aboutsummaryrefslogtreecommitdiffstats
path: root/native/src
diff options
context:
space:
mode:
Diffstat (limited to 'native/src')
-rw-r--r--native/src/defines.h6
-rw-r--r--native/src/unigram_dictionary.cpp54
-rw-r--r--native/src/unigram_dictionary.h12
3 files changed, 46 insertions, 26 deletions
diff --git a/native/src/defines.h b/native/src/defines.h
index b0dbffa0f..21215343e 100644
--- a/native/src/defines.h
+++ b/native/src/defines.h
@@ -232,9 +232,9 @@ static inline void prof_out(void) {
#define FIRST_WORD_INDEX 0
-// 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
+// TODO: Reduce this constant if possible; check the maximum number of digraphs in the same
+// word in the dictionary for languages with digraphs, like German and French
+#define DEFAULT_MAX_DIGRAPH_SEARCH_DEPTH 5
// Minimum suggest depth for one word for all cases except for missing space suggestions.
#define MIN_SUGGEST_DEPTH 1
diff --git a/native/src/unigram_dictionary.cpp b/native/src/unigram_dictionary.cpp
index 6c2c89231..563541dce 100644
--- a/native/src/unigram_dictionary.cpp
+++ b/native/src/unigram_dictionary.cpp
@@ -30,9 +30,9 @@
namespace latinime {
const UnigramDictionary::digraph_t UnigramDictionary::GERMAN_UMLAUT_DIGRAPHS[] =
- { { 'a', 'e' },
- { 'o', 'e' },
- { 'u', 'e' } };
+ { { 'a', 'e', 0x00E4 }, // U+00E4 : LATIN SMALL LETTER A WITH DIAERESIS
+ { 'o', 'e', 0x00F6 }, // U+00F6 : LATIN SMALL LETTER O WITH DIAERESIS
+ { 'u', 'e', 0x00FC } }; // U+00FC : LATIN SMALL LETTER U WITH DIAERESIS
// TODO: check the header
UnigramDictionary::UnigramDictionary(const uint8_t* const streamStart, int typedLetterMultiplier,
@@ -44,7 +44,7 @@ UnigramDictionary::UnigramDictionary(const uint8_t* const streamStart, int typed
// TODO : remove this variable.
ROOT_POS(0),
BYTES_IN_ONE_CHAR(MAX_PROXIMITY_CHARS * sizeof(int)),
- MAX_UMLAUT_SEARCH_DEPTH(DEFAULT_MAX_UMLAUT_SEARCH_DEPTH) {
+ MAX_DIGRAPH_SEARCH_DEPTH(DEFAULT_MAX_DIGRAPH_SEARCH_DEPTH) {
if (DEBUG_DICT) {
AKLOGI("UnigramDictionary - constructor");
}
@@ -64,7 +64,9 @@ static inline void addWord(
queue->push(frequency, word, length);
}
-bool UnigramDictionary::isDigraph(const int *codes, const int i, const int codesSize) const {
+// Return the replacement code point for a digraph, or 0 if none.
+int UnigramDictionary::getDigraphReplacement(const int *codes, const int i, const int codesSize,
+ const digraph_t* const digraphs, const unsigned int digraphsSize) const {
// There can't be a digraph if we don't have at least 2 characters to examine
if (i + 2 > codesSize) return false;
@@ -72,15 +74,18 @@ bool UnigramDictionary::isDigraph(const int *codes, const int i, const int codes
// Search for the first char of some digraph
int lastDigraphIndex = -1;
const int thisChar = codes[i * MAX_PROXIMITY_CHARS];
- for (lastDigraphIndex = sizeof(GERMAN_UMLAUT_DIGRAPHS) / sizeof(GERMAN_UMLAUT_DIGRAPHS[0]) - 1;
- lastDigraphIndex >= 0; --lastDigraphIndex) {
- if (thisChar == GERMAN_UMLAUT_DIGRAPHS[lastDigraphIndex].first) break;
+ for (lastDigraphIndex = digraphsSize - 1; lastDigraphIndex >= 0; --lastDigraphIndex) {
+ if (thisChar == digraphs[lastDigraphIndex].first) break;
}
// No match: return early
- if (lastDigraphIndex < 0) return false;
+ if (lastDigraphIndex < 0) return 0;
// It's an interesting digraph if the second char matches too.
- return GERMAN_UMLAUT_DIGRAPHS[lastDigraphIndex].second == codes[(i + 1) * MAX_PROXIMITY_CHARS];
+ if (digraphs[lastDigraphIndex].second == codes[(i + 1) * MAX_PROXIMITY_CHARS]) {
+ return digraphs[lastDigraphIndex].replacement;
+ } else {
+ return 0;
+ }
}
// Mostly the same arguments as the non-recursive version, except:
@@ -94,28 +99,36 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
int *xCoordinatesBuffer, int *yCoordinatesBuffer,
const int codesBufferSize, const int flags, const int *codesSrc,
const int codesRemain, const int currentDepth, int *codesDest, Correction *correction,
- WordsPriorityQueuePool *queuePool) {
+ WordsPriorityQueuePool *queuePool,
+ const digraph_t* const digraphs, const unsigned int digraphsSize) {
const int startIndex = (codesDest - codesBuffer) / MAX_PROXIMITY_CHARS;
- if (currentDepth < MAX_UMLAUT_SEARCH_DEPTH) {
+ if (currentDepth < MAX_DIGRAPH_SEARCH_DEPTH) {
for (int i = 0; i < codesRemain; ++i) {
xCoordinatesBuffer[startIndex + i] = xcoordinates[codesBufferSize - codesRemain + i];
yCoordinatesBuffer[startIndex + i] = ycoordinates[codesBufferSize - codesRemain + i];
- if (isDigraph(codesSrc, i, codesRemain)) {
+ const int replacementCodePoint =
+ getDigraphReplacement(codesSrc, i, codesRemain, digraphs, digraphsSize);
+ if (0 != replacementCodePoint) {
// Found a digraph. We will try both spellings. eg. the word is "pruefen"
- // 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"
+ // Copy the word up to the first char of the digraph, including proximity chars,
+ // and overwrite the primary code with the replacement code point. Then, continue
+ // processing on the remaining part of the word, skipping the second char of the
+ // digraph.
+ // In our example, copy "pru", replace "u" with the version with the diaeresis 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);
+ codesDest[(i - 1) * (BYTES_IN_ONE_CHAR / sizeof(codesDest[0]))] =
+ replacementCodePoint;
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
codesBuffer, xCoordinatesBuffer, yCoordinatesBuffer, codesBufferSize, flags,
codesSrc + (i + 1) * MAX_PROXIMITY_CHARS, codesRemain - i - 1,
currentDepth + 1, codesDest + i * MAX_PROXIMITY_CHARS, correction,
- queuePool);
+ queuePool, digraphs, digraphsSize);
// Copy the second char of the digraph in place, then continue processing on
// the remaining part of the word.
@@ -125,7 +138,8 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
codesBuffer, xCoordinatesBuffer, yCoordinatesBuffer, codesBufferSize, flags,
codesSrc + i * MAX_PROXIMITY_CHARS, codesRemain - i, currentDepth + 1,
- codesDest + i * MAX_PROXIMITY_CHARS, correction, queuePool);
+ codesDest + i * MAX_PROXIMITY_CHARS, correction, queuePool,
+ digraphs, digraphsSize);
return;
}
}
@@ -164,7 +178,9 @@ int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo,
int yCoordinatesBuffer[codesSize];
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
xCoordinatesBuffer, yCoordinatesBuffer,
- codesSize, flags, codes, codesSize, 0, codesBuffer, masterCorrection, queuePool);
+ codesSize, flags, codes, codesSize, 0, codesBuffer, masterCorrection, queuePool,
+ GERMAN_UMLAUT_DIGRAPHS,
+ sizeof(GERMAN_UMLAUT_DIGRAPHS) / sizeof(GERMAN_UMLAUT_DIGRAPHS[0]));
} else { // Normal processing
getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, codesSize, flags,
masterCorrection, queuePool);
diff --git a/native/src/unigram_dictionary.h b/native/src/unigram_dictionary.h
index 12b68d71b..b85913ffb 100644
--- a/native/src/unigram_dictionary.h
+++ b/native/src/unigram_dictionary.h
@@ -29,6 +29,8 @@ namespace latinime {
class TerminalAttributes;
class UnigramDictionary {
+ typedef struct { int first; int second; int replacement; } digraph_t;
+
public:
// Mask and flags for children address type selection.
static const int MASK_GROUP_ADDRESS_TYPE = 0xC0;
@@ -86,13 +88,15 @@ class UnigramDictionary {
void getWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
const int *ycoordinates, const int *codes, const int inputLength,
const int flags, Correction *correction, WordsPriorityQueuePool *queuePool);
- bool isDigraph(const int *codes, const int i, const int codesSize) const;
+ int getDigraphReplacement(const int *codes, const int i, const int codesSize,
+ const digraph_t* const digraphs, const unsigned int digraphsSize) const;
void getWordWithDigraphSuggestionsRec(ProximityInfo *proximityInfo,
const int *xcoordinates, const int* ycoordinates, const int *codesBuffer,
int *xCoordinatesBuffer, int *yCoordinatesBuffer,
const int codesBufferSize, const int flags, const int* codesSrc,
const int codesRemain, const int currentDepth, int* codesDest, Correction *correction,
- WordsPriorityQueuePool* queuePool);
+ WordsPriorityQueuePool* queuePool, const digraph_t* const digraphs,
+ const unsigned int digraphsSize);
void initSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
const int *ycoordinates, const int *codes, const int codesSize, Correction *correction);
void getOneWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
@@ -145,7 +149,7 @@ class UnigramDictionary {
const int FULL_WORD_MULTIPLIER;
const int ROOT_POS;
const unsigned int BYTES_IN_ONE_CHAR;
- const int MAX_UMLAUT_SEARCH_DEPTH;
+ const int MAX_DIGRAPH_SEARCH_DEPTH;
// Flags for special processing
// Those *must* match the flags in BinaryDictionary.Flags.ALL_FLAGS in BinaryDictionary.java
@@ -155,7 +159,7 @@ class UnigramDictionary {
REQUIRES_GERMAN_UMLAUT_PROCESSING = 0x1,
USE_FULL_EDIT_DISTANCE = 0x2
};
- static const struct digraph_t { int first; int second; } GERMAN_UMLAUT_DIGRAPHS[];
+ static const digraph_t GERMAN_UMLAUT_DIGRAPHS[];
// Still bundled members
unsigned short mWord[MAX_WORD_LENGTH_INTERNAL];// TODO: remove
generated by cgit v1.2.3-83-g751a (git 2.47.1) at 2025-05-23 19:48:42 +0000