blob: 502bf4790719b45b626d78051638983262799366 [file] [log] [blame]
satok30088252010-12-01 21:22:15 +09001/*
2 * Copyright (C) 2010 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17#ifndef LATINIME_UNIGRAM_DICTIONARY_H
18#define LATINIME_UNIGRAM_DICTIONARY_H
19
Jean Chalard8950ce62012-05-07 16:28:30 +090020#include <map>
Jean Chalard293ece02011-06-16 20:55:16 +090021#include <stdint.h>
satoke808e432010-12-02 14:53:24 +090022#include "defines.h"
23
satok30088252010-12-01 21:22:15 +090024namespace latinime {
25
Ken Wakasaf1008c52012-07-31 17:56:40 +090026class Correction;
27class ProximityInfo;
Jean Chalardcf9dbbd2011-12-26 15:16:59 +090028class TerminalAttributes;
Ken Wakasaf1008c52012-07-31 17:56:40 +090029class WordsPriorityQueuePool;
30
satok30088252010-12-01 21:22:15 +090031class UnigramDictionary {
Jean Chalardd3043382012-03-21 18:38:25 +090032 typedef struct { int first; int second; int replacement; } digraph_t;
Jean Chalard6c300612012-03-06 19:54:03 +090033
Ken Wakasae12e9b52012-01-06 12:24:38 +090034 public:
satok4d355982011-12-15 14:53:19 +090035 // Error tolerances
36 static const int DEFAULT_MAX_ERRORS = 2;
37 static const int MAX_ERRORS_FOR_TWO_WORDS = 1;
38
Satoshi Kataoka6cbe2042012-05-30 17:28:34 +090039 static const int FLAG_MULTIPLE_SUGGEST_ABORT = 0;
40 static const int FLAG_MULTIPLE_SUGGEST_SKIP = 1;
41 static const int FLAG_MULTIPLE_SUGGEST_CONTINUE = 2;
Ken Wakasa5db594a2013-01-12 01:18:00 +090042 UnigramDictionary(const uint8_t *const streamStart, const unsigned int flags);
Ken Wakasa1e614932012-10-29 18:06:22 +090043 int getFrequency(const int *const inWord, const int length) const;
44 int getBigramPosition(int pos, int *word, int offset, int length) const;
Ken Wakasaf2789812012-09-04 12:49:46 +090045 int getSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
Ken Wakasa5db594a2013-01-12 01:18:00 +090046 const int *ycoordinates, const int *inputCodePoints, const int inputSize,
Ken Wakasaf2789812012-09-04 12:49:46 +090047 const std::map<int, int> *bigramMap, const uint8_t *bigramFilter,
Ken Wakasa1e614932012-10-29 18:06:22 +090048 const bool useFullEditDistance, int *outWords, int *frequencies,
Ken Wakasaf2789812012-09-04 12:49:46 +090049 int *outputTypes) const;
satok2df30602011-07-15 13:49:00 +090050 virtual ~UnigramDictionary();
satok30088252010-12-01 21:22:15 +090051
Ken Wakasae12e9b52012-01-06 12:24:38 +090052 private:
satok1bc038c2012-06-14 11:25:50 -070053 DISALLOW_IMPLICIT_CONSTRUCTORS(UnigramDictionary);
satok1d7eaf82011-07-13 10:32:02 +090054 void getWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
Ken Wakasa5db594a2013-01-12 01:18:00 +090055 const int *ycoordinates, const int *inputCodePoints, const int inputSize,
Jean Chalard8950ce62012-05-07 16:28:30 +090056 const std::map<int, int> *bigramMap, const uint8_t *bigramFilter,
57 const bool useFullEditDistance, Correction *correction,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090058 WordsPriorityQueuePool *queuePool) const;
Ken Wakasa5db594a2013-01-12 01:18:00 +090059 int getDigraphReplacement(const int *codes, const int i, const int inputSize,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090060 const digraph_t *const digraphs, const unsigned int digraphsSize) const;
Ken Wakasaa10b1a82013-01-08 17:23:43 +090061 void getWordWithDigraphSuggestionsRec(ProximityInfo *proximityInfo, const int *xcoordinates,
62 const int *ycoordinates, const int *codesBuffer, int *xCoordinatesBuffer,
63 int *yCoordinatesBuffer, const int codesBufferSize, const std::map<int, int> *bigramMap,
64 const uint8_t *bigramFilter, const bool useFullEditDistance, const int *codesSrc,
65 const int codesRemain, const int currentDepth, int *codesDest, Correction *correction,
66 WordsPriorityQueuePool *queuePool, const digraph_t *const digraphs,
67 const unsigned int digraphsSize) const;
satok1d7eaf82011-07-13 10:32:02 +090068 void initSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
Ken Wakasa5db594a2013-01-12 01:18:00 +090069 const int *ycoordinates, const int *codes, const int inputSize,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090070 Correction *correction) const;
satok744dab62011-12-15 22:29:05 +090071 void getOneWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
Jean Chalard8950ce62012-05-07 16:28:30 +090072 const int *ycoordinates, const int *codes, const std::map<int, int> *bigramMap,
Satoshi Kataoka687a2442012-08-23 15:46:43 +090073 const uint8_t *bigramFilter, const bool useFullEditDistance, const int inputSize,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090074 Correction *correction, WordsPriorityQueuePool *queuePool) const;
Jean Chalard4d9b2022012-04-23 19:25:28 +090075 void getSuggestionCandidates(
Satoshi Kataoka687a2442012-08-23 15:46:43 +090076 const bool useFullEditDistance, const int inputSize,
Jean Chalard8950ce62012-05-07 16:28:30 +090077 const std::map<int, int> *bigramMap, const uint8_t *bigramFilter,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090078 Correction *correction, WordsPriorityQueuePool *queuePool, const bool doAutoCompletion,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090079 const int maxErrors, const int currentWordIndex) const;
Ken Wakasaa10b1a82013-01-08 17:23:43 +090080 void getSplitMultipleWordsSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
81 const int *ycoordinates, const int *codes, const bool useFullEditDistance,
82 const int inputSize, Correction *correction, WordsPriorityQueuePool *queuePool,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090083 const bool hasAutoCorrectionCandidate) const;
Ken Wakasaa10b1a82013-01-08 17:23:43 +090084 void onTerminal(const int freq, const TerminalAttributes &terminalAttributes,
satok8330b482012-01-23 16:52:37 +090085 Correction *correction, WordsPriorityQueuePool *queuePool, const bool addToMasterQueue,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090086 const int currentWordIndex) const;
satok662fe692010-12-08 17:05:39 +090087 // Process a node by considering proximity, missing and excessive character
Jean Chalard8950ce62012-05-07 16:28:30 +090088 bool processCurrentNode(const int initialPos, const std::map<int, int> *bigramMap,
89 const uint8_t *bigramFilter, Correction *correction, int *newCount,
90 int *newChildPosition, int *nextSiblingPosition, WordsPriorityQueuePool *queuePool,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090091 const int currentWordIndex) const;
Satoshi Kataoka687a2442012-08-23 15:46:43 +090092 int getMostFrequentWordLike(const int startInputIndex, const int inputSize,
Ken Wakasa1e614932012-10-29 18:06:22 +090093 Correction *correction, int *word) const;
94 int getMostFrequentWordLikeInner(const int *const inWord, const int inputSize,
95 int *outWord) const;
Ken Wakasaa10b1a82013-01-08 17:23:43 +090096 int getSubStringSuggestion(ProximityInfo *proximityInfo, const int *xcoordinates,
97 const int *ycoordinates, const int *codes, const bool useFullEditDistance,
98 Correction *correction, WordsPriorityQueuePool *queuePool, const int inputSize,
satok3c09bb12012-01-26 18:36:19 +090099 const bool hasAutoCorrectionCandidate, const int currentWordIndex,
Ken Wakasaa10b1a82013-01-08 17:23:43 +0900100 const int inputWordStartPos, const int inputWordLength, const int outputWordStartPos,
101 const bool isSpaceProximity, int *freqArray, int *wordLengthArray, int *outputWord,
102 int *outputWordLength) const;
Ken Wakasa1e614932012-10-29 18:06:22 +0900103 void getMultiWordsSuggestionRec(ProximityInfo *proximityInfo, const int *xcoordinates,
104 const int *ycoordinates, const int *codes, const bool useFullEditDistance,
105 const int inputSize, Correction *correction, WordsPriorityQueuePool *queuePool,
satok1f6b52e2012-01-30 13:53:58 +0900106 const bool hasAutoCorrectionCandidate, const int startPos, const int startWordIndex,
Ken Wakasa0bbb9172012-07-25 17:51:43 +0900107 const int outputWordLength, int *freqArray, int *wordLengthArray,
Ken Wakasa1e614932012-10-29 18:06:22 +0900108 int *outputWord) const;
Jean Chalard293ece02011-06-16 20:55:16 +0900109
Ken Wakasa0bbb9172012-07-25 17:51:43 +0900110 const uint8_t *const DICT_ROOT;
Ken Wakasae90b3332011-01-07 15:01:51 +0900111 const int ROOT_POS;
Jean Chalard6c300612012-03-06 19:54:03 +0900112 const int MAX_DIGRAPH_SEARCH_DEPTH;
Jean Chalardcd274b12012-04-06 18:26:00 +0900113 const int FLAGS;
Jean Chalardc2bbc6a2011-02-25 17:56:53 +0900114
Jean Chalard6c300612012-03-06 19:54:03 +0900115 static const digraph_t GERMAN_UMLAUT_DIGRAPHS[];
Jean Chalardcc78d032012-03-23 16:48:49 +0900116 static const digraph_t FRENCH_LIGATURES_DIGRAPHS[];
satok30088252010-12-01 21:22:15 +0900117};
Ken Wakasace9e52a2011-06-18 13:09:55 +0900118} // namespace latinime
satok30088252010-12-01 21:22:15 +0900119#endif // LATINIME_UNIGRAM_DICTIONARY_H