aboutsummaryrefslogtreecommitdiffstats
path: root/native/jni/src
diff options
context:
space:
mode:
Diffstat (limited to 'native/jni/src')
-rw-r--r--native/jni/src/digraph_utils.cpp74
-rw-r--r--native/jni/src/digraph_utils.h19
-rw-r--r--native/jni/src/suggest/core/dicnode/dic_node.h19
-rw-r--r--native/jni/src/suggest/core/dicnode/dic_node_state_scoring.h23
-rw-r--r--native/jni/src/suggest/core/suggest.cpp28
-rw-r--r--native/jni/src/suggest/core/suggest.h1
6 files changed, 139 insertions, 25 deletions
diff --git a/native/jni/src/digraph_utils.cpp b/native/jni/src/digraph_utils.cpp
index 8781c5077..6a1ab0271 100644
--- a/native/jni/src/digraph_utils.cpp
+++ b/native/jni/src/digraph_utils.cpp
@@ -27,39 +27,47 @@ const DigraphUtils::digraph_t DigraphUtils::GERMAN_UMLAUT_DIGRAPHS[] =
const DigraphUtils::digraph_t DigraphUtils::FRENCH_LIGATURES_DIGRAPHS[] =
{ { 'a', 'e', 0x00E6 }, // U+00E6 : LATIN SMALL LETTER AE
{ 'o', 'e', 0x0153 } }; // U+0153 : LATIN SMALL LIGATURE OE
+const DigraphUtils::DigraphType DigraphUtils::USED_DIGRAPH_TYPES[] =
+ { DIGRAPH_TYPE_GERMAN_UMLAUT, DIGRAPH_TYPE_FRENCH_LIGATURES };
/* static */ bool DigraphUtils::hasDigraphForCodePoint(
const int dictFlags, const int compositeGlyphCodePoint) {
- if (DigraphUtils::getDigraphForCodePoint(dictFlags, compositeGlyphCodePoint)) {
+ const DigraphUtils::DigraphType digraphType = getDigraphTypeForDictionary(dictFlags);
+ if (DigraphUtils::getDigraphForDigraphTypeAndCodePoint(digraphType, compositeGlyphCodePoint)) {
return true;
}
return false;
}
-// Retrieves the set of all digraphs associated with the given dictionary.
-// Returns the size of the digraph array, or 0 if none exist.
-/* static */ int DigraphUtils::getAllDigraphsForDictionaryAndReturnSize(
- const int dictFlags, const DigraphUtils::digraph_t **digraphs) {
+// Returns the digraph type associated with the given dictionary.
+/* static */ DigraphUtils::DigraphType DigraphUtils::getDigraphTypeForDictionary(
+ const int dictFlags) {
if (BinaryFormat::REQUIRES_GERMAN_UMLAUT_PROCESSING & dictFlags) {
- *digraphs = DigraphUtils::GERMAN_UMLAUT_DIGRAPHS;
- return NELEMS(DigraphUtils::GERMAN_UMLAUT_DIGRAPHS);
+ return DIGRAPH_TYPE_GERMAN_UMLAUT;
}
if (BinaryFormat::REQUIRES_FRENCH_LIGATURES_PROCESSING & dictFlags) {
- *digraphs = DigraphUtils::FRENCH_LIGATURES_DIGRAPHS;
- return NELEMS(DigraphUtils::FRENCH_LIGATURES_DIGRAPHS);
+ return DIGRAPH_TYPE_FRENCH_LIGATURES;
}
- return 0;
+ return DIGRAPH_TYPE_NONE;
+}
+
+// Retrieves the set of all digraphs associated with the given dictionary flags.
+// Returns the size of the digraph array, or 0 if none exist.
+/* static */ int DigraphUtils::getAllDigraphsForDictionaryAndReturnSize(
+ const int dictFlags, const DigraphUtils::digraph_t **const digraphs) {
+ const DigraphUtils::DigraphType digraphType = getDigraphTypeForDictionary(dictFlags);
+ return getAllDigraphsForDigraphTypeAndReturnSize(digraphType, digraphs);
}
// Returns the digraph codepoint for the given composite glyph codepoint and digraph codepoint index
// (which specifies the first or second codepoint in the digraph).
-/* static */ int DigraphUtils::getDigraphCodePointForIndex(const int dictFlags,
- const int compositeGlyphCodePoint, const DigraphCodePointIndex digraphCodePointIndex) {
+/* static */ int DigraphUtils::getDigraphCodePointForIndex(const int compositeGlyphCodePoint,
+ const DigraphCodePointIndex digraphCodePointIndex) {
if (digraphCodePointIndex == NOT_A_DIGRAPH_INDEX) {
return NOT_A_CODE_POINT;
}
- const DigraphUtils::digraph_t *digraph =
- DigraphUtils::getDigraphForCodePoint(dictFlags, compositeGlyphCodePoint);
+ const DigraphUtils::digraph_t *const digraph =
+ DigraphUtils::getDigraphForCodePoint(compositeGlyphCodePoint);
if (!digraph) {
return NOT_A_CODE_POINT;
}
@@ -72,16 +80,48 @@ const DigraphUtils::digraph_t DigraphUtils::FRENCH_LIGATURES_DIGRAPHS[] =
return NOT_A_CODE_POINT;
}
+// Retrieves the set of all digraphs associated with the given digraph type.
+// Returns the size of the digraph array, or 0 if none exist.
+/* static */ int DigraphUtils::getAllDigraphsForDigraphTypeAndReturnSize(
+ const DigraphUtils::DigraphType digraphType,
+ const DigraphUtils::digraph_t **const digraphs) {
+ if (digraphType == DigraphUtils::DIGRAPH_TYPE_GERMAN_UMLAUT) {
+ *digraphs = GERMAN_UMLAUT_DIGRAPHS;
+ return NELEMS(GERMAN_UMLAUT_DIGRAPHS);
+ }
+ if (digraphType == DIGRAPH_TYPE_FRENCH_LIGATURES) {
+ *digraphs = FRENCH_LIGATURES_DIGRAPHS;
+ return NELEMS(FRENCH_LIGATURES_DIGRAPHS);
+ }
+ return 0;
+}
+
/**
* Returns the digraph for the input composite glyph codepoint, or 0 if none exists.
- * dictFlags: the dictionary flags needed to determine which digraphs are supported.
* compositeGlyphCodePoint: the method returns the digraph corresponding to this codepoint.
*/
/* static */ const DigraphUtils::digraph_t *DigraphUtils::getDigraphForCodePoint(
- const int dictFlags, const int compositeGlyphCodePoint) {
+ const int compositeGlyphCodePoint) {
+ for (size_t i = 0; i < NELEMS(USED_DIGRAPH_TYPES); i++) {
+ const DigraphUtils::digraph_t *const digraph = getDigraphForDigraphTypeAndCodePoint(
+ USED_DIGRAPH_TYPES[i], compositeGlyphCodePoint);
+ if (digraph) {
+ return digraph;
+ }
+ }
+ return 0;
+}
+
+/**
+ * Returns the digraph for the input composite glyph codepoint, or 0 if none exists.
+ * digraphType: the type of digraphs supported.
+ * compositeGlyphCodePoint: the method returns the digraph corresponding to this codepoint.
+ */
+/* static */ const DigraphUtils::digraph_t *DigraphUtils::getDigraphForDigraphTypeAndCodePoint(
+ const DigraphUtils::DigraphType digraphType, const int compositeGlyphCodePoint) {
const DigraphUtils::digraph_t *digraphs = 0;
const int digraphsSize =
- DigraphUtils::getAllDigraphsForDictionaryAndReturnSize(dictFlags, &digraphs);
+ DigraphUtils::getAllDigraphsForDictionaryAndReturnSize(digraphType, &digraphs);
for (int i = 0; i < digraphsSize; i++) {
if (digraphs[i].compositeGlyph == compositeGlyphCodePoint) {
return &digraphs[i];
diff --git a/native/jni/src/digraph_utils.h b/native/jni/src/digraph_utils.h
index 6e364b67a..94435228e 100644
--- a/native/jni/src/digraph_utils.h
+++ b/native/jni/src/digraph_utils.h
@@ -27,21 +27,34 @@ class DigraphUtils {
SECOND_DIGRAPH_CODEPOINT
} DigraphCodePointIndex;
+ typedef enum {
+ DIGRAPH_TYPE_NONE,
+ DIGRAPH_TYPE_GERMAN_UMLAUT,
+ DIGRAPH_TYPE_FRENCH_LIGATURES
+ } DigraphType;
+
typedef struct { int first; int second; int compositeGlyph; } digraph_t;
static bool hasDigraphForCodePoint(const int dictFlags, const int compositeGlyphCodePoint);
static int getAllDigraphsForDictionaryAndReturnSize(
- const int dictFlags, const digraph_t **digraphs);
+ const int dictFlags, const digraph_t **const digraphs);
static int getDigraphCodePointForIndex(const int dictFlags, const int compositeGlyphCodePoint,
const DigraphCodePointIndex digraphCodePointIndex);
+ static int getDigraphCodePointForIndex(const int compositeGlyphCodePoint,
+ const DigraphCodePointIndex digraphCodePointIndex);
private:
DISALLOW_IMPLICIT_CONSTRUCTORS(DigraphUtils);
- static const digraph_t *getDigraphForCodePoint(
- const int dictFlags, const int compositeGlyphCodePoint);
+ static DigraphType getDigraphTypeForDictionary(const int dictFlags);
+ static int getAllDigraphsForDigraphTypeAndReturnSize(
+ const DigraphType digraphType, const digraph_t **const digraphs);
+ static const digraph_t *getDigraphForCodePoint(const int compositeGlyphCodePoint);
+ static const digraph_t *getDigraphForDigraphTypeAndCodePoint(
+ const DigraphType digraphType, const int compositeGlyphCodePoint);
static const digraph_t GERMAN_UMLAUT_DIGRAPHS[];
static const digraph_t FRENCH_LIGATURES_DIGRAPHS[];
+ static const DigraphType USED_DIGRAPH_TYPES[];
};
} // namespace latinime
#endif // DIGRAPH_UTILS_H
diff --git a/native/jni/src/suggest/core/dicnode/dic_node.h b/native/jni/src/suggest/core/dicnode/dic_node.h
index cde7b99a7..32faae52c 100644
--- a/native/jni/src/suggest/core/dicnode/dic_node.h
+++ b/native/jni/src/suggest/core/dicnode/dic_node.h
@@ -23,6 +23,7 @@
#include "dic_node_profiler.h"
#include "dic_node_properties.h"
#include "dic_node_release_listener.h"
+#include "digraph_utils.h"
#if DEBUG_DICT
#define LOGI_SHOW_ADD_COST_PROP \
@@ -399,8 +400,15 @@ class DicNode {
// TODO: Remove //
//////////////////////
// TODO: Remove once touch path is merged into ProximityInfoState
+ // Note: Returned codepoint may be a digraph codepoint if the node is in a composite glyph.
int getNodeCodePoint() const {
- return mDicNodeProperties.getNodeCodePoint();
+ const int codePoint = mDicNodeProperties.getNodeCodePoint();
+ const DigraphUtils::DigraphCodePointIndex digraphIndex =
+ mDicNodeState.mDicNodeStateScoring.getDigraphIndex();
+ if (digraphIndex == DigraphUtils::NOT_A_DIGRAPH_INDEX) {
+ return codePoint;
+ }
+ return DigraphUtils::getDigraphCodePointForIndex(codePoint, digraphIndex);
}
////////////////////////////////
@@ -452,6 +460,15 @@ class DicNode {
mDicNodeState.mDicNodeStateScoring.setDoubleLetterLevel(doubleLetterLevel);
}
+ bool isInDigraph() const {
+ return mDicNodeState.mDicNodeStateScoring.getDigraphIndex()
+ != DigraphUtils::NOT_A_DIGRAPH_INDEX;
+ }
+
+ void advanceDigraphIndex() {
+ mDicNodeState.mDicNodeStateScoring.advanceDigraphIndex();
+ }
+
uint8_t getFlags() const {
return mDicNodeProperties.getFlags();
}
diff --git a/native/jni/src/suggest/core/dicnode/dic_node_state_scoring.h b/native/jni/src/suggest/core/dicnode/dic_node_state_scoring.h
index 8e816329f..8902d3122 100644
--- a/native/jni/src/suggest/core/dicnode/dic_node_state_scoring.h
+++ b/native/jni/src/suggest/core/dicnode/dic_node_state_scoring.h
@@ -20,6 +20,7 @@
#include <stdint.h>
#include "defines.h"
+#include "digraph_utils.h"
namespace latinime {
@@ -27,6 +28,7 @@ class DicNodeStateScoring {
public:
AK_FORCE_INLINE DicNodeStateScoring()
: mDoubleLetterLevel(NOT_A_DOUBLE_LETTER),
+ mDigraphIndex(DigraphUtils::NOT_A_DIGRAPH_INDEX),
mEditCorrectionCount(0), mProximityCorrectionCount(0),
mNormalizedCompoundDistance(0.0f), mSpatialDistance(0.0f), mLanguageDistance(0.0f),
mTotalPrevWordsLanguageCost(0.0f), mRawLength(0.0f) {
@@ -43,6 +45,7 @@ class DicNodeStateScoring {
mTotalPrevWordsLanguageCost = 0.0f;
mRawLength = 0.0f;
mDoubleLetterLevel = NOT_A_DOUBLE_LETTER;
+ mDigraphIndex = DigraphUtils::NOT_A_DIGRAPH_INDEX;
}
AK_FORCE_INLINE void init(const DicNodeStateScoring *const scoring) {
@@ -54,6 +57,7 @@ class DicNodeStateScoring {
mTotalPrevWordsLanguageCost = scoring->mTotalPrevWordsLanguageCost;
mRawLength = scoring->mRawLength;
mDoubleLetterLevel = scoring->mDoubleLetterLevel;
+ mDigraphIndex = scoring->mDigraphIndex;
}
void addCost(const float spatialCost, const float languageCost, const bool doNormalization,
@@ -126,6 +130,24 @@ class DicNodeStateScoring {
}
}
+ DigraphUtils::DigraphCodePointIndex getDigraphIndex() const {
+ return mDigraphIndex;
+ }
+
+ void advanceDigraphIndex() {
+ switch(mDigraphIndex) {
+ case DigraphUtils::NOT_A_DIGRAPH_INDEX:
+ mDigraphIndex = DigraphUtils::FIRST_DIGRAPH_CODEPOINT;
+ break;
+ case DigraphUtils::FIRST_DIGRAPH_CODEPOINT:
+ mDigraphIndex = DigraphUtils::SECOND_DIGRAPH_CODEPOINT;
+ break;
+ case DigraphUtils::SECOND_DIGRAPH_CODEPOINT:
+ mDigraphIndex = DigraphUtils::NOT_A_DIGRAPH_INDEX;
+ break;
+ }
+ }
+
float getTotalPrevWordsLanguageCost() const {
return mTotalPrevWordsLanguageCost;
}
@@ -135,6 +157,7 @@ class DicNodeStateScoring {
// Use a default copy constructor and an assign operator because shallow copies are ok
// for this class
DoubleLetterLevel mDoubleLetterLevel;
+ DigraphUtils::DigraphCodePointIndex mDigraphIndex;
int16_t mEditCorrectionCount;
int16_t mProximityCorrectionCount;
diff --git a/native/jni/src/suggest/core/suggest.cpp b/native/jni/src/suggest/core/suggest.cpp
index 764c37240..63bb20004 100644
--- a/native/jni/src/suggest/core/suggest.cpp
+++ b/native/jni/src/suggest/core/suggest.cpp
@@ -18,6 +18,7 @@
#include "char_utils.h"
#include "dictionary.h"
+#include "digraph_utils.h"
#include "proximity_info.h"
#include "suggest/core/dicnode/dic_node.h"
#include "suggest/core/dicnode/dic_node_priority_queue.h"
@@ -221,7 +222,7 @@ int Suggest::outputSuggestions(DicTraverseSession *traverseSession, int *frequen
void Suggest::expandCurrentDicNodes(DicTraverseSession *traverseSession) const {
const int inputSize = traverseSession->getInputSize();
DicNodeVector childDicNodes(TRAVERSAL->getDefaultExpandDicNodeSize());
- DicNode omissionDicNode;
+ DicNode correctionDicNode;
// TODO: Find more efficient caching
const bool shouldDepthLevelCache = TRAVERSAL->shouldDepthLevelCache(traverseSession);
@@ -257,7 +258,10 @@ void Suggest::expandCurrentDicNodes(DicTraverseSession *traverseSession) const {
dicNode.setCached();
}
- if (isLookAheadCorrection) {
+ if (dicNode.isInDigraph()) {
+ // Finish digraph handling if the node is in the middle of a digraph expansion.
+ processDicNodeAsDigraph(traverseSession, &dicNode);
+ } else if (isLookAheadCorrection) {
// The algorithm maintains a small set of "deferred" nodes that have not consumed the
// latest touch point yet. These are needed to apply look-ahead correction operations
// that require special handling of the latest touch point. For example, with insertions
@@ -291,12 +295,18 @@ void Suggest::expandCurrentDicNodes(DicTraverseSession *traverseSession) const {
processDicNodeAsMatch(traverseSession, childDicNode);
continue;
}
+ if (DigraphUtils::hasDigraphForCodePoint(traverseSession->getDictFlags(),
+ childDicNode->getNodeCodePoint())) {
+ correctionDicNode.initByCopy(childDicNode);
+ correctionDicNode.advanceDigraphIndex();
+ processDicNodeAsDigraph(traverseSession, &correctionDicNode);
+ }
if (allowsErrorCorrections
&& TRAVERSAL->isOmission(traverseSession, &dicNode, childDicNode)) {
// TODO: (Gesture) Change weight between omission and substitution errors
// TODO: (Gesture) Terminal node should not be handled as omission
- omissionDicNode.initByCopy(childDicNode);
- processDicNodeAsOmission(traverseSession, &omissionDicNode);
+ correctionDicNode.initByCopy(childDicNode);
+ processDicNodeAsOmission(traverseSession, &correctionDicNode);
}
const ProximityType proximityType = TRAVERSAL->getProximityType(
traverseSession, &dicNode, childDicNode);
@@ -400,6 +410,16 @@ void Suggest::processDicNodeAsSubstitution(DicTraverseSession *traverseSession,
processExpandedDicNode(traverseSession, childDicNode);
}
+// Process the node codepoint as a digraph. This means that composite glyphs like the German
+// u-umlaut is expanded to the transliteration "ue". Note that this happens in parallel with
+// the normal non-digraph traversal, so both "uber" and "ueber" can be corrected to "[u-umlaut]ber".
+void Suggest::processDicNodeAsDigraph(DicTraverseSession *traverseSession,
+ DicNode *childDicNode) const {
+ weightChildNode(traverseSession, childDicNode);
+ childDicNode->advanceDigraphIndex();
+ processExpandedDicNode(traverseSession, childDicNode);
+}
+
/**
* Handle the dicNode as an omission error (e.g., ths => this). Skip the current letter and consider
* matches for all possible next letters. Note that just skipping the current letter without any
diff --git a/native/jni/src/suggest/core/suggest.h b/native/jni/src/suggest/core/suggest.h
index 6c09b9467..136c4e548 100644
--- a/native/jni/src/suggest/core/suggest.h
+++ b/native/jni/src/suggest/core/suggest.h
@@ -64,6 +64,7 @@ class Suggest : public SuggestInterface {
void generateFeatures(
DicTraverseSession *traverseSession, DicNode *dicNode, float *features) const;
void processDicNodeAsOmission(DicTraverseSession *traverseSession, DicNode *dicNode) const;
+ void processDicNodeAsDigraph(DicTraverseSession *traverseSession, DicNode *dicNode) const;
void processDicNodeAsTransposition(DicTraverseSession *traverseSession,
DicNode *dicNode) const;
void processDicNodeAsInsertion(DicTraverseSession *traverseSession, DicNode *dicNode) const;
generated by cgit v1.2.3-83-g751a (git 2.47.1) at 2025-05-09 04:33:45 +0000