blob: 2c66222106506faf2dac2af7a6328c42a8ac949f [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 Wakasa0bbb9172012-07-25 17:51:43 +090042 UnigramDictionary(const uint8_t *const streamStart, int typedLetterMultipler,
Jean Chalardcd274b12012-04-06 18:26:00 +090043 int fullWordMultiplier, int maxWordLength, int maxWords, const unsigned int flags);
Ken Wakasa0bbb9172012-07-25 17:51:43 +090044 int getFrequency(const int32_t *const inWord, const int length) const;
Jean Chalard581335c2011-06-17 12:45:17 +090045 int getBigramPosition(int pos, unsigned short *word, int offset, int length) const;
satokb1ed1d42012-06-14 16:35:23 -070046 int getSuggestions(
47 ProximityInfo *proximityInfo, const int *xcoordinates, const int *ycoordinates,
Jean Chalard8950ce62012-05-07 16:28:30 +090048 const int *codes, const int codesSize, const std::map<int, int> *bigramMap,
49 const uint8_t *bigramFilter, const bool useFullEditDistance, unsigned short *outWords,
Jean Chalard6931df92012-07-12 12:55:48 +090050 int *frequencies, int *outputTypes) const;
satok2df30602011-07-15 13:49:00 +090051 virtual ~UnigramDictionary();
satok30088252010-12-01 21:22:15 +090052
Ken Wakasae12e9b52012-01-06 12:24:38 +090053 private:
satok1bc038c2012-06-14 11:25:50 -070054 DISALLOW_IMPLICIT_CONSTRUCTORS(UnigramDictionary);
satok1d7eaf82011-07-13 10:32:02 +090055 void getWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
Satoshi Kataoka687a2442012-08-23 15:46:43 +090056 const int *ycoordinates, const int *codes, const int inputSize,
Jean Chalard8950ce62012-05-07 16:28:30 +090057 const std::map<int, int> *bigramMap, const uint8_t *bigramFilter,
58 const bool useFullEditDistance, Correction *correction,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090059 WordsPriorityQueuePool *queuePool) const;
Jean Chalardd3043382012-03-21 18:38:25 +090060 int getDigraphReplacement(const int *codes, const int i, const int codesSize,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090061 const digraph_t *const digraphs, const unsigned int digraphsSize) const;
satok1d7eaf82011-07-13 10:32:02 +090062 void getWordWithDigraphSuggestionsRec(ProximityInfo *proximityInfo,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090063 const int *xcoordinates, const int *ycoordinates, const int *codesBuffer,
Jean Chalard4d9b2022012-04-23 19:25:28 +090064 int *xCoordinatesBuffer, int *yCoordinatesBuffer, const int codesBufferSize,
Jean Chalard8950ce62012-05-07 16:28:30 +090065 const std::map<int, int> *bigramMap, const uint8_t *bigramFilter,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090066 const bool useFullEditDistance, const int *codesSrc, const int codesRemain,
67 const int currentDepth, int *codesDest, Correction *correction,
68 WordsPriorityQueuePool *queuePool, const digraph_t *const digraphs,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090069 const unsigned int digraphsSize) const;
satok1d7eaf82011-07-13 10:32:02 +090070 void initSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090071 const int *ycoordinates, const int *codes, const int codesSize,
72 Correction *correction) const;
satok744dab62011-12-15 22:29:05 +090073 void getOneWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
Jean Chalard8950ce62012-05-07 16:28:30 +090074 const int *ycoordinates, const int *codes, const std::map<int, int> *bigramMap,
Satoshi Kataoka687a2442012-08-23 15:46:43 +090075 const uint8_t *bigramFilter, const bool useFullEditDistance, const int inputSize,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090076 Correction *correction, WordsPriorityQueuePool *queuePool) const;
Jean Chalard4d9b2022012-04-23 19:25:28 +090077 void getSuggestionCandidates(
Satoshi Kataoka687a2442012-08-23 15:46:43 +090078 const bool useFullEditDistance, const int inputSize,
Jean Chalard8950ce62012-05-07 16:28:30 +090079 const std::map<int, int> *bigramMap, const uint8_t *bigramFilter,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090080 Correction *correction, WordsPriorityQueuePool *queuePool, const bool doAutoCompletion,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090081 const int maxErrors, const int currentWordIndex) const;
satoka85f4922012-01-30 18:18:30 +090082 void getSplitMultipleWordsSuggestions(ProximityInfo *proximityInfo,
satok744dab62011-12-15 22:29:05 +090083 const int *xcoordinates, const int *ycoordinates, const int *codes,
Satoshi Kataoka687a2442012-08-23 15:46:43 +090084 const bool useFullEditDistance, const int inputSize,
Ken Wakasa0bbb9172012-07-25 17:51:43 +090085 Correction *correction, WordsPriorityQueuePool *queuePool,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090086 const bool hasAutoCorrectionCandidate) const;
Jean Chalardcf9dbbd2011-12-26 15:16:59 +090087 void onTerminal(const int freq, const TerminalAttributes& terminalAttributes,
satok8330b482012-01-23 16:52:37 +090088 Correction *correction, WordsPriorityQueuePool *queuePool, const bool addToMasterQueue,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090089 const int currentWordIndex) const;
satok662fe692010-12-08 17:05:39 +090090 // Process a node by considering proximity, missing and excessive character
Jean Chalard8950ce62012-05-07 16:28:30 +090091 bool processCurrentNode(const int initialPos, const std::map<int, int> *bigramMap,
92 const uint8_t *bigramFilter, Correction *correction, int *newCount,
93 int *newChildPosition, int *nextSiblingPosition, WordsPriorityQueuePool *queuePool,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090094 const int currentWordIndex) const;
Satoshi Kataoka687a2442012-08-23 15:46:43 +090095 int getMostFrequentWordLike(const int startInputIndex, const int inputSize,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090096 Correction *correction, unsigned short *word) const;
Ken Wakasa0bbb9172012-07-25 17:51:43 +090097 int getMostFrequentWordLikeInner(const uint16_t *const inWord, const int length,
Satoshi Kataoka6bc051d2012-06-08 19:52:19 +090098 short unsigned int *outWord) const;
Satoshi Kataoka6cbe2042012-05-30 17:28:34 +090099 int getSubStringSuggestion(
satok7409d152012-01-26 16:13:25 +0900100 ProximityInfo *proximityInfo, const int *xcoordinates, const int *ycoordinates,
satok3c09bb12012-01-26 18:36:19 +0900101 const int *codes, const bool useFullEditDistance, Correction *correction,
Satoshi Kataoka687a2442012-08-23 15:46:43 +0900102 WordsPriorityQueuePool *queuePool, const int inputSize,
satok3c09bb12012-01-26 18:36:19 +0900103 const bool hasAutoCorrectionCandidate, const int currentWordIndex,
104 const int inputWordStartPos, const int inputWordLength,
satok99557162012-01-26 22:49:13 +0900105 const int outputWordStartPos, const bool isSpaceProximity, int *freqArray,
Ken Wakasa0bbb9172012-07-25 17:51:43 +0900106 int *wordLengthArray, unsigned short *outputWord, int *outputWordLength) const;
satok1f6b52e2012-01-30 13:53:58 +0900107 void getMultiWordsSuggestionRec(ProximityInfo *proximityInfo,
108 const int *xcoordinates, const int *ycoordinates, const int *codes,
Satoshi Kataoka687a2442012-08-23 15:46:43 +0900109 const bool useFullEditDistance, const int inputSize,
Ken Wakasa0bbb9172012-07-25 17:51:43 +0900110 Correction *correction, WordsPriorityQueuePool *queuePool,
satok1f6b52e2012-01-30 13:53:58 +0900111 const bool hasAutoCorrectionCandidate, const int startPos, const int startWordIndex,
Ken Wakasa0bbb9172012-07-25 17:51:43 +0900112 const int outputWordLength, int *freqArray, int *wordLengthArray,
113 unsigned short *outputWord) const;
Jean Chalard293ece02011-06-16 20:55:16 +0900114
Ken Wakasa0bbb9172012-07-25 17:51:43 +0900115 const uint8_t *const DICT_ROOT;
satok30088252010-12-01 21:22:15 +0900116 const int MAX_WORD_LENGTH;
Ken Wakasae90b3332011-01-07 15:01:51 +0900117 const int MAX_WORDS;
satok18c28f42010-12-02 18:11:54 +0900118 const int TYPED_LETTER_MULTIPLIER;
119 const int FULL_WORD_MULTIPLIER;
Ken Wakasae90b3332011-01-07 15:01:51 +0900120 const int ROOT_POS;
Jean Chalardc2bbc6a2011-02-25 17:56:53 +0900121 const unsigned int BYTES_IN_ONE_CHAR;
Jean Chalard6c300612012-03-06 19:54:03 +0900122 const int MAX_DIGRAPH_SEARCH_DEPTH;
Jean Chalardcd274b12012-04-06 18:26:00 +0900123 const int FLAGS;
Jean Chalardc2bbc6a2011-02-25 17:56:53 +0900124
Jean Chalard6c300612012-03-06 19:54:03 +0900125 static const digraph_t GERMAN_UMLAUT_DIGRAPHS[];
Jean Chalardcc78d032012-03-23 16:48:49 +0900126 static const digraph_t FRENCH_LIGATURES_DIGRAPHS[];
satok30088252010-12-01 21:22:15 +0900127};
Ken Wakasace9e52a2011-06-18 13:09:55 +0900128} // namespace latinime
satok30088252010-12-01 21:22:15 +0900129#endif // LATINIME_UNIGRAM_DICTIONARY_H