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

Merge "Fix a bug with umlaut processing." into honeycomb-mr1

main
Jean Chalard 2011-03-03 20:08:42 -08:00 committed by Android (Google) Code Review
parent d628e72958 a787dba83b
commit fe18612ced
3 changed files with 37 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
native/src/defines.h
View File

@ -148,6 +148,10 @@ static void prof_out(void) {
#define MAX_DEPTH_MULTIPLIER 3
// TODO: Reduce this constant if possible; check the maximum number of umlauts in the same German
// word in the dictionary
#define DEFAULT_MAX_UMLAUT_SEARCH_DEPTH 5
// Minimum suggest depth for one word for all cases except for missing space suggestions.
#define MIN_SUGGEST_DEPTH 1
#define MIN_USER_TYPED_LENGTH_FOR_MISSING_SPACE_SUGGESTION 3

30
native/src/unigram_dictionary.cpp
View File

@ -41,7 +41,8 @@ UnigramDictionary::UnigramDictionary(const unsigned char *dict, int typedLetterM
MAX_PROXIMITY_CHARS(maxProximityChars), IS_LATEST_DICT_VERSION(isLatestDictVersion),
TYPED_LETTER_MULTIPLIER(typedLetterMultiplier), FULL_WORD_MULTIPLIER(fullWordMultiplier),
ROOT_POS(isLatestDictVersion ? DICTIONARY_HEADER_SIZE : 0),
BYTES_IN_ONE_CHAR(MAX_PROXIMITY_CHARS * sizeof(*mInputCodes)) {
BYTES_IN_ONE_CHAR(MAX_PROXIMITY_CHARS * sizeof(*mInputCodes)),
MAX_UMLAUT_SEARCH_DEPTH(DEFAULT_MAX_UMLAUT_SEARCH_DEPTH) {
if (DEBUG_DICT) LOGI("UnigramDictionary - constructor");
}
@ -80,8 +81,9 @@ bool UnigramDictionary::isDigraph(const int* codes, const int i, const int codes
void UnigramDictionary::getWordWithDigraphSuggestionsRec(const ProximityInfo *proximityInfo,
const int *xcoordinates, const int* ycoordinates, const int *codesBuffer,
const int codesBufferSize, const int flags, const int* codesSrc, const int codesRemain,
int* codesDest, unsigned short* outWords, int* frequencies) {
int currentDepth, int* codesDest, unsigned short* outWords, int* frequencies) {
if (currentDepth < MAX_UMLAUT_SEARCH_DEPTH) {
for (int i = 0; i < codesRemain; ++i) {
if (isDigraph(codesSrc, i, codesRemain)) {
// Found a digraph. We will try both spellings. eg. the word is "pruefen"
@ -89,23 +91,29 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(const ProximityInfo *pr
// Copy the word up to the first char of the digraph, then continue processing
// on the remaining part of the word, skipping the second char of the digraph.
// In our example, copy "pru" and continue running on "fen"
// Make i the index of the second char of the digraph for simplicity. Forgetting
// to do that results in an infinite recursion so take care!
++i;
memcpy(codesDest, codesSrc, i * BYTES_IN_ONE_CHAR);
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
codesBufferSize, flags, codesSrc + (i + 1) * MAX_PROXIMITY_CHARS,
codesRemain - i - 1, codesDest + i * MAX_PROXIMITY_CHARS,
outWords, frequencies);
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
codesBuffer, codesBufferSize, flags,
codesSrc + (i + 1) * MAX_PROXIMITY_CHARS, codesRemain - i - 1,
currentDepth + 1, codesDest + i * MAX_PROXIMITY_CHARS, outWords,
frequencies);
// Copy the second char of the digraph in place, then continue processing on
// the remaining part of the word.
// In our example, after "pru" in the buffer copy the "e", and continue running on "fen"
// In our example, after "pru" in the buffer copy the "e", and continue on "fen"
memcpy(codesDest + i * MAX_PROXIMITY_CHARS, codesSrc + i * MAX_PROXIMITY_CHARS,
BYTES_IN_ONE_CHAR);
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
codesBufferSize, flags, codesSrc + i * MAX_PROXIMITY_CHARS, codesRemain - i,
codesDest + i * MAX_PROXIMITY_CHARS, outWords, frequencies);
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
codesBuffer, codesBufferSize, flags, codesSrc + i * MAX_PROXIMITY_CHARS,
codesRemain - i, currentDepth + 1, codesDest + i * MAX_PROXIMITY_CHARS,
outWords, frequencies);
return;
}
}
}
// If we come here, we hit the end of the word: let's check it against the dictionary.
// In our example, we'll come here once for "prufen" and then once for "pruefen".
@ -128,7 +136,7 @@ int UnigramDictionary::getSuggestions(const ProximityInfo *proximityInfo, const
{ // Incrementally tune the word and try all possibilities
int codesBuffer[getCodesBufferSize(codes, codesSize, MAX_PROXIMITY_CHARS)];
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
codesSize, flags, codes, codesSize, codesBuffer, outWords, frequencies);
codesSize, flags, codes, codesSize, 0, codesBuffer, outWords, frequencies);
} else { // Normal processing
getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, codesSize,
outWords, frequencies);

3
native/src/unigram_dictionary.h
View File

@ -46,7 +46,7 @@ private:
void getWordWithDigraphSuggestionsRec(const ProximityInfo *proximityInfo,
const int *xcoordinates, const int* ycoordinates, const int *codesBuffer,
const int codesBufferSize, const int flags, const int* codesSrc, const int codesRemain,
int* codesDest, unsigned short* outWords, int* frequencies);
int currentDepth, int* codesDest, unsigned short* outWords, int* frequencies);
void initSuggestions(const int *codes, const int codesSize, unsigned short *outWords,
int *frequencies);
void getSuggestionCandidates(const int skipPos, const int excessivePos,
@ -108,6 +108,7 @@ private:
const int FULL_WORD_MULTIPLIER;
const int ROOT_POS;
const unsigned int BYTES_IN_ONE_CHAR;
const unsigned int MAX_UMLAUT_SEARCH_DEPTH;
// Flags for special processing
// Those *must* match the flags in BinaryDictionary.Flags.ALL_FLAGS in BinaryDictionary.java