blob: 3d3007ce09a76d96532ed2ca014ce17e44f72a63 [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
satoke808e432010-12-02 14:53:24 +090020#include "defines.h"
satok8fbd5522011-02-22 17:28:55 +090021#include "proximity_info.h"
satoke808e432010-12-02 14:53:24 +090022
satok30088252010-12-01 21:22:15 +090023namespace latinime {
24
satok30088252010-12-01 21:22:15 +090025class UnigramDictionary {
Jean Chalard8dc754a2011-01-27 14:20:22 +090026
27 typedef enum { // Used as a return value for character comparison
28 SAME_OR_ACCENTED_OR_CAPITALIZED_CHAR, // Same char, possibly with different case or accent
29 NEAR_PROXIMITY_CHAR, // It is a char located nearby on the keyboard
30 UNRELATED_CHAR // It is an unrelated char
31 } ProximityType;
32
satok30088252010-12-01 21:22:15 +090033public:
satoke808e432010-12-02 14:53:24 +090034 UnigramDictionary(const unsigned char *dict, int typedLetterMultipler, int fullWordMultiplier,
satok662fe692010-12-08 17:05:39 +090035 int maxWordLength, int maxWords, int maxProximityChars, const bool isLatestDictVersion);
Jean Chalardc2bbc6a2011-02-25 17:56:53 +090036 int getSuggestions(const ProximityInfo *proximityInfo, const int *xcoordinates,
37 const int *ycoordinates, const int *codes, const int codesSize, const int flags,
38 unsigned short *outWords, int *frequencies);
satok30088252010-12-01 21:22:15 +090039 ~UnigramDictionary();
40
41private:
Jean Chalardc2bbc6a2011-02-25 17:56:53 +090042 void getWordSuggestions(const ProximityInfo *proximityInfo, const int *xcoordinates,
43 const int *ycoordinates, const int *codes, const int codesSize,
44 unsigned short *outWords, int *frequencies);
45 bool isDigraph(const int* codes, const int i, const int codesSize) const;
46 void getWordWithDigraphSuggestionsRec(const ProximityInfo *proximityInfo,
47 const int *xcoordinates, const int* ycoordinates, const int *codesBuffer,
48 const int codesBufferSize, const int flags, const int* codesSrc, const int codesRemain,
satok3c4bb772011-03-04 22:50:19 -080049 const int currentDepth, int* codesDest, unsigned short* outWords, int* frequencies);
Jean Chalardc2bbc6a2011-02-25 17:56:53 +090050 void initSuggestions(const int *codes, const int codesSize, unsigned short *outWords,
51 int *frequencies);
satok54fe9e02010-12-13 14:42:35 +090052 void getSuggestionCandidates(const int skipPos, const int excessivePos,
satoka3d78f62010-12-09 22:08:33 +090053 const int transposedPos, int *nextLetters, const int nextLettersSize,
54 const int maxDepth);
satok30088252010-12-01 21:22:15 +090055 void getVersionNumber();
56 bool checkIfDictVersionIsLatest();
57 int getAddress(int *pos);
satok30088252010-12-01 21:22:15 +090058 int getFreq(int *pos);
satok30088252010-12-01 21:22:15 +090059 int wideStrLen(unsigned short *str);
satok30088252010-12-01 21:22:15 +090060 bool sameAsTyped(unsigned short *word, int length);
satok30088252010-12-01 21:22:15 +090061 bool addWord(unsigned short *word, int length, int frequency);
Jean Chalardf5f834a2011-02-22 15:12:46 +090062 unsigned short toBaseLowerCase(unsigned short c);
satok68319262010-12-03 19:38:08 +090063 void getWordsRec(const int childrenCount, const int pos, const int depth, const int maxDepth,
64 const bool traverseAllNodes, const int snr, const int inputIndex, const int diffs,
satoka3d78f62010-12-09 22:08:33 +090065 const int skipPos, const int excessivePos, const int transposedPos, int *nextLetters,
66 const int nextLettersSize);
satok817e5172011-03-04 06:06:45 -080067 bool getSplitTwoWordsSuggestion(const int inputLength,
68 const int firstWordStartPos, const int firstWordLength,
69 const int secondWordStartPos, const int secondWordLength);
satok662fe692010-12-08 17:05:39 +090070 bool getMissingSpaceWords(const int inputLength, const int missingSpacePos);
satok817e5172011-03-04 06:06:45 -080071 bool getMistypedSpaceWords(const int inputLength, const int spaceProximityPos);
satokd2997922010-12-07 13:08:39 +090072 // Keep getWordsOld for comparing performance between getWords and getWordsOld
73 void getWordsOld(const int initialPos, const int inputLength, const int skipPos,
satoka3d78f62010-12-09 22:08:33 +090074 const int excessivePos, const int transposedPos, int *nextLetters,
75 const int nextLettersSize);
satok30088252010-12-01 21:22:15 +090076 void registerNextLetter(unsigned short c, int *nextLetters, int nextLettersSize);
satok58c49b92011-01-27 03:23:39 +090077 int calculateFinalFreq(const int inputIndex, const int depth, const int snr, const int skipPos,
Jean Chalard07a84062011-03-03 10:22:10 +090078 const int excessivePos, const int transposedPos, const int freq,
79 const bool sameLength) const;
satok715514d2010-12-02 20:19:59 +090080 void onTerminalWhenUserTypedLengthIsGreaterThanInputLength(unsigned short *word,
satok54fe9e02010-12-13 14:42:35 +090081 const int inputIndex, const int depth, const int snr, int *nextLetters,
satoka3d78f62010-12-09 22:08:33 +090082 const int nextLettersSize, const int skipPos, const int excessivePos,
83 const int transposedPos, const int freq);
satok54fe9e02010-12-13 14:42:35 +090084 void onTerminalWhenUserTypedLengthIsSameAsInputLength(unsigned short *word,
85 const int inputIndex, const int depth, const int snr, const int skipPos,
Jean Chalard8dc754a2011-01-27 14:20:22 +090086 const int excessivePos, const int transposedPos, const int freq);
satok28bd03b2010-12-03 16:39:16 +090087 bool needsToSkipCurrentNode(const unsigned short c,
satok68319262010-12-03 19:38:08 +090088 const int inputIndex, const int skipPos, const int depth);
Jean Chalard8dc754a2011-01-27 14:20:22 +090089 ProximityType getMatchedProximityId(const int *currentChars, const unsigned short c,
90 const int skipPos, const int excessivePos, const int transposedPos);
satok662fe692010-12-08 17:05:39 +090091 // Process a node by considering proximity, missing and excessive character
satok48e432c2010-12-06 17:38:58 +090092 bool processCurrentNode(const int pos, const int depth,
satokcdbbea72010-12-08 16:04:16 +090093 const int maxDepth, const bool traverseAllNodes, const int snr, int inputIndex,
satoka3d78f62010-12-09 22:08:33 +090094 const int diffs, const int skipPos, const int excessivePos, const int transposedPos,
95 int *nextLetters, const int nextLettersSize, int *newCount, int *newChildPosition,
satokcdbbea72010-12-08 16:04:16 +090096 bool *newTraverseAllNodes, int *newSnr, int*newInputIndex, int *newDiffs,
97 int *nextSiblingPosition);
satokaee09dc2010-12-09 19:21:51 +090098 int getBestWordFreq(const int startInputIndex, const int inputLength, unsigned short *word);
satok662fe692010-12-08 17:05:39 +090099 // Process a node by considering missing space
satokaee09dc2010-12-09 19:21:51 +0900100 bool processCurrentNodeForExactMatch(const int firstChildPos,
101 const int startInputIndex, const int depth, unsigned short *word,
102 int *newChildPosition, int *newCount, bool *newTerminal, int *newFreq, int *siblingPos);
Jean Chalard07a84062011-03-03 10:22:10 +0900103 bool existsAdjacentProximityChars(const int inputIndex, const int inputLength) const;
104 inline const int* getInputCharsAt(const int index) const {
satok8fbd5522011-02-22 17:28:55 +0900105 return mInputCodes + (index * MAX_PROXIMITY_CHARS);
106 }
satoke808e432010-12-02 14:53:24 +0900107 const unsigned char *DICT;
satok30088252010-12-01 21:22:15 +0900108 const int MAX_WORD_LENGTH;
Ken Wakasae90b3332011-01-07 15:01:51 +0900109 const int MAX_WORDS;
satok662fe692010-12-08 17:05:39 +0900110 const int MAX_PROXIMITY_CHARS;
satoke808e432010-12-02 14:53:24 +0900111 const bool IS_LATEST_DICT_VERSION;
satok18c28f42010-12-02 18:11:54 +0900112 const int TYPED_LETTER_MULTIPLIER;
113 const int FULL_WORD_MULTIPLIER;
Ken Wakasae90b3332011-01-07 15:01:51 +0900114 const int ROOT_POS;
Jean Chalardc2bbc6a2011-02-25 17:56:53 +0900115 const unsigned int BYTES_IN_ONE_CHAR;
satok3c4bb772011-03-04 22:50:19 -0800116 const int MAX_UMLAUT_SEARCH_DEPTH;
Jean Chalardc2bbc6a2011-02-25 17:56:53 +0900117
118 // Flags for special processing
119 // Those *must* match the flags in BinaryDictionary.Flags.ALL_FLAGS in BinaryDictionary.java
120 // or something very bad (like, the apocalypse) will happen.
121 // Please update both at the same time.
122 enum {
123 REQUIRES_GERMAN_UMLAUT_PROCESSING = 0x1
124 };
125 static const struct digraph_t { int first; int second; } GERMAN_UMLAUT_DIGRAPHS[];
satok30088252010-12-01 21:22:15 +0900126
satok30088252010-12-01 21:22:15 +0900127 int *mFrequencies;
satok30088252010-12-01 21:22:15 +0900128 unsigned short *mOutputChars;
Jean Chalardc2bbc6a2011-02-25 17:56:53 +0900129 const int *mInputCodes;
satok30088252010-12-01 21:22:15 +0900130 int mInputLength;
satok715514d2010-12-02 20:19:59 +0900131 // MAX_WORD_LENGTH_INTERNAL must be bigger than MAX_WORD_LENGTH
132 unsigned short mWord[MAX_WORD_LENGTH_INTERNAL];
satok30088252010-12-01 21:22:15 +0900133 int mMaxEditDistance;
satokd2997922010-12-07 13:08:39 +0900134
135 int mStackChildCount[MAX_WORD_LENGTH_INTERNAL];
136 bool mStackTraverseAll[MAX_WORD_LENGTH_INTERNAL];
137 int mStackNodeFreq[MAX_WORD_LENGTH_INTERNAL];
138 int mStackInputIndex[MAX_WORD_LENGTH_INTERNAL];
139 int mStackDiffs[MAX_WORD_LENGTH_INTERNAL];
140 int mStackSiblingPos[MAX_WORD_LENGTH_INTERNAL];
Tadashi G. Takaoka887f11e2011-02-10 20:53:58 +0900141 int mNextLettersFrequency[NEXT_LETTERS_SIZE];
satok30088252010-12-01 21:22:15 +0900142};
143
144// ----------------------------------------------------------------------------
145
146}; // namespace latinime
147
148#endif // LATINIME_UNIGRAM_DICTIONARY_H