Merge "Preserve a class for testing"
diff --git a/java/res/values-af/strings.xml b/java/res/values-af/strings.xml
index f5a6245..2892940 100644
--- a/java/res/values-af/strings.xml
+++ b/java/res/values-af/strings.xml
@@ -135,8 +135,7 @@
<string name="hint_add_to_dictionary" msgid="573678656946085380">"Raak weer om te stoor"</string>
<string name="has_dictionary" msgid="6071847973466625007">"Woordeboek beskikbaar"</string>
<string name="prefs_enable_log" msgid="6620424505072963557">"Aktiveer gebruikerterugvoer"</string>
- <!-- no translation found for prefs_description_log (7525225584555429211) -->
- <skip />
+ <string name="prefs_description_log" msgid="7525225584555429211">"Help om hierdie invoermetode-redigeerder te verbeter deur gebruikstatistiek en omvalverslae outomaties te stuur"</string>
<string name="keyboard_layout" msgid="8451164783510487501">"Sleutelbordtema"</string>
<string name="subtype_en_GB" msgid="88170601942311355">"Engels (VK)"</string>
<string name="subtype_en_US" msgid="6160452336634534239">"Engels (VS)"</string>
@@ -163,12 +162,9 @@
<string name="not_now" msgid="6172462888202790482">"Nie nou nie"</string>
<string name="custom_input_style_already_exists" msgid="8008728952215449707">"Dieselfde invoerstyl bestaan reeds: <xliff:g id="INPUT_STYLE_NAME">%s</xliff:g>"</string>
<string name="prefs_usability_study_mode" msgid="1261130555134595254">"Bruikbaarheidstudie-modus"</string>
- <!-- no translation found for prefs_key_longpress_timeout_settings (6102240298932897873) -->
- <skip />
- <!-- no translation found for prefs_keypress_vibration_duration_settings (7918341459947439226) -->
- <skip />
- <!-- no translation found for prefs_keypress_sound_volume_settings (6027007337036891623) -->
- <skip />
+ <string name="prefs_key_longpress_timeout_settings" msgid="6102240298932897873">"Vertraging van sleutellangdruk"</string>
+ <string name="prefs_keypress_vibration_duration_settings" msgid="7918341459947439226">"Sleuteldruk se vibrasie-tydsduur"</string>
+ <string name="prefs_keypress_sound_volume_settings" msgid="6027007337036891623">"Sleuteldruk se klankvolume"</string>
<string name="prefs_read_external_dictionary" msgid="2588931418575013067">"Lees eksterne woordeboeklêer"</string>
<string name="read_external_dictionary_no_files_message" msgid="4947420942224623792">"Geen woordeboeklêers in die aflaaiselsvouer nie"</string>
<string name="read_external_dictionary_multiple_files_title" msgid="7637749044265808628">"Kies \'n woordeboeklêer om te installeer"</string>
diff --git a/java/res/values-el/strings.xml b/java/res/values-el/strings.xml
index 41cf098..fad1cf5 100644
--- a/java/res/values-el/strings.xml
+++ b/java/res/values-el/strings.xml
@@ -135,8 +135,7 @@
<string name="hint_add_to_dictionary" msgid="573678656946085380">"Αγγίξτε ξανά για αποθήκευση"</string>
<string name="has_dictionary" msgid="6071847973466625007">"Λεξικό διαθέσιμο"</string>
<string name="prefs_enable_log" msgid="6620424505072963557">"Ενεργοποίηση σχολίων χρηστών"</string>
- <!-- no translation found for prefs_description_log (7525225584555429211) -->
- <skip />
+ <string name="prefs_description_log" msgid="7525225584555429211">"Βοηθήστε μας να βελτιώσουμε αυτό το πρόγραμμα επεξεργασίας μεθόδου εισόδου, στέλνοντας αυτόματα στατιστικά στοιχεία και αναφορές σφαλμάτων."</string>
<string name="keyboard_layout" msgid="8451164783510487501">"Θέμα πληκτρολογίου"</string>
<string name="subtype_en_GB" msgid="88170601942311355">"Αγγλικά (Η.Β.)"</string>
<string name="subtype_en_US" msgid="6160452336634534239">"Αγγλικά (Η.Π.Α)"</string>
@@ -163,12 +162,9 @@
<string name="not_now" msgid="6172462888202790482">"Όχι τώρα"</string>
<string name="custom_input_style_already_exists" msgid="8008728952215449707">"Το ίδιο στυλ εισόδου υπάρχει ήδη: <xliff:g id="INPUT_STYLE_NAME">%s</xliff:g>"</string>
<string name="prefs_usability_study_mode" msgid="1261130555134595254">"Λειτουργία μελέτης χρηστικότητας"</string>
- <!-- no translation found for prefs_key_longpress_timeout_settings (6102240298932897873) -->
- <skip />
- <!-- no translation found for prefs_keypress_vibration_duration_settings (7918341459947439226) -->
- <skip />
- <!-- no translation found for prefs_keypress_sound_volume_settings (6027007337036891623) -->
- <skip />
+ <string name="prefs_key_longpress_timeout_settings" msgid="6102240298932897873">"Καθυστέρηση παρατεταμένου πατήματος πλήκτρου"</string>
+ <string name="prefs_keypress_vibration_duration_settings" msgid="7918341459947439226">"Διάρκεια δόνησης πατήμ. πλήκτ."</string>
+ <string name="prefs_keypress_sound_volume_settings" msgid="6027007337036891623">"Ένταση ήχου πατήματος πλήκτρου"</string>
<string name="prefs_read_external_dictionary" msgid="2588931418575013067">"Ανάγνωση εξωτερικού αρχείου λεξικού"</string>
<string name="read_external_dictionary_no_files_message" msgid="4947420942224623792">"Δεν υπάρχουν αρχεία λεξικού στο φάκελο \"Λήψεις\""</string>
<string name="read_external_dictionary_multiple_files_title" msgid="7637749044265808628">"Επιλογή αρχείου λεξικού για εγκατάσταση"</string>
diff --git a/java/res/values-en-rGB/strings.xml b/java/res/values-en-rGB/strings.xml
index 3aab692..c0b9ede 100644
--- a/java/res/values-en-rGB/strings.xml
+++ b/java/res/values-en-rGB/strings.xml
@@ -135,8 +135,7 @@
<string name="hint_add_to_dictionary" msgid="573678656946085380">"Touch again to save"</string>
<string name="has_dictionary" msgid="6071847973466625007">"Dictionary available"</string>
<string name="prefs_enable_log" msgid="6620424505072963557">"Enable user feedback"</string>
- <!-- no translation found for prefs_description_log (7525225584555429211) -->
- <skip />
+ <string name="prefs_description_log" msgid="7525225584555429211">"Help improve this input method editor by automatically sending usage statistics and crash reports"</string>
<string name="keyboard_layout" msgid="8451164783510487501">"Keyboard theme"</string>
<string name="subtype_en_GB" msgid="88170601942311355">"English (UK)"</string>
<string name="subtype_en_US" msgid="6160452336634534239">"English (US)"</string>
@@ -163,12 +162,9 @@
<string name="not_now" msgid="6172462888202790482">"Not now"</string>
<string name="custom_input_style_already_exists" msgid="8008728952215449707">"The same input style already exists: <xliff:g id="INPUT_STYLE_NAME">%s</xliff:g>"</string>
<string name="prefs_usability_study_mode" msgid="1261130555134595254">"Usability study mode"</string>
- <!-- no translation found for prefs_key_longpress_timeout_settings (6102240298932897873) -->
- <skip />
- <!-- no translation found for prefs_keypress_vibration_duration_settings (7918341459947439226) -->
- <skip />
- <!-- no translation found for prefs_keypress_sound_volume_settings (6027007337036891623) -->
- <skip />
+ <string name="prefs_key_longpress_timeout_settings" msgid="6102240298932897873">"Key long press delay"</string>
+ <string name="prefs_keypress_vibration_duration_settings" msgid="7918341459947439226">"Keypress vibration duration"</string>
+ <string name="prefs_keypress_sound_volume_settings" msgid="6027007337036891623">"Keypress sound volume"</string>
<string name="prefs_read_external_dictionary" msgid="2588931418575013067">"Read external dictionary file"</string>
<string name="read_external_dictionary_no_files_message" msgid="4947420942224623792">"No dictionary files in the Downloads folder"</string>
<string name="read_external_dictionary_multiple_files_title" msgid="7637749044265808628">"Select a dictionary file to install"</string>
diff --git a/java/res/values-it/strings.xml b/java/res/values-it/strings.xml
index 028a7a2..c64543a 100644
--- a/java/res/values-it/strings.xml
+++ b/java/res/values-it/strings.xml
@@ -135,8 +135,7 @@
<string name="hint_add_to_dictionary" msgid="573678656946085380">"Tocca di nuovo per salvare"</string>
<string name="has_dictionary" msgid="6071847973466625007">"Dizionario disponibile"</string>
<string name="prefs_enable_log" msgid="6620424505072963557">"Attiva commenti degli utenti"</string>
- <!-- no translation found for prefs_description_log (7525225584555429211) -->
- <skip />
+ <string name="prefs_description_log" msgid="7525225584555429211">"Contribuisci a migliorare l\'editor del metodo di immissione inviando automaticamente statistiche sull\'utilizzo e rapporti sugli arresti anomali"</string>
<string name="keyboard_layout" msgid="8451164783510487501">"Tema della tastiera"</string>
<string name="subtype_en_GB" msgid="88170601942311355">"Inglese (UK)"</string>
<string name="subtype_en_US" msgid="6160452336634534239">"Inglese (USA)"</string>
@@ -163,12 +162,9 @@
<string name="not_now" msgid="6172462888202790482">"Non ora"</string>
<string name="custom_input_style_already_exists" msgid="8008728952215449707">"Esiste già uno stile di inuput uguale: <xliff:g id="INPUT_STYLE_NAME">%s</xliff:g>"</string>
<string name="prefs_usability_study_mode" msgid="1261130555134595254">"Modalità Studio sull\'usabilità"</string>
- <!-- no translation found for prefs_key_longpress_timeout_settings (6102240298932897873) -->
- <skip />
- <!-- no translation found for prefs_keypress_vibration_duration_settings (7918341459947439226) -->
- <skip />
- <!-- no translation found for prefs_keypress_sound_volume_settings (6027007337036891623) -->
- <skip />
+ <string name="prefs_key_longpress_timeout_settings" msgid="6102240298932897873">"Ritardo pressione lunga tasti"</string>
+ <string name="prefs_keypress_vibration_duration_settings" msgid="7918341459947439226">"Durata vibraz. pressione tasto"</string>
+ <string name="prefs_keypress_sound_volume_settings" msgid="6027007337036891623">"Volume audio a pressione tasto"</string>
<string name="prefs_read_external_dictionary" msgid="2588931418575013067">"Leggi file dizionario esterno"</string>
<string name="read_external_dictionary_no_files_message" msgid="4947420942224623792">"Nessun file di dizionario nella cartella Download"</string>
<string name="read_external_dictionary_multiple_files_title" msgid="7637749044265808628">"Seleziona un file di dizionario da installare"</string>
diff --git a/java/src/com/android/inputmethod/research/ResearchLogger.java b/java/src/com/android/inputmethod/research/ResearchLogger.java
index e0bd37c..5aaced0 100644
--- a/java/src/com/android/inputmethod/research/ResearchLogger.java
+++ b/java/src/com/android/inputmethod/research/ResearchLogger.java
@@ -81,6 +81,7 @@
import java.util.ArrayList;
import java.util.List;
import java.util.Random;
+import java.util.regex.Pattern;
/**
* Logs the use of the LatinIME keyboard.
@@ -1065,7 +1066,7 @@
new LogStatement("LatinImeOnStartInputViewInternal", false, false, "uuid",
"packageName", "inputType", "imeOptions", "fieldId", "display", "model",
"prefs", "versionCode", "versionName", "outputFormatVersion", "logEverything",
- "isUsingDevelopmentOnlyDiagnosticsDebug");
+ "isDevTeamBuild");
public static void latinIME_onStartInputViewInternal(final EditorInfo editorInfo,
final SharedPreferences prefs) {
final ResearchLogger researchLogger = getInstance();
@@ -1087,13 +1088,29 @@
Integer.toHexString(editorInfo.imeOptions), editorInfo.fieldId,
Build.DISPLAY, Build.MODEL, prefs, versionCode, versionName,
OUTPUT_FORMAT_VERSION, IS_LOGGING_EVERYTHING,
- ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS_DEBUG);
- } catch (NameNotFoundException e) {
- e.printStackTrace();
+ researchLogger.isDevTeamBuild());
+ } catch (final NameNotFoundException e) {
+ Log.e(TAG, "NameNotFound", e);
}
}
}
+ // TODO: Update this heuristic pattern to something more reliable. Developer builds tend to
+ // have the developer name and year embedded.
+ private static final Pattern developerBuildRegex = Pattern.compile("[A-Za-z]\\.20[1-9]");
+ private boolean isDevTeamBuild() {
+ try {
+ final PackageInfo packageInfo;
+ packageInfo = mLatinIME.getPackageManager().getPackageInfo(mLatinIME.getPackageName(),
+ 0);
+ final String versionName = packageInfo.versionName;
+ return !(developerBuildRegex.matcher(versionName).find());
+ } catch (final NameNotFoundException e) {
+ Log.e(TAG, "Could not determine package name", e);
+ return false;
+ }
+ }
+
/**
* Log a change in preferences.
*
diff --git a/native/jni/src/digraph_utils.cpp b/native/jni/src/digraph_utils.cpp
index 8781c50..6a1ab02 100644
--- a/native/jni/src/digraph_utils.cpp
+++ b/native/jni/src/digraph_utils.cpp
@@ -27,39 +27,47 @@
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 @@
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 6e364b6..9443522 100644
--- a/native/jni/src/digraph_utils.h
+++ b/native/jni/src/digraph_utils.h
@@ -27,21 +27,34 @@
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 cde7b99..32faae5 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 @@
// 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 @@
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 8e81632..8902d31 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 @@
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 @@
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 @@
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 @@
}
}
+ 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 @@
// 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 764c372..63bb200 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 @@
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 @@
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 @@
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 @@
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 6c09b94..136c4e5 100644
--- a/native/jni/src/suggest/core/suggest.h
+++ b/native/jni/src/suggest/core/suggest.h
@@ -64,6 +64,7 @@
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;