nyancrimew
/
LatinIME
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Enhance the safety net in multiple word suggestions 

Bug: 6576793

Change-Id: I97100b482d3cf4e1b417c197b5d1e971a72b998d
main
Satoshi Kataoka 2012-05-31 15:04:58 +09:00
parent a2cfb2e3b0
commit 67e3cc8488
2 changed files with 45 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
native/jni/src/defines.h
View File

@ -234,11 +234,12 @@ static inline void prof_out(void) {
#define SUB_QUEUE_MAX_WORDS 1
#define SUB_QUEUE_MAX_COUNT 10
#define SUB_QUEUE_MIN_WORD_LENGTH 4
#define MULTIPLE_WORDS_SUGGESTION_MAX_WORDS 10
// TODO: Extend this limitation
#define MULTIPLE_WORDS_SUGGESTION_MAX_WORDS 5
// TODO: Remove this limitation
#define MULTIPLE_WORDS_SUGGESTION_MAX_WORD_LENGTH 12
// TODO: Remove this limitation
#define MULTIPLE_WORDS_SUGGESTION_MAX_TOTAL_TRAVERSE_COUNT 110
#define MULTIPLE_WORDS_SUGGESTION_MAX_TOTAL_TRAVERSE_COUNT 45
#define MULTIPLE_WORDS_DEMOTION_RATE 80
#define MIN_INPUT_LENGTH_FOR_THREE_OR_MORE_WORDS_CORRECTION 6

45
native/jni/src/unigram_dictionary.cpp
View File

@ -430,6 +430,48 @@ int UnigramDictionary::getSubStringSuggestion(
const int inputWordStartPos, const int inputWordLength,
const int outputWordStartPos, const bool isSpaceProximity, int *freqArray,
int*wordLengthArray, unsigned short* outputWord, int *outputWordLength) {
if (inputWordLength > MULTIPLE_WORDS_SUGGESTION_MAX_WORD_LENGTH) {
return FLAG_MULTIPLE_SUGGEST_ABORT;
}
/////////////////////////////////////////////
// safety net for multiple word suggestion //
// TODO: Remove this safety net //
/////////////////////////////////////////////
int smallWordCount = 0;
int singleLetterWordCount = 0;
if (inputWordLength == 1) {
++singleLetterWordCount;
}
if (inputWordLength <= 2) {
// small word == single letter or 2-letter word
++smallWordCount;
}
for (int i = 0; i < currentWordIndex; ++i) {
const int length = wordLengthArray[i];
if (length == 1) {
++singleLetterWordCount;
// Safety net to avoid suggesting sequential single letter words
if (i < (currentWordIndex - 1)) {
if (wordLengthArray[i + 1] == 1) {
return FLAG_MULTIPLE_SUGGEST_ABORT;
}
} else if (inputWordLength == 1) {
return FLAG_MULTIPLE_SUGGEST_ABORT;
}
}
if (length <= 2) {
++smallWordCount;
}
// Safety net to avoid suggesting multiple words with many (4 or more, for now) small words
if (singleLetterWordCount >= 3 || smallWordCount >= 4) {
return FLAG_MULTIPLE_SUGGEST_ABORT;
}
}
//////////////////////////////////////////////
// TODO: Remove the safety net above //
//////////////////////////////////////////////
unsigned short* tempOutputWord = 0;
int nextWordLength = 0;
// TODO: Optimize init suggestion
@ -555,9 +597,6 @@ void UnigramDictionary::getMultiWordsSuggestionRec(ProximityInfo *proximityInfo,
// Current word
int inputWordStartPos = startInputPos;
int inputWordLength = i - startInputPos;
if (inputWordLength > MULTIPLE_WORDS_SUGGESTION_MAX_WORD_LENGTH) {
break;
}
const int suggestionFlag = getSubStringSuggestion(proximityInfo, xcoordinates, ycoordinates,
codes, useFullEditDistance, correction, queuePool, inputLength,
hasAutoCorrectionCandidate, startWordIndex, inputWordStartPos, inputWordLength,