xref: /aosp_15_r20/external/libtextclassifier/native/annotator/datetime/regex-parser.h (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 #ifndef LIBTEXTCLASSIFIER_ANNOTATOR_DATETIME_REGEX_PARSER_H_
18*993b0882SAndroid Build Coastguard Worker #define LIBTEXTCLASSIFIER_ANNOTATOR_DATETIME_REGEX_PARSER_H_
19*993b0882SAndroid Build Coastguard Worker 
20*993b0882SAndroid Build Coastguard Worker #include <memory>
21*993b0882SAndroid Build Coastguard Worker #include <string>
22*993b0882SAndroid Build Coastguard Worker #include <unordered_map>
23*993b0882SAndroid Build Coastguard Worker #include <unordered_set>
24*993b0882SAndroid Build Coastguard Worker #include <vector>
25*993b0882SAndroid Build Coastguard Worker 
26*993b0882SAndroid Build Coastguard Worker #include "annotator/datetime/extractor.h"
27*993b0882SAndroid Build Coastguard Worker #include "annotator/datetime/parser.h"
28*993b0882SAndroid Build Coastguard Worker #include "annotator/model_generated.h"
29*993b0882SAndroid Build Coastguard Worker #include "annotator/types.h"
30*993b0882SAndroid Build Coastguard Worker #include "utils/base/integral_types.h"
31*993b0882SAndroid Build Coastguard Worker #include "utils/base/statusor.h"
32*993b0882SAndroid Build Coastguard Worker #include "utils/calendar/calendar.h"
33*993b0882SAndroid Build Coastguard Worker #include "utils/strings/stringpiece.h"
34*993b0882SAndroid Build Coastguard Worker #include "utils/utf8/unicodetext.h"
35*993b0882SAndroid Build Coastguard Worker #include "utils/utf8/unilib.h"
36*993b0882SAndroid Build Coastguard Worker #include "utils/zlib/zlib.h"
37*993b0882SAndroid Build Coastguard Worker 
38*993b0882SAndroid Build Coastguard Worker namespace libtextclassifier3 {
39*993b0882SAndroid Build Coastguard Worker 
40*993b0882SAndroid Build Coastguard Worker // Parses datetime expressions in the input and resolves them to actual absolute
41*993b0882SAndroid Build Coastguard Worker // time.
42*993b0882SAndroid Build Coastguard Worker class RegexDatetimeParser : public DatetimeParser {
43*993b0882SAndroid Build Coastguard Worker  public:
44*993b0882SAndroid Build Coastguard Worker   static std::unique_ptr<DatetimeParser> Instance(
45*993b0882SAndroid Build Coastguard Worker       const DatetimeModel* model, const UniLib* unilib,
46*993b0882SAndroid Build Coastguard Worker       const CalendarLib* calendarlib, ZlibDecompressor* decompressor);
47*993b0882SAndroid Build Coastguard Worker 
48*993b0882SAndroid Build Coastguard Worker   // Parses the dates in 'input' and fills result. Makes sure that the results
49*993b0882SAndroid Build Coastguard Worker   // do not overlap.
50*993b0882SAndroid Build Coastguard Worker   // If 'anchor_start_end' is true the extracted results need to start at the
51*993b0882SAndroid Build Coastguard Worker   // beginning of 'input' and end at the end of it.
52*993b0882SAndroid Build Coastguard Worker   StatusOr<std::vector<DatetimeParseResultSpan>> Parse(
53*993b0882SAndroid Build Coastguard Worker       const std::string& input, int64 reference_time_ms_utc,
54*993b0882SAndroid Build Coastguard Worker       const std::string& reference_timezone, const LocaleList& locale_list,
55*993b0882SAndroid Build Coastguard Worker       ModeFlag mode, AnnotationUsecase annotation_usecase,
56*993b0882SAndroid Build Coastguard Worker       bool anchor_start_end) const override;
57*993b0882SAndroid Build Coastguard Worker 
58*993b0882SAndroid Build Coastguard Worker   // Same as above but takes UnicodeText.
59*993b0882SAndroid Build Coastguard Worker   StatusOr<std::vector<DatetimeParseResultSpan>> Parse(
60*993b0882SAndroid Build Coastguard Worker       const UnicodeText& input, int64 reference_time_ms_utc,
61*993b0882SAndroid Build Coastguard Worker       const std::string& reference_timezone, const LocaleList& locale_list,
62*993b0882SAndroid Build Coastguard Worker       ModeFlag mode, AnnotationUsecase annotation_usecase,
63*993b0882SAndroid Build Coastguard Worker       bool anchor_start_end) const override;
64*993b0882SAndroid Build Coastguard Worker 
65*993b0882SAndroid Build Coastguard Worker  protected:
66*993b0882SAndroid Build Coastguard Worker   explicit RegexDatetimeParser(const DatetimeModel* model, const UniLib* unilib,
67*993b0882SAndroid Build Coastguard Worker                                const CalendarLib* calendarlib,
68*993b0882SAndroid Build Coastguard Worker                                ZlibDecompressor* decompressor);
69*993b0882SAndroid Build Coastguard Worker 
70*993b0882SAndroid Build Coastguard Worker   // Returns a list of locale ids for given locale spec string (collection of
71*993b0882SAndroid Build Coastguard Worker   // locale names).
72*993b0882SAndroid Build Coastguard Worker   std::vector<int> ParseAndExpandLocales(
73*993b0882SAndroid Build Coastguard Worker       const std::vector<StringPiece>& locales) const;
74*993b0882SAndroid Build Coastguard Worker 
75*993b0882SAndroid Build Coastguard Worker   // Helper function that finds datetime spans, only using the rules associated
76*993b0882SAndroid Build Coastguard Worker   // with the given locales.
77*993b0882SAndroid Build Coastguard Worker   StatusOr<std::vector<DatetimeParseResultSpan>> FindSpansUsingLocales(
78*993b0882SAndroid Build Coastguard Worker       const std::vector<int>& locale_ids, const UnicodeText& input,
79*993b0882SAndroid Build Coastguard Worker       const int64 reference_time_ms_utc, const std::string& reference_timezone,
80*993b0882SAndroid Build Coastguard Worker       ModeFlag mode, AnnotationUsecase annotation_usecase,
81*993b0882SAndroid Build Coastguard Worker       bool anchor_start_end, const std::string& reference_locale,
82*993b0882SAndroid Build Coastguard Worker       std::unordered_set<int>* executed_rules) const;
83*993b0882SAndroid Build Coastguard Worker 
84*993b0882SAndroid Build Coastguard Worker   StatusOr<std::vector<DatetimeParseResultSpan>> ParseWithRule(
85*993b0882SAndroid Build Coastguard Worker       const CompiledRule& rule, const UnicodeText& input,
86*993b0882SAndroid Build Coastguard Worker       int64 reference_time_ms_utc, const std::string& reference_timezone,
87*993b0882SAndroid Build Coastguard Worker       const std::string& reference_locale, const int locale_id,
88*993b0882SAndroid Build Coastguard Worker       bool anchor_start_end) const;
89*993b0882SAndroid Build Coastguard Worker 
90*993b0882SAndroid Build Coastguard Worker   // Converts the current match in 'matcher' into DatetimeParseResult.
91*993b0882SAndroid Build Coastguard Worker   bool ExtractDatetime(const CompiledRule& rule,
92*993b0882SAndroid Build Coastguard Worker                        const UniLib::RegexMatcher& matcher,
93*993b0882SAndroid Build Coastguard Worker                        int64 reference_time_ms_utc,
94*993b0882SAndroid Build Coastguard Worker                        const std::string& reference_timezone,
95*993b0882SAndroid Build Coastguard Worker                        const std::string& reference_locale, int locale_id,
96*993b0882SAndroid Build Coastguard Worker                        std::vector<DatetimeParseResult>* results,
97*993b0882SAndroid Build Coastguard Worker                        CodepointSpan* result_span) const;
98*993b0882SAndroid Build Coastguard Worker 
99*993b0882SAndroid Build Coastguard Worker   // Parse and extract information from current match in 'matcher'.
100*993b0882SAndroid Build Coastguard Worker   StatusOr<std::vector<DatetimeParseResultSpan>> HandleParseMatch(
101*993b0882SAndroid Build Coastguard Worker       const CompiledRule& rule, const UniLib::RegexMatcher& matcher,
102*993b0882SAndroid Build Coastguard Worker       int64 reference_time_ms_utc, const std::string& reference_timezone,
103*993b0882SAndroid Build Coastguard Worker       const std::string& reference_locale, int locale_id) const;
104*993b0882SAndroid Build Coastguard Worker 
105*993b0882SAndroid Build Coastguard Worker  private:
106*993b0882SAndroid Build Coastguard Worker   bool initialized_;
107*993b0882SAndroid Build Coastguard Worker   const UniLib& unilib_;
108*993b0882SAndroid Build Coastguard Worker   const CalendarLib& calendarlib_;
109*993b0882SAndroid Build Coastguard Worker   std::vector<CompiledRule> rules_;
110*993b0882SAndroid Build Coastguard Worker   std::unordered_map<int, std::vector<int>> locale_to_rules_;
111*993b0882SAndroid Build Coastguard Worker   std::vector<std::unique_ptr<const UniLib::RegexPattern>> extractor_rules_;
112*993b0882SAndroid Build Coastguard Worker   std::unordered_map<DatetimeExtractorType, std::unordered_map<int, int>>
113*993b0882SAndroid Build Coastguard Worker       type_and_locale_to_extractor_rule_;
114*993b0882SAndroid Build Coastguard Worker   std::unordered_map<std::string, int> locale_string_to_id_;
115*993b0882SAndroid Build Coastguard Worker   std::vector<int> default_locale_ids_;
116*993b0882SAndroid Build Coastguard Worker   bool use_extractors_for_locating_;
117*993b0882SAndroid Build Coastguard Worker   bool generate_alternative_interpretations_when_ambiguous_;
118*993b0882SAndroid Build Coastguard Worker   bool prefer_future_for_unspecified_date_;
119*993b0882SAndroid Build Coastguard Worker };
120*993b0882SAndroid Build Coastguard Worker 
121*993b0882SAndroid Build Coastguard Worker }  // namespace libtextclassifier3
122*993b0882SAndroid Build Coastguard Worker 
123*993b0882SAndroid Build Coastguard Worker #endif  // LIBTEXTCLASSIFIER_ANNOTATOR_DATETIME_REGEX_PARSER_H_
124