emplace_back(); 79bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return &(parameters_.back()); 80bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 81bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi int parameter_size() const { return parameters_.size(); } 82bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi const Parameter ¶meter(int i) const { 83bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi TC_DCHECK((i >= 0) && (i < parameter_size())); 84bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return parameters_[i]; 85bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 86bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 87bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // Accessors for the sub (i.e., nested) features. Nested features: as in 88bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // offset(1).label. 89bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi FeatureFunctionDescriptor *add_feature() { 90bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi sub_features_.emplace_back(new FeatureFunctionDescriptor()); 91bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return sub_features_.back().get(); 92bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 93bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi int feature_size() const { return sub_features_.size(); } 94bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi const FeatureFunctionDescriptor &feature(int i) const { 95bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi TC_DCHECK((i >= 0) && (i < feature_size())); 96bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return *(sub_features_[i].get()); 97bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 98bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi FeatureFunctionDescriptor *mutable_feature(int i) { 99bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi TC_DCHECK((i >= 0) && (i < feature_size())); 100bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return sub_features_[i].get(); 101bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 102bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 103bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi private: 104bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // See comments for set_type(). 105bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi std::string type_; 106bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 107bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // See comments for set_name(). 108bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi std::string name_; 109bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 110bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // See comments for set_argument(). 111bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi int32 argument_ = 0; 112bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 113bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // See comemnts for add_parameter(). 114bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi std::vector<Parameter> parameters_; 115bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 116bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi // See comments for add_feature(). 117bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi std::vector<std::unique_ptr<FeatureFunctionDescriptor>> sub_features_; 118bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 119bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi TC_DISALLOW_COPY_AND_ASSIGN(FeatureFunctionDescriptor); 120bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi}; 121bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 122bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi// List of FeatureFunctionDescriptors. Used to store the result of parsing the 123bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi// spec for several feature functions. 124bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharificlass FeatureExtractorDescriptor { 125bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi public: 126bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi FeatureExtractorDescriptor() {} 127bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 128bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi int feature_size() const { return features_.size(); } 129bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 130bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi FeatureFunctionDescriptor *add_feature() { 131bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi features_.emplace_back(new FeatureFunctionDescriptor()); 132bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return features_.back().get(); 133bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 134bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 135bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi const FeatureFunctionDescriptor &feature(int i) const { 136bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi TC_DCHECK((i >= 0) && (i < feature_size())); 137bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return *(features_[i].get()); 138bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 139bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 140bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi FeatureFunctionDescriptor *mutable_feature(int i) { 141bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi TC_DCHECK((i >= 0) && (i < feature_size())); 142bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi return features_[i].get(); 143bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi } 144bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 145bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi private: 146bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi std::vector<std::unique_ptr<FeatureFunctionDescriptor>> features_; 147bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 148bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi TC_DISALLOW_COPY_AND_ASSIGN(FeatureExtractorDescriptor); 149bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi}; 150bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 151bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi} // namespace nlp_core 152bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi} // namespace libtextclassifier 153bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi 154bda09f1da39ce38a5ece4757b82a64776e53214cMatt Sharifi#endif // LIBTEXTCLASSIFIER_COMMON_FEATURE_DESCRIPTORS_H_ 155